Google and European Competition Law

Paper for Telecommunications Policy Research Conference 2011, Arlington VA

August 2011

Martin Cave, London School of Economics Howard Williams, Oxford Internet Institute

Introduction

Google has been in the wars for the past year or so. Its status as a new kind of beneficent company has increasingly been questioned, and it has faced more and more criticism in Europe and elsewhere. Critics have been active on two principal fronts – privacy and market power. As the months pass, the latter issue has taken higher salience, with the launch of an investigation by the Federal Trade Commission in the US in June 2011 and an investigation begun in November 2010 by the key European competition authority, the European Commission, of complaints made by three companies. In Europe this is in addition to actions instituted within individual member states, such as France.

The goal of this paper is to consider Google’s interactions to date with European competition law, particularly the Merger Regulations, and to discuss possible bases in the future for impugning Google’s conduct under that law. In some ways this is premature, as competition investigations advance at best at a slow and steady pace. For example, the Microsoft case in Europe, which is often cited as a point of comparison with any action against Google, took nearly a decade from start to finish, even if it did ultimately involve fines on the company of the order of €1 billion. Nonetheless the Google investigation does raise some extremely interesting issues which may help to illuminate the future regulation of search and other high tech activities, whatever the outcome of the current complaints.

The paper is organised as follows. Section 1 gives a very brief outline of the relevant European law, while section 2 discusses some aspects of search and related activities. Section 3 discusses the treatment of two mergers and acquisitions involving Google in Europe, and section 4 seeks to set the current investigation within the framework of competition law. Section 5 notes the results of a completed inquiry in France, and Section 6 gives some brief and tentative conclusions. .

1. European competition law.

The foundation of European competition law is two articles of the Treaty, currently numbered 101 and 102. As Articles of the Treaty, they take precedence over all other instruments, including legislation sectoral rules, and any ‘soft law’, such as regulations, decisions, notices, guidelines etc. Article 101 prohibits agreements among firms. Article 102 prohibits the abuse of dominance. It states: ‘Any abuse by one or more undertakings of a dominant position within the common market or in a substantial part of it shall be prohibited as incompatible with the common market insofar as it may affect trade within Member States. Such abuse may, in particular, consist in a) . . d).....’

The non-exclusive list of examples includes unfair prices or terms, limiting production, discrimination and bundling and the like.

Precedents for the application of Article 102 are created by Judgments of the European Court

of Justice and by the administrative practice of the European Commission, which issues Decisions, some but not all of which are appealed to the Court.

In a case in 1978, the Court defined the key term ‘dominance’In Article 102 as

‘a position of economic strength enjoyed by an undertaking which enables it to prevent effective competition being maintained on the relevant market by affording it the power to behave to an appreciable extent independently of its competitors, its customers and ultimately of its consumers.’1

In a subsequent case, the Court stated that a firm in a dominant position

‘has a special responsibility not to allow its conduct to impair undistorted competition in the common market’.2

Since 1990, mergers with a European (‘Community’) dimension are handled by the European Commission within the framework of the European Community Merger Regulation or ECMR. Whether a merger or concentration has a Community dimension depends upon various turnover thresholds. When the Commission has jurisdiction it has to decide within specified time periods whether a merger could significantly impede effective competition in the common market or a substantial part of it. It asks, in particular, whether the merger could create or strengthen a dominant position. The Commission can approve a merger, or block it, or allow it to proceed after commitments offered by the parties.

2. Google, search and related activities

In this section we make some brief observations about Google and also about the nature of search activity.3 Their purpose is to open up discussion of the nature of the dominance issues which search can raise. Subject to the exception discussed in Section 5, European competition authorities have not opined on these matters. We are also aware of but have not analysed the more recent flurry of activity by economists in response to the recently announced investigations.

A. Google

During its first decade Google4 was an extraordinary success story driven by innovation, imagination, fortuitous timing and good luck. Google was started, famously, in a garage by two Stanford PhD students, Sergei Brin and Larry Page.5 Its initial intellectual property was an algorithm which calculated the quality of any web page, and thus permitted the ranking (known as PageRank) of any page. Performed at very high speed, the procedure generated an instantaneous ordered list of where to look for material on the burgeoning web. With the growth of the web in recent years, such prescient anticipation of the need for scale has proved invaluable.

