Assessing the long run competitive effects of digital ecosystem mergers
Jasper van den Boom1 & Peerawat Samranchit2
Abstract This paper focuses on the effects of complementarity and economies of scope, which are prominent characteristics observed in digital ecosystems, on entry foreclosure. A simple economic model shows that a merger with these characteristics drives down an entrant’s profit, providing a sufficient condition for a higher entry barrier. Thus, some mergers that offer short-run benefits for consumers have anti-competitive effects in the long run. The paper also includes a study of five major cases before the European Commission involving one of the Big Five ecosystems (Apple, Amazon, Facebook, Google and Microsoft). These cases demonstrate that long-run effects may be under-assessed in mergers involving digital ecosystems. The paper recommends that competition authorities adopt a distinct approach for the assessment of mergers involving digital ecosystems. A formalistic definition of a digital ecosystem can be used to identify which merging entities fall within its scope. When a merger falls within the scope of the specific regime, authorities can apply alternate standards of assessment that focus more on the long-run effects of complementarities and economies of scope that permeate throughout the ecosystem, rather than focusing on the short-run efficiencies. Moreover, undertakings could be subjected to a revised burden of proof and flexible measures related to potential long-run foreclosure effects, which can ensure that the market remains competitive and the risk of market tipping is mitigated.
Introduction This paper studies the roles of complementarity and economies of scope in creating long-run anti- competitive effects of conglomerate mergers involving digital ecosystems. The contribution of this paper is the insights it produces into the combined effects of these two characteristics in relation to foreclosure and subsequent long-run harms. It argues that these effects may not be given sufficient weight by the European Commission (hereafter, the Commission) when deciding on these mergers. This paper formulates a novel theory of harm supported by an economic model and an ex-post review of decisions by the Commission to demonstrate the theory empirically. The paper concludes with policy suggestions on how
1 Jasper van den Boom is a PhD-candidate at Tilburg University’s Tilburg Law School, at the department for Law, Technology, Markets and Society and a member of the Tilburg Law and Economics Center (TILEC), can be contacted at [email protected] 2 Peerawat Samranchit is a PhD-candidate at Tilburg University’s Tilburg School of Economics and Management (TiSEM) at the Economics department and a member of Tilburg Law and Economics Center (TILEC), can be contacted at [email protected]
1 to account for complementarity and economies of scope in digital conglomerate mergers and how to mitigate potential long-run harms to competition.
The past decades have seen a large number of mergers involving digital platforms with high market shares in their core markets with extensive digital ecosystems. Companies such as Google, Apple, Facebook and Microsoft have engaged in numerous partnerships and acquisitions.3 Acquisitions of smaller players allow these undertakings to acquire know-how and applications that synergize with their existing ecosystems’ products and services. This allows them to expand their ecosystem and enter into new business areas.
Mergers related to ecosystems often fall in the category of conglomerate mergers, as the acquired products or services have neither a specific horizontal nor a vertical relation to one another but often provide complementary functions. Traditionally, conglomerate mergers are viewed as less likely to produce anti- competitive effects and, in fact, to create efficiencies. This paper attempts to challenge this notion, particularly in digital ecosystems where complementarity and economies of scope are likely to be strong. It argues that a more substantial foreclosure effect is observed in digital markets, especially in the long run.4
It is important to first clearly define how an ecosystem should be understood for the purpose of this paper. This paper identifies digital ecosystems as follows:
A digital ecosystem is a collection of products and services that are interoperable within the platform. These products and services are complementary and exhibit economies of scope. They adhere to standards set by the platform central to the ecosystem. The platform exhibits some degree of governance over the products or services within its ecosystem.
3 Recent examples can be found in Google acquisition of companies such as Nestlab, Skybox Imagining and Deepmind, Microsoft’s acquisition of aQuantative, Skype Technologies, Nokia Devices and LinkedIn and Amazon’s acquisitions of Lovefilm. Double Helix Games and Twitch (Dolata, 2017). Of course, this is only a small selection of the 770 acquisitions made by the ‘Big Five’ in the past three decades. See Visualizing Tech Giants’ Billion Dollar Acquisitions (CBN Insights Research Briefs, 5 May 2020), available at < https://www.cbinsights.com/research/tech- giants-billion-dollar-acquisitions-infographic/>. 4 Comments and observations on the long-run anti-competitive effects of mergers have been made in the works of Bourreau & A. De Streel, ‘Digital Conglomerate Mergers and EU Competition Policy’ (2019), p. 30 – 32; J. Crémer; Y-A De Montjoye & H. Schweitzer, ‘Competition Policy for the Digital Era’ Special Advisory Report for the European Union (2019); Digital Competition Expert Panel, ‘Unlocking Digital Competition’, Report of the Digital Competition Expert Panel (2019), p. 11-12, 31-32, 40, 91-101 (hereafter: The Furman Report); Motta M. and Peitz M., ‘Big tech mergers’, Information Economics and Policy (2020) https://doi.org/10.1016/j.infoecopol.2020.100868; this writing aims to provide further technical and empirical analysis on this topic.
2
While the term digital ecosystem does not yet have a formal definition, this economic definition of an ecosystem helps understand how it creates value. The definition, however, fits well within the frameworks of modern conceptualizations of the digital ecosystem used in policy and academia.5
This definition provides several intuitions concerning the workings of ecosystems. First, complementarity means that the value of products when consumed together is greater than the summation of the values when these products are consumed separately. Several products and services provided by digital ecosystems are intrinsically complementary, such as Microsoft OS/Office, Android/Google Drive, and smartphones/health- tracking devices. Second, economies of scope refer to the cost-saving from supplying multiples products at the same time. In other words, it is cheaper for a firm to provide a set of products together than to supply each product separately. This is usually due to common inputs, such as software integrations, personnel, and algorithms. Bourreau and de Streel (2019) make a similar argument that these two characteristics are the main reason why traditional conglomerates are different from digital ones.6
Data is one of the most important sources of complementarity and economies of scope in the digital industry. With a larger ecosystem, the firm has more access to more extensive sets of data. A firm can combine them to make more accurate personalized advertisements, provide better services due to a better understanding of consumers or higher quality. Moreover, data can also be a shareable input. A bigger set of data may allow firms to develop new products and services at a lower cost.7
Lianos I & Carballa B., ‘Economic Power and New Business Models in Competition Law and Economics: Ontology and New Metrics’, CLES Research Paper Series 3/2021 (2021), provides an extensive definition of the ecosystem concept, this paper argues that “the concept of ecosystem reflects the emergence of business environments marked by modularity in production, co-evolution, and decisional complexity, where innovation must be coordinated across different hierarchies, markets, and industries. They form “intentional communities” of economic actors who to a large extent co-evolve their goods and services with aligned visions and “whose individual business activities share in some large measure the fate of the whole community”. Furthermore, the paper argues that ecosystems are defined by the existence of non-generic complementarities.; Alexiadis P. & De Streel A., ‘Designing an Intervention Standard for EU Digital Platforms’ EUI Working Paper Series RSCAS 2020/14 (2020), refers to the works of M. Bourreau and A. de Streel, Digital Conglomerates and EU Competition Policy, March 2019, pp 12-13 and Koca, Product Release Strategies in the Digital Economy, PhD Thesis, Imperial College London, 2018, arguing that “product ecosystems exist when products bought together by a customer generate synergies between those products. In turn, those synergies might facilitate the leveraging of market power between products and/or services”. This definition refers to the presence of complementarity between ecosystem services; Policy reports such as Crémer et al (2019) and the Stigler Center Study of the Economy and the State, ‘Stigler Committee on Digital Platforms’, Final Report (2019), do not clearly define digital platforms but merely refer to them as a form of conglomeration that produces economies of scope and where data is shared. These reports focus in particular on the function of ecosystem to entrench the dominant position of large digital platform operators. While an overarching definition of digital ecosystems does not exist, it is clear that economies of scope, complementarity, interoperability and a shared goal or regime are central tenets of ecosystem creation. 6 Bourreau M. & De Streel, A., ‘Digital Conglomerates and EU Competition Law’ (2019) 7 ibid, p. 11-12
3
Due to complementarity and economies of scope, acquisitions related to ecosystems may happen under different incentives than traditional conglomerate mergers. In particular, these two characteristics provide a synergy that favours a bigger ecosystem. It can generate higher values from its products and services at a lower cost. Hence, the acquisition of additional products or services makes an ecosystem more efficient. While this seems to be beneficial, it can be problematic in the long run. That is, a potential entrant may find it harder to compete with an expansive incumbent’s ecosystem. In other words, greater efficiency in the short run may eliminate competition in the long run.
The first section of this paper formulates the theory of harm based on the combined effects of economies of scope and complementarities in digital ecosystem mergers. The main mechanism is that conglomerate mergers create a competitive advantage for the acquiring ecosystem that competitors cannot duplicate. As such, they diminish the profitability of entry into the market. Subsequently, the section explains how this foreclosure ultimately leads to (empirically observable) long-run exploitative abuses. We also provide an overview of relevant existing theories of harm to clarify how our theory of harm differs from existing theories of harm formulated by other authors.
The second section contains a simple economic model that allows us to incorporate both complementarity and economies of scope. It shows that both characteristics provide a sufficient condition for an increase in the entry barrier post-merger. The stronger the synergy is, the more likely entry foreclosure will occur. The incumbent ecosystem has an incentive to merge to increase its efficiency and to possibly deter entry.
Furthermore, we analyse merger decisions by two types of competition authorities. First, a myopic competition authority assesses merger cases without taking into account the long-run dynamic of entry. Note that this does not mean it is short-sighted. It could be because of requirements in merger control that limit the timeframe. Second, a foresighted competition authority takes into account the full consideration of long-run entry. The model shows that the myopic competition authority may allow a merger when it should not and prevent a merger when it should be allowed. It also highlights that the short-run efficiency argument is not enough to clear a merger. Short-run gains can be detrimental in the long run.
Section 3 studies five mergers involving digital ecosystems that Commission has assessed. These cases are Google/DoubleClick (2008), Microsoft/Skype (2011), Facebook/WhatsApp (2014), Microsoft/LinkedIn (2016), and Apple/Shazam (2018). 8 We focus on the considerations made by the Commission when
8 Case No. Comp/M.4731, ‘Google/DoubleClick – C(2008) 927 final’, Commission Decision of 11 March 2008 (Google/Doubleclick); Case M. 6281, ‘Microsoft/Skype C(2011) 7279 Final, Commission Decision of 07/10/2011 (Microsoft/Skype); Case M.7217, ‘Facebook/WhatsApp C(2014) 7239 Final, Commission Decision of 3.10.2014 (Facebook/WhatsApp); Case M.8124, ‘Microsoft/LinkedIn’ C(2016) 8404 Final, Commission Decision of 6.12.2016 (Microsoft/LinkedIn); Case M.8788, ‘Apple/Shazam – C(2018) 5748 final’, Commission Decision of 6 September 2018 (Apple/Shazam)
4 deciding on the merger case concerning complementarity, economies of scope and the potential long-run effects of the merger, in particular. Subsequently, the case studies use the knowledge of hindsight to assess what happened in the relevant markets post-merger. Finally, the studies try to explain why certain foreclosure effects or practical consequences of the merger were not fully appreciated at the time of the merger decision and how this translates into (potential) consumer harm.
In the final section, the paper puts forward policy suggestions that allow competition authorities to take more action against potential long-run effects resulting from conglomerate mergers involving digital ecosystems. Specifically, this paper argues that in order to capture the full effects that permeate throughout the ecosystem, the competition authority may have to rely on a stricter merger control assessment when large digital ecosystems are involved. Firstly, a more formal definition may be required to identify who falls in the scope of this regime, rather than relying on market definitions as used in traditional competition law cases. Secondly, competition authorities must take into account the conjoint effects of economies of scope and complementarities as potential sources of long-run harm. Without disregarding the short-run efficiencies, the competition authority must take seriously the potential harms it sees in the long run and take more proactive actions to mitigate these. The last suggestion aims to deal with the uncertainty that arises from imposing remedies for potential long-run harms, by suggesting the use of flexible commitments that only trigger when certain conditions are met and/or a readjusted burden of proof to impose commitments.
Section 1 – Theories of harm
Many merger cases in the digital market can be viewed as conglomerate mergers. 9 Traditionally, conglomerate mergers are considered to have only a limited impact on competition. They are generally viewed as creating efficiency. This paper intends to re-examine such a notion. The first part of this section proposes a theory of harm that could arise from complementarity and economies of scope. It discusses how these characteristics can lead to foreclosure in the long run. We also identify four conditions that foreclosure is more likely to occur. In the second part of this section, we discuss related literature and existing theories of harm related to digital ecosystems proposed elsewhere.
