According to the authors, the European Commission in its Google Shopping case did not outline which theory of foreclosure justified its finding of infringement. Likewise, there is no consensus in the literature about which theory of harm may best justify the decision.
In the light of this, the authors seek to develop such an economic and legal theory of harm in this paper, which can be found here. They argue that, by tying its search and shopping platforms, Google became able to serve customers with whom it would have not dealt otherwise. However, this may divert trade from potentially more efficient vertical platforms. By tying its shopping search to its general search service through visual prominence, Google can attract additional advertisers on its search platform that would otherwise have possibly advertised on competing search platforms. Thus, the effect of tying is a restriction on competition in vertical search that deserves antitrust scrutiny.
The paper is structured as follows:
Section 2 reviews the investigations against Google in the Google Shopping case.
This was already described in the paper above, so I am not going to repeat it here. It is worthwhile emphasising, however, that the main point made by the authors here is that the sanctioned conduct is more than a simple passive refusal of access by Google. Instead, the conduct consisted in actively favouring Google Shopping by a variety of means (i.e. visual prominence, higher ranking, and immunity from adjustment algorithms).
Section 3 reviews a number of potential theories of harm underpinning this decision.
The authors consider that the Commission, in contrast with its Android decision, did not reveal the exact legal theory of harm under which it sanctioned Google. They thus advance a number of possibilities:
- The first possible theory of harm concerns discrimination. Article 102 TFEU makes it abusive for a dominant company to apply “dissimilar conditions to equivalent transactions with other trading parties, thereby placing them at a competitive disadvantage.” The decision seems to refer to this when it points out that Art. 102 is concerned not only with abusive practices that harm consumers directly, but also with those conducts that harm consumers indirectly through their impact on an effective competition structure. However, this is a statement of general principles and not necessarily connected to the concrete conduct at issue. Furthermore, the authors identify a number of problems with such an approach. First, it is not clear what is meant by ‘competitive disadvantage’ in the provision, and applying this notion in this case would be highly problematic. Attempting to evaluate what discrimination means in a horizontal search page is inherently fraught given that the search algorithm is by definition a means to discriminate, rank and pick winners based on some selected metrics. Second, the provision has often been interpreted as an explicit recognition of fairness considerations toward competitors more than a concern for welfare-reducing conduct.
- The second possible theory of harm concerns ‘essential facilities’. In order for this doctrine to apply, Google’s general search platform would have to be defined as an essential facility to which competing vertical search engines and websites need access in order to effectively compete. In effect, the Court of Justice has identified narrow set of circumstances under which the essential facilities doctrine may be applicable, namely when the refusal to access the facility is likely to prevent any competition in the market, access is indispensable, and access is denied without any objective justification. Applied in the context of online search, this would require proof that: (a) achieving certain ranking results is objectively necessary in order for vertical platforms to compete effectively; (b) Google’s strategy of favouring its services through visual prominence is likely to eliminate effective competition in a given vertical search market; (c) the refusal results in consumer harm. These conditions set a high threshold, and many commentators have argued that refusal to deal and the essential facility doctrine are not applicable at all to online search. In any event, there is no explicit reference or endorsement of the doctrine in the decision. Instead, the Commission concludes that the doctrine is not relevant in a situation such as this, where Google is not required to either transfer any asset or enter into agreements with one or more competing comparison-shopping services.
- A third, preferable theory is that Google’s conduct amounts to an unconventional form of tying. Definitions of tying generally include the practice of not only requiring buyers of the tying good to purchase the tied good, but also of inducing buyers to purchase the tied good. Thus, the kind of psychological inducement to click on Google’s vertical search services through visual prominence that the Commission identifies may be best characterised as an instance of tying vertical search to general search. Since the price of search is zero, and discounts are thus not available, visual prominence becomes the means by which Google induces selection of its tied good (Google Shopping).
Section 4 highlights how Google’s conduct may amount to anticompetitive tying under European competition law.
Tying in Europe has four elements: (i) dominance in the tying product market; (ii) the tying and tied goods must be separate products; (iii) lack of consumer choice/coercion; and (iv) anticompetitive foreclosure. There are also possible defences that can be advanced by the dominant firm based on objective justifications or efficiencies.
While ultimately an empirical question, it is plausible that Google has market power in general search. Since Google Search is a multisided platform, Google might aim to extract more profits from one side of the market (advertisers) by charging very low prices to a different side of the market (searchers). Given the zero price of search, an exercise of market power over searchers could take the form of quality reduction, since Google might be able to degrade the search experience without losing a significant number of customers.
Concerning tying, two products are distinct if, in the absence of tying or bundling, a substantial number of customers would purchase or would have purchased the tying product without also buying the tied product from the same supplier, thereby allowing standalone production of both the tying and the tied product. The authors dismiss arguments that specialised search provides a subset of the functions of a general search engine, and that general and specialised search do not differ in functionality. Instead, general search and specialised shopping search are two distinct products, as indicated by the existence of independent providers of specialised search.
