COMPETITION AMONG SEARCH ENGINES AND THE ANTITRUST CASE AGAINST GOOGLE

Search engines such as Google or Bing are multi-billion-dollar businesses. At the same time, the market for online search is highly concentrated around the globe. While Google is the clear market leader in virtually all Western countries, Baidu in China, Yandex in Russia and to a lesser degree Yahoo! in Japan have dominant positions in these countries.

However, while it appears to be relatively easy to understand that large customer bases may be more attractive for advertising companies, this becomes less clear on second sight. If online advertising is charged on a pay-per-click basis, an online site that induces 10 000 clicks may be as attractive as ten smaller sites that induce 1000 clicks each (see Manne and Wright, 2011). Nevertheless, large search engines may still be more attractive than smaller ones, as (1) there can be a fixed cost per web page associated with monitoring advertising campaigns and (2) larger search engines may be better able to place targeted advertising, as they have access to a larger base of historical search data and past ‘clicking behavior’. These two features can make larger search engines more attractive than smaller ones. In addition, Google has traditionally created (by means of contract) some artificial incompatibility between advertising campaigns on Google and other search engines, but this incompatibility issue has been largely resolved following investigations by the Federal Trade Commission (see FTC, 2013). Furthermore, since space is limited on web pages, a given web page that induces 10 000 clicks for a given ad generates more revenue per page than a page that only induces 1000 clicks.

It is much less clear, however, how important a search engine’s size is for search engine users. While it is plausible that access to a large set of (historical) search data and con­sumer clicking behavior is beneficial to improve an engine’s search results, there is some debate about how much data are needed to further refine the search mechanism before the marginal benefit of additional data exceeds the additional cost of processing it (see Manne and Wright, 2011). In fact, the literature is divided about whether it is Google’s sheer size that allows it to maintain its market position (e.g., Pollock, 2010; Crane, 2012) or whether it is its superior innovation (see Manne and Wright, 2011; Bork and Sidak, 2012). Overall, it appears that Google’s superior ability to place advertising, based on its analysis of large datasets of customer clicking behavior, and the fixed cost of placing and monitoring advertising, gives rise to indirect network effects from users to advertisers, which ultimately makes Google a two-sided market.⁵

In any case, switching costs between search engines are very modest for consumers, as the past has shown. When Google entered the market in 1998, Altavista was the leading search engine with Yahoo! closely following in second place in Western countries. Still Google managed not only to enter the market, but also to offer superior quality so that it even leapfrogged its competitors. Similarly, Rambler was the leading Russian search engine in the late 1990s before Yandex surpassed it. Many commentators therefore suggest that Google’s success is also a result of its superior quality (e.g., Evans, 2008; Devine, 2008; Argenton and Prufer, 2012).

Overall, the quality of search engines can be approximated by ‘expected time a user needs to obtain a satisfactory result’. The time needed to find a satisfactory result in turn depends on several factors (Argenton and Prufer, 2012), including search algorithm quality, hardware quality, and data quality, where data quality refers both to data freely available on the Internet and search engine-specific data that has been collected during previous search processes.

While the existence of a superior search engine is, of course, not a policy concern for competition authorities in itself, there have also been numerous complaints that Google abuses its dominant position, especially to favor its own subsidiaries (such as Google Maps, YouTube or Google Shopping) over competing platforms. More precisely the alle­gation is that Google biases its search results so that links to its own subsidiaries appear ahead of links to rival sites even though a rival’s site may be a better fit for what the user is searching for.

In addition, other allegations have concerned Google’s (unlicensed) use of content generated by competing specialized search engines (so-called vertical search engines) as well as the strategic incompatibility that Google introduced for third parties between advertising campaigns on Google platforms and other web pages. Both issues have largely been resolved through commitments accepted by the Federal Trade Commission (FTC, 2013) in January 2013. A fourth antitrust investigation concerns the exclusive or default contracts Google has used to incentivize mobile handset manufacturers (through exclusive contracts, rebates or potentially predatory prices) to use Google’s Android operating system and Google as the default search engine (see, e.g., Manne and Wright, 2011; Bork and Sidak, 2012). Whether these contracts give rise to market foreclosure is, at the time of writing in 2015, being investigated by the European Commission and other competition agencies.

Regarding the most prominent allegation of search bias, in January 2013 the Federal Trade Commission (FTC, 2013) decided not to initiate formal proceedings. In contrast, after more than five years of investigation and analysis (which started in early 2010) and following a lengthy discussion of various commitments that Google had offered to undertake in order to mitigate the alleged search bias problem, in April 2015 the European Commission issued a formal statement of objections (SO) against Google. The SO outlines the European Commission’s preliminary view that Google is:

[...] abusing a dominant position, in breach of EU antitrust rules, by systematically favoring its own comparison shopping product in its general search results pages in the European Economic Area (EEA).

More specifically, the Commission’s main preliminary conclusions are that (1) ‘Google systematically positions and prominently displays its comparison shopping service in its general search results pages, irrespective of its merits’; (2) ‘Google does not apply to its own comparison shopping service the system of penalties, which it applies to other comparison shopping services on the basis of defined parameters, and which can lead to the lowering of the rank in which they appear in Google’s general search results pages’; and that (3) ‘Google’s conduct has a negative impact on consumers and innovation. It means that users do not necessarily see the most relevant comparison shopping results in response to their queries, and that incentives to innovate from rivals are lowered as they know that however good their product, they will not benefit from the same prominence as Google’s product’. The SO suggests that:

Google should treat its own comparison shopping service and those of rivals in the same way. This would not interfere with either the algorithms Google applies or how it designs its search results pages. It would, however, mean that when Google shows comparison shopping services in response to a user’s query, the most relevant service or services would be selected to appear in Google’s search results pages. (European Commission, 2015)

In addition, further proceedings are underway at US state level as well as in India, Argentina and South Korea.