1.1. Theory of harm from complementarity and economies of scope We propose that complementarity and economies of scope lead to long-run foreclosure, limiting the degree of competition. To see this, suppose that there are an incumbent ecosystem and a potential entrant. The
9 Because mergers that we are interested in are related to complementarity, they differ from portfolio effects in a fundamental way. This is because portfolio effects concern with weak substitute; see Neven, D., ‘The analysis of conglomerate effects in EU merger control’, Handbook of Antitrust Economics, MIT Press, Cambridge, MA (2008)
5 entrant has to pay an entry cost to enter. The incumbent has an opportunity to acquire a stand-alone firm prior to the entry decision by the entrant. If the incumbent acquires the stand-alone firm, its ecosystem will become more efficient. It generates a higher value, due to complementarity, which increases the utility to consumers. At the same time, the incumbent’s cost does not increase substantially because of economies of scope. Without both complementarity and economies of scope, it is not necessarily true that the generated value will be greater than the increase in cost. When the incumbent becomes more efficient post-merger, the profit that the entrant will earn if it enters diminishes. It becomes less likely that the profit will be enough to compensate for the entry cost. In other words, the merger between the incumbent and the stand-alone firm raises the barrier to entry. The incumbent could potentially use the merger to prevent the market structure from changing from monopoly to duopoly.
Accordingly, a competition authority who decides if the merger should be allowed has to consider the dynamic effect on the market structure. If the competition authority is myopic, meaning that it does not consider the possibility of entry, it may allow the merger when it should not be allowed. It is possible that the competition authority allows the merger because of the efficiency argument. However, this efficiency could be the reason to prevent entry in the long run. The consumers are benefited in the short run, but they are harmed in the long run.
We identify four circumstances that the foreclosure effect is more likely to be pronounced. The first circumstance is when the degree of complementarity is high. That is when consumers derive more utility when consuming a larger bundle of goods and services. Second, when economies of scope are strong, the incumbent with a larger ecosystem attains a more comparative advantage over the entrant. Third, when the ecosystem of the incumbent is already expansive, it is unlikely that the incumbent can develop an ecosystem that can compete with the incumbent. Fourth, if competition between the incumbent and the entrant is more likely to be intense, the entrant has a lower incentive to enter because competition will dilute its profit.
Sections 2 and 3 provide supports to our proposed theory of harm from two perspectives. Section 2 develops an economic model to analyse the situation theoretically. Furthermore, we also look from a legal perspective. Section 3 empirically reviews merger cases related to digital ecosystems decided by the Commission.
1.2 Other theories of harm and related literature The concerns about anti-competitive effects from conglomerate mergers are not new. The Commission expressed these concerns in several landmark cases even before the digital era, such as Tetra Laval/Sidel
6 and General Electric/Honeywell.10 While the Commission has laid out several theories of harm from conglomerate mergers, Neven (2008) argues that the primary anti-competitive effect comes from tying and bundling, leading to foreclosure.11
There is long-standing economic literature on tying and bundling. In the early economic analysis, several Chicago School economists theorize that a monopolist in one market cannot profitably leverage its market power in its monopolized market to another competitive market. This assertion is famously known, among competition policy scholars, as the single-monopoly-profit theory. 12 However, the Chicago School’s argument relies on a crucial assumption that tying or bundling must not create externality that affects market structure.13
Subsequent studies demonstrate that a monopolist may have an incentive to bundle if it leads to the foreclosure of its competitors. The seminal work by Whinston (1990) shows that a monopolist has an incentive to tie to foreclose entry when products are independent, but not when the products are complementary. His model assumes that the monopolist has monopoly power in market A, while it faces a potential entrant only in market B.14 Furthermore, Carlton and Waldman (2002) argue that a monopolist has an incentive to bundle complementary products when the monopolist faces a potential entrant in market A as well. In their model, both products A and B are perfect complements. By bundling, the entrant in market B is automatically foreclosed. Without the entrant into market B, the potential entrant in market A cannot sell its product neither since consumers cannot buy product B from elsewhere.15 Thus, the monopolist has an additional incentive to prevent entry in market B. Even though the strategy seems unprofitable when considering only from market B perspective, the monopolist might use the strategy to keep its market power in market A.16 Note that the settings in Whinston (1990) and Carlton and Waldman
10 Case COMP/M.2220, General Electric/Honeywell, Commission Decision of 3 July 2001; Case Comp/M.2416, Tetra Laval/Sidel, Commission Decision of 1 January 2003; 11 Neven (2008) 12 To see the main argument of this theory, suppose there are two products, say, A and B where both products must be consumed together. A monopolist monopolizes market A, while there is perfect competition in market B. The cost of production for both products is 1 euro. Suppose further that each consumer is willing to pay 10 euro for a bundle of product A and product B. So, the maximum profit that a firm can extract is 8 euro per bundle. Since market B is perfectly competitive, the price for product B will be driven down to cost. Then, the monopolist can charge 9 euro for product A. As a result, with tying, the monopolist earns the net profit of 8 euro for each bundle which is also equal to the maximum profit possible. Hence, tying cannot possibly increase the monopolist’s profit. 13 See - Elhauge, E., ‘Tying, bundled discounts, and the death of the single monopoly profit theory’, Harv. L. Rev., 123 (2009) for the summary of conditions where the single-monopoly-profit theory does not hold. 14 Whinston M., ‘Tying, Foreclosure and Exclusion’ American Economic Review Vol. 80/4, pp. 837-859 (1990) 15 Carlton D.W. & Waldman M., ‘The Strategic Use of Tying to Preserve and Create Market Power in Evolving Industries’. The RAND Journal of Economics 33, 194–220 (2002) 16 This rationale was observed in the case US v. Microsoft, where Microsoft tied its Windows OS with Internet Explorer (IE) to prevent Netscape from entering the OS market.
7
(2002) have the commitment issue: the incumbent has an incentive to de-bundle if the entrant does enter.17 Nalebuff (2004) shows that the incumbent does not have the commitment issue when it can be used as a price discrimination device.18 The rise of digital platforms with two-sided markets and the network effect provides new incentives for the monopolist to use tying or bundling strategy leading to foreclosure. When negative pricing is not practical,19 Jeon and Choi (2020) point out that a platform can use a tying strategy to mimic a negative price. A competitor who sells only one product will not be able to compete.20
In addition to tying and bundling, there is also a situation called “killer acquisitions,” where an incumbent acquires an entrant to stop it from entering the market. In Motta and Peitz (2020), an entrant could become a competitor if it successfully develops its product. Nevertheless, the product can be developed if the entrant has enough resources (e.g., funding or data).21 They show that the incumbent has an incentive to acquire the entrant to maintain its monopoly position, where the incumbent may or may not keep developing the acquired product. Cunningham et al. (2021) estimate that 5.3 percent to 7.4 percent of acquisitions in the U.S. pharmaceutical market are killer acquisitions.22
Existing economic literature tends to focus the analysis on horizontal mergers where the efficiency gain comes from economies of scale or a lower marginal cost. And, when applicable, the anti-competitive effect comes from the exit of existing competitors. In addition to the papers that are already mentioned, other examples are Farrell and Shapiro (1990), Gowrisankaran (1999), Motta (2004), Nocke and Whinston (2010), and Varma et al. (2020). Mergers related to digital ecosystems normally exhibit both economies of scope and complementarity.23 The economies of scope can be related to a lower cost of production that is relatively similar to existing literature. However, the role of complementarity is not evident in the horizontal merger settings.
17 Whinston (n. 23); Carlton & Waldman (n. 24) 18 Nalebuff, B., ‘Bundling as an entry barrier’, The Quarterly Journal of Economics, 119(1), 159-187 (2004) 19 A negative price may not be practical due to a moral hazard problem. For example, suppose that consumers can buy a product at a negative price on an e-commerce platform. To get free money, some consumers may order them even though they do not derive utility from them. 20 Jeon D.S. & Choi J.P., ‘A leverage theory of tying in two-sided markets with non-negative price constraints’, American Economic Journal: Microeconomics (2020) 21 Motta & Peitz (2020) (n. 4) 22 Cunningham, C., Ederer, F., & Ma, S., ‘Killer acquisitions’, Journal of Political Economy, 129(3), 649-702 (2021) 23 Farrell, J. & Shapiro C., ‘Horizontal Mergers: An Equilibrium Analysis’, American Economic Review, 80(1), 107- 126 (1990); Gowrisankaran, G., ‘A dynamic model of endogenous horizontal mergers’The RAND Journal of Economics, pp.56-83 (1999); Motta, M., ‘Competition policy: theory and practice’, Cambridge University Press (2004); Nocke, V. and Whinston, M.D., ‘Dynamic merger review’, Journal of Political Economy, 118(6), pp.1200- 1251 (2010); Varma, G.D. and De Stefano, M., Entry Deterrence, Concentration, and Merger Policy (August 8, 2020). Available at SSRN: https://ssrn.com/abstract=3626734 or http://dx.doi.org/10.2139/ssrn.3626734; Bourreau & De Streel (2019) (n. 6)
8
Another interesting theory of harm is related to the one-stop-shopping feature, proposed by Rhodes and Zhou (2019). When there is a search cost, consumers prefer to go to a larger ecosystem because they can buy several products and services in one go. Such an advantage favours a large ecosystem. It is a feature that a small competitor cannot duplicate.24 This may lead to exclusion. Interestingly, this harm is what we observe in the merger between Google and DoubleClick, which we review in Section 3.
Finally, the harm may not manifest as a foreclosure in the absence of competition. It may come from a reduction in quality. Anderson and Coate (2005) analyse the length of advertisements in television broadcasting. They show that when the degree of competition is lower, there will be more advertisements. While their situation is not exactly related to digital platforms, it shows a general result that firms with more market power could decrease the quality of their products or services.25 In fact, we also observe the reduction in quality following Google/DoubleClick merger, where Google subsequently altered its contracts with advertisers. More details can be found in Section 3.
Section 2: A Model of Entry Barrier by Leveraging Ecosystems We develop a simple model that simultaneously incorporates complementarity and economies of scope to illustrate our theory of harm developed in Section 1.1. It shows that these two characteristics provide a sufficient condition for a higher entry barrier. As such, a merger involving a digital ecosystem has a greater potential to lessens competition in the long run. Consequently, a competition authority has to take into account the effect of changes in market structure in its investigations.
2.1. Model Suppose there are two ecosystems – an incumbent (퐼) and a potential entrant (퐸). The incumbent’s ecosystem consists of a set of products and services denoted by Θ퐼, with an intrinsic value 푣(Θ퐼). The marginal cost of supplying this set of products is 푐(Θ퐼). Similarly, the entrant’s ecosystem offers the set of products and services Θ퐸, with a value 푣(Θ퐸) and the marginal cost 푐(Θ퐸). Define 푠(Θ푖) ≡ 푣(Θ푖) − 푐(Θ푖) 26 as the net surplus of ecosystem 푖. Each ecosystem sets its price 푝푖 for 푖 ∈ {퐼, 퐸}. If the potential entrant wants to enter, it incurs an entry cost, denoted by 퐹.
24 Rhodes, A. and Zhou, J., ‘Consumer search and retail market structure’, Management Science, 65(6), pp. 2607- 2623 (2019) 25 Anderson, S.P. and Coate, S., ‘Market provision of broadcasting: A welfare analysis’, The review of Economic studies, 72(4), pp. 947-972 (2005) 26 In reality, digital ecosystems may set different prices for different products and services and sell them separately. Consumers may only buy a subset of products and services. Nevertheless, we assume that each ecosystem sets only one price: that is, it bundles all of their products together. This greatly simplifies the analysis. In addition, as argued by Neven (2008), the biggest concern for conglomerate mergers is also tying and bundling. The price here can be explicit (retail price) or implicit (consumer data that the firm collected).
9
The ecosystems are have two characteristics – complementarity and economies of scope. For complementarity, the value of the combination of the products is greater than the combination of the values of separate products. Economies of scope allow the ecosystems to supply a bundle of the products at a lower marginal cost than to supply each of them separately. Assumption 1 formalizes these two characteristics.
Assumption 1: Let 훩 and 훩′ be disjoint sets of products and services (훩 ∩ 훩′ = ∅) . Then, by complementary and economies of scope, respectively, we have
푣(Θ ∪ Θ′) > 푣(Θ) + 푣(Θ′), and (1) 푐(Θ ∪ Θ′) < 푐(Θ) + 푐(Θ′). (2)
Consumers are heterogeneous in their preferences of the ecosystems à la Hotelling. A unit mass of consumers is uniformly located on a unit-length line. The locations of the incumbent and the entrant (if enters) are at points 0 and 1, respectively. For simplicity, the location of the incumbent is assumed to remain the same even though the entrant does not enter. The utilities of a consumer at point 푑 on the line for the incumbent (푢퐼) and the entrant (푢퐸), respectively, are
푢퐼(Θ퐼, 푝퐼) = 푣(Θ퐼) − 푡푑 − 푝퐼 and 푢퐸(Θ퐸, 푝퐸) = 푣(Θ퐸) − 푡(1 − 푑) − 푝퐸, (3) where 푡 is a disutility parameter. In other words, holding other factors constant, consumers who are located closer to the left prefer the incumbent’s ecosystem, and vice versa.
