You are on page 1of 19

Central European University

Network effects in two-sided markets

0 Challenges in merger regulation
Term paper for the course Competition policy: Economics in Applying European Competition Law Author: Bertalan Papp

In my paper I present the underlying economic theory of two-sided markets and its implications to the assessment of mergers involving indirect network effects. The major case of the Google/DoubleClick merger raised a particular concern regarding the potential nonhorizontal foreclosure of competitors, an allegation which was

grounded in two-sided market theory. Despite the sound theoretical substantiation of the concern, the competition authorities involved (the EC and the FTC) eventually cleared the case. The case nevertheless provides a fine example of how the two-sided nature of a market may alter the tools required for proper competitive assessment. As an aggravating factor, this nature was taken into proper consideration by the authorities; however they concluded that the necessary conditions for it to become harmful were not fulfilled.

Table of contents
I. II. III. Introduction .................................................................................. 1 Two-sided market theory.............................................................. 2 Google/DoubleClick merger ........................................................... 5

Online advertising industry structure ................................................... 5 Competitive environment................................................................... 8 Theories of harm .............................................................................. 9 The role of network effects ............................................................... 10 Merger assessment .......................................................................... 11 IV. Conclusion ................................................................................. 13



The Google/DoubleClick merger was the first major case of concentration subsequent to the European Commissions adoption of Non-Horizontal Merger Guidelines. It covered horizontal, vertical and conglomerate merger effects also. Following a thorough investigation at the same time and in cooperation with the Federal Trade Commission, the EC has cleared the merger arriving at the same conclusion as the FTC, i.e. the merger is unlikely to significantly impede competition on the online advertising market.

Among some other concerns, the case raised the particular question of unilateral non-horizontal merger effects. Potential foreclosure either in the ad serving or the ad intermediation market, stemming from indirect network effects internalized by the merged entity, may restrict competition and enable the merged entity to raise prices above competitive level to the detriment of consumers. The assessment of these levels however must depend on firm consideration of the proper relevant market on which the infringement might occur, a requirement that is difficult to fulfill in dynamic markets like online advertising and internet publishing content.

The market definition must provide solid grounds for the theory of harm, in particular because of the further analytical difficulties arising from the twosided nature of the market. Unless these difficulties are addressed with well1

suited theories and the methods based on them, the outcome of the investigation eventually might harm consumer welfare and allocative efficiency. The paper focuses on the potential exclusion strategies the merged entity might use to raise rivals cost, thus barriers to entry or expansion by leveraging market power from the ad serving markets to the market of ad intermediaries, or vice versa. The resulting foreclosure might tip the market toward monopolization due to indirect network effects. There are three main conditions for this to happen, which were extensively analyzed and finally dismissed by the authorities. Section II provides a brief overview of twosided market theory; Section III summarizes the Google/DoubleClick merger case according to the authorities investigation with a focus on the role of network effects in merger assessment; and Section IV concludes.

II. Two-sided market theory

In the following I give an outline of the economics of two-sided markets in order to provide a theoretical background for the subsequent review of the competition policy analysis. According to the general definition we can identify a two-sided in general terms multi-sided market by the particular characteristic of having at least two separate groups of

consumers, whose utility associated with the usage of the platform that 2

connects them is positively affected by the presence of consumers on the other side of the market (see Rochet Tirole, 2003; Roson, 2005; Evans Schmalensee, 2005; Armstrong, 2006). Put it simply, the groups provide each other with network benefits. Due to these indirect network externalities their demand and consumption not only depends on the internal

characteristics of the product, but also on their valuation of the other groups extent. The markets with network externalities present are mostly organized by such platforms that enable two sides to interact with each other. An important consequence of this market structure is that the price set to a respective group influences both the demand of the consumers directly concerned, and the demand of the consumers at the other side of the platform who face a separate price. This character of the pricing decision results in a unique optimization problem for the owner of the platform, who sets the respective prices: beside the level of prices, their relative size, that is their structure also matters in getting both sides of the market on board (Rochet Tirole, 2003). Competitive prices of platforms therefore even in theory might diverge from marginal costs, taking on zero or even negative values. The evolution of price structures is rather determined by the own-price elasticity of the consumer groups, and the underlying direction and effect of indirect network externalities conveyed in these elasticities (Evans Schlamensee, 2005). As a consequence, pricing strategies may result in a disproportionately higher 3