Whether Google has in fact biased its search results in favor of its own subsidiaries is difficult to determine from the outside, as Google’s search algorithm is naturally a busi­ness secret. Moreover, personalized search results (based on one’s own search history and one’s cookies) imply that different individuals may obtain different search results for the same keyword search.

The European Commission’s statement of objection also reveals how difficult it is to delineate the relevant product market. While the European Commission defines a distinct product market for general search, it is far from clear how consumers would substitute if Google were to charge for search requests. While some users may switch to other general search engines, other users may use Wikipedia, Amazon, IMDb, LinkedIn, Twitter and other websites with search functions for their searches for general information, books, movies, people and so on. In fact, the Commission appears to hold the view that users search for specific web pages, while one may also argue that users rather search for information. In its market delineation the Commission therefore exclusively focuses on technical aspects (how and which websites are crawled and listed), but does not analyze consumer behavior. Hence, it is completely unclear whether the Commission’s market delineation is appropriate or not. Similarly, the Commission argues that comparison shopping services constitute a different product market from specialized search services, online retailers, and merchant platforms/market places. Again, the Commission bases its view entirely on technical and functional aspects, arguing that comparison shopping services constitute a market in their own right, since consumers cannot directly purchase the product from these sites. This means, of course, that Google shopping would belong to a different and more competitive market, if Google were to include one-click-purchase options in its ads. From an economic perspective, this is not immediately plausible, as a further vertical integration would imply that Google is no longer dominant with regard to eBay and Amazon. Again, user behavior has not been analyzed to delineate the market, which is somewhat troublesome.

In addition, the European Commission’s statement of objections appears to suggest that, in the absence of an objective justification, discrimination and favoring one’s own services is abusive by its very nature for dominant companies. Alternatively, the Commission may hold the view that Google Shopping is an essential facility or bottle­neck for online retailers, even though the Commission does not use the term. In any case, given the presence of numerous market places such as eBay and Amazon, the view may be difficult to sustain.

Finally, the Commission argues that the success and growth of Google’s services do not reflect its relative quality and attractiveness for users. The statement of objections suggests that Google’s success is not the outcome of competition on the merits, although little evidence is provided to substantiate this claim. Given these shortcomings, an inter­esting question - also from a political economy of antitrust perspective - is why the Commission chose Google Shopping as its showcase and not any of the other services for which market definition may be much less contentious. For now, it is interesting to see how the case will evolve.

Even ignoring the practical problems of proving potential abuse, the next question concerns potential remedies to prevent any anticompetitive search biases in the future. A number of scholars have suggested mandating search neutrality (e.g., Pollock, 2010; Edelman, 2011; Ammori and Pelican, 2012; Crane, 2012; Manne and Wright, 2012). As has been pointed out though, search neutrality is, first of all, difficult to operational­ize and, second, may inhibit further innovation, thereby harming consumers in the end (see Grimmelmann, 2011; Bork and Sidak, 2012; Crane, 2012). As a consequence, some policy-makers have proposed to unbundle or separate Google’s search business from its content business. However, consequences for innovation and consumers may be even more adverse than with search neutrality requirements, as unbundling would imply that search engines would no longer be allowed to answer questions themselves, but only provide links to answers. Moreover, an unbundling remedy would become completely untenable in other cases as, for example, Amazon may also be considered the largest search engine for books. Obviously, unbundling does not make any sense here. Given that many websites have search functions and may be considered search markets, unbundling requirements become highly problematic.

Another suggestion has been to require Google to reveal its search algorithm, but such a measure would appear disproportionate, as has been argued in the literature, as it concerns the heart of Google’s business and the main element of competitive rivalry (e.g., Argenton and Prufer, 2012; Bork and Sidak, 2012). Others have proposed regulating Google’s search algorithm and changes in it. However, practically, this is not without problems either. For example, Google changed its algorithm 665 times in 2012 alone.⁶ Consequently, there is a risk that regulation will either be too slow or even retard innovation.

Instead, Argenton and Prufer (2012) have suggested that Google should be required to share its specific search engine data to foster competition in search engine markets. This suggestion is based on the assumption that it is very difficult for competing search engines to catch up or even overtake Google, due to their lack of online search data to develop better search engine algorithms. Hence access to (historical) search data may help to enable Google’s competitors to develop better search algorithms, thereby increasing competitive pressures in the market for search engines.

Another option, which is more light-handed, would be to mandate that Google colors the background or the links to its own subsidiaries in a manner similar to sponsored links. Once consumers clearly realize that some search results point to Google websites or services, they can better evaluate the quality of the results and, if they are not satis­fied, switch to some other search engine. Increased transparency should resolve most of the problems associated with any potential discriminatory search bias in vertical search.⁷

The European Commission has been in discussions with Google since 2011 on how the concerns can be alleviated through binding commitments. In 2014,⁸ Google proposed a threefold remedy for its current and future specialized services. First, users would be informed by a label indicating Google’s own services. Second, Google services would be graphically separated from general search results. Third, Google would prominently display three links to three rival specialized search services in a format that is visu­ally comparable to that of links to its own services. Joaquin Almunia, then European Commissioner for Competition, stated that the objective of the Commission is not to interfere in Google’s search algorithm but to ensure that rivals can compete fairly with Google. The Commission, at that time, stated that the concessions ‘are far-reaching and have the clear potential to restore a level playing-field in the important markets of online search and advertising’.⁹ However, with the change of Commissioners, following the appointment of the new Commission in 2014, the European Commission has changed its view and issued the statement of objections, mentioned above.

9.4