Additional to two ecosystems, there is a stand-alone firm that is a monopolist in market 퐴. The product supplied by this firm has the value 푣(퐴) and the associated marginal cost 푐(퐴). Assume that 푠(퐴) ≡ 푣(퐴) − 푐(퐴) > 0. For simplicity, all consumers have the same preference towards firm 퐴. They receive a gross utility 푣(퐴) if they consume the product. Firm 퐴 sets the price 푝퐴 if it operates. The incumbent has an option to acquire firm 퐴 before the entrant decides to enter. The incumbent makes a take-it-or-leave-it offer to firm 퐴. If the merger is successful, the incumbent’s set of products and services becomes Θ퐼 ∪ A.
The timing of the game is as follows. (1) The incumbent decides whether to acquire firm 퐴. (2) The entrant decides if it will enter with the entry cost 퐹. (3) All active firms set their prices simultaneously. (4) Each consumer decides which ecosystem to join (single homing) and whether to buy product 퐴 if firm 퐴 is active. Then, the payoffs are realized. The information is perfect. All parties observe the decisions made in previous stages. The solution concept is the subgame perfect Nash equilibrium (SPNE).
Before we start solving the model, it will be useful to show that the combination of complementarity and economies of scope is sufficient to guarantee that a bigger ecosystem could generate more surplus than a smaller one. Lemma 1 formalizes this result.
10
Lemma 1: If 푠(훩′) > 0, complementarity and economies of scope guarantee that the surplus of the merged ecosystem – 푠(훩 ∪ 훩′) – is larger than the surplus pre-merger – 푠(훩). That is,
푠(훩 ∪ 훩′) > 푠(훩).
Proof: See the appendix.
Intuitively, the result is derived from the characteristics that the merged entity can create higher value at a lower cost. On the contrary, if one of them is not true, it is possible that the surplus post-merger will be lower.
2.2. Solving the Model The game is solved by using backward induction. Before going into the competition between both ecosystems, we first analyse the behaviour of firm 퐴 whenever it is not acquired by the incumbent. Given that all consumers value product 퐴 at 푣(퐴), firm 퐴, who is a monopolist, will charge the price equals to the consumers’ valuation, regardless of the entry decision. All consumers are willing to buy. This observation is summarized in the following lemma.
Lemma 2: When firm 퐴 is active, it charges the monopoly price 푝퐴 = 푣(퐴). All consumers are willing to buy. The profit to firm 퐴 is 휋퐴 = 푠(퐴) ≡ 푣(퐴) − 푐(퐴). At the last stage of the game, there are four subgames. These include whether the incumbent acquires firm 퐴 and whether the entrant enters. Consider the two subgames where the entrant does not enter, while firm 퐴 is or is not acquired. In these subgames, the incumbent is a monopolist. Let the set of incumbent’s ′ ′ ′ products and services is Θ퐼, where Θ퐼 = Θ퐼 if the incumbent does not merge with firm 퐴 and Θ퐼 = Θ퐼 ∪ 퐴 ′ if they do. Given the price 푝퐼, the consumers will participate in the incumbent’s ecosystem if 푣퐼(Θ퐼) − 푡푑 −
푝퐼 ≥ 0, or
(푣 (Θ′)−푝 ) 푑 ≤ max { 퐼 퐼 퐼 , 1}. 푡
The incumbent chooses 푝퐼 to maximize its profit. By solving the first-order conditions, it is straightforward 푚 ′ to get the equilibrium price under monopoly 푝퐼 given Θ퐼, which is
′ ′ 푣(훩퐼) + 푐(훩퐼) ′ 푚 ′ 푖푓 푠(훩퐼) < 2푡 푝퐼 (훩퐼) = { 2 ′ ′ 푣퐼(훩퐼) − 푡 푖푓 푠(훩퐼) ≥ 2푡. The corresponding profit is
′ 2 푠(훩퐼) ′ 푚 ′ 푖푓 푠(훩퐼) < 2푡 휋퐼 (훩퐼) = { 4푡 ′ ′ 푠(훩퐼) − 푡 푖푓 푠(훩퐼) ≥ 2푡.
11
Now, we analyse the other two remaining subgames that the entrant enters. Given the prices set by the ecosystems, each consumer decides which platform she will participate. That is, each consumer compares the utility from joining the entrant’s ecosystem 푢퐸(Θ퐸, 푝퐸) with 푢퐼(Θ퐼, 푝퐼) if the incumbent and firm 퐴 did not merge or 푢퐼(Θ퐼 ∪ 퐴, 푝퐼) if they merge. As usual, the ecosystems set their prices to maximize their profits.
Note that under duopoly, the incumbent and the entrant must have an incentive to operate. They must earn 푑 a positive profit margin, i.e., 푝푖 − 푐푖 ≥ 0 for 푖 ∈ {퐼, 퐸}. In other words, the value of the entrant’s ecosystem must be sufficiently large to compete with the incumbent, for both cases whether the merger occurs or not. Similarly, the incumbent, merged or unmerged, must also be able to compete with the entrant’s ecosystem if it enters. These conditions are satisfied when the values of the two ecosystems are not too far apart from each other. Accordingly, we make the following assumption.
Assumption 2: The difference between the surpluses of the incumbent’s ecosystem and the entrant’s ecosystem is not too far apart such that both of them can profitably operate under duopoly, i.e.,
푠(훩퐼 ∪ 퐴) − 푠(훩퐸) ≤ 3푡, and
푠(훩퐸) − 푠(훩퐼) ≤ 3푡.
When the incumbent does not merge with firm 퐴, the surplus of the incumbent’s ecosystem 푠(Θ퐼) is lower than 푠(Θ퐼 ∪ A) according to Lemma 1. Hence, if the entrant is able to compete with the merged incumbent, it will also be able to compete with the non-merged one. A similar argument is applicable to the second condition. Each ecosystem sets its price to maximize its profit taken the price of another ecosystem as given. Under Assumption 2, the equilibrium price for ecosystem 푖, for 푖 ∈ {퐼, 퐸}, is
2푐(훩푖) + 푐(훩푗) + 푣(훩푖) − 푣(훩푗) 푝푑(훩 , 훩 ) = 푡 + 푖 푖 푗 3 where 푖 ≠ 푗. The duopoly profit excluding the entry cost (when applicable) is
2 (3푡 + 푠(훩푖) − 푠(훩푗)) 휋푑(훩 , 훩 ) = . 푖 푖 푗 18푡 Intuitively, the profit of each ecosystem depends positively on its surplus, while it depends negatively on other ecosystem’s surplus. Furthermore, notice that the profits of both ecosystems increase with the disutility parameter 푡. When 푡 is low, consumers who are located far away from ecosystem 푖 still derive high utilities from the ecosystem. If 푝푖 is low enough, these consumers will still participate in ecosystem
12
푖. Ecosystem 푖 has a higher incentive to compete more aggressively to get these consumers. Accordingly, ecosystem 푗 responds by keeping 푝푗 low as well.
Moving up to stage two when the entrant decides whether to enter. Given the entrant’s equilibrium profit. 푑 ′ The entrant will enter when its operating profit is greater than the entry cost: 휋퐸 (Θ퐸, Θ퐼) − 퐹 ≥ 0. This gives the first proposition.
′ Proposition 1: Given the set of products and services of the incumbent 훩퐼, the entrant will enter if and only if
2 (3푡 − 푠(Θ′) + 푠(Θ )) 퐹 ≤ 퐹̅(Θ′, Θ ) ≡ 퐼 퐸 . 퐼 퐸 18푡
′ The entry cost 퐹 must be lower than the threshold 퐹̅(Θ퐼, Θ퐸) for the entrant to enter. In this context, 퐹̅(⋅) ′ represents the barrier to entry. The lower this threshold is, the higher the barrier. Notice that 퐹̅(Θ퐼, Θ퐸) is ′ decreasing in the surplus of the incumbent’s ecosystem 푠(Θ퐼). According to Lemma 1, if the incumbent merges with firm 퐴, the surplus of the merged ecosystem, through complementarity and economies of scope, will be higher than before the merger, i.e., 푠(Θ퐼 ∪ 퐴) > 푠(Θ퐼). Thus, 퐹̅(Θ퐼 ∪ 퐴, Θ퐸) < 퐹̅(Θ퐼, Θ퐸). After the merger, it is harder for the entrant to compete with the incumbent. The entrant’s profit will be lower, so the entry cost must be sufficiently low to incentivize the entry.
Finally, we move to the first stage when the incumbent decides whether to acquire firm 퐴. Recall Lemma 2, if firm 퐴 is active, it will make the profit of 푠(퐴). So, the incumbent can acquire firm 퐴 if and only if it pays 푠(퐴). The incumbent is willing to pay if it gets an additional profit higher than the acquisition cost.
This is divided into three cases. First, when 퐹 < 퐹̅(Θ퐼 ∪ 퐴, Θ퐸) < 퐹̅(Θ퐼, Θ퐸), the merger does not prevent entry. The entrant will enter regardless of the merger decision. Hence, the incumbent compares the duopoly 푑 푑 profits pre- and post-merger – 휋퐼 (Θ퐼 ∪ 퐴, Θ퐸) and 휋퐼 (ΘI, ΘE). Second, when 퐹̅(Θ퐼 ∪ 퐴, Θ퐸) ≤ 퐹 ≤
퐹̅(Θ퐼, Θ퐸), the merger will block the entry. The incumbent will acquire firm 퐴 if its monopoly profit after 푚 the merger (휋퐼 (Θ퐼 ∪ 퐴)) minus the acquisition cost is greater than the duopoly profit without the merger.
Third, when the entry cost is high such that 퐹̅(Θ퐼 ∪ 퐴, Θ퐸) < 퐹̅(Θ퐼, Θ퐸) < 퐹, the entrant never enters. The 푑 푑 incumbent compares the monopoly profits between the two options, i.e., 휋퐼 (Θ퐼 ∪ 퐴, ΘE) and 휋퐼 (Θ퐼, Θ퐸). The incumbent’s equilibrium strategy in the first stage is summarized in the following proposition.
Proposition 2: In the first stage of the game, the incumbent acquires firm 퐴 by offering 푠(퐴) provided that the increase in profit is greater than 푠(퐴). The condition is divided into three cases:
i. When 퐹 < 퐹̅(훩퐼 ∪ 퐴, 훩퐸) < 퐹̅(훩퐼, 훩퐸), the entrant always enters. The incumbent acquires firm 푑 푑 퐴 푤ℎ푒푛 휋퐼 (훩퐼 ∪ 퐴, 훩퐸) − 휋퐼 (훩퐼, 훩퐸) ≥ 푠(퐴).
13
ii. When 퐹̅(훩퐼 ∪ 퐴, 훩퐸) ≤ 퐹 ≤ 퐹̅(훩퐼, 훩퐸), the acquisition prevents the entry. Otherwise, the entrant 푚 푑 will enter. As such, the incumbent acquires firm 퐴 when 휋퐼 (훩퐼 ∪ 퐴) − 휋퐼 (훩퐼, 훩퐸) ≥ 푠(퐴).
iii. When 퐹̅(훩퐼 ∪ 퐴, 훩퐸) < 퐹̅(훩퐼, 훩퐸) < 퐹, the entrant never enters. The incumbent acquires firm 퐴 푚 푚 if 휋퐼 (훩퐼 ∪ 퐴) − 휋퐼 (훩퐼) ≥ 푠(퐴).
It is intuitive that when the synergy is large, that is 푠(Θ퐼 ∪ 퐴) is much greater than s(Θ퐼), the merger is more likely to occur. However, the merger does not always happen. There are two effects that contribute to this – the demand and the price effects. First, in some cases, e.g., duopoly or monopoly with large 푡, the incumbent does not serve the whole market even after the merger. Nevertheless, the incumbent has to pay the acquisition cost to compensate firm 퐴 who sells to all consumers. Second, it is possible that the price after the merger increases less than the cost of acquisition. Combining both effects, the cost of acquisition could be higher than an additional profit when the synergy from the merger is not sufficiently high.
Case ii. in Proposition 2, where the merger blocks entry, is the most problematic case. This is because the incumbent can unilaterally alter the market structure in its favour. When the monopoly profit is high compared to the duopoly profit, the incumbent has an incentive to acquire firm 퐴 to prevent entry and maintain its monopoly power.
The model analysed so far assumes that the entrant makes the entry decision with the entry cost 퐹. Yet, there is an alternative interpretation. The model can be used to describe a situation where the incumbent and the entrant are already in the market. 퐹, in this case, is a fixed operating cost. With this interpretation, 푑 ′ the entrant is forced to exit the market when 퐹 > 휋퐸 (Θ퐸, Θ퐼). So, when the incumbent acquires firm 퐴, it lowers the threshold similar to Proposition 1. The incumbent can potentially force the entrant out of business by acquiring firm 퐴. This alternative interpretation is relevant to the Google/DoubleClick case, which we will discuss in Section 3.a.
