اثرات شبکه، ساختار بازار و عملکرد صنعت

This paper analyzes oligopolistic markets with network externalities. Exploiting a minimal complementarity structure on the model primitives that allows for pure network goods, we prove existence of non-trivial fulfilled-expectations equilibrium. We formalize the concept of industry viability, investigate its determinants, and show that it improves with more firms in the market and/or by technological progress. These results enlighten some well-known conclusions from case studies in the management strategy literature. We also characterize the effects of market structure on industry performance, which depart substantially from ordinary markets. The approach relies on lattice-theoretic methods, supplemented with basic insights from nonsmooth analysis.

It has often been observed that the nature of competition is qualitatively different in network industries. The presence of interlinkages in consumersʼ purchasing decisions induces demand-side economies of scale that may strongly affect market behavior and performance. When such effects prevail, be they of the snob or bandwagon type, purchase decisions are influenced by buyersʼ expectations, leading to behavior not encompassed by traditional demand theory (Veblen [38], Leibenstein [22]). From an industrial organization perspective, these distinctive features raise new questions and impose some methodological challenges. In their pioneering work on markets with network effects, Katz and Shapiro [19] proposed the concept of fulfilled expectations Cournot equilibrium (FECE), which was adopted by some of the early literature. This has led to a number of results that distinguish network markets from ordinary ones.1 The purpose of the present paper is to provide a thorough theoretical investigation of markets with homogeneous goods and network externalities. We consider oligopolistic competition amongst firms in a market characterized by positive (direct) network effects when the products of the firms are perfectly compatible, so that the relevant network is industry-wide. This is motivated by both positive and normative considerations. In terms of the former, several important industries fit the perfect compatibility framework, in particular those in the telecommunications sector, such as fax, telephone, the Internet, but also many classical industries such as compact discs, fashion and entertainment.2 More important are the normative grounds, which stem mainly from the critical problem of industry take-off that new network goods are confronted with. A single (industry-wide) network is a crucial element in surmounting the take-off hurdle, or at least in avoiding potentially long delays before achieving success (Shapiro and Varian [31]). Indeed, the business strategy literature has concluded, through a number of detailed case studies dealing with the emergence of particular industries in the last thirty years, that interconnection amongst all the firms in a network industry (i.e., a single network) is probably the most important ingredient for success in launching a new network product (Rohlfs [29]). Thus a good understanding of the single network case will shed quite some light on the incentives for compatibility faced by firms and consumers in the case of firm-specific networks. We shall return to this key point several times below. In contrast to the extant literature, this paper considers general demand functions with non-separable network effects, a critical feature if one wishes to capture pure network goods (those with no stand-alone value, such as most telecommunication products), and the so-called feature of demand-side increasing returns (see assumption (A5)(A5)). With pure network goods, the trivial outcome of zero output is always a self-fulfilling equilibrium, since there will be no actual demand if the market expectation is that there will be no eventual sales (in other words, nobody wishes to be the only person around owning a phone, say). In view of this, the industry will fail to take off at all if this is the only equilibrium, but might also end up coordinating on this worst possible outcome when other equilibria are present. In a nutshell, this is the so-called industry viability problem, a general treatment of which is the central concern of this paper. To this end, an important pre-requisite is a good understanding of the issues of existence and multiplicity of FECE, which can clearly discriminate between the trivial FECE and the non-trivial ones, in terms of meaningful conditions imposed on primitives of the oligopoly model. Another aim of the paper is to provide an extensive inquiry into the effects of market structure (or exogenous entry) on market performance. Throughout, the paper takes a comparative perspective in that results are contrasted with their Cournot counterparts, in an attempt to shed light on the distinctive features of network industries. The underlying approach is to impart minimal complementarity structure to the model at hand, which achieves the twin goals of ensuring the existence of a fulfilled expectations Cournot equilibrium while at the same time allowing clear-cut predictions on the comparative statics of market performance with respect to the number of firms. The critical structure is imposed in the form of two economically meaningful complementarity conditions on the primitives that guarantee the key properties that, along a given firmʼs best response, industry output increases in rivalsʼ total output as well as in the expected network size. In terms of methodology, the existence and comparative statics parts rely on lattice-theoretic techniques, but these need to be supplemented by basic novel insights from nonsmooth analysis, in particular for the viability analysis.3 We next provide an overview of our findings, coupled with a literature review. While existence of FECE follows from the monotonicity structure via a double application of Tarskiʼs fixed point theorem, this is of limited interest, as the underlying equilibrium may a priori be the trivial one. To complete the analysis, we derive two sets of conditions, each of which ensures the existence of a non-trivial equilibrium. These conditions have clear economic interpretations; they amount to requiring relatively strong network effects near the origin or away from the