Entry into platform markets subject to strong network effects and high switching costs can occur in two ways. First, by offering drastically new functionality (i.e. through Schumpeterian innovation). Second, through “platform envelopment” whereby a provider in one platform market – the origin market – enters another platform market – the target market – and combines its own functionality with that of the target in a multi-platform bundle that leverages shared user relationships and/or common components. Envelopers capture market share by foreclosing an incumbent’s access to users; in doing so, they harness the network effects that previously had protected the incumbent.
The paper (which is quite long and dense, and so calls for a long review) is organised as follows.
Section II provides a brief overview of the economics of platforms.
Platform businesses are characterised by the existence of within-group and/or cross-group network effects. Platforms featuring within-group network effects typically have one-sided effects. For example, a communication platform (e.g. a messaging app) is more valuable to consumers when the number of consumers that can be reached through that platform is greater. Platforms with cross-group network effects are examples of two- or multi-sided platforms – i.e. the value of the platform to (at least) one side of the market increases with the participation levels in another side of the market. In some cases, a platform market may be two-sided despite one side experiencing mainly within-group network effects and the other side cross-group network effects. This is the case, for instance, of social-network advertising platforms.
The existence of within- and cross-group externalities reflect membership and usage externalities. A (positive) membership externality exists when the value received by a member of one side increases with the number of members on the same or another side. A (positive) usage externality exists when the members of a group benefit when members of the same or the other group intensify their use of the platform. Because for multi‐sided platforms the demand by one customer group depends on the demand by the other customer group, they need to have both sides on board to create value for either side. In order to keep both sides on board, platforms will choose their price structure considering both the price elasticity of each side and the magnitude of the externalities or network effects linking both sides. Hence, the profit maximising price on one side may fall below its marginal cost. It may be zero or even negative.
Section III discusses the logic of platform envelopment.
Due to network effects, competition in platform markets tends to be characterised by winner-take-all battles for dominance. Platform envelopment refers to the entry of a platform with market power in an origin market into another platform’s market (the target market). Some authors (namely Eisenmann et al.) consider that platform envelopment has had a major role in shaping the ecosystem of platform-mediated network industries, and have identified at least 40 examples of this in recent years, beginning with Microsoft entering into the streaming media and browser markets, up to Google and Facebook recently entering multiple adjacent markets to their core business. While envelopment is most commonly successful when the products in the origin and target platform markets are complements (e.g. operating systems and applications), it can also occur in cases where the products are weak substitutes or even unrelated.
Normally, though not always, the enveloper ties its own services in the origin market with those offered in the target market. The authors distinguish between three types of common tying strategies employed in envelopment: bundling, virtual bundling and self-preferencing. As to the first type, the enveloper might decide to sell the products or services to its overlapping customer base in the origin and target platforms as a bundle, either exclusively or alternatively. Bundling is accomplished either by product design (e.g. video and photo editing) or contractually. Virtual bundling occurs when each product is sold separately, but price coordination might still achieve some or all the effects of tying (e.g. when consumers prefer to concentrate their purchases of interoperable software from a single supplier due to bundled discounts). The third strategy, self-preferencing, consists of affecting the terms of trade in the target market so that the enveloper might enter the target market by bending the origin platform’s rules in order to provide a better outcome for its own products or services.
Section IV focuses on platform envelopment strategies that combine data from origin and target markets.
By entering multiple platforms with overlapping user bases, the enveloper may be able to collect and combine data about their common users by tracking their behaviour across platforms. There are three channels through which user data may provide static benefits that may encourage multi-platform entry: it may lead to the more efficient provision of products in the target market, it may reduce asymmetries of information to which new entrants are usually subject to, and it may allow the enveloper to provide a better product or service in the origin market or extract more surplus there by reverse economies of scope and improved price-discrimination. From a dynamic standpoint, this data envelopment may help monopolise the target platform market, capture all data generated in such a platform, and entrench its dominance in the origin platform.
This theory of harm thus relies on the idea that a platform operating in the target market represents a credible potential competitor for the enveloper in the origin market, and that achieving control of the user base in the target market can be used to improve the service offered in the origin market or to engage in efficient price discrimination. The enveloper can pre-empt this threat by entering the target market and pricing low enough to acquire dominance in it.
Section VI summarises the potential procompetitive and anticompetitive effects of ‘data tying’ strategies, while Section VII explores ways to limit their possible detrimental effects.
In addition, given how challenging antitrust enforcement is likely to prove, one may want to look beyond it. To bring successful cases, competition agencies will have to gather enough evidence of (a) coercion in data extraction and (b) likely effects effectively to overcome the presumption of legality. This is a challenging prospect, particularly in platform markets. Remedying anticompetitive platform envelopment is not an easy job either. A mere “cease-and-desist order” will not restore competition. Structural remedies, such as “divestments” or “line of business prohibitions,” may destroy considerable value. Other remedies, such as “data portability,” “data sharing” or “privacy regulation” require continuous monitoring by personnel with sophisticated understanding of the platform markets at issue. These are the sort of problems best dealt through ex-ante regulation. Competition agencies may want to focus instead on preventing envelopment ex ante by stopping mergers across platforms and preventing acquisitions that may have the objective of eliminating the threat posed by disruptive innovators (i.e. so-called “killer acquisitions”). Other tools, such as ex-post market investigations, or ex ante mandatory data sharing and data portability rules, should also be considered. Another possibility would be to enhance the privacy protection duties incumbent on platforms with market power, or directly to limit the ability of dominant, multi-platform conglomerates to combine user data across platforms. All these interventions would cap the data superiority of dominant platforms with presence in several markets; open the (origin) platforms where the data is monetised to competition; and eliminate incentives by dominant platforms to enter other (target) platform markets in order to capture additional and complementary user data.
This is a very interesting, thoughtful and detailed paper. It is 52 pages long, but it is hard to see how it could be much shorter. In any event, and in order to keep the size of this review manageable/ I decided not to cover: (i) the very interesting discussion of the economic incentives that underpin the adoption of platform envelopment strategies – including efficiencies, but also the possibility to engage in price discrimination and foreclosure. In my view, this discussion could be published as a paper on its; (ii) a discussion on the economics of data; (iii) the whole of Section V, which contains a description of the application of the theory of harm developed by the authors to online advertising platforms (i.e. Facebook and Google).
The paper is also easy to follow – and I say this as a lawyer with limited economic expertise. In particular, I found the discussion of how platform envelopment works and when it might occur enlightening and comprehensive.
I found myself thinking at times whether platform envelopment should not be compared to other conglomerate theories of harm, and I think that it would have been useful for the authors to make that broader link. This is particularly so because I feel that the prominence of conglomerate theories is likely to increase as it becomes clearer that they are well suited to deal with moligopolistic market structures (to borrow from Petit). As such, it would be useful to start thinking about whether we can develop some overarching principles governing the application of conglomerate theories of harm in this field – as a paper I reviewed a few weeks ago did.