2.3. When should the merger be allowed? This section discusses whether the merger between the incumbent and firm 퐴 should be allowed. We focus on a case that a competition authority has the power to either allow or prevent the merger. We assume that the competition authority bases its decision on the effect of the merger on consumers. In particular, the merger will be allowed if it increases the consumer surplus. While the discussion on whether the consumer surplus standard is the most appropriate one is warranted, it seems to be the standard adopted by some competition authorities, including the European Commission and the Federal Trade Commission.
We consider two types of competition authority – myopic and foresighted. The myopic competition authority does not take into account the potential effect of the merger on market structure. It does not consider that the entry might occur if the merger is prohibited, while the entry could be prevented otherwise.
14
Note that this does not necessarily mean the competition authority is short-sighted. It could be due to a requirement of a merger control that a competition authority has to assess potential impacts of mergers for a short period of time into the future. As we will discuss in the case law, the Commission tends to look into the future for around two to three years. Though, the integration of products and services in digital ecosystems could take three to five years after a merger. As a result, the long-run effect of entry foreclosure may not be properly included in merger assessments by design. In contrast, the foresighted competition authority takes the long-term effect of entry into the assessment.
From our model’s perspective, the short-run effect refers to the effect on the consumer surplus after the merger assuming that the entrant does not enter. That is, the incumbent remains the monopolist regardless of the merger decision. On the other hand, the long-run effect is the consequence on the consumer surplus given the entrant has an opportunity to decide if it will enter or not.
Before analysing how different types of competition authorities will decide, we calculate the consumer surplus for each possible outcome. First, notice that when firm 퐴 is not merged with the incumbent, it can extract all of the surpluses from consumers. Therefore, the consumer surplus from consuming product 퐴 when firm 퐴 is a stand-alone firm is always zero. The consumer surplus depends on the eventual market structure and the appropriate sets of products and services of the incumbent and the entrant.
Lemma 3: The consumer surpluses under monopoly (퐶푆푚) and duopoly (퐶푆푑) given the sets of products ′ and services 훩퐼 and 훩퐸, respectively, are
′ 2 푠(훩퐼) ′ 푖푓 푠(훩퐼) < 2푡 퐶푆푚(훩′) = { 8푡 퐼 푡 푖푓 푠(훩′) ≥ 2푡, 2 퐼
1 2 퐶푆푑(훩′, 훩 ) = [(푠(훩′) − 푠(훩 )) + 9푡(2(푠(훩′) + 푠(훩 )) − 5푡)]. 퐼 퐸 36푡 퐼 퐸 퐼 퐸 Proof: See the appendix.
When the incumbent remains the monopolist, the merger weakly increases the consumer surplus. If the market is not covered prior to the merger (푠(Θ퐼) < 2푡), the merger strictly increases the consumer surplus. 푚 푚 This is because the price increases less than the created value, i.e., 푝퐼 (Θ퐼 ∪ 퐴) − 푝퐼 (Θ퐼) < 푣(Θ퐼 ∪ 퐴) −
푣(Θ퐼), and more consumers are buying from the incumbent. On the other hand, if the market is already covered (푠(Θ퐼) ≥ 2푡), the merger does not affect the consumer surplus. The incumbent can increase the price equal to the created value, while the number of consumers is still bounded at one.
When the market structure shifts from monopoly to duopoly, the consumer surplus increases in most cases, as the entrant brings in the competition. However, this is not always true. The entry can lead to a lower
15 consumer surplus when (1) the value of the entrant’s ecosystem (푣(Θ퐸)) is significantly lower than the incumbent’s ecosystem (푣(Θ퐼′)) and (2) the disutility parameter (푡) is high. The intuition is that when ′ 푣(Θ퐸) is much lower than 푣(Θ퐼), the entrant could not create enough competitive pressure on the incumbent. Accordingly, the incumbent, who knows that it will lose some demand from consumers who have strong preferences for the entrant anyway, increases the price to extract more surplus from its consumers. Only consumers whose locations are very close to 1 gain more surplus. However, as 푣(Θ퐸) is too low, it is not enough to compensate for the decrease in the surplus from consumers on the left. (See Figure 1 in Appendix A2 for the illustration.) Additionally, when 푡 is high, each firm has more market power over their respective groups of consumers. This exacerbates the situation as it allows both firms to keep their prices high. In Appendix A2, we provide a numerical example of a case that the entry decreases the consumer surplus.
Having calculated the consumer surplus, we can analyse how different types of competition authority will decide. Suppose that the competition authority will allow the merger when it strictly increases the consumer surplus. Proposition 3 summarizes the myopic competition authority’s decision.
Proposition 3: The myopic competition authority will allow the merger if the surplus of the incumbent before the merger – 푠(훩퐼) – is sufficiently low such that 푠(훩퐼) < 2푡. Proof: See the appendix.
The myopic competition authority will allow the merger when the market is not yet covered. That is, the initial value of the incumbent’s ecosystem is not high enough. The merger, resulting in synergy, will lead to more consumers buying from the incumbent. The benefit of the generated surplus (푠(Θ퐼 ∪ 퐴) − 푠(Θ퐼)) is partially passed on to consumers since the price does not fully increase. On the contrary, when the market is covered before the merger, this generated surplus is not passed on to consumers as the incumbent can fully increase the price. Notice that the competition authority’s decision does not depend on the entry cost. This is by construction since the competition authority is not taking into account the long-term perspective.
Next, we consider the foresighted competition authority, who take into account the long-term perspective of the change in market structure in its assessment. Its decision is summarized in the following proposition.
Proposition 4: The foresighted competition authority’s decision is as follows:
i. When 퐹 < 퐹̅(훩퐼 ∪ 퐴, 훩퐸) < 퐹̅(훩퐼, 훩퐸), the merger is always allowed.
푚 푑 ii. When 퐹̅(훩퐼 ∪ 퐴, 훩퐸) ≤ 퐹 ≤ 퐹̅(훩퐼, 훩퐸), the merger is allowed if 퐶푆 (훩퐼 ∪ 퐴) > 퐶푆 (훩퐼, 훩퐸).
iii. When 퐹̅(훩퐼 ∪ 퐴, 훩퐸) < 퐹̅(훩퐼, 훩퐸) < 퐹, the decision coincides with the myopic competition authority.
16
Proof: See the appendix
Comparing Propositions 3 and 4, both types of competition authorities always reach the same conclusion when the entry cost is sufficiently high such that the entrant will never enter. This is intuitive as the market structures in the short term and the long term are always identical. However, when the entry cost is low or moderate, the market structure in the long-run and the short-run may be different. Different types of competition authorities will reach different conclusions. In fact, it is possible for the myopic competition authority to generate both false positive and false negative.
The myopic competition authority will clear the merger when it should not when s(Θ퐼) < 2t. There is an efficiency argument that the merger increases the consumer surplus when viewing from the short-term 푚 푚 lens since 퐶푆 (Θ퐼 ∪ 퐴) > 퐶푆 (Θ퐼) . However, the merger may increase the entry barrier so that
퐹̅(Θ퐼 ∪ 퐴, Θ퐸) < 퐹 < 퐹̅(Θ퐼, Θ퐸); the entrant can no longer enter. This prevents the consumer surplus from 푑 increasing to 퐶푆 (Θ퐼, Θ퐸) in the long run. This case highlights the limitation of the efficiency argument of digital ecosystem mergers. That is, efficiency gains are not sufficient to clear a merger case. The competition authority must make sure that these gains have no adverse consequence thereafter.
A merger case that should be allowed can also be prevented by the myopic competition authority. In opposite to the previous case, this situation could occur when s(Θ퐼) ≥ 2t. From a short-term perspective, the merger does not increase the consumer surplus (recall the discussion below Lemma 3). However, if the merger does not prevent entry, i.e., 퐹 < 퐹̅(Θ퐼 ∪ 퐴, Θ퐸), allowing the merger can increase the consumer 푑 푑 surplus in the long run since 퐶푆 (Θ퐼 ∪ 퐴, Θ퐸) > 퐶푆 (Θ퐼, Θ퐸).
Return to the foresighted competition authority’s decision, as in Proposition 4, the criteria used depend on the exact value of entry cost 퐹. When the entry cost is extremely low (case i.), the market structure will eventually become a duopoly. The competition authority should always allow the merger. Because of duopoly, the incumbent is unable to extract all generated surplus from the merger. Part of the generated surplus is passed on to the consumers. In the polar opposite case where the entry cost is too high so that the incumbent will always be the monopolist, the result is the same as the myopic competition authority. That is, the merger should be allowed when the incumbent does not cover the market prior to the merger. The merger allows the incumbent to expand its market share, so that more consumers can enjoy the incumbent’s ecosystem post-merger.
Finally, when the entry cost is in an intermediate region, the merger does prevent the entry. The competition authority should allow the merger if and only if the consumer surplus under monopoly post-merger is greater than the consumer surplus under duopoly without the merger. As discussed earlier, this situation can happen in some rare cases. It requires that the value of the entrant’s ecosystem is relatively small, and
17 the disutility parameter 푡 is high. These two conditions imply that the entrant does not create enough competitive pressure on the incumbent. So, the duopoly structure allows the incumbent to charge a higher price to the consumers who have a strong preference for the incumbent.
Two key lessons for competition policy can be drawn from the model. First, any conglomerate merger that exhibits complementarity and economies of scope inevitably increases the entry barrier. The competition authority should assess how likely a potential entrant will be foreclosed following the merger. That is, a long-run effect on the market structure has to be analysed. Second, the short-run efficiency argument is not sufficient to pass a merger. The competition authority should make sure that there will be no adverse consequence in the long run.
Section 3 - Ex post review of merger decisions involving digital ecosystems In this section we discuss the landmark cases before the European Commission involving digital ecosystem mergers: Google/DoubleClick (2008), Microsoft/Skype (2011), Facebook/WhatsApp (2014), Microsoft/LinkedIn (2016) and Apple/Shazam (2018). For each of these cases, we will look into the harms to competition resulting from the merger due to the complementarity between the acquired service and the other ecosystem services and the economies of scope between these services. We focus in particular on the long-run harms resulting from these mergers; even mergers that are seemingly efficient and benefit consumers in the short-run, may harm consumers by foreclosing the market in the long run.
3.1. Google/DoubleClick Google/DoubleClick will serve as a starting point for this paper to analyse the Commission’s assessment of acquisitions involving a digital ecosystem. With this acquisition, Google combined its proprietary key search advertising and ad placing technologies with DoubleClick’s key display advertising and ad serving technologies.27 The Commission started its merger assessment by noting that the different activities from the two undertakings the Commission noted the high levels of complementarity between the services and the overlapping customer base. However, the Commission argued that there was no direct competition between the two undertakings. As such, it classified the merger as conglomerate rather than horizontal. The Commission looked at three theories of harm that may result in foreclosure: (i) foreclosure scenarios based on DoubleClick’s position in ad serving; (ii) foreclosure scenarios based on Google’s market position in
27 Note: keyword advertising is the use of sponsored results in any kind of search that blends in with the organic results, while display advertising comes in audio-visual forms (pictures, videos or spoken ads) that describe and market the product. Ad placing technology refers to Google’s system for placing (targeted) advertisements on websites and ad serving technology refers to the auctioning of advertisements and deciding which advertiser gets published.
18 search advertising and online ad intermediation and; (iii) foreclosure scenarios based on the combination of Google and DoubleClick’s datasets.28
The Commission found that, despite the parties’ leadership position in their respective markets - they did not possess any superior technologies that in each other’s markets that would allow them to enter each other’s markets to durably compete. In fact, the Commission viewed Yahoo and Microsoft as having a competitive advantage over Google and DoubleClick concerning integrated advertising technologies, while Google and DoubleClick were only leading in their respective stand-alone product markets.29 As such, the first two scenarios were unlikely to materialize. Concerning the combination of datasets, the Commission noted that privacy and confidentiality was important to publishers and advertiser. Therefore, they would have no incentive to accept changes in Google’s terms and condition to allow more data combination or re- usage.30 With the lack of ability to foreclose competition by leveraging its position from one market, and the perceived countervailing buyer power of advertisers and publishers, the Commission cleared the merger without requiring any commitments.
