Google Search. Shopping for an appropriate abuse standard

Now that Google and the Commission are at daggers drawn in Luxembourg over the 27 June decision in Google Search,[1] it seems like high time to make our educated guesses about how the recently disclosed arguments in the tech giant’s September appeal[2] will come into play. Therefore, this post is a sort of an update (or a plug-in) to my recent paper EU Competition Law Needs to Install a Plug-in,[3] which for better or for worse was submitted one day after the Commission’s decision was adopted and it was published five days before we had the first news of Google’s bringing the case before the General Court.

As anticipated, Google’s appeal will revolve around the claim that the theory of damage behind its conduct was that of an essential facilities case, while the Commission found it abusive below the refusal to deal threshold. As a matter of fact, Nicholas Banasevic (head of the unit responsible for the Google case) has made clear that the June decision is ‘a very detailed effects-based decision‘ in a ‘plain and simple leveraging case,’ without there being any need to ‘apply another “label” to it, including abuses such as “refusal to supply” rivals.’[4]

Banasevic’s assertion implies that the Silicon Valley company’s leveraging on its dominance in general search to artificially favour its own products in an adjacent market would suffice to be found abusive, as far as the (actual or potential) anticompetitive effect is supported by a fair deal of evidence. No wonder, Google’s reaction would be trying to bring the discussion to the field of legal standard by pleading that the Commission’s move away from the refusal to supply test constitutes an attempt to lowering the abuse threshold below the essential facilities requirements of indispensability and removal of effective competition.

Against this backdrop, it must be recalled that the long-drawn Google Shopping battle, the first chapter of which was brought to a close by the June decision, touches upon a number of de lege lata and de lege ferenda questions. Indeed, it tables both the debate over the adjustments required by competition law of the European Union to be able to deal with the challenges of the digital environment and the traditional controversy about the limits of competition authorities’ say in dominant companies’ business models.

From the de lege ferenda point of view, the decision reveals the reluctance of the Commission to embrace the paradigm of dynamic competition in the digital economy that has already gained a foothold in merger control.[5] In particular, the multi-sided character of Google Search does not seem to be reckoned with to a sufficient extent either in market definition (limited to the ‘general Internet search’ side of the platform) or in market power appraisal. Therefore, the competitive pressure exerted on Google’s search advertising by other web portals (including non-search advertising through, for instance, social networks like Facebook or merchant platforms such as Amazon Marketplace) is disregarded.

One would also miss a more decisive role to be played by the third side of the search engine, on which Google provides content indexing services to webpage publishers. After all, such services are the source of organic results and, therefore, the junction between the general search engine (i.e. allegedly dominated market) and the comparison shopping services (i.e. related market in which the abuse would have had effects). In fact, the definition of a hypothetical market for content indexing, in which rival price comparison websites are Google’s customers, could be crucial for determining whether there are alternative avenues for such competitors to give access to their comparison shopping services. That is, if other horizontal or vertical search engines, social networks or mobile apps could be considered part of this market, the ability of Google Search’s alleged discrimination to generate a foreclosure effect in the related markets for comparison shopping services would be questioned by the availability of alternative showcases.

Furthermore, excessive reliance on invariably high market shares since 2008 (above 90% in certain cases) to support Google’s dominance fails to explain the absence of switching cost for users (illustrated by the sentence ‘competition is one click away‘). In a setting of cost-free change, cybernaut loyalty to Google Search could only be explained by either user inertia or the absence of a search engine performing comparably to Google’s. However, on the one hand, irrational consumer behaviour (albeit already considered as one of the various indicia for the existence and abuse of a dominant position[6]) does not seem apt to serve as the fundamental pillar for the economic theory behind Google’s dominance. On the other hand, if consumers are assumed to behave rationally, the stability of Google’s market position would precisely suggest that it competes on the merits, as far as the degradation of its organic search results in order to foster its own vertical search services would have resulted in the loss of rational users.[7]

To hurdle those obstacles, the Commission resorts to the traditional monolithic construction of network effects as barriers to entry. Nonetheless, under the paradigm of digital market contestability, network effects would not have raised inexpugnable entry barriers. Quite to the contrary, they would have contributed to Google Search being ousted by rival horizontal search engines if the latter had developed a better performing algorithm or the if former had degraded its own in a way that less relevant organic results are retrieved.

Equally noteworthy is the insistence by the Commission on the fact that merchant platforms[8] like Amazon Marketplace or eBay compete on the markets for price comparison services. In this concern, the decision clarifies that Google’s conduct would have been also capable of distorting competition in a hypothetical broader market in which merchant platforms competed with price comparison websites, given that the latter would be Google Shopping’s closest rivals. The Commission could be trying to conjure away the phantasm of the appreciability threshold for anticompetitive effects in adjacent markets.

