TY - JOUR
AU1 - Cave,, Martin
AB - Key Points The emergence of two-sided platforms challenges competition law to adapt its ‘software’—the practical way in which it addresses its decision problem. The manner in which this can be done is illustrated by an inquiry into a merger between two food-ordering platforms. Issues faced include market definition, single-homing, price–concentration analysis, the strength of the indirect network effects and the likelihood of the market tipping. The illustration suggests that there are practicable ways of adapting existing approaches to deal with the above problems. I. Introduction In writing about digital transformation, Jean Tirole discusses the challenges competition policy faces when confronting two-sided markets, and ‘the review of the software of competition policy’ which is required.1 I understand his phrase to refer to the analytical procedures or operating system which bridges the gap between the ‘hardware’ of the statutes and the ‘apps’, or individual case decisions. This operating system for competition policy is usually carefully and conservatively managed by competition authorities, but the radical nature of multi-sidedness imposes major challenges. A view occasionally expressed is that the challenge is so severe that competition law is essentially disempowered. In the limit, an authority, having nothing reliable to say about a proposed two-sided merger, for example, could only conclude: ‘if the market is two-sided, then everything is permitted’. Yet this is more a counsel of despair than practicable advice for an authority which typically has a statutory duty to make binary merger decisions, usually on the basis of the balance of probabilities, in the presence of a variety of complicating factors. Even in one-sided markets, there remain several unresolved disputes about the appropriate analytical framework to adopt2, in addition to the inevitable uncertainties about how individual markets have developed, are developing and will develop. In short, competition authorities have little choice but to ‘keep calm and carry on’. This is likely in the first instance to involve developing software patches to superimpose on the existing operating system. Possible examples of this are variations on the SSNIP test to take account of zero-price services and the rather more complex attempts to adapt current UPP or GUPPI tests. But the key point is that such patches do have to be applied, to avoid what has vividly been described as ‘the classic drunk’s mistake—looking under the street light for your lost car keys just because the light is better there’.3 The purpose of this article is to discuss how these challenges are or might be addressed in practice. I propose to do so by reference to a two-sided platform merger inquiry which was undertaken by the UK Competition and Markets Authority in 2017.4 It was a relatively inconspicuous transaction in which the activities of the combining entities were confined to a national (at most) geographical area. It, therefore, provides a relatively simple context in which to explore the new lines of code which analysis of two-sided markets may require. The case is included to provide a concrete illustration of particular points; the paper is not ‘about’ the case. I propose to focus upon five aspects of the analysis of platforms, each of which is likely to play a substantial role in any competition case.5 They are: - market definition; - the implications of single- and multi-homing; - price–concentration analysis; - the strength of the indirect network effects; - the probability of the market tipping. There is a substantial academic literature on each of these topics which I do not intend to discuss or cite in detail. Instead my discussion will ‘jump off’ from a source which is closer to practice: a 2018 report from the OECD entitled Rethinking Antitrust Tools for Multi-sided Platforms,6 which comprises a summary chapter and a series of papers by past or present antitrust officials and others, prepared for a meeting in 2017. As well as describing the nature of the five issues selected and the OECD’ s proposals and observations, I will summarise material relating to the five issues in the published report (‘the Report').7 Since I was one of the group involved in supervising the performance of the analysis and drawing conclusions from it, the reader should not expect a wholly dispassionate assessment. II. The mise en scène The CMA’s current published Merger Guidelines,8 which contain four paragraphs expressly on two-sided markets, were first published in 2010. In relation to market definition they state (footnotes omitted): 5.2.20. [Abbreviated] In the presence of indirect network effects, prices charged to each set of customers take account of the need to get both sets ‘on board’. It may therefore be difficult to conduct a hypothetical monopolist test because: (i) there is no single price to both sets of customers to which to apply a SSNIP; (ii) the effect of a SSNIP on the demand of one set of customers may be exacerbated by indirect network effects; and (iii) the constraints on the merger firms’ products may come not only from other two-sided intermediaries but also from ‘one-sided’ firms serving one set of customers. In relation to different groups of customers, 5.7.16. [Abbreviated] Where there are direct network effects (i.e. within one group of customers), a merger may improve the competitive offering to all customers. Where network effects are indirect and involve two distinct groups of customers, a merger may improve the competitive offering to some customers but make it worse for others. In relation to tipping, 5.8.6-7 [Abbreviated] Direct or indirect network effects can make the market prone to ‘tipping’. Tipping may create switching costs for existing customers, and often means that in the presence of one network it may not be possible for a network configured in a different way to be viable. In markets characterised by network effects, a likely entrant will need to take the risk of developing new infrastructure but may not succeed in creating the necessary demand to make this profitable. In assessing tipping, the Authorities will consider whether or not customers would be willing to switch to a new supplier. Customers’ willingness will depend on the costs and benefits to them of switching. To assess these, the Authorities may take into account: the (actual or perceived) cost of switching, brand loyalty, the length of existing contracts and the closeness of relationships between suppliers and customers. In addition, customers may be willing to sponsor and/ or facilitate new entry. This material represents the visible part of the official software. But, as with an iceberg, nine-tenths of it is out of sight, in the form either of unpublished internal guidance, which sets out procedures, lists possible lines to take, and notes or even indexes past cases. However, the even more important and in some cases unrevealed part of it resides in the institutional or individual memories of the competition agency’s staff: where to start in an inquiry; what analytical techniques have worked in which type of case; what approaches were abandoned under criticism of the parties to the investigation; what data collection difficulties arose; what argumentation survived appeal (and what did not). It is this material which is in very short supply in the case of two-sided market mergers. The circumstances of the merger under discussion can for our purpose (which is strictly limited as described above) be quickly summarised. The two merging firms are food-ordering platforms, linking restaurants and customers. The share of this market which they account for is about 80 per cent, and the increment associated with the merger is less than about 10 per cent. The wider market place also includes firms which both operate an ordering platform and provide food delivery (‘food-ordering and logistics companies'), and restaurants or restaurant chains which themselves take orders and deliver food. It is notable that a monetary transaction joins both sides of the market place. The platform is usually rewarded by a percentage of the customers’ bill. This is naturally smaller in the case of food-ordering platforms, where restaurants organise delivery themselves, than in the case of food-ordering and logistics companies, which also provide delivery. In summary, the Authority concluded that9: - some of the evidence indicated that the smaller firm imposes a limited competitive constraint on the larger, and there were doubts about its recovery programme and about how long it would maintain itself in the market place; - the overall market is dynamic and evolving, with the prospect of significant growth from food-ordering and logistics companies, one of which has exerted a greater competitive pressure on the larger acquiring firm than did the smaller acquiree; this effect is likely to increase; - where only the two merger parties compete, the competitive impact of the acquiree on the acquirer is particularly small. Accordingly, although it was recognised that the evidence did not point in one direction, the proposed merger was found not to lead to a substantial lessening of competition. I now discuss how the four issues identified above were addressed in the analysis. III. Market definition The observations on market definition in the CMA’s merger guidelines, dated 2010, are noted in the previous section. Since that date, a significant academic literature has grown up, summarised in the OECD volume cited above. An important distinction proposed in that literature is between two-sided ‘transaction’ and two-sided ‘non-transaction’ markets. In the former, a monetary transaction takes place between the two sides; in the latter, there is no such transaction or it is invisible to the platform, even if there is an interaction between the two sides. The ‘transaction’ can take many forms. For example, in the case of an advertiser-supported platform, it may arise when a user clicks on an advertisement, and thus triggers a payment by the advertiser to the platform.10 It has been argued that in a two-sided transaction market the products purchased by the two sides must stand in a one-to-one relation