The merger’s effects on competition were arguably stronger than the Commission had foreseen in its assessment. Post-merger, Google quickly captured a large share of the market and became dominant in digital advertising. It’s newly acquired services allowed it to provide a one-stop-shop service for digital advertising which included the auctioning, targeting and publishing of both keyword and display advertising. DoubleClick’s technologies for ad publishing and tracking provided Google with know-how that strengthened its search and advertising divisions as well. 31 Moreover, as noted by Google in its notification of the merger to the SEC, the combination of the service gave it access to new customers through improved accessibility. This allowed Google to serve more customers in the long tail of advertising: Google’s advertising solution proved particularly popular under small (and often local) entrepreneurs. These small players together created a valuable share of the market.32
Ultimately, Google was able to capture an increasingly large market share while Yahoo and Microsoft’s market share diminished. The latter became a niche player in online advertising and the first left the market
28 Google/DoubleClick, par. 215-288 29 Ibid., par. 222-284 30 Ibid., par. 359-366 31 DoubleClick’s technologies involved targeting on the basis of cookies and was significantly more efficient than Google’s targeting technologies, see Srinisvan D., ‘Why Google Dominates Advertising Markets - Competition Policy Should Lean on the Principles of Financial Market Regulation’, Stanford Technology Law Review Vol 24.1 (2020); Why Did Google Buy DoubleClick (Towards Data Science, 6 May 2020) available at:
19 completely. The risk of long-run harms exhibited itself in 2016: while advertisers originally enjoyed the efficiencies of the merger, it no longer had the incentive to stop Google from changing its terms and conditions to allow for the combination and re-use of its data. The risk of the creation of super profiles as noted by the Commission became reality, with little to no pushback from advertisers. 33 Moreover, blockaded entry has been observed in recent reports that note that Google is able to extract economic rents without much competitive threat and without entry into to the advertising market. 34 In conclusion, consumers enjoyed novel access to efficient advertising; the long-run effects were reduced consumer choice and countervailing buyer power and higher prices.
We argue that the unexpected effects of the merger are explained by (a) the high level of complementarity between two services and (b) the economies of scope derived from operating both services jointly. In the context of Google/DoubleClick’s merger, the high levels of complementarity between all activities surrounding advertisement facilitated the creation of a bundled one-stop-shop service. Moreover, the technologies and data acquired with DoubleClick created economies of scope with Google’s existing activities in advertising; access to more data and better technologies led to better targeted advertising and Google’s provision of both display and keyword advertising allowed it to develop new ‘richer’ search advertisement services: Google Shopping and the use of display elements in search. 35 In Google/DoubleClick the Commission only looked at the potential for competition between the two undertakings, but did not pay mind to the effects of combining two high complementary technologies of companies in a leadership position within their respective markets.
3.2. Microsoft/Skype Three years after Google/DoubleClick, the Commission assessed the Microsoft/Skype merger. Microsoft’s acquisition of Skype’s video calling consumer communication service (CCS), provided it with entry into the CCS market. The Commission’s review of this merger predominantly revolved around the possibility for Microsoft to integrate Skype’s CCS functionalities with its proprietary enterprise communications software Lync. Lync offered a combination of instant messaging, video-calling and complex functions such
33 Oracle’s submission to the ACCC’s submission for the Digital Platform Inquiry provides an in-depth insight into how Google’s policy concerning publishers has changed in attachment B of their submission, which can be found here
20 as automatic call distribution, call type detection, call authorization codes, centralized administration and monitoring as well as centralized management and maintenance.36
The Commission’s concern was potential foreclosure effects if Microsoft integrated the Skype and Lync‘s functionalities and/or networks.37 However, the Commission dismissed these concerns on the basis that Skype was not a ’must have’ software and that access to the Skype network for enterprises was not dependent on using Lync software. Instead, professional clients could use both Skype and Lync next to one another. Moreover, network effects in the CCS video-calling were not considered to be particularly strong, as most consumers only use Skype to communicate with a small group of friends and family.
Four years after the merger, Microsoft rebranded Lync as Skype-for-business and integrated the consumer communications and enterprise communications networks with one another. As Commission noted in their decision, by integrating the networks, Microsoft offers a communication tool that consumers are familiar with, but that lives up to the standards of quality and safety that are required for enterprise communications tools.38
We argue that the complementarity between ECS and CCS services are significantly weaker than those between keyword advertising and display advertising as observed in Google/DoubleClick, since the products are not as close to one another. The consumer demands and methods for attracting potential customers are too different between these products to create strong complementarities. We argue this due to the differences in uses between Skype as a CCS and Skype-for-business as an ECS, which minimizes overlapping use. Skype-for-business is predominantly used for professional communications related to internal or external communications of businesses. Skype on the other hand is used for personal communications. As such, even when a user uses both programs, there is little to no benefits that spill-over from using it both personally and professionally. As such, synergies derived from the complementary functions between the two are low. Alternatively, businesses will likely discourage employees from communication professionally with their personal accounts, that are less secure or of a lower quality than professional accounts.39
In the broader context of the ecosystem and product development however, the acquisition gave rise to significant scope economies. Regarding product development, Microsoft launched Microsoft Teams in 2016. This software product runs on Skype technologies, integrates the features of Lync and provides novel
36 Case M. 6281, ‘Microsoft/Skype C(2011) 7279 Final, Commission Decision of 07/10/2011 (Microsoft/Skype), par. 181 37 Ibid., par. 191-213 38 Ibid., par. 173-176 39 Ibid.,, par. 14, 35
21 features such as document sharing and group chats. Microsoft’s acquisition of Skype technologies helped it develop a novel and more comprehensive product.40
Finally, Microsoft has attempted to create complementarities between its different services by tying its enterprise communications software and other productivity software. In particular through Microsoft Outlook, where it is possible by default to plan a Skype meeting, but this possibility is not offered for other meetings. With recent updates to Teams, notifications for meetings will automatically redirect users to Teams, even if there is an invitation to conference via another enterprise communications tool in the invitation itself. These issues were not addressed in Microsoft/Skype, but were addressed later in Microsoft/LinkedIn, so they will be addressed there.
We conclude that Microsoft/Skype does not give rise to the same competitive harms as Google/DoubleClick due to the absence of strong complementarities between the relevant markets. While there are significant economies of scope between the acquired service and proprietary technology, the ECS market has remained open to entry and competition. This is showcased by the stable market share of other ECS providers such as Cisco and new entry by Zoom. It seems that both some levels of economies of scope and complementarity are required to create independent long-run foreclosure effects. In the absence of one of these conditions, foreclosure effects are only created through anti-competitive leveraging practices such as tying and bundling, which the Commission already assesses in its decisions.
3.3. Facebook/WhatsApp The Facebook/WhatsApp case demonstrates a merger where complementarities between the proprietary service and the acquired service are significantly higher than in Microsoft/Skype. This merger involved two services that operated in the CCS market: WhatsApp as a number-dependent consumer communications service and Facebook Messenger as a number-independent service. In this case, there was limited assessment of the complementarities between these two services as the Commission deemed this merger horizontal. In its assessment, the Commission argues that potential anti-competitive effects are unlikely as the services had different qualities (Facebook had a richer environment, one is number-dependent and the other number-independent and the privacy policies were different) and consumers showed a tendency to multi-home between the different services. Therefore, the Commission cleared the merger as there was no elimination of a potential competitor and no significant strengthening of market power.41
40 Microsoft Replacing Skype for Business with Teams (The Seattle Times, 28 September 2017), can be accessed at: https://phys.org/news/2017-09-microsoft-skype-business-teams.html 41 Case M.7217, ‘Facebook/WhatsApp C(2014) 7239 Final, Commission Decision of 3.10.2014, par. 1-107
22
Had the Commission however looked at the complementarities between the two services and the economies of scope derived from this merger, its conclusion may have been different. The Commission notes that instant messaging CCS compete on several parameters: reliability of the service, privacy and security, the size of the network and price. When it is accepted that WhatsApp and Facebook Messenger do not compete with one another due to their differences, the Commission could have looked at foreclosure effects similar to portfolio effects. By employing the two largest complementary networks, Facebook offers two free and ad free services with high reliability, large networks and which differ in privacy and security.42 Looking at the competitive parameters, the acquisition of these products alone provided Facebook with a dominant position in the supposedly separate markets. Moreover, operating both services creates scope economies in the form of know-how and shared inputs and resources (technologies, server space etc.). By focusing on the horizontal effects, these conglomerate effects remained underassessed in the decision.
The importance of these conglomerate effects was demonstrated post-merger. During the investigation of the merger, Facebook had argued that there were technical obstacles to integrating Facebook Messenger and WhatsApp and to match user identities. However, WhatsApp announced changes to their privacy policies allowing for the linking of identities in August 2016, leading the Commission to address a statement of objections to Facebook. The European Commission fined Facebook EUR 110 million in 2017 for providing misleading information.43
After the merger, entry into the consumer communications market for instant messaging is seemingly foreclosed. There are still niche operators in the CCS IM market, but they do not have a credible strategy to monetize their services. Moreover, due to a combination of network effects and the quality difference between Facebook-ecosystem products and stand-alone products, countervailing buyer power seems limited. This was once again demonstrated with Facebook’s recent announcement of changes in its privacy policy. This evoked a negative response among consumers which threatened to switch to Signal. However, even in light of this protest by consumers the switch to Signal has remained limited in numbers.44 It should be noted that if these niche operators were completely absent, the changes to privacy policy may have been
42 WhatsApp data was not to be used by Facebook and WhatsApp offered end-to-end encryption while Facebook Messenger does not; see Facebook/WhatsApp, par. 87-91 43 European Commission, ‘Mergers: Commission fines Facebook €110 million for providing misleading information about WhatsApp takeover’, Press Release of 18 May 2017 44 ‘Facebook ordered not to apply controversial WhatsApp T&Cs in Germany’ (Tech Crunch, 11 May 2021), can be accessed < https://techcrunch.com/2021/05/11/facebook-ordered-not-to-apply-controversial-whatsapp-tcs-in- germany/>; ‘WhatsApp to Force Users to Accept Changes of Terms to Service’ (The Guardian, 14 May 2021) < https://www.theguardian.com/technology/2021/may/14/whatsapp-to-force-users-to-accept-changes-to-terms-of- service>; WhatsApp clarifies it’s not giving all your data to Facebook after surge in Signal and Telegram users (The Verge, 12 January 2021), can be accessed: < https://www.theverge.com/2021/1/12/22226792/whatsapp-privacy- policy-response-signal-telegram-controversy-clarification>
23 even further reaching.45 We argue that Facebook’s dominant position and users’ reliance or return to the platform can be contributed to the combination of high levels of complementarity and scope economies has led to competitive foreclosure in the absence of leveraging practices, combined with network effects in these markets. Facebook and WhatsApp were able to provide higher quality services that provided users synergies and efficiencies in the short-run, but which ultimately led to long-run consumer harms including the degradation of quality in the form of diminished privacy and data protection.
We also observe that the foreclosure effect in the CCS market post-Facebook/WhatsApp was greater than in the ECS market post-Microsoft/Skype. This is likely partly due to the higher levels of complementarity between two similar CCS products than between differentiated CCS and ECS products. There are however other factors that may play a role. For instance, consumer demands in the ECS market are more heterogeneous than in the CCS market. Users in the ECS markets look for products with more heterogeneous functions and have different demands surrounding the quality and functionalities of a product (for instance, they may want to have communications tools that are heavily focused on stability and high quality, data protection or to broadcast to larger audiences). In order to acquire the optimal product, enterprises have a higher willingness to pay than end consumers. As such, the foreclosure effect that arises in the CCS market by proliferating free products may not occur in ECS markets. It is arguable that the foreclosure effects of conglomerate mergers absent tying are stronger in more homogeneous markets, especially those where price competition is not possible.
3.4. Microsoft/LinkedIn The Microsoft/LinkedIn merger marks a change of course for the Commission. Not only is this the first of the studied cases where the Commission has required commitments to clear the merger, it is also the first case where the Commission conducts a more elaborate assessment of the potential effects of combining datasets. In their Merger Decision, the Commission identifies a Professional Social Network (PSN) market for LinkedIn, which differs from other social media by virtue of its nature and content. The Commission argues that switching costs for PSN services are high, as creating and maintaining a professional profile requires significant effort and attention.46
45 ‘Facebook admits defeat over controversial WhatsApp privacy policy’ (BGR, 7 May 2021), it should be noted that admitting defeat is an overstatement, Facebook rather gave users the option to opt out over an otherwise automatically implemented change of terms and services. 46 Case M.8124, ‘Microsoft/LinkedIn’ C(2016) 8404 Final, Commission Decision of 6.12.2016; (i) PC operating systems, (ii) productivity software, (iii) customer relationship management (CRM) software, (iv) sales intelligence solutions, (v) online communications services, (vi) professional social networking, (vii) online recruitment and online advertising services.46
24
The Commission notes several issues that may arise from the merger: first, the Commission focuses on the possibility to pre-install LinkedIn on the Windows OS. This would trigger consumers’ pre-installation bias and cement LinkedIn’s position as the dominant PSN service provider. The Commission considered this to be an anti-competitive form of leveraging and has addressed this by prohibiting it in its commitments.
Secondly, the Commission expressed concerns about the integration of LinkedIn into other productivity software such as Outlook. Were Microsoft to integrate LinkedIn functionalities into its productivity software but deny access to APIs that allowed third-party competitors to do so, this would again provide LinkedIn with a stronger competitive position through leveraging. The Commission again dealt with these forms of leveraging through the commitments required for the clearance of this merger.