origin. Although the model is static in nature, we construct an explicit learning dynamics, mapping consumersʼ expectation of the network size to the corresponding Cournot equilibrium industry outputs. This tatonnement-type dynamics shall serve a dual purpose. It provides a natural theoretical foundation for an equilibrium concept that might be viewed as too demanding in its implicit simultaneous determination of both firmsʼ behavior and the correct size of the market. The dynamics also serves as a convenient tool to analyze the viability of the industry. In fact, it has tacitly been the basis of earlier informal discussions of the viability issue in the literature. Studies of telecommunications markets, such as Rohlfs [28] and Economides and Himmelberg [14], often suggest that network industries typically have three equilibria. Under this natural dynamics, the two extreme equilibria are stable in expectations and the middle equilibrium (usually called critical mass) is unstable. The argument behind this structure is quite simple for pure network goods: If consumersʼ initial expectation is below the critical mass, so that few buyers are expected to acquire the good, then the good will be of little value to consumers and few of them will end up buying it. These low sales in turn further depress consumersʼ expectations through the above dynamics, and the market unravels towards the trivial (or no-trade) equilibrium, giving rise to a failure to take off for the industry. However, if expectations are higher to start with and network effects are relatively strong, higher equilibria will also be possible. This argument is often used to explain the start-up problem in network industries, or the difficulties faced by incumbent firms in attempting to generate enough expectations to achieve critical mass. In this setting, due to increasing returns on the demand side and to the need for expectations, multiple equilibria and path dependence (the notion that early events can have significant long run effects) are the norm, rather than the exception. An important aim of the present paper is to shed light on the role of market structure as a determinant of the viability of a network industry, a novel and fundamental issue that, surprisingly, has not been addressed in the theoretical literature. We find that the presence of more firms in the market always enhances industry viability, by lowering the critical mass needed to avoid the trivial equilibrium. The same conclusion holds for exogenous technological progress. These two effects provide a plausible explanation of several recorded failures and successes in attempts to launch new network goods, as reported in some detail in other sections, in particular with regard to the history of the fax industry. Indeed, Rohlfs [29] forcefully argues that interconnection between suppliers of a network good is a critical feature that is at the heart of past successful new industry launches, sometimes in conjunction with technological improvements.4 Rohlfsʼ detailed case studies provide strong evidence for the policy relevance of our theoretical results on viability. Regarding market performance, the basic structure leads to an industry output that increases in the number of firms, n, as in standard Cournot competition. As this also implies an increase in the equilibrium network size, market price need not decrease with more competition, i.e., quasi-competitiveness need not hold here. The most drastic departure from standard oligopoly lies in the effects of entry on per-firm profits. Whenever per-firm output and market price increase (decrease) with n, per-firm profits increase (decrease) in n as well (see Economides [13]). The conclusion that competition may increase each firmʼs profit is quite provocative and leads to several important implications, both from theoretical and policy-oriented perspectives. The effects of entry on social welfare and consumer surplus also display some distinctive features relative to standard Cournot competition. Demand-side economies of scale broaden the conditions under which social welfare increases with more entry, but they may have the opposite effect on consumer surplus whenever the marginal increase of price due to a higher network size increases with output. Our results build on the perception already prevalent in the literature that standard results on the workings of competition can easily be reversed in network industries. Since, for each dimension of market performance, the conventional intuitive outcome and its opposite can both hold in robust ways, it is highly desirable to arrive at a clear understanding of the respective specific market characteristics under which these two outcomes prevail. 5 As a consequence, a number of policy issues will need revisiting in network industries, whenever market characteristics are such that unconventional outcomes prevail. There is more scope for pro-competitive cooperation or coordination by firms in network markets. There will be a pronounced tendency towards less entry deterrence activities; a higher propensity for licensing, probably coupled with lower royalty rates or licensing fees; less patenting or a relatively more permissive attitude towards patent infringement by a firmʼs rivals; and more joint ventures for research and development towards common standards, improved product performance and lower production costs. Proper reaction to these new incentives for coordinated action by market competitors might well require a significant overhaul of existing antitrust policy (Shapiro [30] and Katz and Shapiro [20]). Another noteworthy aspect of this paper is that it offers three explicit examples with easy closed-form solutions to illustrate in a simple way some of the key conclusions. In particular, Example 1 captures with closed-form solutions most of the relevant features often associated with the telecommunications industry in the literature, as well as our new results on viability. The paper is organized as follows. Section 2 presents the model, the equilibrium concept and the assumptions. Section 3 deals with existence of equilibrium. Section 4 formalizes the concept of industry viability and its determinants. Section 5 analyzes market performance as a function of the number of firms. Section 6 concludes, and Section 7 contains all the proofs.