1 Case 27/76 United Brands v Commission [1978] ECR 207, [1978]1 CMLR 429. 2 Case 322/81 Michelin v Commission [1983] ECR 3461, [1985] 1 CMLR 282. 3 This section draws on M Cave and H Williams, The Perils of Dominance: Exploring the Economic of Search in the Information Society, ICOMP, 2011 4 Google is a common misspelling for Googol, or 10 to the power of 100. 5 K. Auletta, Googled: the End of the World as We Know It (2009) is a readable and perceptive source for the early, and later, years of Google.

News of the availability of the search engine took off. By the start of 2000, 7 million daily searches were being undertaken, compared with half a million a year previously. The computing power required grew correspondingly, as did the costs. The key need was to devise a corresponding revenue stream. ISPs could bundle search into a paid-for subscription. For Google and other free-standing search engines, the solution lay in generating advertising revenue.

After trial and error, Google hit upon the long term solution and the basis of its current business strategy. The search results page is divided into two areas, one (normally on the left) containing the “organic” search results, ordered by the adopted algorithm, the other (normally across the top and on the right) containing a list of paid advertisements.6 Advertisers bid on search query terms, known as keywords. The price in question is that paid when a user clicks on the advertisement- the cost per click. The price is based on what was bid by the bidder ranked below, making it a so-called second price auction. However, as the search engine’s revenue is the price per click times the number of clicks, this encourages it to put popular advertisers up at the top. It also wants the advertisements to be attractive to searchers. These two additional ‘quality scores’, together with the bid price per click, determine the ultimate rankings.

There is, however, concern that the factors used by Google to condition the results by adding quality scores have a potential to undermine the economic elegance of the second price auction. Thus whilst the quality score may be useful in inhibiting gaming by firms in terms of search results and their placing there may be distortionary impacts on the prices paid in the auctions. Further there is a concern that Google, at its simplest, is able to manipulate the results in its strategic interest. The working of the Google search engine has spawned a new service activity, search engine optimisation, and invoked considerable critical attention. Whilst the perceived integrity and usefulness to consumers of the search results is critical to the use of Google search tools there is growing concern that the results can be manipulated, for example the recent work by Ben Edelman.7

The basic business model of Google can be understood in terms of its ability to monetise search terms through the provision of online advertising services. The sheer volume8 and resulting scale effects flowing from the success of Google in search delivers significant advantages over its competitors or those wanting to enter the search or search advertising market. Most notably these scale effects allow Google to be able to secure higher revenues for equivalent search terms compared to other search engines, calibrate its auction process and monetise new search terms considerably more rapidly than its competitors and to extract greater profits from the ‘long tail’. In delivering “free” services in the ICT and telecommunications markets (such as search, gmail, Google docs) Google reinforces its market position in search.

6 See D S Evans, ‘The economics of the online advertising industry,’ Review of Network Economics, 2008, Vol 7, Issue 3. 7 For several recent articles examining the bias in search results see http://www.benedelman.org/ 8 One measure of the scale of information processed by Google is that it processes approximately 1Petabyte of information every hour – all the books in the Library of Congress is the equivalent of 1.5% of a petabyte

B. Search – Inside the Black Box

Search has become integrated into our daily lives; it has become the interface between ourselves and professional and social use of the Internet. But what actually goes on inside a search engine? Eric Schmidt recently call search a “black box” – an artifact which we commonly understand as something which allows the onlooker to see what goes in and what goes out but nothing else.

It is commonly understood that at the heart of a search engine is an algorithm which relies on a variety of inputs and decision rules to produce outputs. The input is the word or phrase keyed in by the user. The output is what the user sees on the screen, which is usually made up of three types of results:

• a ranked list of ‘natural’ results in the form of hyperlinks and a short text, • a list of adverts (also called ‘sponsored links’), and • results which comprise links to other products and services provided by the search engine or its partners (such as maps, video, etc)

The first thing to understand about what is going on inside a search engine is that these three different kinds of results are not generated by the same algorithm. On the contrary, each is generated by a distinct algorithm designed for very different purposes. In order for them to be categorized more clearly, they can be described as a ‘search algorithm’, a ‘search advertising algorithm’ and a ‘product placement algorithm’.