price set to the group with the relatively lower price elasticity (profit center). This higher price however only ensures the presence of the other group with the relatively higher elasticity (loss leader), by at least in part paying for their usage of the intermediary platform (Rochet Tirole, 2003). From a competition policy point of view, network effects and two-sided markets are particularly interesting, because the involved industries show a tendency toward concentration (Evans - Schlamensee, 2007; Filistrucchi, 2011). Despite the economies of scale gained through concentration, a merger may lead to higher market power and consequently a more independent conduct, therefore higher price margins. It is not

straightforward to tell however, whether this is due to the increased demand that is associated with a higher level of utility from consumption (which can be derived both from the usage and the utilization of the platform, i.e. the extent of the separate groups); or an exploitation of excessive market power. The tools for market definition and competitive assessment therefore must be tailored to the inherent characteristics of two-sided markets otherwise the investigation would possibly arrive at false conclusions (Evans Noel, 2007).

III. Google/DoubleClick merger

Online advertising industry structure
In the following I draft the industry structure for online advertising in time of the merger, based on the information that the investigation of both the Federal Trade Commission and the European Commission has revealed (Baye et al., 2008; Brockhoff et al., 2008, European Commission, 2008; Federal Trade Commission, 2008; RBB, 2008). As the two authorities final verdicts were in harmony, based on their jointly concluded evaluation I present not only the industry structure, but later its performance and the merger assessment also. Online advertising is assumed to constitute a different market from advertising in other types of media due to its unique level of targeting efficiency. The industry consists of providers of advertising space, i.e. online publishers, and the advertisers, who possibly alongside with other offline types purchase online ad space to reach consumers (see Figure 1 for an outline of the industry structure). Google is a major provider of online advertising space, who also provides online advertising intermediation services, therefore is deemed as an integrated publisher. At the time of the merger DoubleClick was a leading provider of standalone advertisement serving technology: a service, which ensures that the advertising appears in compliance with the agreed upon terms on the purchased space. Ad serving 5

has an additional parallel function: advertiser-side and publisher-side services provide the customers of ad service with the measurement and tracking of information, i.e. the monitoring of the performance of the advertisement.

Figure 1. Online advertising industry structure Online advertisements however are rather differentiated: the two broad categories are simple text ads and display ads. There is differentiation also with regards to the targeting of the advertisement: it can be put to place based on search queries; according to the context of the publisher environment on which it appears; the targeted users behavioral track 6

record; or other characteristics like geographic location or the actual time. Google specializes in text ads targeted according to search queries on its search platform and contextual targeting of text ads on third party publishers sites. DoubleClick on the other hand typically provides ad serving technology as an input for the placement of display advertisements. Not only advertisements but the advertisement space available is also differentiated by its quality, and consequently by its distribution channel: the more valuable online space of large publishers, i.e. the premium inventory is usually distributed directly and mostly without the intermediation of ad networks, while the less valuable, i.e. remnant inventory is usually sold through these intermediaries. Beside ad networks acting as distribution channels, ad services acting as tools are also different according to the type of the advertisement: display ads require differentiated service technology relative to simple text ads. The substitution between these differentiated forms of advertising is assessed to be limited, therefore separate markets for text and display ads are suggested. It is more so, because a significant part of online advertising sales is attributable to large publishers (premium inventory) and is contracted directly with large advertisers. Only the remaining part of advertising space (remnant inventory of large publishers or small publishers ad space) is made use of through ad networks. The exact market definition however was eventually left open by the EC, mainly due to the relatively 7

new and dynamically developing industry context. On the other hand the FTC concluded that both search ads and non-search ads, and direct sales and intermediated sales are constituting a separate relevant market respectively. As for geographic markets the advertising market is nationally and linguistically limited, while the two-sided market for intermediation is concluded to be at least Europe-wide. Although Google and DoubleClick were not direct competitors due to these differentiation considerations and the limited nature of substitutability, at the level of distribution, Google competes with such intermediaries to which DoubleClicks ad serving technology is an input.

Competitive environment

Google is an integrated player in the online advertising industry, possessing its own advertising intermediation network (AdSense/AdWords) that

provides search and contextual ad services to purchasers of Googles advertising space and also for other advertisements. The main competitors of Googles leading position in online search advertising are Yahoo! and Microsoft. On the market for intermediation these companies has active own subsidiaries.