This paper has provided a thorough theoretical analysis of a static model of oligopolistic competition with non-additive network effects. A minimal complementarity structure on the model leads to industry output increasing in the rivalsʼ output that a firm faces and in the expected network size, thus yielding in one broad stroke existence of symmetric equilibrium as well as some key characterization results with a comparative statics flavor. The so-called start up problem is extensively investigated, in terms of basic properties of the market primitives, and the strength of network effects. In particular, industry viability, a key concept in network markets for which we provide novel theoretical foundations, is shown to be enhanced by higher numbers of competitors in the market as well as by technological progress. The central feature here is a simple learning/adjustment dynamics that also serves as a theoretical foundation for the solution concept of fulfilled expectations Cournot equilibrium. We elaborate in some detail on the natural tendency for multiple equilibria, path dependence, and the importance of initial market expectations, features that emerge due to the presence of demand-side increasing returns driven by non-additive network effects. As to the effects of market structure, sufficient conditions are derived for each dimension of market performance to increase or decrease with more competition. The tendency for counterintuitive effects, which is extensively characterized, is much stronger than in ordinary markets. Most notably, price and per-firm profits can both increase with the number of firms, with the latter effect having no counterpart in ordinary markets even under scale economies (Amir and Lambson [4]). Along with the need, often critical, for firms to join hands to successfully launch new network products, these results underscore the co-opetitive nature of the FECE concept: Firms are partners in setting expectations and building consumer base, but (business-stealing) competitors in serving that base. Several instructive examples with closed-form solutions are constructed, one of which reflects exactly the prototypical three-equilibrium configuration that is broadly thought to capture the essence of the viability issue through expectation dynamics in telecommunications industries. In terms of policy implications, by identifying precise and tight conditions for the various possible effects to hold, our results provide solid theoretical foundations for some well-known policy prescriptions that need revisiting for network markets (Shapiro [30]). The main departure from ordinary markets is the emergence of a start up problem, with potentially serious consequences for both firms and society. A successful launch of a new product with a small stand-alone value depends on various factors, including the usual ones, such as intrinsic quality, production costs, reputational aspects, and government participation. In addition, as the case studies reported in the business strategy literature confirm, interconnection amongst competitors or agreement on compatibility is quite often a critical determinant of success (Shapiro and Varian [31] and Rohlfs [29]). The results on viability provide a solid theoretical grounding to lend insight to the lessons on the start up problem gleaned from these case studies, in ways that apply to both successes and failures. One implication for product development is that, as a way out of the start up trap, firms ought to bundle multiple functions in network products, in order to ensure a sufficient stand-alone value. Our results on market performance also largely confirm Rohlfs [29] clear-cut conclusion that interconnection is most often a win-win proposition for both firms and society.25 When the effects of more competition can lead to multiple reversals of conventional intuition, the usual trade-offs between consumer surplus and producer surplus are no longer the norm, and many pillars of conventional wisdom about suitable public policy for such industries need re-examining. The presence of network effects might have unusual implications on firmsʼ attitudes towards intellectual property rights and entry deterrence. Firms in possession of patents will have a much higher than usual incentive to engage in licensing to their rivals on rather generous terms (Shepard [32], Shapiro [30]). Pooling of patents held by different firms is also to be expected. In terms of antitrust implications, various forms of pro-competitive cooperation amongst direct rivals should be allowed or even encouraged. This is particularly true concerning the often difficult and costly process of establishing a common standard needed for a new network industry to succeed. In terms of public policy, government participation in network industry start ups can be crucial due to the major role it can play in terms of influencing market expectations (the S0S0 variable) upwards. In addition, the interconnection process can raise such thorny and complex issues for the private actors that a positive coordinating role for government agencies often arises. Even initial subsidies might play a very constructive role. In one of his most instructive case studies, Rohlfs [29] reports that the unprecedented success of the ultimate network industry – the Internet – is largely due to the direct role played by the U.S. government via its DoD and NSF temporary subsidy programs, in terms of ensuring global interconnection.26 Rohlfsʼ detailed account of this glorious episode of government intervention suggests that, without it, interconnection for the Internet – something usually taken for granted – could easily have failed or been substantially delayed. The present analysis paves the way for further promising research in a number of interesting directions, including (i) the role of marketing in the start up problem, (ii) the scope for divisionalization in network industries in terms of both start up and profit incentives of firms, and (iii) the comparison with the case where firms possess commitment power in setting output levels.