A search algorithm is not neutral but is intended to produce natural search results which are the best answer to what the user is looking forward. And like all algorithms the best natural results are defined by a set of decision rules which, in the final analysis are exogenous to the algorithm. As noted by a Google executive, Marissa Meyer9,

“Search engines use algorithms and equations to produce order and organisation online where manual effort cannot. These algorithms embody rules that decide which information is “best”, and how to measure it. Clearly defining which of any product or service is best is subjective”

The object, therefore, is to find the best result. This is not the same as the most obvious result, since any search engine can find the obvious results. If you type BBC into a search engine, the first result you expect to see is a link to the BBC homepage. However, the really clever aspect of search engines is that they can find results that are much less frequently searched for – the obscure and rarely searched for result. This is often known as the ‘long tail’. The long tail is not easy to capture. It requires not only lots of users asking lots of questions10, it requires constant updates to the search algorithm.

9 Do not neutralise the web’s endless search, Marissa Meyer, Financial Times, July 14 2010.

10 One measure of the search is the number of new searches each day that a search engine has never experienced before; “a quarter of all daily searches on Google have never been seen before” Meyer, ibid and Eric Schmidt, November 2010, put the number closer to 15% - but each was stressing the importance and volume of new search terms.

Hence the nature of a search engine is dynamic. A Google employee has described it thus11;

“Google ranking is a collection of algorithms used to seek out relevant and useful results for a user’s query. … Our algorithms use hundreds of different signals to pick the top results for any given query. Signals are indicators of relevance, and they include items as simple as the words on a webpage or more complex calculations such as the authoritativeness of other sites linking to any given page. Those signals and our algorithms are in constant flux, and are constantly being improved. On average, we make one or two changes to them every day.”

These mechanical indications combined with historical data, as well as trial and error, are the prime drivers of how search engines seek to find the ‘right’ result. However, it is critical to understand that humans also play a significant role – in the end the decision rules are defined by management decisions. Examples of rules generated by humans include valuing the relevancy of a result to a given query by reference to the length of the website, or the number exclamation marks, or the number of other websites which hyperlink to it or the webpage loading speed. It has even been suggested that the relative success of adverts on a given site (measured by the CTR, or click-through rate) can influence search results.

These human-designed features of search algorithms can be used in a variety of ways to influence search outcomes. For example, they can be used to identify certain values that are characteristic of types of site or online business, such as vertical search or online maps. Those types of site can then be promoted or demoted as a group or even individually. Moreover, given the importance of search advertising revenues to online search, it is interesting, to say the least, that natural search results might be, and in principle can be, skewed towards sites which are most profitable from an advertising point of view. In any event, it is important to note that this is a process in which there is a considerable human element that goes into determining the output from a given search input.

Secondly there is the ‘search advertising algorithm’. This algorithm helps set the prices that advertisers pay for the ‘sponsored links’ that appear at the side and above the ‘natural’ search results. Under the terms of the auction process for AdWords, buyers of search advertising enter a sealed bid (which other buyers do not see it) for the maximum amount they are willing to pay for the keyword they wish to bid on. The amount the winning bidder actually pays is the higher of one cent more than the next bidder, or one cent more than the minimum cost-per-click (CPC) set by Google. The search advertising algorithm determines at what level Google sets the minimum cost-per-click.

The minimum CPC or reserve price is set individually for each bidder and is made up of the following factors (Source: Google):

- the historical clickthrough rate (CTR) of the keyword and the matched ad on Google - your account history, which is measured by the CTR of all the ads and keywords in your account - the historical CTR of the display URLs in the ad group - the quality of your landing page - the relevance of the keyword to the ads in its ad group - your account’s performance in the geographical region where the ad will be shown - other relevant factors

11 This stuff is tough, Amit Singhal, Google Fellow, 25 February 2010, Google European Public Policy Blog

In other words, Google’s reserve price for each individual AdWords bidder is set on the basis of its search advertising algorithm.

Finally there is what we refer to as the ‘product placement algorithm’. This algorithm is taking on more significance as search engines, and Google in particular, create a ‘stack’ or bundle of related products and services based on search. The mechanisms by which this product placement algorithm works are captured in the following quotation:12

“People often ask how we rank our “own” content, like maps, news or images. In the case of images or news, it’s not actually Google’s content, but rather snippets and links to content offered by publishers. We’re merely grouping particular types of content together to make things easier for users.