Although the debate about the absence of a de minimis rule in abuse of dominance appeared to have been settled long ago,[9] the High Court of Justice of England and Wales has recently interpreted the ruling by the Court of Justice in Post Danmark II in the sense that it cannot be extrapolated to those cases in which the anticompetitive effect is deployed in a non-dominated adjacent market (e.g. specialised price comparison services).[10] According to the High Court, the holding in Post Danmark II that, ‘since the structure of competition on the market has already been weakened by the presence of the dominant undertaking, any further weakening of the structure of competition may constitute an abuse of a dominant position[11] can only apply to abuses producing effects in the same market in which the dominant position is held.

From the de lege lata perspective, the abusive nature of self-preferencing below the threshold of the essential facilities doctrine involves serious obstacles that the Commission tries to bridge over by replacing the requirements of indispensability and (likely) elimination of effective competition with the thorough demonstration of the potential[12] and actual[13] anticompetitive foreclosure brought about by Google’s conduct.

The first analytical strain would be the competitive advantage supposedly granted by Google’s search engine to its specialised price comparison services being equated to an anticompetitive foreclosure of its rival comparison services (without need to demonstrate that such advantage is indispensable for the latter to compete effectively so that, if it is not shared with them, effective competition in price comparison would be likely removed). One would see it as tantamount to imposing on Google the positive duty to help its rivals compete against it. This is confirmed by the unwonted principle of equal treatment expressly recognised in the decision.

However, a reduction of the abuse standard that makes prima facie illegal any potential foreclosure of competitors could be seen as challenging the concept of competition law as a safeguard of the competitive process, not competitors. Ultimately, the protection of rivals against any competitive disadvantage on which rivals might be placed vis-à-vis the dominant company could jeopardise the competitive process itself if it reaches the point of denying the dominant company’s right to avail of its competitive advantage to foreclose competitors (i.e. its right to compete on the merits). The same purpose of protecting the dominant firm’s right to compete on the merits seems to make the game-changing ruling in Intel factor breathing new life into the ‘as efficient competitor’ test so that the efficiency of competitors is factored in to raise the standard of abusive foreclosure.[14]

Even if the requirements of indispensability and elimination of effective competition were to be dispensed with, it seems still necessary to prove an actual or likely prejudice to competition that includes a reduction of consumer welfare in the long run (e.g. less innovation or choice in the market).[15] Nonetheless, the finding of a search bias (i.e. the design of search algorithms so as to retrieve discriminatory results) to the detriment of consumers is an inherently subjective and complex analysis, given that algorithms are in permanent dynamic adjustment to user queries. Also, there is a high risk of confusion between natural demotion of less relevant rival links and discriminatory treatment.

In this respect, the US Federal Trade Commission, which decided to close its parallel investigation into Google Search on 3 January 2013, found that the main goal of the changes introduced by Google in its algorithm was replying more efficiently to user queries by providing them with the relevant information directly on the general result pages. The conclusion across the pond was that, even if such changes had the effect of actually demoting certain rival comparison shopping services on general organic result pages, they also resulted in an overall increase of the quality of these results and, thus, of net consumer welfare.

This reasoning of the US Federal Trade Commission shows the appropriateness of revisiting the debate over the transplant of the efficiency defence available in anticompetitive collusion into the field of abuse of a dominant position,[16] recently revived by Grand Chamber.[17] This is a must in the digital ecosystem, in which companies enjoying big market positions play a crucial role for welfare by internalising externalities. From this point of view, the display of the products or services sought by users directly on the general result pages, instead of redirecting them to specialised third-party search services, could justify the presumed abuse committed by Google if equivalent conditions to those in Article 101(3) TFEU were fulfilled.

The Commission tries to avoid the difficulties in legal standard by placing the emphasis on the great amount of evidence[18] pointing at the potential of Google’s conduct to modify the traffic patterns to rival price comparison webpages and the concomitant increase in Google Shopping’s market share (without the correlation between both phenomena being quite clear). The causal relationship between the suspected anticompetitive effect and Google’s presumed dominant position in the market for horizontal search is equally far from clear. In front of the General Court, the proof of such causal link by the Commission will face not only the necessarily subjective demonstration of the search bias and the complex and evolutionary nature of algorithms, but also the myriad of alternative explanations for the increase in Google Shopping’s traffic and the loss of traffic towards rival price comparison services.

Thirdly, whether the Google Search case could have or should have been treated as a refusal to supply has been discussed at great length,[19] ranging from Nazzini’s claim that it is not on its face a refusal to supply[20] to Vesterdorf’s argument that the sole theory of antitrust liability applicable is the refusal to supply theory,[21] through Ibáñez Colomo’s presenting the case as possibly the definitive nail in the coffin of the refusal to supply doctrine.[22] I have suggested in my paper moving from the orbit of Article 102(b) TFEU towards that of Article 102(c) TFEU by invoking the gatekeeper condition of Google’s as a way of exorcising the refusal to supply phantasm.[23]