Third, the Commission looked into the possible effect of Microsoft’s acquisition of Sales Intelligence Solution service (SIS services) technologies held by LinkedIn with Microsoft’s existing Customer Relationship Management Software.47 According to the decision, SIS services complement CRM software as they provide useful insights which can increase productivity and effectiveness of sales forces. This complementarity was confirmed as industry reports noted that users appreciated sales intelligence solutions that connected directly to their CRM programs and the numerous partnerships that already existed between SIS and CRM service providers.48 While several notifying parties expressed concerns that Microsoft may have the ability and incentive to foreclosure competitors by tying its SIS and CRM software, a high level of uncertainty remained regarding the profitability of such an endeavour. Ultimately, the Commission decided that the merger did not give rise to serious concerns, as there were still stronger – stand-alone – competitors in the SIS market such as Salesforce, SAP and Oracle.49
The assessment of effects on the SIS and CRM markets also connects closely to the Commission’s increased assessment of the effects of merging data and datasets. As both SIS and CRM rely heavily on access to data, the Commission investigated whether Microsoft’s acquisition of LinkedIn data would allow it to deny access to that data for others, leading to a form of input foreclosure. Pre-merger, LinkedIn did not share its full dataset with any third parties: its full dataset was the driver for its in-house SIS service, while CRM providers were allowed access to a subset of LinkedIn’s data.50 The data in this case was special, as datasets
47 Microsoft/Linkedin, par. 29; CRM software solutions help companies of various industry sectors manage their customer interactions by organising, automating, and synchronising data from various sources, such as sales, marketing, customer database, customer service and technical functions. CRM software solutions collate sets of data and display them in a user-friendly manner. This enables companies, in particular the sales department, to improve customer relationships, to better manage accounts, to enhance sales effectiveness, to optimise data quality, and to mitigate regulatory compliance risks. 48 Microsoft/LinkedIn, par. 206 - 217 49 Ibid par. 218- 245 50 Ibid., par. 246-250
25 from the SIS and CRM market demonstrate non-generic complementarities with one another. As CRM software is used to manage the ongoing sales processes with existing customers, access to SIS data may help the seller create better insights into additional demands by their customer. A proper CRM and SIS integration may thus better help the seller identify what and when their client wants to make additional purchases. For this reason, both are heavily data-dependent: the company needs information on the client’s business and current activities (CRM software) and potential expansions and interests (SIS software).51 Privileged access to such data may thus give rise to an integrated service of two highly complementary services, both technologically and data-wise.
While notifying parties expressed views that LinkedIn may share its full dataset with third parties to monetize its service, the Commission argued that this was unlikely due to potential privacy concerns by users. Rather, the Commission expressed optimism concerning Microsoft’s acquisition of LinkedIn – and Microsoft’s subsequent access to data – could lead to pro-competitive efficiencies as it allowed for the development of new products and improving existing products on the market. 52 The Commission’s concerns for competition on the basis of data focused squarely on the potential input foreclosure that occurred if Microsoft denied all access to the data. These concerns were ultimately dismissed due to the strong position of competitors and the prevalence of other sources of data for competing providers.53
The Commission’s assessment in Microsoft/LinkedIn showed increased scrutiny of the potential anti- competitive effects of tying or bundling different software services. Through its remedies, the Commission addressed many short-run concerns including pre-installation, the integration of LinkedIn functionalities with other productivity software and access to APIs. While the Commission took note of the acquisition of highly complementary services (CRM and SIS services) and the economies of scope from merging datasets, it still viewed them simply as pro-competitive efficiencies and was not concerned with the long run effects.
This paper agrees with the assessment of the Commission that Microsoft’s position in the relevant markets in this case was not strong enough to warrant commitments at the time of the merger, as it competed with a dominant undertaking. However, developments in the CRM and SIS markets should be closely monitored over the coming years to see if ecosystem foreclosure effects start to occur. However, it is worth monitoring the development in the upcoming years.
At the time of the merger, Microsoft announced six post-merger integrations: (i) integrating the LinkedIn profile into all productive software apps; (ii) the development of intelligent news feeds based on Microsoft
51 Ibid., par. 210 et seq. 52 Ibid., par. 249 53 Ibid., par. 251-277
26 and LinkedIn data; (iii) integration of the LinkedIn professional network into Cortana; (iv) integration of the Microsoft Dynamics CRM and all other CRM services to directly connect with LinkedIn’s Sales Navigator (its SIS service); (v) using LinkedIn software to develop extensive profiles on employees for management in productivity measurement software and (vi) the use of LinkedIn software for professional training software.54 At the end of 2019, Microsoft announced that two of six integrations were ready to launch: the integration between LinkedIn Sales Navigator and Microsoft Dynamics and integrations between LinkedIn profiles and all Office apps.55
It is likely that the Commission commitments are the reason these integrations are delayed, and therefore the effects of such integration are yet unknown. It is however clear that digital platforms employ long-run strategies to maximize growth and that the commitments with a time horizon of 5 years may delay the implementation of certain integrations and leveraging techniques, but do not address behaviour that may harm competition in the long-run. Moreover, the Commission decision shows that while the Commission recognizes the role of complementarity and data-driven economies of scope, its assessment of the harms to competition remain strongly focused on anti-competitive practices post-merger at the time of this assessment.
3.5. Apple/Shazam Apple/Shazam concerned the acquisition of Shazam into the Apple ecosystem. Shazam is a provider of Audio Recognition technology that is also active in online advertising. The Commission assessed both the potential competitive effects of the merger of these services and the potential effects of integrating their datasets. The Commission again notes limited horizontal overlap and mostly considers these two undertakings as providers of complementary services, thus focusing on conglomerate effects. The Commission looks both at the role of data and at the integration of functionalities in five separate product market.56 The Commission identifies several theories of harm, that include the denial of access to Shazam functionalities and Apple’s potential so self-preference by recommending its Apple Music services to gain access to songs looked up on Shazam. Concerning these potential harms, the Commission applies a test similar to Microsoft/LinkedIn, it does not focus on the complementarities of having these different
54 Information from Microsoft’s Investor Presentation concerning the merger, accessible here: https://view.officeapps.live.com/op/view.aspx?src=https://c.s-microsoft.com/en- us/CMSFiles/InvestorPresentation.pptx?version=a19a08fd-6017-7f22-64c3-2f9aec438edf 55 LinkedIn CEO explains why Microsoft integrations have been so slow since $26B acquisition (GeekWire, 30 December 2019)
27 technologies, but rather focuses on how integration, technical tying or refusal of access would impact competition. In this case, the Commission decides that the abilities and incentives for Apple to foreclose competitors through these types of behaviour is limited or absent.57
The most important aspect of this case however is the Commission’s in-depth treatment of data-driven effects. First, the Commission looks into the possibility to leverage commercially sensitive data. The most innovative part of the Apple/Shazam decision, however, is the Commission’s assessment of the possibility that Apple gains a ‘big data advantage’, through its exploitation of Shazam’s data to develop new functionalities or improve existing functionalities. With this assessment, the Commission has introduced a new test to assess potential complementarity and economies of scope arising from integrating the acquired product and data into the products that are currently offered by Apple.58
This novel test lies more in line with the debate on the effects of big data. The Commission considers whether the acquired data should be seen as an important input for the provision of their services. In order to answer this question, the Commission looked at the nature of the data and the four V’s (variety, velocity, volume and value) to determine whether Apple can use the data to improve its music streaming services.59 In this case, the nature of Shazam data did not create competitive concerns. This is because data on actual music consumption is more valuable than music users have attempted to recognize, nor did Apple gain a significant competitive advantage over competitors such as Spotify following the four V’s of data. The data collected by Shazam were not more comprehensive than other datasets available in the market and were generated at lower speed and with lower user engagement.60
The review of the Apple/Shazam merger shows that the Commission again increases its scrutiny over mergers involving digital ecosystems. The Commission expands its test to better account for economies of scope that can be derived from data to improve existing products or develop new products. In light of our proposed theory of harm, this is a necessary step to capture the effects of mergers involving digital ecosystems. It is unfortunate however that the Commission does not take the same approach in assessing the potential economies of scope from the newly acquired technologies. Here, two issues can be identified. First, the Commission viewed the acquisition of Shazam’s ACR technologies solely in light of potential anti-competitive behaviour post-merger. It did not fully consider how complementarities between the acquired and existing technologies may afford Apple with a competitive advantage that cannot be overcome by competitors. This aligns with the second point; the Commission reviewed the acquisition of ACR
57 Ibid., par. 231-312 58 Ibid., para. 231-292; 313 et seq. 59 Ibid., para 313-325 60 Ibid., para 326-329
28 technologies in light of complementarities with Apple’s music streaming services and online advertising endeavours. However, there is no assessment of the effects of acquiring ACR technologies that permeate throughout the rest of the ecosystem. While the Commission touches on the potential role in ACR technologies to develop Apple’s voice assistant Siri, it does not engage in an in-depth review of potential foreclosure effects in the digital assistant market.61 The importance of having in competitive advantage in voice-driven technologies should not be underestimated, as this has a wider impact on Apple’s position in markets such as the smartphone and smartphone OS market, personal computer markets or other markets that involve smart devices.
The long-run effects of the Apple/Shazam merger remain unknown for now, as the merger took place recently. However, in studying the previous mergers the Commission’s assessment may raise concerns. First, as discussed above, the Commission does not use a holistic approach where is investigates the effects of acquiring both the technologies and data that possess high levels of complementarity and economies of scope with pre-existing services. Rather, it limits its assessment to the complementarity of functionalities solely to exclude the horizontal dimension and limits its assessment to potential leveraging behaviour post- merger. The Commission also does not focus on the economies of scope derived from acquiring new technologies but limits its assessment of scope economies to datasets. We argue that the Commission should focus on the potential complementarities and economies of scope for each acquired dataset and technology on the entirety of the ecosystem, rather than viewing these phenomena as separate issues in separate markets. It is because of these effects that relying on market definition does not reflect the realities of competition in digital markets.
Second, reliance on the merging entity’s privacy policy has proven problematic in Google/DoubleClick, where the privacy policy was ultimately changed by Google in order to strengthen its position in the market. These concerns are exacerbated by Apple’s recent decision to forbid data-collection on the iOS platforms. While this move seemingly enhances data-protection, concerns have been expressed that this new policy is a bid to expand Apple’s position in digital advertising. As such, Apple may become inclined to collect more data from its own ecosystem while excluding others.62
Third, and finally, the Commission does identify potential long-run issues. These include Apple’s bid to become a leading player in the market for voice-driven digital assistants and the potential role for Shazam as the provider of an ACR software. However, it does not act on it as the level of uncertainty for long-run
61 Ibid., para 341-348 62 Apple’s Move to Block User Tracking Spawns New Digital Ad Strategies (Wall Street Journal, 26 March 2021), can be accessed: https://www.wsj.com/articles/apples-move-to-block-user-tracking-spawns-new-digital-ad- strategies-11616751001; Apple to boost ads business as iPhone changes hurt Facebook (ARStechnica, 22 April 2021)
29 harms remains too high to impose commitments. In the upcoming section, we will propose means of dealing with this uncertainty while safeguarding future competitiveness of the market
3.6. Conclusions from the ex-post review of the cases Several conclusions can be drawn from the studied cases. First, the study observes that in acquisitions where both economies of scope and complementarity are high, the foreclosure effect becomes more pronounced over a relatively short period of time. In Google/DoubleClick and Facebook/WhatsApp, where the acquisition happened in very closely related markets that rely on similar technologies and data, the foreclosure effects in the advertising and messaging markets could be observed relatively quickly after integration in the ecosystem.
In Microsoft/Skype, where the economies of scope between the acquired messaging technologies and the existing services were strong, but complementarity between the CCS market and professional services was weak, foreclosure did not occur in a similar way. However, the acquisition of know-how led to the creation of the novel product Teams that finds stronger integration with other ecosystem functionalities and may create long-run foreclosure effects if it becomes the default for professional communications. Similarly, in Apple/Shazam, strong economies of scope can be noted as the existing products can be expanded with improved audio recognition and voice-driven functionalities. In this case, the newly acquired technologies provide shared inputs across a wide range of products existing in the ecosystem. However, complementarity between audio recognition software and Apple’s other activities in different mobile, computer and audio- streaming markets are not as strong as those observed in Google/DoubleClick or Facebook/WhatsApp.
Finally, Microsoft/LinkedIn exhibits strong complementarity but weak economies of scope. The integration of a professional networking service with Microsoft’s other professional and productivity software may lead to increased user synergies, as having access to an ‘online resume’ through Office apps or having a number-independent professional messaging service may help users extract more utility. Moreover, the Commission noted in their decision that the CRM and SIS software demonstrate complementarities, both through their integration and datasets. However, the economies of scope derived from acquiring LinkedIn are seemingly weaker, as Microsoft’s access to LinkedIn functionalities or profile data does not create demonstrable scope economies with its other productivity software, which generally do not demonstrate a high level of reliance on access to personal data or direct network externalities. With the Commission’s decision to limit the integration of LinkedIn functionalities into Microsoft’s other productivity software, the integration of functionalities and foreclosure effects have been delayed. However, it remains to be seen whether the commitments were sufficient to prevent Microsoft from entrenching its position as the provider of productivity software and its ecosystem further. The table hereunder shows the difference in the presence of economies of scope and complementarity.