In other cases, we might show you a Google Map when you search for an address. But our users expect that, and we make a point of including competing map services in our search results (go ahead, search for “maps” in Google). And sometimes users just want quick answers. If you type “100 US dollars in British pounds,” for example, you probably want to know that it’s “£63.9p”—not just see links to currency conversion websites.”

This seems to mean that Google does not apply its usual search algorithm to its own products and services. Instead, it seems to apply a distinct and separate ‘product placement algorithm’ so that the most relevant of its own products and services (and sometimes it seems those of its partners) are placed in the leading results. This forms part of the complaint to the European Commission.

The user experience is the outcome of the interplay between these ‘algorithms’. In order to maximize the value proposition there is clearly great potential for a search engine to run its ‘search algorithm’ and ‘product placement algorithm’ together. The former can be used to weed out content that Google prefers to see in a lower ranking, whilst the latter is used to place Google’s own product and services high up the list.

As with the link between the ‘search algorithm’ and ‘product placement algorithm’, there is also a link between the ‘search algorithm’ and ‘search advertising algorithm’. The lower the ranking a site obtains under the search algorithm, the more likely it is to have need to the recourse of search advertising to get the attention of users. Moreover, the lower the ranking it obtains, the higher it will have to bid to obtain the keywords it needs to get to the attention of viewers.

It is vital for web users and online businesses to know how search results are determined – what does go on inside the black box? Whilst there is general consensus that search results should filter out spam and other unwanted material, web users and businesses need to be aware whether the search results they are presented are deliberately skewed to favour certain commercial services, e.g. those provided by the search engine itself.

12 Amit Singhal, WSJ, 17 September 2010

C. Market power in search and related areas

Internet search is a classic two-sided market in which the search engine is an intermediary between those searching for information and those placing advertisements. The price which searchers pay is exposure to advertisements - which may be a cost or a benefit13 .The possibility that they are an imposition discourages too many, or too intrusive, advertisements. Clearly, the most popular search engine is likely to earn the largest revenues and deprive its rivals of funds with which to compete. The European Commission has cited data showing that Google has a market share of search in Europe of around 95%14 and officials have stated that the market share figures give a good indication of market power.

The economist Daniel Spulber has undertaken an analysis of what he calls the ‘circular flow of information managed by a search firm’.15 What distinguishes the search engine from many other firms at the centre of two-sided markets is the amount of information the search engine has about its customers and advertisers – the volume and detail of this information may be the key to interpreting Larry Page’s observation about search being the be-all and end-all.

Searchers reveal the product categories in which they are interested by their search terms, by the content (including with Google Books, their choice of literature) which they view, by clicking on advertisements and by their on-line purchasing arrangements. Search engines convey information to advertisers on search terms and keywords and how much value is attached to these words and terms through the auctions16. The advertisers can observe the success of their own advertising, in terms of generating click-throughs and purchases, and accordingly make adjustments to the bids they place for key words. They also disclose to the search firm what the keywords and search terms the advertisers are interested in and, by their bids, the value of these keywords and search terms. New data are constantly being generated and processed, and the search firm is the spider in the middle of the web. The search engine is able to set the terms of trade based on access to all the information in a “game” where it holds all the cards.

The economics of advertiser-supported search revolve around fixed costs, economies of scale and features of the demand side. The fixed cost element is common to almost all activities and need not detain us long. A search firm of any size will have to implement certain procedures, such as upgrading its software, and these will vary less than proportionately to scale. The volume of computer power and their co-ordination required to undertake large scale search beggars belief. It is likely that concentrating these activities in a few large sites scattered around the globe, consuming as much electricity as a small town, is the least cost procedure.

In this case, platform economies of scale are augmented by advertiser economies of scale, leading the latter to prefer large campaigns which are cheaper to set up and monitor than a series of smaller ones. It may also be that platform economies of scale are reinforced through

13 It is probably reasonable to assume that the ad cost also features in the price of the goods for those who do purchase 14 Joaquín Almunia Vice President of the European Commission responsible for competition policy Competition in Digital Media and the Internet UCL Jevons Lecture London, 7 July 2010, SPEECH/10/365.