with each other. This creates, it is argued, a link between the two sides so strong that it cannot sensibly be disregarded. In a single market involving both sides, the total price is likely to be positive; this may overcome the problem with the application of the SSNIP test that it cannot be applied with respect to a zero price—although other ways around this difficulty have been proposed including the replacement of SSNIP by a test involving a Small but Significant Non-transitory Decrease in Quality (or SSNDQ). The UK Merger Guidelines11 generally note that the process of market definition is generally bound to overlap and be inseparable from the assessment of competitive effects. In the present case, the approach to two-sidedness which is endorsed is effectively as set out above. It is noteworthy that this approach is not universally accepted, however; and the OECD’s somewhat agnostic summary observations on market definition in two-sided markets can be summarised as follows: - in some cases a formal market definition can be dispensed with; - in the absence of indirect network effects, separate market definition exercises on each side will suffice; - where there are indirect network effects they can usually be taken into account in the framework of either a separate or combined market definition.12 IV. Single and multi-homing Much of the analysis of platform competition revolves around whether restaurants and customers confine themselves to one platform (single-home) or choose between multiple platforms before they make a purchase (multi-home). The usual interpretation of a multi-homer in economic or competitive discussions implies more than simple registration on several platforms. It requires the agent universally (or at least frequently) to scour more than one platform prior to making a choice. Thus, if a household alternated between Italian and Indian cuisine, and had favourite restaurants for each which were on different platforms, it would be an oscillating single-homer, not a multi-homer. One (still imperfect) way of capturing this nuance in survey work is to define a multi-homing household not by its platform registrations but by the criterion of having made purchases on several platforms in the past × weeks. If every party is truly a multi-homer, then competition is feasible (subject to a variety of other problems such as strong direct networks effects). If both sides single home, then platforms compete on both sides. Every buyer or seller captured can then be exploited. The more interesting case arises when one side single-homes and the other does not—a competitive bottleneck.13 The platform then tries to recruit as many single homers as possible, which can then be sold on to the highest bidder on the other side. This puts a premium on converting a multi-homing agent—for example a restaurant—into a single homer. In the case of retail customers, this can be accomplished by loyalty rewards. These possibilities are illustrated in the Report in the 2×2 diagram shown in Figure 1. Figure 1: Single- and multi-homing and the effect on platform competition Restaurants Single-homing Multi-homing Consumers Single-homing Platforms compete on the consumer side and on the restaurant side Platforms compete on the consumer side; platforms face little direct competition on the restaurant side, as they provide access to separate sets of consumers Multi-homing Platforms compete on the restaurant side; platforms may face little direct competition on the consumer side, as they provide access to separate sets of restaurants, though this depends on the extent to which consumers consider the different sets of restaurants to be close substitutes14 Platforms compete on both sides and may try to push restaurants (or consumers) towards single-homing, eg through exclusivity (restaurant side) or loyalty rewards (consumer side) Restaurants Single-homing Multi-homing Consumers Single-homing Platforms compete on the consumer side and on the restaurant side Platforms compete on the consumer side; platforms face little direct competition on the restaurant side, as they provide access to separate sets of consumers Multi-homing Platforms compete on the restaurant side; platforms may face little direct competition on the consumer side, as they provide access to separate sets of restaurants, though this depends on the extent to which consumers consider the different sets of restaurants to be close substitutes14 Platforms compete on both sides and may try to push restaurants (or consumers) towards single-homing, eg through exclusivity (restaurant side) or loyalty rewards (consumer side) Source: The Report, Appendix E, page E2. Open in new tab Figure 1: Single- and multi-homing and the effect on platform competition Restaurants Single-homing Multi-homing Consumers Single-homing Platforms compete on the consumer side and on the restaurant side Platforms compete on the consumer side; platforms face little direct competition on the restaurant side, as they provide access to separate sets of consumers Multi-homing Platforms compete on the restaurant side; platforms may face little direct