Industrial Organization
Mobile Communications

The economics of telecommunication networks are of considerable interest, especially with the prospect of more intensive deregulation. A joint CEPR workshop with the Centre de Recherche en Economie et Statistique (CREST) in Paris on 9/10 July addressed an area of particular interest, 'The Economics of Mobile Communications'. The workshop was organized by Leonard Waverman (University of Toronto and CREST) and Patrick Rey (CREST, CNRS and CEPR). The workshop formed part of CEPR's research programme on 'Market Structure, Industrial Organization and Competition Policy', supported by the Commission of the European Communities under its Human Capital and Mobility programme. Additional financial support was provided by France Télécom and Société Française du Radiotéléphone.

Patrick Rey opened the workshop with his paper 'Capacity Constraints, Mergers and Collusion', written with Olivier Compte and Frédéric Jenny. This paper analyses the possibility of tacit collusion in a model of Bertrand-Edgeworth price competition with capacity constraints, according to the distribution of capacities. It first analyses a particular class of collusive equilibria where firms have symmetric pricing policies, and then discusses the implications for the analysis of mergers and of asymmetries in the distribution of capacities. The model shows that tacit collusion is easiest to sustain when the aggregate capacity of small firms is close to the market size, and becomes more difficult when capacity exceeds the market size. The asymmetry not only affects the sustainability of tacit collusion, but also the market shares that firms can achieve. A greater dispersion of capacities tends to make collusion harder to sustain. Nancy Gallini (University of Toronto and CREST) pointed out that the analysis sheds new light on the use of the Herfindahl index for evaluating mergers: here, a more symmetric configuration, implying a lower index, facilitates collusion. Robert Crandall (Brookings Institution) discussed the determinants of capacity constraints in the mobile telecoms market, distinguishing those constraints linked to bandwidths, often exogenous, from those linked to physical network installations, partially endogenous in the long-run.

Gianni De Fraja (University of York) presented 'Pricing and Entry in Regulated Industries: The Role of Regulatory Design'. This paper focuses on the optimal regulatory power when, as long as the incumbent continues to be regulated, the chosen regulatory policy exerts a crucial influence on the entrant's expected profitability, and hence the prospective appeal of entry. Entry is beneficial because it widens the market, but it also harms the incumbent firm, directing demand towards the goods supplied by the entrant. Consequently, the price imposed on the regulated firm should be made dependent on whether entry has occurred. Hence, entry in a regulated industry can have counter-intuitive consequences. Michael Riordan (Boston University) discussed the model's features: the assumption of complete information, the cost structure, the classic trade-off for differentiated oligopoly models (economies of scale versus product variety), and the regulator's commitment to subsidies but not to price. He then analysed the difference between the levels of regulated prices (above or below the Ramsey price) and of unregulated prices, first in the case of no commitment, then in the commitment case. Finally, he suggested some ideas for telecom regulation, including different ways to commit on price caps and entry.

Mark Armstrong (University of Southampton) presented 'The Access Pricing Problem', written with John Vickers, analysing the problem that arises when a vertically integrated dominant firm controls the supply of a key input to its competitors. Drawing on the Baumol-Willig efficient component pricing rule (ECPR), the paper examines the 'opportunity costs' of providing access. The original ECPR rule is based on a model with a homogeneous product: where inputs are combined in fixed proportions, bypass is not possible, and the entrant is a price taker. This paper generalizes the model to allow for variable proportions and bypass, multi-dimensional input and output vectors, and endogenous and uncertain entry and exit decisions. It shows that with an adequately adapted rule, the integrated firm still has an incentive to deter entry (or induce exit) of rivals, but these incentives do not diverge from social welfare.

Bruno Jullien (CREST and CEPREMAP) presented 'Regulation, Competition and Entry', written with Kai-Uwe Kühn, on the interaction between price regulation and entry, when transfers cannot be used. The paper shows that permitting ex post entry may be a powerful instrument of regulation. In some circumstances, the mere threat of using the information provided through entry is sufficient to achieve full efficiency in the market. In general, however, the regulator will permit some degree of inefficient entry in order to increase allocative efficiency and reduce informational rents. The paper also allows the regulator to choose ex ante the size of the contestable market by dividing a global market into two sub-markets and analysing the optimal size. Furthermore, it introduces access charges to be paid by the entrant to the incumbent: by setting access charges that deviate from marginal costs, the regulator can improve on both price inefficiencies and incentive costs. Lastly, the paper analyses the optimal access charge, showing that allowing for access charges extends the range of inefficient entry.