Furthermore, tying can exist if customers are merely induced to use a product, even if there is no coercion. Inducement can occur place by exploiting status quo biases – as in the Microsoft and Google Android cases. Although Google correctly argues that nothing forces users to click on links for additional services such as Google Shopping, and search users can click on other lower links or bypass Google by accessing websites directly, users are induced by prominent visualization not to do so.
Finally, search bias has the potential to foreclose national markets for comparison shopping services (and also for the general search market), leading to such consequences as a reduction of innovation, lower quality of search, and reduced consumer access to relevant services.
Section 5 argues that foreclosure in this context may be profitable.
It has been claimed that the single monopoly profit theorem prevents anti-competitive foreclosure in contexts such as that of the Google Shopping case. The single-monopoly-profit theorem shows that, in a vertical chain of production, the vertically integrated monopolist can earn monopoly profit only in one of the markets. The authors refer to the Bork and Sidak paper I reviewed a long time ago. According to these authors, the single monopoly profit theorem holds in both horizontal and vertical contexts, but seemingly for different reasons: upstream firms that tie downstream products cannot make more than a single monopolist’s profit, while firms that tie goods horizontally generally reduce their profits relative to the untied provision of the goods. In both scenarios, Google would not have incentives to foreclose.
The authors dismiss this argument.
– First, they argue that Google may have incentives to foreclose. If the value of Google’s general search to searchers falls because of tying and the lower quality of search results, there is a potential for fewer general searches. The extent of the reduction in searches depends on the elasticity of demand for search with respect to search quality. It is not the case, however, that the fall in profits associated with degrading the quality of general search necessarily reduces overall profits in multisided markets. When the price for one side of the market is zero, tying that market to another zero-price market in that same platform merely expands the set of buyers from whom the monopolist extracts surplus. Without tying, the monopolist does not deal at all with paying customers in the tied market (or does so at no profit). With tying, however, the general search monopolist extends its monopoly power from tying to tied market, which allows it to extract value from the tied market’s paying customers (e.g. advertisers).
Using Google as an example, the absence of a price for search limits vertical search rivals’ ability to bribe searchers to use their platforms. As a result, total profits to Google from searchers and advertisers can be higher with tying than without. Although tying may cause a reduction in profit from general search advertising as a result of quality degradation leading to a reduced number of searches, it allows Google to gain advertisers in Google Shopping with whom it may not otherwise have dealt at all. In other words, it is plausible that the losses in general search revenues will be offset by gaining access anti-competitively to a different set of advertisers in specialized search.
– Secondly, there are two kinds of potential social losses from the extension of market power from general into specialized search. Search users may lose surplus from the degradation of the general search experience that is associated with the artificial prioritization of Google’s specialised search over the purely organic display of general search results. Second, specialised search advertisers must buy advertising from Google at supra-competitive prices rather than in a competitive market for specialised search.
Section 6 concludes.
This paper develops a theory of harm according to which Google’s practice of inducing users on its general search page to rely on Google’s specialized search services represents a form of tying between horizontal and vertical search. The paper has shown how, in theory, this conduct is capable of satisfying all the legal requirements for tying. However, this theory of tying in the context of Google depends on two important caveats. First, the theory depends to a certain extent on the complementarity of the tying and tied two-sided platforms. However, as long as there are some Google searchers that rely on a general search before using a specialised search services, Google may grow its profits from specialised search advertisers by inducing these searchers to rely on its own specialised search engine. Second, the theory of tying in two-sided markets in this article assumes that each platform – i.e. general and specialist search platforms – can be supplied competitively. This is not necessarily the case, especially if there are network externalities in the provision of each platform, which is an empirical question.
This is a theoretically demanding yet comprehensive attempt to explain how certain self-preference mechanisms in digital platforms can have anticompetitive effects. I do not really have the expertise to comment on the paper, and I am unable to offer my opinion on the Commission’s decision. Nonetheless, I can make a few comments.
Firstly, discrimination as an abuse has been subject to a clarification in Meo, reviewed here. However, this clarification did not really touch on self-preference mechanisms, and, as such, its relevance to Google Shopping is limited. Secondly, it was not clear to me why general and specialised search are distinct markets; then again, this may be because the authors, at various points, emphasise that the fulfillment of the various requirements set out by their theory of harm is an empirical matter. A last question concerns the consequences of finding that visual prominence can lead to foreclosure. In effect, the authors have to argued this because it is impossible to demonstrate that search users were coerced, or had no alternative, to using the tied product (specialised search), which is usually a pre-requisite for a finding of unlawful tying. However, if mere display that induces search users the buying of a related product can amount to tying, this would significantly lower the threshold at which supplying additional services– e.g. Google Maps – becomes anticompetitive.