competition on the consumer side, as they provide access to separate sets of restaurants, though this depends on the extent to which consumers consider the different sets of restaurants to be close substitutes14 Platforms compete on both sides and may try to push restaurants (or consumers) towards single-homing, eg through exclusivity (restaurant side) or loyalty rewards (consumer side) Restaurants Single-homing Multi-homing Consumers Single-homing Platforms compete on the consumer side and on the restaurant side Platforms compete on the consumer side; platforms face little direct competition on the restaurant side, as they provide access to separate sets of consumers Multi-homing Platforms compete on the restaurant side; platforms may face little direct competition on the consumer side, as they provide access to separate sets of restaurants, though this depends on the extent to which consumers consider the different sets of restaurants to be close substitutes14 Platforms compete on both sides and may try to push restaurants (or consumers) towards single-homing, eg through exclusivity (restaurant side) or loyalty rewards (consumer side) Source: The Report, Appendix E, page E2. Open in new tab The accompanying footnotes point out some important special features. Thus, whereas restaurants make ‘once and for a period’ decisions as to which platforms to use, a multi-homing customer makes a series of discrete decisions over time. A platform can, therefore, try to lure a new restaurant to take its service by suggesting that its (the platform’s) better service would divert customers to it. These and other factors complicate the process of drawing inferences concerning the degree of constraint which one platform can provide on another in each side of the market. In terms of the degree of multi-homing (and subject to the definitional and measurement problems discussed above), the Authority reached the following conclusions: – in relation to multi-homing across the two merger parties, restaurants attached to the smaller were likely to be listed on the larger, but not vice versa; – there is significantly more multi-homing across the merger parties than between the larger merger party and the largest food-ordering and logistics platform; – on the consumer side, where the larger merger party has a predominant share of the pure food-ordering platforms business, that party has a higher share of single homers than do restaurants. The Report concludes: ‘We therefore expect competition between the parties to be mainly focussed on the consumer side'.15 This conclusion is an input into the final decision about the nature of the competitive constraints placed on one platform and another, together with a variety of other factors, including the strength of indirect network effects—the subject of the next section. Finally, the OECD, in its discussion of the more general question of assessing market power on platforms, counsels caution over what the nature of homing can tell us about the substitutability of platforms.16 V. Price–concentration analysis It is possible in principle to adapt to a multi-sided context the GUPPI and UPP approaches widely used in single-sided markets, but the adaptation process is complex, because a change in any one price affects all four quantities. It has been shown that knowledge of six diversion ratios is required. It is also necessary to consider the possibility that a platform merger can increase prices to one side without that side suffering. As Evans and Schmalansee note: ‘suppose that Open Table [a restaurant booking platform] proposed a merger with a competitor and it is determined that the merged firm would likely increase prices to restaurants. It does not follow that the merger is undesirable however. Restaurants would likely have access to more consumers, and that might make up for the price increase. And if restaurants single-home and the merged firm does not take the radical step of charging consumers to make reservations, consumers would clearly be better off; they would still face a zero charge and could access more restaurants on a single platform'.17 In the merger under discussion here, the CMA performed a price–concentration analysis which took account of the two-sidedness of the market, by examining the effect of changes in the number of restaurants on platforms in different geographical areas on the value of consumer orders made on the platforms of the two merger parties.18 In particular, the results showed that – a change in the number of restaurants available on one of the merger partner platforms had a negative impact on the value of orders placed on the other. But these effects diminish, in absolute terms, over time, with the smaller partner imposing no discernible constraint on the larger in the most recent year for which data were available; – expansion of the number of restaurants on the two largest food-ordering and logistics firms had a negative effect on orders on both the merger partners. These (and other) results fed into the overall conclusion summarised in section II. VI. The strength of the indirect network externalities This