Whether or not the refusal to supply legal test is applicable, the Commission’s finding of an exclusionary abuse on the grounds of the mere foreclosure effect brought about by the discrimination of (not even efficient) competitors could come in for criticism in light of the legal certainty principle, which requires that the dominant company be able to assess the lawfulness of its own conduct.[24] It is true that such principle does not preclude competition authorities from adjusting the analytical framework for abuse within the ample terms of Article 102 TFEU to new behaviours not fitting in the categories predefined by case law. However, one could also argue that competition authorities are bound to act consistently with the jurisprudential requirements for the different types of abuse found in the past and, in particular, to avoid lowering the legal standard to the point that the abusive character of some behaviours becomes uncertain.[25]

From this perspective, it would even be possible to wonder whether the imposition of an unprecedented fine in relation to a behaviour that has not led to sanctions in the past (by relying on legal grounds and jurisprudential support that are far from clear and settled) would run counter Articles 49(1) of the Charter of Fundamental Rights of the European Union and 7 of the European Convention of Human Rights. These provisions forbid holding an individual guilty of an offence for a conduct that did not qualified as such under law at the time when it was committed.

[1] Decision of the European Commission of 27 June 2017 in case AT.39740 Google Search (Shopping).

[2] Case T-612/17 Google and Alphabet v Commission.

[3] Pablo Solano Díaz, ‘EU Competition Law Needs to Install a Plug-in’ (2017) 40(3) World Competition 393 (https://www.kluwerlawonline.com/abstract.php?area=Journals&id=WOCO2017026).

[4] Matthew Newman & Mary Eccles, ‘EU Google Decision gives “guiding principles” to comply with antitrust order’, MLex (http://www.mlex.com/GlobalAntitrust/DetailView.aspx?cid=929939&siteid=190&rdir=1).

[5] See Judgement of the General Court of 11 December 2013 in case T-79/12 Cisco and Messagenet v Commission.

[6] Judgement of the Court of 17 September 2007 in case T-201/04 Microsoft v Commission, §1052.

[7] Renato Nazzini, ‘Google and the (Ever-stretching) Boundaries of Article 102’ (2015) 6 Journal of European Competition Law & Practice 301 (https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2965420).

[8] Platforms that, in addition to offering a price comparison tool, allow to complete the purchase on site.

[9] See, among others, Judgments of the Court of 13 February 1979 in case 85/76 Hoffmann-La Roche v Commission, and of 6 October 2015 in case C‑23/14 Post Danmark v Konkurrencerådet [Post Danmark II], §§72-73.

[10] Judgement of the High Court of Justice of England and Wales of 12 February 2016 in case Streetmap.eu Limited v Google Inc. et al. [2016] EWHC 253 (Ch), §§95-98.

[11] Judgment of the Court in Post Danmark II, supra note 10.

[12] According to the decision in Google Search, real-world consumer behaviour, surveys and eye-tracking studies would have demonstrated that the ten highest-ranking generic search results on page 1 together generally receive approximately 95% of all clicks on generic search results (with the top search result receiving about 35% of all the clicks). The first result on page 2 of Google’s search results receives only about 1% of all clicks.

[13] According to the decision in Google Search, the suspected demotions applied in Google’s generic search algorithms would have in fact provoked drops of traffic to certain rival websites of 92% in Germany, 85% in the United Kingdom and 80% in France.

[14] Judgment of the Court of 6 September 2017 in case C-413/14 P Intel v Commission, §§142-143.

[15] See, among others, Judgement of the Court of 27 March 2012 in case C-209/10 Post Danmark v Konkurrencerådet [Post Danmark I], §44, and Communication from the Commission of 24 February 2009 on Guidance on its enforcement priorities in applying Article 82 of the EC Treaty to abusive exclusionary conduct by dominant undertakings [OJ C 45], §19.

[16] Judgement of the Court in Post Danmark I, supra note 15, §41.

[17] Judgement of the Court in Intel v Commission, supra note 14, §140.

[18] Evidence includes contemporary documents from both Google and other market operators, 5.2 terabytes of real information on about 1.7 billion user queries, surveys and experimental models, financial and traffic data on the importance of result visibility on Google’s general search pages and an extensive market investigation addressed to several hundred customers and competitors.

[19] Alison Jones & Brenda Sufrin, EU Competition Law: Text, Cases, and Materials (6th edition, OUP 2016) 554.

[20] Renato Nazzini, supra note 8.

[21] Bo Vesterdorf, ‘Theories of Self-Preferencing and Duty to Deal – Two Sides of the Same Coin’ (2015) 1(1) Competition Law & Policy Debate 4.

[22] Pablo Ibáñez Colomo, ‘How to distinguish between tying and refusal to deal cases (hint: it’s not just words)’, Chillin’ Competition (https://chillingcompetition.com/2015/04/24/how-to-distinguish-between-tying-and-refusal-to-deal-cases-hint-its-not-just-words/).

[23] Pablo Solano Díaz, supra note 3.

[24] Judgment of the Court of 14 October 2010 in case C-280/08 P Deutsche Telekom v Commission, §202.

[25] Renato Nazzini, supra note 8.

Leave a Reply

Your email address will not be published. Required fields are marked *