HANDBOOK OF TELECOMMUNICATIONS ECONOMICSVOLUME 1This Page Intentionally Left BlankHANDBOOK OFTELECOMMUNICATIONSECONOMICSVOLUME 1STRUCTURE, REGULATION AND COMPETITIONEdited byMARTIN E. CAVEUniversity of WarwickSUMIT K. MAJUMDARImperial College of Science, Technology and MedicineUniversity of LondonandINGO VOGELSANGBoston University2002ELSEVIERAMSTERDAM . BOSTON . LONDON . NEW YORK . OXFORD . PARISSAN DIEGO . SAN FRANCISCO . SINGAPORE . SYDNEY . TOKYOElsevierThe Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, UKRadarweg 29, PO Box 211, 1000 AE Amsterdam, The NetherlandsFirst edition 2002Reprinted 2005, 2006 Copyright 2002 Elsevier Ltd. All rights reservedNo part of this publication may be reproduced, stored in a retrieval systemor transmitted in any form or by any means electronic, mechanical, photocopying,recording or otherwise without the prior written permission of the publisherPermissions may be sought directly from Elseviers Science & Technology RightsDepartment in Oxford, UK: phone (44) (0) 1865 843830; fax (44) (0) 1865 853333;email: permissions@elsevier.com. Alternatively you can submit your request online byvisiting the Elsevier web site at http://elsevier.com/locate/permissions, and selectingObtaining permission to use Elsevier materialNoticeNo responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be madeBritish Library Cataloguing in Publication DataA catalogue record for this book is available from the British LibraryLibrary of Congress Cataloging-in-Publication DataA catalog record for this book is available from the Library of CongressISBN13: 978-0-444-50389-3ISBN10: 0-444-50389-7ISSN- 1569-4054Printed and bound in The Netherlands06 07 08 09 10 10 9 8 7 6 5 4 3 For information on all Elsevier publicationsvisit our website at books.elsevier.comLIST OF CONTRIBUTORSDr Mark Armstrong Economics GroupNufeld CollegeOxford UniversityOxford, UKProfessor Gerald W. Brock Graduate Telecommunication ProgramThe George Washington UniversityWashington, USAProfessor Martin E. Cave Centre for Management under RegulationUniversity of WarwickCoventry, UKProfessor Peter Cramton Department of EconomicsUniversity of MarylandCollege Park, USAProfessor Melvyn A. Fuss Department of EconomicsUniversity of TorontoToronto, CanadaProfessor Jerry Hausman Department of EconomicsMassachusetts Institute of TechnologyCambridge, USAProfessor David L. Kaserman College of BusinessAuburn UniversityAuburn, USAProfessor Stanley J. Liebowitz School of ManagementUniversity of Texas at DallasRichardson, USAProfessor Sumit K. Majumdar The Management SchoolImperial College of Science,Technology and MedicineUniversity of LondonLondon, UKProfessor John W. Mayo McDonough School of BusinessGeorgetown UniversityWashington, USAProfessor Stephen E. Margolis College of ManagementNorth Carolina State UniversityRaleigh, USAProfessor Eli M. Noam School of BusinessColumbia UniversityNew York, USAProfessor Michael H. Riordan Department of EconomicsGraduate School of BusinessNew York, USAProfessor Daniel F. Spulber Department of Management and StrategyJ.L. Kellogg Graduate School ofManagement, Northwestern UniversityEvanston, USADr. William W. Sharkey Federal Communications CommissionWashington DC, USAProfessor David E. M. Sappington Lanzilloti-McKethanEminent ScholarDepartment of EconomicsUniversity of FloridaGainesville, USAProfessor Lester D. Taylor Department of EconomicsUniversity of Arizona, USAProfessor Ingo Vogelsang Department of EconomicsBoston UniversityBoston, USAProfessor Glen A. Woroch Department of EconomicsUniversity of CaliforniaBerkeley, USAProfessor Leonard Waverman London Business SchoolLondon, UKVI List of ContributorsCONTENTS