section analyses how the question of assessing the strength of indirect network externalities can be addressed.19 These effects arise when the value of a platform to parties on one side depends upon the number of customers or providers on the other side.20 Thus, in this instance, the relevant question are: (i) Does the value which a restaurant derives from being on a platform increase with the number of customers on the other side of the platform? And (ii), does the value to customers of being on a platform grow with the number of restaurants on the platform? At first sight the answer to both questions appear to be ‘yes’. But there are nuances in the answer which need to be taken into account In each case, there is a notional function connecting the number of customers or restaurants, and the benefit accruing to the other side of the market. If the graph of this function is completely flat, there is no indirect network externality. If it rises constantly, there is a positive indirect network externality across the whole range. But it could take any shape, including rising up to a ‘critical mass,’ flattening out, and then falling. Thus, the benefit that a customer might derive from access to restaurants via a food-ordering platform might first rise with the number of restaurants, then remain constant when each customer is satisfied with the choice, but then fall as the process of scrolling through the list of restaurants becomes tedious.21 In a merger inquiry, the direction and force of indirect network effects can be important in assessing the relative strength of competitors in their present state, and in forming conjectures about how the market place is going to develop in the future. In the merger inquiry under examination here, the strength of the indirect network effects was investigated in four ways: – internal evidence from the merging parties’ documents – evidence received from other online food-ordering platforms – econometric evidence, and – a survey of consumers. I focus here upon the last piece of evidence, as the first two are very case-specific, and the third has been noted in the previous section. The consumer survey carried out as part of inquiry permitted the analysis of answers to a diversion question asked in various localities: ‘If your platform of choice were not available, what alternative would you choose instead?’22 The results were complex and somewhat mixed: – respondents were more likely to switch to a named food-ordering and logistics operator, the more local restaurants that operator had; – switching behaviour between the merging partners in the event of a required diversion did not support the same result; – at best this diversion test does not expressly address the question of whether more restaurants on one side attract more consumers on the other. The observations may have other explanations. – The conclusion drawn from this evidence was that it painted a mixed picture of their actual strength. In addressing the likely effect of the merger, the report concluded—taking into account all factors including its analysis of the strength of the indirect network externalities, that in fact the acquiree was imposing a limited competitive constraint upon the acquirer. VII. Market tipping One of the reasons for wanting to know the strength of network externalities is that, as noted in the UK merger guidelines summarised in section II, this variable may influence the probability of the market tipping. Merging firms may then argue that if the market will tip anyway, a merger will not entail a substantial lessening of competition; to quote Karl Marx, it merely ‘lessens or shortens the birth pangs.’ The report examines this aspect at some length. It discusses a number of factors which might support or refute the tipping hypothesis. Thus, suppose that one class of agents is reluctant to be listed with other classes: for example, an expensive restaurant might not want to be associated with cheaper ones. This would argue against tipping. But no evidence of this preference for separation was found in respect of the merger parties on either the restaurant or the customer side. But there is evidence that other providers including the food-ordering and logistics operators have different restaurant clienteles. As for the other conditions, these are, conventionally, the degree of platform differentiation (discussed above), the strength of the indirect network externalities (which on the consumer side were found not to be especially strong) and customers’ homing practices. In relation to the last factor, the key point is that, if one side can multi-home at a comparatively modest cost, then a major force leading to tipping—the obligation on all agents to choose only the most advantageous platform—evaporates. In these circumstances, platforms can co-exist even if they are fairly undifferentiated, and in equilibrium they can be of different sizes. This is important since, as noted above, both restaurants and customers exhibit a substantial degree of multi-homing. There is scope for the collection of some informal empirical evidence on market outcomes. For