15 Daniel Spulber, ‘The map of commerce: internet search, competition, and the circular flow of information’, Journal of Competition Law and Economics, 5 (4), 633-682., 16 During his participation to the Council of Foreign Relations debate on disruptive technologies, Eric Schmidt said that each day 15% of the queries Google see each day are new, search terms that Google has not seen before. New York November 3rd 2010. (video source: http://www.youtube.com/watch?v=eJAMD5p5tQo)

contracts that limit the use of other, competing platforms; for example, Google often adopts contractual restrictions on advertisers to prevent efficient running of a campaign on multiple platforms.

As noted in the above section, on the supply side, the search engine company optimises the results it presents not just in terms of the natural research results but also in terms of optimising advertising and product placements. Hence the search engine as a platform is not neutral, but has its own optimising function. A view on the practical applications of this optimisation can be seen in the arguments made by some for search neutrality.17

The demand side is more interesting. As Evans notes, platforms such as Google’s offering search advertisements generate indirect network effects.18 These are not the same as those which accrue to mobile networks which offer cheaper rates for on-network calls, since price- based signals are not relevant on the consumer side in advertiser supported regimes. Instead, they come into being because the thicker the market (the greater the amount of activity on the platform), the greater the number of feasible trades and the greater the chance that consumers’ heterogeneous tastes can be satisfied. As Evans further notes, there is a positive feedback loop in operation involving both sides of the market.19 The power of the data grows with accumulation- both within a given market horizontally and, as we shall see, as other related advertising markets are penetrated

Can a small search firm leapfrog over another larger one? It could in principle if it were more efficient. Or it could provide search advertisement services to other search sites. But, as Evans shows, a larger search firm is likely to be better placed to do this, and hence to outbid a smaller one for third party traffic. Also, in the case of Google it is has, through its wide product portfolio across fixed and mobile internet services, created a multitude of default options that effectively lock consumers into Google; for example, Android and Chrome default consumers to Google services.

Evans’ conclusion is that while there are factors at work (most notably the feedback loop) which might lead one to suppose that this is a case of ‘winner-take-all” competition, there remain others which go in the opposite direction. This suggests the need for a case by case analysis of any competition problems in the relevant markets.

3. Google mergers in Europe

As Section 1 showed, both Article 102 and the Merger Regulations are concerned with dominance, the former as a precondition for intervention, the latter as an outcome of a concentration. But dominance is evaluated in relation to some set of goods or services. Market definition is thus a key element in a European competition investigation; the alternative option of focussing directly on constraints is not available.

In this section, we review two merger proceedings undertaken by the European authorities. (Other famous acquisitions, such as those of AdMob and IDA travel did not meet the

17 See http://www.searchneutrality.org/ 18 D S Evans, ‘The economics of the online advertising industry’, Review of Network Economics, 2008. 19 Ibid., p 374.

European thresholds for an investigation.) Paradoxically, one of them – the tie-up between Microsoft and Yahoo! did not involve Google, even though the latter’s existence had a major impact on the outcome. The other merger was Google’s take-over of DoubleClick. This was a merger between vertically-related (or perhaps, diagonally related firms), rather than between firms in the same horizontal market.

The European Commission has published non-horizontal merger guidelines in 2008 which are quite hospitable to such concentrations, while they recognise the possibility of foreclosure.20 A three step process is proposed, asking - does the merged firm have the ability to foreclose competitors? - does the merged firm have the incentive to engage in business conduct which forecloses competitors? - what is the overall impact on competition – ie on consumers?

In the event, Google/Double Click was the first application of the non-horizontal merger guidelines.

As Auletta explained Google’s wish to acquire DoubleClick as follows:21

‘DoubleClick was as dominant in its arena, placing display advertising- as Google was in placing text ads. DoubleClick provides the digital platform which allows sites like MySpace to sell online ads and advertisers and ad agencies to buy them, with DoubleClick culling from its data base the information that targets the ads.’

In other words, DoubleClick was in the ad server market: it ensured, once an advertiser had bought space, that the correct advertisement appeared in (was served to) the right place at the right time. The tools which accomplish this task can also be employed to measure the performance of the advertisement, for example by tracking the behaviour of the user, via recording of click-through rates etc.

The Commission was inclined to identify a separate market for search advertising, as distinct from online advertising or even a wider market. But because the outcome did not depend on this decision, it was able to avoid making a formal decision.22 It also addressed the question of whether direct ad sales and intermediated sales were in the same market. Again, it could pass on this question. DoubleClick’s activity, display ad serving technology, was found to comprise a market separate from ad serving technology for text ads.