OF VOLUME 1 OF THE HANDBOOKList of Contributors vChapter 1Structure, Regulation and Competition in the Telecommunications IndustryMARTIN E. CAVE, SUMIT K. MAJUMDAR and INGO VOGELSANG1. Introduction 32. Economic Characteristics 52.1. Historical Overview 52.2. Network Effects 92.3. Customer Demand Analysis 112.4. Econometric Cost Functions 132.5. Representation of Technology and Production 163. Regulation 193.1. Price Regulation 213.2. Theory of Access Pricing and Interconnection 223.3. Interconnection Practices 243.4. Universal Service 253.5. Interaction Among Regulatory Institutions 274. Competition 294.1. Competition Policy 294.2. Long Distance Competition 324.3. Mobile Telephone 344.4. Economics of Spectrum Auctions 354.5. Local Network Competition 365. Conclusion 38References 39SECTION I STRUCTUREChapter 2Historical OverviewGERALD W. BROCK1. Introduction 442. Development of Regulated Monopoly Telecommunication 462.1. The Early Telegraph and Telephone Industry 462.2. Regulation and Monopoly 503. The Shrinking Boundary of Regulated Monopoly 543.1. Competition in Customer Premises Equipment 543.2. Competition in Long Distance Service 583.3. The Divestiture 624. Efforts to Develop Interconnected Competition 654.1. The Competitive Unregulated Internet 654.2. The Telecommunications Act of 1996 and LocalTelephone Competition 705. Conclusion 73References 74Chapter 3Network EffectsSTANLEY J. LIEBOWITZ and STEPHEN E. MARGOLIS1. Network Externalities 762. Classications 773. Impacts of Network Effects 793.1. Choosing Network Size 803.2. Choosing Among Competing Networks 834. Features of Modern Network Effects Models 865. The Empirical Importance of Network Effects 885.1. Network Effects in Principle 885.2. Measuring Network Effects 896. Policy Implications 927. Conclusions 93References 94Chapter 4Customer Demand AnalysisLESTER D. TAYLOR1. Introduction 982. Telecommunications Demand Analysis in the 1970s and 1980s 983. The General Nature of Telecommunications Demand 1014. Theoretical Considerations 1054.1. A Generic Model of Access Demand 1054.2. Models of Toll Demand 1064.3. Point-to-Point Toll Demand 1085. Empirical Studies 1095.1. Studies of Residential Access Demand and Local Usage;Residential Access Demand: Taylor and Kridel, 1990 1095.2. Bypass via EAS: Kridel, 1988 113VIII Contents of Volume 15.3. Choice of Class of Local Service: Train, McFadden andBen Akiva, 1987 1165.4. Studies of Toll Demand, Point-to-Point Toll Demand: Larson,Lehman, and Weisman, 1990 1175.5. Toll Demand Models Estimated from a Sample ofResidential Telephone Bills: Rappoport and Taylor, 1997 1195.6. Competitive Own- and Cross-Price Elasticities in theIntraLATA Toll Market: Taylor, 1996 1225.7. Two-Stage Budgeting and the Total Bill Effect inToll Demand: Zona and Jacob, 1990 1246. An Overview of What We Know About TelecommunicationsDemand 1267. Challenges for the Future 136References 138Chapter 5Econometric Cost FunctionsMELVYN A. FUSS and LEONARD WAVERMAN1. Introduction 1441.1. Monopoly or Competition 1441.2. Regulation and Price-Caps: The Need for Assessing Productivity 1451.3. Efciency Measurement and the Use of Alternative Approaches 1481.4. General Comments 1502. The Concepts of Economies of Scale, Scope and Subadditivity 1512.1. Economies of Scale and Scope 1513. Technological Change 1583.1. Estimating Marginal Costs 1613.2. Revenue Weights versus Cost Weights in Price-Caps Formulas 1624. A Selective Review of the Evidence 1645. Conclusion 170Appendix 171References 175Chapter 6Representation of Technology and ProductionWILLIAM W. SHARKEY1. Introduction 1802. Network Design Principles 1812.1. Telecommunications Network Elements 