example, the relation between the shares of orders won by the two merger parties can be examined; these proved to be fairly stable in recent years. Or the development of the same market place has developed can be tracked in other jurisdictions—which in this case reveals a variety of somewhat divergent results.23 It is not hard to ‘explain’ each observation in terms of a variety of contingent features of each jurisdiction, but this kind of reasoning can contribute to the evaluation of a claimed broad generalisation. Overall, the Authority reached the conclusion that that there is no compelling evidence that the survival of only a single food-ordering platform is pre-ordained. VIII. Conclusion In this paper I have sought to illustrate, based on a concrete example, how some of the issues raised by two-sided markets are being addressed in the context of mergers. For the avoidance of doubt, I am not proposing this example as a model for other inquiries to follow, but as an illustration of the practical complexities of adapting the ‘software of competition policy’ to new challenges. But are they really that new? It is now widely recognised that competition policy had been speaking in terms of platforms before the terminology was introduced, since analogue mergers between two-sided platforms have existed for centuries—for example in the forms of the advertiser-supported press or broadcasting media, modern shopping malls (and their ancient precursors), and airports which bring together passengers, airlines and retail outlets in a multi-sided market. The difference is that theoretical economists such as Rochet and Tirole,24 and more empirical economists—notably Evans and Schmalensee25—turned their minds to the problem. This occurred during the early stages of the dizzying growth of the digital platforms which now head the list of the world’s largest corporations. Some consensus is emerging among economists as to how to analyse general platform issues, although new complexities and special cases continue to be revealed.26 What seems to be less developed at the explicit level is the ‘software’ of the analytical process—what data (in the widest sense of the term) to collect, how to process it and how to interpret and ascribe weights to the results. This should ultimately allow some generalisations, subject to the inevitable fact that any really existing market will stubbornly refuse to conform to any theoretical stereotype. In terms of the specific issues considered above, it is difficult to get a sense of the state of play in merger cases in general, because competition authorities differ in the degree to which they disclose the nature of the ‘software’ which they use. But based on the discussion above, my impression is that, where the legislation does not directly or indirectly bind the authority to a particular approach to market definition, the OECD document’s agnostic approach is apposite: this roughly says that the choice between combined and separate definitions for the two sides of the market should not make much difference, provided the linkages between the two sides are adequately taken into account. The impact of single or multi-homing is a matter of some complexity. While the consequences in extreme cases can easily be rationalised, in reality a case is likely to generate either ambiguous (because of definitional problems) or mixed results. More theoretical and empirical results on this question are highly desirable. In relation to price–concentration analysis, there is not at present a fully-developed patch to deal with two-sidedness. We are as bereft of a simple, easily applicable and widely accepted approach as was the case in all competition analysis before the emergence of UPP as an organising framework. The fall back is to ensure that the estimations employed expressly take account of the two-sidedness of the market. The two remaining issues are linked. Indirect network effects are a defining feature of two-sided markets. Their presence and strength (in either direction or in both directions) are key to how any competition investigation should be approached. It is also important to bear in mind that their marginal indirect network effect may be inconstant. In the merger considered here, a variety of evidence was brought to bear on these questions. Other investigations will no doubt unearth many more approaches. Finally market tipping, or the question of the sustainability of competition, is an important one in the context of mergers in particular. It looks as if, theoretically, a variety of different equilibria are possible, and it is reasonable to hope that more empirical enlightenment on this highly platform-specific question will soon be provided by empirical industrial organisation economists. In the meantime, a cautious view of the inevitability of tipping seems appropriate. To sum up, the task which competition authorities face in dealing with two-sided mergers is complicated, but probably not insuperable. Footnotes 1 Jean Tirole, Economics for the Common Good (Princeton University Press 2017) 392. 