30
Economies of scope
Strong Weak
Google/DoubleClick
Strong Facebook/WhatsApp Complementarity Microsoft/LinkedIn
Microsoft/Skype Weak Apple/Shazam
Several lessons can also be drawn from the case studies concerning the current methods of assessment employed by the Commission in relation to conglomerate ecosystem mergers.
First, we observe that the combination of strong economies of scope and high levels of complementarity between the acquired service and the existing services in the market leads to a higher chance that foreclosure in the market occurs and that later entry is deterred. These effects should be viewed together and in context of the entire ecosystem rather than selected markets.
Secondly, the effects that permeate throughout the ecosystem are hard to fully capture due to the expansiveness of digital ecosystems. As such, defining all relevant markets in a merger can be extremely challenging. In order to step away from the market definition, the Commission may have to adopt a distinct approach for assessing digital ecosystems that could rely on more formalistic considerations.
Third, The Commission has identified several specific long-run harms in their assessments that have ultimately materialized, including the creation of super profiles, the merging of datasets or functionalities or the changing of privacy policies. In the studied cases however, the Commission has not required any commitments from the undertakings. We argue that a more pro-active approach may be required to prevent long-run harms. In the upcoming section, we will argue that the use flexible remedies may help prevent long-run harms while taking into account the uncertainty that is paired with assessing developments in the market over a long time horizon.
31
Section 4: Policy Suggestions In this section we propose three policy changes in the assessment of mergers that involve digital ecosystem. First, we propose how the Commission can take into account the combined effects of complementarities and economies of scope that permeate throughout the ecosystem by applying a specific merger policy for digital ecosystem acquisitions. Secondly, we argue how the Commission can use formalistic definitions to determine to which undertakings this regime applies. Thirdly, we recommend several changes to the assessment procedure, including the use of a revised burden of proof and the use of flexible commitments to mitigate long-run harms.
4.1 The role of complementarity and economies of scope Our first recommendation is to adopt an assessment that assesses the complementarities and economies of scope from both the acquired technologies and data jointly and in light of the wider ecosystem to capture the effects of digital conglomerate mergers accurately. We argue that this approach should be applied specifically to digital ecosystems, as the effects of complementarities and economies of scope are considered to be less pronounced in traditional markets than in digital markets. As such, the Commission may have to formulate a distinct approach to assessing mergers involving digital ecosystems.
In such a distinct approach, the Commission can also tackle difficulties surrounding market definition in digital markets and capturing ecosystem effects. Lianos & Carballa (2021) describe extensively how the contemporary approach to market definition has issues capturing the different effects that arise in complex networks and ecosystems. This is not limited exclusively to digital ecosystems, but also to complex transaction markets such as credit card markets. Lianos & Carballa argue that while reliance on aftermarket approaches or a stronger focus on indirect network externalities may resolve issues in the latter, the digital ecosystem is more complex as digital ecosystems involve feedback loops between various user groups on each side of the platform.63 Following our theory of harm, the modularity of digital products and the possibility to re-use complementary technologies and data throughout an unspecified number of ecosystem products creates further complexities that cannot be captured by focusing on specific pre-defined markets related to the merger.
4.2 The formalistic approach to digital ecosystems This is related to our second recommendation, for which we argue that both the issue of market definition and the required ecosystem-specific approach can be accommodated by formulating a specific set of (potentially ex-ante) rules for digital ecosystems which allow for a more formalistic approach to determine who is inside of the scope of the specific merger-regime and who is outside of the scope of this regime. By
63 Lianos & Carballa (2021), p. 14
32 adopting a formalistic approach, the Commission can define which ecosystems should be viewed as having ‘cross-market significance’ or ‘ecosystem power.’, Member States such as Germany are currently developing an approach that relies on similar concepts, allowing a more comprehensive analysis of ecosystem effects. 64 A similar approach could be formulated for the European level that exists in concurrence with the Digital Markets Act rules. While the latter focuses on particular prohibitions for core platform services, the first may focus specifically on cross-market power and the expansion of ecosystems, including through acquisitions.
4.3 Three changes in the assessment procedure Once the Commission identifies an ecosystem that falls within the scope of the specific merger regime, it can apply different rules towards clearing the merger. In our third recommendation we argue that three standards of assessment should apply specifically to these mergers.
Fort the first change to the standards of assessment, we argue that the presumption that conglomerate mergers are unlikely to lead to anti-competitive effects, but rather result in pro-competitive effects should not apply. In light of the economic model and the observed effects in the case studies, it can be argued that complementarity and economies of scope in digital ecosystems have effects similar to horizontal mergers. The envelopment of complementary – yet similar – services results in the potential foreclosure of indirect entry into core markets or preventing the development of competing ecosystems. Moreover, the merger may provide the digital ecosystem with a competitive advantage that potential entrants cannot overcome. In this light, conglomerate mergers should be viewed as having a similar potential to raise barriers to entry or foreclose competition as horizontal mergers, at least when digital ecosystems are involved.
For the second change to the standards of assessment, we argue that as the foreclosure effect arises due to efficiencies produced by the merger, a revised burden of proof may be required in assessing mergers involving digital ecosystems to address long-run harms, even when short-run efficiencies are identified. Specifically, we argue that the presence of efficiencies in the short run is insufficient to clear the merger (absent commitments) unless the undertaking can convincingly argue that there is no risk of long-run foreclosure.
64 For instance, inspiration can be drawn from the German approach. In §19a of the Gesetz gegen Wettbewerbsbeschränkungen (GWB/Act against Restraints of Competition), the German Federal Cartel Office relies on indicators such as dominance in individual market powers (which is an indicator but not required to find cross market significance), vertical integration, financial strength and access to competition-relevant data. Under §19a (1), the Federal Cartel Office may prohibit any undertaking from engaging in certain practices which are particularly harmful to competition; Busch C., ‘Regulation Of Digital Platforms As Infrastructures For Services Of General Interest’, WISO DISKURS Friedrich Ebert Stiftung 09/2021 (2021), p. 17-19
33
The third change to the standards of assessment aims to address the high level of uncertainty that is paired with assessing long-run effects. It may not be desirable to rely on limit potentially efficient behaviours at the outset of the merger. Instead, we suggest that the Commission rely on flexible commitments, which trigger when the foreseen harms materialize. In the case studies, it has been observed that the Commission often identifies potential long-run harms in its decision yet does not act on it at that time.65 The uncertainty for harms is due to the long-time horizon and the existence of remedying circumstances such as the presence of competition or countervailing buyer power. However, once the harms materialize the Commission has no option to stop the undertaking’s anti-competitive behaviour as no commitments have been asked. We argue that flexible remedies should apply specifically for these situations, allowing the Commission to determine specific remedies for harms that were identified at the time of the merger, but could not be addressed under strict pre-defined commitments at the time. This may help remedy the uncertainty involved in long-run harms, while ensuring that competition in the market remains protected over a longer time span.
Section 5: Conclusion This paper looks at the long-run effects of conglomerate mergers involving digital ecosystems on potential entry. Given the two characteristics of digital ecosystems, i.e., complementarity and economies of scope, the possibility of foreclosure is more likely than traditional conglomerate mergers. That is, the anti- competitive effects are more conspicuous than conglomerate mergers before the digital era. With the fact that we have observed ubiquitous mergers by big-tech companies in the past decade, the reassessment of merger control on conglomerate mergers involving digital ecosystems is warranted.
We develop a simple economic model that allows us to incorporate both complementary and economies of scope. These characteristics guarantee that when an incumbent ecosystem acquires a stand-alone firm, the entry barrier is indeed increased. The more potent synergy between the ecosystem and the stand-alone firm leads to a higher entry barrier. As such, this provides an additional incentive for the incumbent to merge. In the short run, it increases the efficiency of its ecosystem. In the long run, it may be able to prevent entry and preserve its monopoly power. As such, in allowing a merger case, a competition authority should consider the dynamic aspect of the merger as well. If it fails to incorporate the dynamic consideration, the merger might be cleared when it should be prevented, and vice versa.
The case studies demonstrate that the long-term effects derived from the digital ecosystem perspective were not given fair weight. While some effects were hard to foresee at the time, other long-run problems were
65 For instance, the Commission identifies the potential for Google to revise its privacy policy and increase its data collection; Skype’s integration with Lync to create Microsoft Teams; Facebook’s changes to the WhatsApp Privacy policy; Microsoft’s integrations in the CRM and SIS markets and Apple’s potential to use Shazam for the further development of its voice assistant software throughout the case studies.
34 identified by the Commission but were dismissed. The changing of privacy policies by Google and Facebook, the value changes of LinkedIn’s data, Apple’s dominant position in smartwatches and Skype’s integration into Lync were all neglected in the assessment. Moreover, the Commission tends to pay significant amounts of attention to issues arising from network effects, while it did not pay any mind to economies of scope until Apple/Shazam, and only implicitly discusses complementarity under concerns on tying and bundling. In order to assess long-run effects, the Commission must look into the potential foreclosure effects of these efficiencies as well.
There are several issues that must be addressed with potential policy changes. First, the assessment must take into account the full range of scope economies and complementarities jointly throughout the entirety of the ecosystem. This complex assessment of effects is necessary to fully reflect the consequences of a merger on competition. However, as these effects are hard to capture across the ecosystem when relying on market definition, an alternative policy may have to be developed for digital ecosystems. To this extent, we suggest that the identification of ecosystems that fall within such a policy are based on formalistic definitions and that undertakings that meet certain criteria (such as activity in a large number of digital markets, turnover and end users across different digital markets) fall within the scope of the regime. Once this regime applies, we argue that relying on a revised burden of proof can improve the assessment of mergers involving digital ecosystems. Here, the Commission will only have to make plausible that the merger is liable to harm competition in the long-run and that it is up to the undertaking to alleviate these concerns. In cases where the Commission remains unconvinced of the arguments brought forward by the undertaking, we recommend the use of flexible measures. These measures will only apply to specific concerns voiced by the Commission. Once these specific scenario’s put forward by the Commission materialize (such as the creation of ‘super profiles’ or the merging of certain technologies) the flexible measures will trigger and the Commission can decide on suitable remedies to impose.
These recommendations aim to ensure that digital ecosystem mergers are treated differently according to their unique characteristics while leaving merger control of traditional mergers unscathed. It also allows for a manageable workload in assessing the complex long run effects for the Commission, while taking into account the uncertainty that is paired with the assessment of long-run harms to competition.
35
Appendix
A1 Proofs of Lemmas and Propositions
A1.1 Proof of Lemma 1
By definition, we have s({Θ, Θ′}) = 푣({Θ, Θ′}) − 푐({Θ, Θ′}). Due to complementarity and economies of scope, 푣(Θ ∪ Θ′) > 푣(Θ) + 푣(Θ′) and 푐({Θ, Θ′}) < 푐(Θ) + 푐(Θ′). Hence,
′ ′ 푠(Θ ∪ Θ′) > (푣(Θ) − 푐(Θ)) + ⏟(푣 ( Θ ) − 푐 ( Θ ) ) > 푣(Θ) − 푐(Θ) ≡ 푠(Θ). >0
A1.4 Proof of Lemma 3
푠(Θ′) For the monopoly outcome, when s(Θ′) < 2t, the market is not covered. Only consumers whose 푑 ≤ 퐼 I 2푡 buy from the monopolist. Thus, the consumer surplus is
′ 푠(Θ퐼) 2푡 푣(Θ′) − 푐(Θ′) ∫ (푣(Θ′) − 푡푑 − 퐼 퐼 ) 푑푑. 퐼 2 0
′ When s(ΘI) ≥ 2t, all consumers buy from the monopolist. Accordingly, the consumer surplus can be calculated as
1 ′ ′ ∫(푣(Θ퐼) − 푡푑 − (푣(Θ퐼) − 푡)) 푑푑. 0
By straightforward calculations, we get the first part of Lemma 3.
For the duopoly outcome where the entrant is active in the market, recall that consumers whose 푑 ≤ ∗ ∗ 푑(푝퐼 , 푝퐸) will buy from the incumbent, while others will buy from the entrant. So, the consumer surplus is
∗ ∗ 푑(푝퐼 ,푝퐸) 1 ∗ ′ ∗ ∗ 퐶푆 (Θ′퐼, Θ퐸) = ∫ (푣(Θ퐼) − 푡푑 − 푝퐼 )푑푑 + ∫ (푣(Θ퐸) − 푡(1 − 푑) − 푝퐸)푑푑 , ∗ ∗ 0 푑(푝퐼 ,푝퐸) This yields the second part of the proposition.