The key issues were non-horizontal. Complainants and some respondents to the questionnaire issued by the commission expressed concerns that DoubleClick’s offerings might be redesigned to make them more costly in use with advertising competing with Google, leading ultimately to a tipping effect which would marginalise competing networks. The

20 S Bishop and M Walker, The Economics of EC Competition Law: Concepts, Application and Measurement, 2010, ch 8. 21 Auletta, op. cit. in fn. 5, p. 174. 22 This is standard practice in merger inquiries. The FTC concluded: ‘the advertising space sold by search engines is not a substitute for the space sold directly or indirectly by [online] publishers or vice versa. Or, to put it in terms of merger analysis, the evidence shows that sale of search advertising does not act as a significant constraint on the prices or quality of other online advertising sold directly or indirectly or vice versa’. Statement of the Federal Trade Commission Concerning Google/DoubleClick (FTC File No. 071-0171), p. 3.

Commission did not find sufficient evidence of direct and indirect network effects for it to be likely that this would occur. Conversely, Google might use its market power in search advertising to make customers also use DoubleClick’s services. The Commission concluded that this strategy would not be profitable.

A further non-horizontal issue considered was a diagonal one. Suppose that DoubleClick increased the cost of display advertising; this might cause customers to switch to text advertising, to Google’s benefit. The commission found this hypothesis unconvincing. Accordingly it cleared the merger.23

The second decision we consider does not involve Google directly, but its presence is powerfully felt nonetheless. Famously, the story began with an attempt by Microsoft to buy Yahoo! In contradistinction to Google’s general opposition to intervention, Google petitioned the government to block the bid.24 Going further, Google announced that it would become the selling agent for a large part of Yahoo!’s search advertisements. But when large numbers of customers objected, and the Department of Justice indicated that it would intervene, reportedly just three hours before the Department was to file anti-trust charges, Google dropped the plan to partner with Yahoo.25

Microsoft then re-entered the fray. Its proposal to acquire the on-line algorithmic search and search advertising businesses of Yahoo! Was notified to the Commission in January 2010. In the course of investigating this concentration, the Commission compiled market share data. It was noted that:26

“The internet search markets in the EEA (European Economic Area) countries are very concentrated. Google is by far the platform that enjoys the highest market shares. In 2009, Google has market shares above [90-100]% in most national markets, whereas Microsoft and Yahoo both have market shares below [5-10]%.”

According to the Commission, Google’s shares of paid search data in September 2009, the latest available when the Commission was making its decision, were as follows:27

90-100%: Austria, Belgium, Germany, Denmark, Finland, Spain, France, Ireland, Italy, Netherlands, Sweden.

80-90%: UK, Europe. 70-80%: world

60-70%: USA.

On the one hand, the two companies involved had a small combined market share which was significantly lower than that of their main competitor, Google, and historically these companies had encountered serious difficulties in competing effectively against Google as

23 Julia Brockhoff et al. ‘Google/DoubleClick: the first test for the Commission’s non-horizontal merger guidelines.’ Competition Policy Newsletter, no 2 2008, pp54-5. See also Bishop and Walker, op. cit. in fn 19, pp 449-451. 24 Thereby illustrating, according to the wry Auletta (op. cit. in fn. 5, p.246), Emerson’s dictum that ‘a foolish consistency is the hobgoblin of little minds.’ In other words Google did what any other company would have done on receipt of advice from their lawyers concerning how to maximise shareholder value. 25 DOJ press Release, Yahoo! Inc and Google Inc Abandon their Advertising Agreement, Nov 5, 2008. 26 Case no. COMP/M.5727-MICOSOFT/YAHOO1SEARCHBUSINESS, para 112. . 27 Ibid. para 114.

separate entities. On the other hand, the concentration was a merger between the second and third players, in a market where barriers to entry appeared to be high. Overall, the market investigation confirmed that scale is an important factor in order to be an effective competitor in this sector. The market investigation further confirmed that the proposed transaction was perceived as having pro-competition effects, as it would create a stronger competitor to Google. Accordingly the Commission cleared it on 18 February 2010.28