2 A topical example of such a disagreement is provided by the issue of what effect mergers will have on innovation. 3 David S. Evans and Richard Schmalansee, ‘The anti-trust analysis of multi-sided platform businesses’ (2015) The Oxford Handbook of International Antitrust Economics 430. 4 For a theoretical treatment, see Joao Correia-da-Silva, Bruno Jullien, Yassine Lefouilli, and Joana Pinho, ‘Horizontal Mergers Between Multi-sided Platforms: Insights From Cournot Competition’ (August 2018) Working Paper TSE-946 Toulouse School of Economics. 5 This means that I ruthlessly leave out every other aspect of the inquiry. 6 OECD, Rethinking Antitrust Tools for Multi-sided Platforms; 2018. Available at: http://www.oecd.org/daf/competition/Rethinking-antitrust-tools-for-multi-sided-platforms-2018.pdf 7 CMA, A report on the Anticipated Acquisition by JUST EAT plc of Hungryhouse Holdings Limited, November 2017, available at https://assets.publishing.service.gov.uk/media/5a0d6521ed915d0ade60db7e/justeat-hungryhouse-final-report.pdf & https://assets.publishing.service.gov.uk/media/5a0d64fee5274a0ee5a1f22a/appendices-glossary-final-report-justeat-hungryhouse.pdf 8 CMA Merger Assessment Guidelines, first published 2010, Available at https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/284449/OFT1254.pdf 9 Report, 6-7. 10 See David Evans, Attention Platforms, the Value of Content, and Public Policy (January 2019) 5. 11 JUST EAT (n 8) 4. 12 OECD (n 7), 15. I omit the recommendations on SSNIP tests from a conviction that their principal contribution is a conceptual framework rather than a means of conducting empirical analysis. 13 See Mark Armstrong, ‘Competition in two-sided markets’ 37 (3) Rand Journal of Economics 668-691. 14 This framework is a simplification of how competition between platforms is likely to work in practice and is most applicable to competition between horizontally undifferentiated platforms. More detail is provided in Appendix E [of the Report]. 15 Report, para 6.77. 16 OECD, 19 and 21. 17 (n 4), 428. 18 See Report, Annex F, F3–F13. 19 A useful survey of platforms and network effects in general, including tipping, can be found in a survey paper by Paul Belleflamme and Martin Peitz, ‘Platforms and network effects’ (2017) in Louis Corchon and Marco Marini, Edward Elgar (eds.) Handbook of Game Theory and Industrial Organisation 286–317. 20 Note that simply counting numbers on each side of a platform a fairly primitive simplification, given the heterogeneity of the parties on the same side of certain platforms. In the present context geographical heterogeneity is important. It is therefore to conduct any empirical work on the basis of local markets. 21 But note that the platform can deal with the latter problem by various means, including reducing the radius of the coverage when the number of restaurants exhibited starts to grow too fast. 22 There are always doubts about the reliability of answers to hypothetical diversion questions. In another well-known two-sided platform merger, a natural experiment came partly to the rescue. Massimo Motta records that: “However, sometimes there might be unexpected events that may help an authority make some more quantitative assessments of the substitutability among products. For instance [in connection with the Facebook/WhatsApp merger], one day WhatsApp had a widespread four-hour service unexpected outage, and what users did in such circumstances may turn out to give hints at how they look at different products. In that particular case, the European Commission found that in the following 24 hours, competing communications apps (such as Telegram and LINE) gained millions of new users, information which was used to suggest low switching costs for users and low entry barriers.” See https://www.competitionpolicyinternational.com/wp-content/uploads/2018/05/CPI-Talks....pdf 23 Report, paras 6.86-6.94. 24 Jean-Charles Rochet and Jean Tirole, ‘Platform competition in two-sided markets’ (2003) 1 (4) Journal of the European Economic Association 990-1029. 25 In numerous works listed in Evans and Schmalensee, (n 4), 443–4. 26 This is not the same as unanimity, which does not prevail with one-sided markets either. Author notes Professor Martin Cave is a Board Member of CERRE. Since October 2018, he is Chair of Ofgem, the UK energy regulator. Prior to that, he was an inquiry chair at the UK Competition and Markets Authority, having formerly been a deputy chair at the UK Competition Commission. He is currently a visiting professor at the London School of Economics, and formerly held chairs at Brunel University, at Warwick University and the London School of Economics. Martin.e.cave@btinternet.com. The views expressed here belong to the author alone. He is grateful for comments from David Evans. © The Author(s) 2019. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - Platform Software Versus the Software of Competition Law
JF - Journal of European Competition Law & Practice
DO - 10.1093/jeclap/lpz036
DA - 2019-12-01
UR - https://www.deepdyve.com/lp/oxford-university-press/platform-software-versus-the-software-of-competition-law-lrJc0ga0oI
SP - 472
VL - 10
IS - 7
DP - DeepDyve
ER -