A1.5 Proof of Proposition 3
36
When the surplus prior to the merger is sufficiently low such that 푠(Θ퐼) < 2푡, the initial surplus is 2 2 s(Θ퐼) /8푡. If the merger is allowed, the consumer surplus increases to s(Θ퐼 ∪ 퐴) /8푡 when 푠(Θ퐼 ∪ 퐴) <
2푡 or 푡/2 when 푠(Θ퐼 ∪ 퐴) ≥ 2푡. In either case, they are more than the initial consumer surplus.
On the contrary, the initial consumer surplus when 푠(Θ퐼) ≥ 2푡 is 푡/2. After the merger, 푠(Θ퐼 ∪ 퐴) is still higher than 2푡. The consumer surplus remains the same at 2푡. Therefore, the competition authority prohibits the merger.
A1.6 Proof of Proposition 4
For the first case where F < F̅(Θ퐼 ∪ 퐴, ΘE) < F̅(ΘI, ΘE), the entry cost is sufficiently low such that the ∗ entrant always enters. Therefore, the competition authority will allow the merger if CS (Θ퐼 ∪ 퐴, Θ퐸) > ∗ CS (ΘI, Θ퐸). By some rearrangement, this condition is equivalent to
푠(Θ퐼 ∪ 퐴) − 푠(Θ퐸) + 푠(ΘI) − 푠(Θ퐸) + 18푡 > 0.
By Assumption 2, both 푠(Θ퐼 ∪ 퐴) − 푠(Θ퐸) and 푠(ΘI) − 푠(Θ퐸) are between −3푡 and 3푡. Hence, the preceding condition is always satisfied. The merger is always allowed.
For the second case where F̅(Θ퐼 ∪ 퐴, ΘE) ≤ F ≤ F̅(ΘI, ΘE) , the entrant will enter if the merger is prohibited, while it will not enter otherwise. Therefore, the merger will be allowed if and only if the m consumer surplus under monopoly CS ({ΘI, A}) is greater than the consumer surplus under duopoly ∗ CS (ΘI, ΘE).
Finally, when F̅(Θ퐼 ∪ 퐴, ΘE) < F̅(ΘI, ΘE) < 퐹, the entrant will never enter regardless of the merger decision. Hence, the market structure in the long-run is always the same as the short-run. The foresighted competition authority uses exactly the same criteria as the myopic competition authority.
A2 An Example of a reduction in consumer surplus from entry
In this appendix, we show an example where the consumer surplus under duopoly is lower than under monopoly. Let the values of the parameters to be as follows: 푣(Θ퐼′) = 10, 푐(Θ퐼′) = 5, 푣(Θ퐸) = ∗ ′ 푚 ′ 6, 푐(Θ퐸) = 3, and 푡 = 2.5. These yield 퐶푆 (Θ퐼, Θ퐸) = 0.92 and 퐶푆 (Θ퐼) = 1.25. The surplus that each consumer receives for both monopoly and duopoly cases are plotted in Figure 1.
Figure 1: Surplus from Each Consumer under Monopoly and Duopoly
37
Note: The values of parameters are 푣퐼 = 10, 푐퐼 = 5, 푣퐸 = 6, 푐퐸 = 3, and 푡 = 2.5.
38
Bibliography
Law and soft law
- Council Regulation (EC) No 139/2004 of 20 January 2004 on the control of concentrations between undertakings (the EC Merger Regulation), - Council Regulation (EC) No 1/2003 of 16 December 2002 on the implementation of the rules on competition laid down in Articles 81 and 82 of the Treaty - Commission Notice – Guidelines on the assessment of horizontal mergers under the Council Regulation on the control of concentrations between undertakings”, DG COMP, 28 January 2004 . par. 71 (b) (Horizontal Merger Guidelines) - Commission Notice – Guidelines on the Assessment of Non-horizontal Mergers under the Council Regulation on the control of concentrations between undertakings, DG COMP, 28 October 2010; Commission Notice – Guidelines on the assessment of horizontal mergers under the Council Regulation on the control of concentrations between undertakings”, DG COMP, 28 January 2004 - Gesetz gegen Wettbewerbsbeschränkungen
Competition law cases
- Case No. Comp/M.4731, ‘Google/DoubleClick – C(2008) 927 final’, Commission Decision of 11 March 2008 (Google/Doubleclick); - Case M. 6281, ‘Microsoft/Skype C(2011) 7279 Final, Commission Decision of 07/10/2011 (Microsoft/Skype); - Case M.7217, ‘Facebook/WhatsApp C(2014) 7239 Final, Commission Decision of 3.10.2014 (Facebook/WhatsApp); - Case M.8124, ‘Microsoft/LinkedIn’ C(2016) 8404 Final, Commission Decision of 6.12.2016 (Microsoft/LinkedIn); - Case M.8788, ‘Apple/Shazam – C(2018) 5748 final’, Commission Decision of 6 September 2018 (Apple/Shazam) - Case COMP/M.2416, ‘Tetra Laval/Sidal, Commission Decision of 13 January 2003 (Tetra Laval/Sidal) - Case COMP/M.2220, ‘General Electric/Honeywell’, Commission Decision of 3 July 2001 (General Electric/Honeywell) - COMP/M.6570 UPS/TNT Express, Commission Decision of 30 January, 2013; (UPS/TNT)
39
- Case COMP/M.7630 FedEx/TNT Express, Commission decision of 8 January, 2016 (FedEx/TNT) - Case AT.39740, Google Search (Shopping) – C2017 4444 final, Official Publication of the European Union (2017) (Google Shopping/Google Search)
Literature
- ACCC, ’Digital Platform Inquiry – Final Report’ (2019) - Anderson C., ‘The Long Tail – How Endless Choice is Creating Unlimited Demand’, Unabridged ed. (2009) - Anderson, S.P. and Coate, S., ‘Market provision of broadcasting: A welfare analysis’, The review of Economic studies, 72(4), pp.947-972 (2005) - Batura O., Van Gorp N., Larouche P., ‘Online Platforms and the EU Digital Single Market’, A response to the call for evidence by the House of Lord’s internal market subcommittee (2015) - Bourreau M. & De Streel A., ‘Digital Conglomerate Mergers and EU Competition Policy’ (2019); - Busch C., ‘Regulation Of Digital Platforms As Infrastructures For Services Of General Interest’, WISO DISKURS Friedrich Ebert Stiftung 09/2021 (2021) - Cardwell D., ‘The Role of Efficiency Defence in EU Merger Control Proceedings Following UPS/TNT, FedEx/TNT and UPS v. Commission’, Journal of European Competition Law & Practice (2017) - Carlton D.W. & Waldman M., ‘The Strategic Use of Tying to Preserve and Create Market Power in Evolving Industries’. The RAND Journal of Economics 33, 194–220 (2002) - Chen Z. & Rey P., ‘A theory of conglomerate mergers’, Mimeo (2018)
- Comisión Federal de Competencia Económica, ‘Rethinking Competition in the Digital Economy’ (2020) - Crémer J.; De Montjoye Y-A & Schweitzer H., ‘Competition Policy for the Digital Era’ Special Advisory Report for the European Union (2019); - Cunningham, C., Ederer, F., & Ma, S., ‘Killer acquisitions’ Journal of Political Economy, 129(3), 649-702 (2021) - Digital Competition Expert Panel, ‘Unlocking Digital Competition’, Report of the Digital Competition Expert Panel (2019) (The Furman Report) - Dolata U., ‘Apple, Amazon, Google, Facebook, Micrsoft - Market Concentration – Competition – Innovation Strategies’, SOI Discussion Paper 2017-01 (2017)
40
- Easterbrook F., ‘The Limits of Antitrust’, University of Chicago Law Occassional Paper, No. 21 (1985) - Eisenmann T, Parker G. & Van Alstyne M., ‘ Platform envelopment’, Strategic Management Journal, 32(12), pp.1270-1285 (2011); - Elhauge, E., ‘Tying, bundled discounts, and the death of the single monopoly profit theory’, Harv. L. Rev., 123 (2009) - Farrell, J. & Shapiro C., ‘Horizontal Mergers: An Equilibrium Analysis’, American Economic Review, 80(1), 107-126 (1990),
- Gerard, D., ‘Merger Control Policy: How to give Meaningful Consideration to Efficiency Claims?’, Common Market Law Review 40 pp. 1367-1412 (2003) - Giannino, M., ‘Microsoft/LinkedIn: What the European Commission Said on the Competition Review of Digital Market Mergers’ (July 19, 2017). Available at SSRN: https://ssrn.com/abstract=3005299 or http://dx.doi.org/10.2139/ssrn.3005299 - Gowrisankaran, G., ‘A dynamic model of endogenous horizontal mergers’ The RAND Journal of Economics, pp.56-83 (1999) - House Judiciary Committee Report: 'Investigation of Competition in Digital Markets', (2020) - Jeon D.S. & Choi J.P., ‘A leverage theory of tying in two-sided markets with non-negative price constraints’, American Economic Journal: Microeconomics (2020) - Lambert T.A., ‘The Limits of Antitrust in the 21st Century’, Regulation (2020), - Lee J., ‘The Google-DoubleClick Merger: Lessons From the Federal Trade Commission's Limitations on Protecting Privacy’, Communication Law and Policy, 25:1, 77-103, (2020) DOI: 10.1080/10811680.2020.1690330 - Lianos I & Carballa B., ‘Economic Power and New Business Models in Competition Law and Economics: Ontology and New Metrics’, CLES Research Paper Series 3/2021 (2021) - Motta, M., ‘Competition policy: theory and practice’, Cambridge University Press (2004) - Motta M. and Peitz M., ‘Big tech mergers’, Information Economics and Policy (2020) https://doi.org/10.1016/j.infoecopol.2020.100868 - Maria Lancieri F. & Sakowski P., ‘Competition in Digital Markets: A Review of Expert Reports’, Stigler Center Working Paper Series No. 303; Stanford Journal of Law, Business, and Finance, (Forthcoming), 2020 - Nalebuff, B., ‘Bundling as an entry barrier’, The Quarterly Journal of Economics, 119(1), 159- 187 (2004). - Neven, D., ‘The analysis of conglomerate effects in EU merger control’, Handbook of Antitrust Economics, MIT Press, Cambridge, MA (2008)
41
- Nocke, V. and Whinston, M.D., ‘Dynamic merger review’, Journal of Political Economy, 118(6), pp.1200-1251 (2010) - OECD, ‘Portfolio Effects in Conglomerate Mergers – DAFFE/COMP(2002)5, Competition Law & Policy (2001) - Patterson D.E. & Shaprio C., ‘Transatlantic Divergence in GE/Honeywell: Causes and Lessons’ 16 Antitrust (2001) - Rhodes, A. and Zhou, J., ‘Consumer search and retail market structure’ Management Science, 65(6), pp.2607-2623 (2019) - Stigler Committee on Digital Platforms, ‘Final Report’, Stigler Center (2020) Srinisvan D., ‘Why Google Dominates Advertising Markets - Competition Policy Should Lean on the Principles of Financial Market Regulation’, Stanford Technology Law Review Vol 24.1 (2020) - Varma, G.D. and De Stefano, M., Entry Deterrence, Concentration, and Merger Policy (August 8, 2020). Available at SSRN: https://ssrn.com/abstract=3626734 or http://dx.doi.org/10.2139/ssrn.3626734 - Whinston M., ‘Tying, Foreclosure and Exclusion’ American Economic Review Vol. 80/4, pp. 837-859 (1990) - R. Whish & D. Bailey, ‘Competition Law’, 8th ed. OUP (2016)
Newspaper articles and online sources
- Facebook admits defeat over controversial WhatsApp privacy policy (BGR, 7 May 2021), - Visualizing Tech Giants’ Billion Dollar Acquisitions (CBN Insights Research Briefs, 5 May 2020), Available at < https://www.cbinsights.com/research/tech-giants-billion-dollar- acquisitions-infographic/>
- Why Did Google Buy DoubleClick’ (Towards Data Science, 6 May 2020) Available at:
- WhatsApp WeChat and Facebook Messenger Apps – Global Usage of Messaging Apps, Penetration and Statistics (Messengerpeople, 30 October 2020), available at: https://www.messengerpeople.com/global-messenger-usage-statistics/;
42
- What Countries are the Biggest WhatsApp Users (Conversocial, 20 February 2020), available at: https://www.conversocial.com/blog/what-countries-are-the-biggest-whatsapp-users - This Deal Helped Turn Google into an Ad Powerhouse. Is that a Problem? (The New York Times, 21 September 2020), available at: https://www.nytimes.com/2020/09/21/technology/google-doubleclick-antitrust-ads.html Microsoft - Replacing Skype for Business with Teams (The Seattle Times, 28 September 2017), can be accessed at: https://phys.org/news/2017-09-microsoft-skype-business-teams.html
43