These two cases presented the Commission with fairly easy tasks. Almost all merger notifications are approved by the Commission. Whish notes that in the entire life of the ECMR, the Commission has blocked only 20 concentrations, of which four were overturned on appeal to the courts. In 6.5% of cases, commitments were accepted.29 It is therefore likely that the Commission might well go down the same route of complex behavioural regulation as the FTC has gone in the ITA Software Inc case.30

4. The Google complaint before the European Commission

In November 2010, the European Commission announced that it was opening an inquiry into allegations of anti-trust violations by Google.31 The complaints were made by search service providers about unfavourable treatment of their services in Google’s unpaid and search results, coupled with an alleged preferential placement of Google’s own services. A further complaint was taken over form the German competition authority in December 2010. Microsoft added its complaint in March 2011.’32

The Commission notes that initiation of proceedings does not imply that the Commission has proof of any infringements. There is no legal deadline to complete inquiries. Possible outcomes of the investigation include the issuing of a Statement of Objections and closing of the investigation with no further action taken. No information has been, or is likely to be, officially released about the investigation before it is completed.

In relation to allegations that Google has abused a dominant position in online search, the preliminary steps are market definition and proof of dominance. This is unlikely to be problematic, though it may be contentious. Google may be approaching another informal threshold used in some observations by the Commission and others, but not by the courts or formally by the Commission – that of super-dominance, for which a position of dominance approaching a monopoly is required. A firm in such a position can be held to a higher standard in respect of its special responsibility to protect the competitive process.33

The most telling of the complaints levelled against Google is that it lowers the ranking of

28 Ibid., paras 251-257. 29 R Whish, Competition Law, 6th edition, p. 819 30 US Department of Justice Press Release 8 April 2011. 31 Press Release IP/10/1624, 30 November 2010. 32 Its general counsel remarked that: ‘having spent more than a decade wearing the shoe on the other foot with the European Commission, the filing of a formal antitrust complaint is not something we take lightly.’ http://blogs.technet.com/b/microsoft_on_the_issues/archive/2011/03/30/adding-our-voice-to-concerns-about- search-in-europe.aspx 33 See Whish, op. cit. in fn. 27 above, pp 184-6.

unpaid search results of competing services providing specialised content (so-called vertical search services) and raises the ranking of its own competing services. Since users tend to click predominantly on the first few entries on the screen, it is reasonable to assume that this is very effective. One of the complainants to the Commission, Foundem, has adduced evidence which shows that a sudden and otherwise unexplained elevation of the ranking of Google’s own price comparison sites had major impact upon their market share. One Google official appears to have acknowledged that this happens.34

The second issue referred to in the Commission’s press release is the allegation that Google lowered the ‘Quality Score’ for sponsored links of competing vertical search providers. By doing so, Google raises the hurdle price which the competing service has to bid to appear at a given rank in the short list of sponsored links at the right of the displayed page of search results.

Two further allegations are made. First, that Google imposes certain types of exclusivity obligations on advertising partners and on computer and software vendors, and that Google imposes portability restrictions on online advertising campaign data.

Several of these complaints raise the fundamental question of whether the conduct is consistent with Google’s obligations as a dominant firm. According to European competition law, can Google say: it is our platform, and we can do what we like with it, or is it precluded from discriminating in favour of its owned sites?

It is likely that this hinges upon the degree of replicability of the platform. This question can be addressed in several ways, including the more conceptual approach developed by Spulber and discussed above. More prosaically, consideration can be given to the mechanics of a de novo entry into search. In the case of universal search, this will involve a very large fixed cost of gaining access to and indexing a very large number of web pages in order to get started.

However, that is not all. The largest search engine will be best placed to undertake a more detailed indexing, which will improve the accuracy of search and benefit both user and advertiser, thus generating higher per unit revenues. Virtuous and correspondingly vicious circles are created.

This arrangement operates to a lesser degree in respect of vertical search engines, which by virtue of their specialised nature require less investment. However, there is likely to be an economy of scope between the operation of universal and vertical search engine, which may create a barrier to the entry of free-standing vertical operators.

If the European Commission accepts this description of the situation with respect to costs and entry, then a number of possible avenues might be open to it. Some of these are canvassed in a recent article by David Wood, which discusses precedents in other sectors such as transport and financial services.35

34 See Why search matters, ICOMP, 2011, p. 10. 35 D Wood, EU Competition Law and the Internet: Present and Past Cases, Competition Law International vol 7 (1), 2011.