DoubleClick is the leading provider of ad serving technologies. Its main competitors are aQuantive/Atlas (advertiser side), 24/7 Real 8

Media/OpenAdStream and AdTech/AOL (publisher side). These companies are concluded to exert effective competitive pressure, supported by historically present switching propensity and declining prices in spite of the increasing demand.

Theories of harm
Although there were various concerns raised related to horizontal merger effects, the investigation concluded that the parties were not engaged in direct competition and did not impose any significant competitive constraint on each other pre-merger. As the focus of my paper is on indirect network effects arising from the two-sided nature of the market, I leave the presentation of horizontal effect assessment and concentrate on nonhorizontal effects. Non-horizontal merger effects may include the foreclosure of rivals by constraining access to key competitive assets. By raising costs directly to those customers who use DoubleClicks advertiser- or publisherside services; or indirectly to intermediary competitors of Googles AdSense platform, the merged entity might be able to restrict or exclude competitors. As for the direct price increase, concern were raised that a potential increase in either the publisher-side or advertiser-side service of DoubleClick would raise the cost of selling or purchasing display advertising space, so that advertisers or publishers may switch from display to text advertising, therefore the usage of Googles integrated product. Another possible way of

leveraging market power from one market to another is to offer bundled or tied services of AdSense and DoubleClick. As a consequence consumers of DoubleClick would at least in part switch from intermediary competitors to AdSense. These theories of harm however were explicitly depending on the presumption of the significant market power of DoubleClick that can be leveraged to the intermediary market.

The role of network effects

To all articulated theories of harm, there was a severe aggravating factor underlying, which had to be analyzed extensively in the course of the investigation; namely the impact of network effects. Due to the presence of indirect network externalities in the two-sided intermediary market the above outlined restrictions to competition may induce a tipping effect. As the success of a platform is mainly driven by the extent of the respective groups to which it provides access, it enables the merged entity to raise rivals cost by depriving them of this access, and therefore of the ability of performing an intermediary function that could compete with the quality of Googles. This is so because both publishers and advertisers appreciate the increased probability and expected value of the matching provided by the intermediary platform. The barriers to entry and expansion constituted by the attractive potential of a network stem from the critical size of the network. In turn it would result in a potential to raise prices unilaterally. This effect might eventually marginalize competing platforms as a result of the concentration, 10

leaving Google with the dominant position in the market for intermediation and sale of search and contextual advertising. This theoretically plausible presumed tipping effect however depends on the ability and incentives of the merged entity to engage in a unilateral price increase or certain exclusionary strategies. The profitable and effective price raise is dependent on three main factors: (i) the presence of effective competition in the display advertisement service market and the associated market power of DoubleClick; (ii) the extent of network externalities prevailing in the market for advertisement intermediation (iii) the price share of display ad service in total cost of display advertising and the closeness or distance of substitution between display and text advertising.

Merger assessment
The proposed theories of harm are refuted mostly based on these factors. As revealed by the investigation, there is effective competition constraining the conduct of DoubleClick in the ad serving market. Both entry and competition exerts sufficient competition constraints to leave no potential for the merged entity to increase its prices. The theory of harm in two-sided markets also raised the question of single- or multi-homing behavior of the publishers and the advertisers group. In case single-homing is typical among costumers, intended foreclosure succeeds with higher probability due to the higher switching costs. It is crucial therefore in assessing potential










participation in either of the two markets is exclusive or binding. The authorities investigation revealed that in case of intermediary services multihoming is more typical as there are no significant initial fixed costs attached to joining another ad network. Therefore it is implied that consumers are both able and willing to participate in more than one intermediary service, moreover they have already demonstrated to do so. In the presence of multi-homing, the indirect network effects coming from the large extent of a customer base either on the publisher or the advertiser side are alleviated; thus Google becoming the dominant player due to the

concentration, and gaining ability to foreclose intermediary competition is deemed to be unlikely. Incentives to foreclose are also concluded to be limited due to the fierce competition constraints exerted by alternative advertisement service

solutions. Although the price of ad serving represents a substantial part of intermediation costs, the investigation revealed that ad serving and intermediation together constitute only a small share of the total advertising costs of advertisers and also of the total advertising profits of purchasers. Therefore even a relatively large increase in the price of ad serving would only lead to negligible alteration in advertising consumption, be it a diversion from display to text advertisements or the diversion within text