'Option Contracts and Vertical Foreclosure' by Albert Ma (Boston University) focuses on whether vertical integration can be profitably used to foreclose a rival from a market. Its answer is affirmative, but while foreclosure can result from vertical integration, consumers can benefit. This is in sharp contrast to most previous work and to the presumption of most antitrust analysis. In particular, the model applies to the telecom market, where downstream firms may be cellular telephone companies in other locations. The model also has a strong bearing on the future of telephone and cable industries, and on the deregulation of the local telephone service market. Ralph Winter (University of Toronto) noted that the foreclosure effects do not depend on the assumption that the goods in the market are sold through 'option contracts'. This is a natural simplifying assumption, but the argument applies more generally to a market where downstream firms sell bundles of services supplied by upstream firms. This strengthens the applicability of the analysis to markets such as mobile communications.

In 'Auction Design', John McMillan (University of California, San Diego) discussed the recent auction organized by the US Federal Communications Commission (FCC): under the hammer were the rights to use the electromagnetic spectrum for new forms of mobile telecoms or personal communications services. The paper explains that the main reason for using the pioneering simultaneous ascending auction was license interdependencies. With a sequential auction, in which items are simply offered one after the other, identical items can actually sell for quite different prices. Moreover, the sequential form might tempt a firm to bid in predatory fashion, driving early prices to unreasonably high levels. More importantly, the sequential form hinders license aggregation and has other pitfalls. Crandall questioned the possibilities of 'real' applications of the auction design, the availability of good instruments to evaluate market performance, and the analysis of multi-product/multi-unit externalities. Jullien stressed that the simultaneous ascending auction does not account for bounded rationality and information asymmetries, aspects that become very important in the auction administrative process.

Ronald Harstad (Rutgers University) presented 'Computationally Manageable Combinatorial Auctions' on the FCC auction design and the computational problem of evaluating combinatorial bids. The paper analyses various structures of permitted combinatorial bids for which the computational problems are manageable even for the simultaneous sale of many assets (that is, nested, cardinality-based and geometric structures). It argues that allowing restricted combinatorial bidding may provide many of the potential advantages of unrestricted combinatorial bids without imposing an impractical computational burden. It also indicates that the manageability of the problem of determining a minimal improving (that is, revenue increasing) combinatorial bid is essentially equivalent to the bid-taker's problem. Thomas Palfrey (Caltech) discussed feasibility and time constraint issues, stressing that there is some lack of economic 'sense' in the nested, cardinally-based and geometric structures. A fixed number of proposed combinations and a maximum number of two dimensions were suggested to improve the computing problem. Riordan noted that the game theory approach prevents the method from being too deterministic.

In 'Access and Interconnection Pricing: How Efficient is the ECPR?', Nicholas Economides (New York University) examined the consequences of relaxing some of the assumptions on which Baumol's ECPR rests. The paper pays special attention to the case where the monopolist charges a price for the complementary component that is above all relevant marginal costs. Exclusion of the rival also prevents the possibility of a lower price for the complementary component, resulting in consumers' deadweight loss; hence, the net social gain from the rival's presence could be positive. The ECPR's exclusion of inefficient rivals may be socially harmful: the market presence of even an inefficient rival could bring net social benefits, by causing the price to fall sufficiently so that the net gains to consumers exceed the inefficiency costs of the rival's production. The extent of price decreases, the size of the rival's cost advantage, and the rival's equilibrium market share are the key determinants of whether its presence in the market would be beneficial. The paper shows that, even in a Cournot duopoly, non-trivial cost inefficiencies by the rival are consistent with a net social gain. The applicability of this model to the telecom market was the focus of subsequent debate. The interconnection fees in the New Zealand telecom market were discussed by McMillan. Waverman discussed the contestability problem, the limit price, and the role of regulation in the setting up of an efficient system. Regarding the access deficit in the telecom market, Armstrong discussed which one, of the entrant's price or of the incumbent's price, is the most useful. It was noted that an ex post price could better cover the access deficit.

Pedro Barros and Antonio Leite (Universidade Nova de Lisboa) presented 'The Good Monopoly: A Case for Joint Ownership of Competing Systems?'. This paper discusses the argument that separate ownership (duopoly) generates a bigger surplus than joint ownership (monopoly), in the case of goods which are substitutes in consumption (but not perfect substitutes) and produced by different firms. This argument has been the basis for policy decisions forbidding the operation of competing telecom networks, notably trunking and GSM cellular, by jointly owned operators. The paper finds circumstances in which such an argument is not true, considering two competing systems and consumers that value more highly the good available in the system in which production costs are higher. In contrast with previous literature, in this model the joint owner of competing systems induces consumers to join one system or the other so that it maximizes profits. Marc Ivaldi (Université des Sciences Sociales de Toulouse) noted that the key assumption of the model is the cost difference (not the price gap) between the GSM and the trunking system, and questioned whether the marginal cost captures the quality effect in the telecom market. The inclusion of consumption externalities in the taste parameter of the consumer was recommended. Some problems in the demand distribution were raised, since in the model, total demand (that is, the number of subscribers) is fixed. Lastly, Jullien discussed how competition between systems and the ownership structure can affect the number of subscribers.