For example, in Sea Containers v. Stena Sealink, the owner of a port refused to allow access to it by vessels competing with its own ferry services. The Commission’s Decision found that36:

“the owner of an essential facility which uses its power in one market in order to protect or strengthen its position in another related market, in particular by refusing to grant access on less favourable terms than those of its own services, and thus imposing a competitive disadvantage on its competitor, infringes Article 102”

A closer factual parallel might be found in computer reservation system (CRS) in the airline industry, where in Sabre/Amadeus, an airline owning such a system was required to make it available to its competitor, and was obliged to apply the principle of neutrality to all users.37

It is, of course, a stretch to apply the concepts of essential facilities developed primarily for traditional network industries to search. From Google’s point of view, this would be a very disappointing outcome.

5. Google in France

Among the member states of the European Union, France has probably been hardest on Google. In October 2010, the French Competition Authority found that Google had applied its rules abusively to terminate the Adwords account of Navx, which provides a GPS navigation service. Google’s behaviour had been non-objective, discriminatory and non- transparent. The Authority accepted commitments from Google including a clarification of the scope of its content policy and a warning system that AdWords accounts were to be suspended.38

In June 2011, Google was sued by a French competitor 1plusVfor €295 million in a French court for alleged anti-competitive conduct. (Such suits come with the territory of successful firms.) But more significantly in December 2010, the same French Competition Authority delivered an opinion on the market for online advertising, sought by the French Government.

The Authority reaches a nuanced conclusion. It recognises that search engines ‘give rise to functionalities that have genuine use and have the potential to create significant value added’. It notes that ‘sponsored links offered by search engines have a specific function in the eyes of advertisers’; that ‘Google has a dominant position in the market for search-based advertising’, but that ‘that dominance is not unlawful in itself: it is the result of extraordinary innovative effort and was supported by significant, sustained investments.’ It concludes that competition law can limit Google’s conduct in response to the competitive scenario described by participants [in the inquiry]’, and ‘does not recommend the enactment of a general regulatory framework’.39

36 Case COMP IV/34.689 37 CaseIV/32.318. 38 Autorité de la Concurrence, Décision no. 10-MC-01 du 3 juin 2010 relative à la demande de mesures conservatoires presentée par la societé Navx; Déecision no. 10-D-30 du 28 octobre 2010 relative à des pratiques mises en oeuvre dans la secteur de la publicité sur Internet. 39 Autorité de la Concurrence, Decision no. 10-A-29 of December 14 2010 regarding competition in online advertising, (English version) paras 398, 401-2, 404-5.

Interestingly, the French telecommunications regulator ARCEP, in its paper on net neutrality, discusses the role of Internet society service vendors or ISVs, which include search engines. It proposes that ISVs must ‘comply with a principle of non-discrimination in different operators ability to access offers, and with principles of objectivity and transparency with respect to users, in terms of the rules employed in cases where the ISV selects and/or ranks content coming from third parties. which is notably the case with search engines.’40

ARCEP says that it will await the opinion of the Competition Authority (summarised above), but some form of regulation of search does not seem ruled out.

6. Conclusion

The goal of this paper is not to predict the outcome of the European Commission’s investigation into Google, nor to speculate whether the outcome in Europe will be the same as or different from that in the United States.41 Instead we have tried to set out some of the European context in terms of prior decisions in cases involving Google and other precedents.

There is unlikely to be much doubt about Google’s vast superiority in market share in search and search advertising in Europe, but there is room for more than cursory debate about its significance and persistent or possibly transitory nature. If competition is, as Google often states ‘just a click away’, then market shares may not signify very much. If, on the other hand, the search market exhibits major barriers to entry, then the existence of a use of a persistently dominant or super-dominant position would carry special responsibilities, which may well include refraining from at least some of the practices alleged against Google by complainants. These might extend to a general obligation on Google not to favour its own upstream and downstream activities in presenting its organic or sponsored search results. The presence or absence of obligations of this kind seems to be the key to the European case.

40 ARCEP, Internet and network neutrality: proposals and recommendation, September 2010 (English version) pp. 44-46. 41 In the Microsoft case, the outcomes were very different.