advertisements from one intermediary network to another. 12

IV. Conclusion
It was the explicit aim of Googles acquisition of DoubleClick to attract additional publishers and increase the sales of its intermediary platform AdSense through the better utilization of their remnant advertisement space inventories. This however, is only the consequence of a business rationale driven by strong economies of scale and synergies stemming from indirect network effects. This would possibly give way to pro-competitive effects in form of lower prices or improved quality of the intermediation service. The question from a competition policy point of view therefore was whether the merged entity became able to foreclose its competitors through leveraging market power from the ad service market; and would this conduct have anti-competitive effect outweighing the potential pro-competitive gains. Owing to the two-sided nature of the intermediary market, the investigation had to take account of unconventional considerations: intermediary services were deemed to be prone to tipping into monopolization. In case of strong evidence was found on both the ability and incentives of the merge entity to engage in anti-competitive foreclosure, this would aggravate the impact a conduct that would enhance the market power of Googles AdSense by raising rivals costs. The risen theories of harmed however relied on strong assumptions regarding the market power of DoubleClick on the ad serving market, i.e. the level of switching costs; the incentivizing effects of indirect 13

network externalities and the significance of the share of costs attached to ad serving and intermediation within total advertising activity. As a result of a thorough investigation taking account of the special twosided nature of the market the authorities concluded that DoubleClick was not in a position to substantially exercise its market power as the switching costs were assessed to be sufficiently low, while ad serving represented only a small share of total advertising costs. As for the indirect network effects, the presence of a typical multi-homing behavior alleviated the potential to monopolize the intermediation market through the tipping as a consequence of the concentration in question. The final verdict concluded that these above factors render the merger unlikely to restrict competition to the detriment of consumers, while providing an instructive example of the role of network effects in merger assessment.


List of references
ARMSTRONG, M. (2006): Competition in two-sided markets. The RAND Journal of Economics, Vol. 37., Issue 3.: pp. 668691. BAYE, M.R. BARENSTEIN, M. HOLT, D.J. IPPOLITO, P. M. LACKO, J.M. LEARY, J.B. PAPPALARDO, J.K. PAUTLER, P.A. VITA, M.G. (2008): Economics of the FTC: The Google/DoubleClick merger, Resale Price Maintenance, Mortgage Disclosures, and Credit Soaring in Auto Insurance. Federal Trade Commission, Bureau of Economics, Washington. BROCKHOFF, J. JEHANNO, B. - POZZATO, V. BUHR, C. EBERL, P. PAPANDROPOULOS, P. (2008): Google/DoubleClick: The first test of the Commissions non-horizontal merger guidelines. Copmetition Policy Newsletter, Directorate-General for Competition, Brussels. EVANS, S. D. NOEL, M. D. (2007): Defining Markets That Involve Multi-sided Platform Businesses: An Empirical Framework With an Application to Gooles Purchase of DoubleClick. AEI-Brookings Joint Center for Regulatory Studies, Working paper 07-18. EVANS, S. D. SCHMALENSEE, R. (2005): The Industrial Organization of Markets with Two-Sided Platforms. NBER Working Papers No. 11603. National Bureau of Economic Research, Cambridge. EUROPEAN COMMISSION (2008): Regulation (EC) No. 139/2004 Merger Procedure Case No. COMP/M.4731 Google/DoubleClick. Commission of the European Communities, Brussels. FEDERAL TRADE COMMISSION (2008): Statement of FTC Concerning

Google/DoubleClick. FTC File no. 071-0170.


FILISTRUCCHI, L. KLEIN, T. J. MICHIELSEN, T. (2011): Assessing Unilateral Merger Effects in a Two-Sided Market: An Application to the Dutch Daily Newspaper Market. Tilec Discussion paper No. 2011-046 / CentER Discussion paper No. 2011114. RBB (2008): Google/DoubleClick: The search for a theory of harm. RBB Brief 26. ROCHET, J.-C. - TIROLE, J. (2003): Platform Competition In Two-Sided Markets. Journal of the European Economic Association, Vol. 1., Issue 4.: pp. 9901029. ROCHET, J.-C. - TIROLE, J. (2004): Two-sided Markets: An Overview. Idei University of Toulouse (Working Paper) ROSON, R. (2005): Two-sided markets: A Tentative Survey. Review of Network Economics, Vol. 4., Issue 2.: 142-160.