1812.2. The Trade-Off Between Switching and Transmission 1862.3. Telecommunications Trafc Theory 1882.4. Optimization in the Local Access Network: Minimum Distanceand Minimum Cost Networks 190Contents of Volume 1 IX2.5. Network Optimization in the Interofce and Intercity Networks 1933. Network Facilities and Technologies 1953.1. Wireline Subscriber Access 1953.2. Alternative Subscriber Access Technologies 1973.3. Switching 1993.4. Interofce Transport Facilities 2003.5. Interconnection Arrangements and Costs 2023.6. Broadband Standards and Technologies 2034. Cost Proxy Models of the Telecommunications Network 2044.1. A Brief Survey of Cost Proxy Models in Telecommunications 2054.2. The Hybrid Cost Proxy Model 2104.3. International Applications of HCPM 2154.4. Cost Proxy Models as a Tool for Applied Empirical Analysis 2175. Conclusion 220References 221SECTION II REGULATIONChapter 7Price RegulationDAVID E. M. SAPPINGTON1. Introduction 2272. Forms of Incentive Regulation 2282.1. Banded Rate of Return Regulation 2282.2. Earnings Sharing Regulation 2282.3. Revenue Sharing Regulation 2302.4. Rate Case Moratoria 2302.5. Price Cap Regulation 2312.6. Partial Deregulation 2322.7. Yardstick Regulation 2332.8. Options 2343. Trends in Incentive Regulation 2364. General Issues in the Design and Implementationof Incentive Regulation 2404.1. Reasons for Incentive Regulation 2404.2. Incentive Regulation in Practice 2434.3. Possible Drawbacks to Incentive Regulation 2455. Designing Price Cap Regulation 2485.1. Setting the X Factor 2485.2. Determining the Length of Time Between Reviews 2515.3. Mid-Stream Corrections: Z Factors 2535.4. Implementing the Price Cap Constraint 253X Contents of Volume 15.5. The Price Cap Constraint in Practice 2605.6. Additional Restrictions on Individual Price Levels 2635.7. The Number of Baskets 2655.8. Service Quality 2665.9. Conclusion 2666. Hybrid Regulatory Plans 2676.1. Earnings Sharing Plans 2676.2. Revenue Sharing Plans 2716.3. Regulatory Options 2726.4. Conclusions 2757. The Impact of Incentive Regulation 2757.1. Prices 2767.2. Operating Costs 2787.3. Network Modernisation 2797.4. Total Factor Productivity 2807.5. Earnings 2827.6. Service Quality 2837.7. Universal Service 2847.8. Summary 2858. Conclusions 285References 287Chapter 8The Theory of Access Pricing and InterconnectionMARK ARMSTRONG1. Introduction 2972. One Way Access Pricing 2982.1. The Effect of Unbalanced Retail Tariffs 2992.2. The Problem of Foreclosure in an Unregulated Market 3052.3. Fixed Retail Prices with No Bypass: The ECPR 3112.4. Fixed Retail Prices and Bypass 3162.5. Ramsey Pricing 3212.6. Unregulated Retail Prices 3252.7. Introducing Dynamic Issues 3312.8. Controversies and Conclusions 3333. Competitive Bottlenecks 3373.1. Mobile Call Termination 3373.2. Access Charges for the Internet 3454. Two Way Access Pricing and Network Interconnection 3494.1. Fixed Subscriber Bases: International Call Termination 3504.2. Interconnection with Competition for Subscribers 356Contents of Volume 1 XI5. Conclusion: Instruments and Objectives 379References 381Chapter 9Interconnection PracticesELI M. NOAM1. Interconnection as the key policy tool of telecommunications 3871.1. Why regulate interconnection 3891.2. Regulation of interconnection and unbundlingin a competitive market 3892. Interconnection as a tool for the creation of monopoly:the U.S. experience 3903. Interconnection as a tool for competitive entry 3913.1. Reforming access charges 3924. Interconnection as a tool for protecting competition 3934.1. Local competition 3934.2. Unbundling 3954.3. Quality 3974.4. Cable television interconnection 3974.5. Mobile interconnection 3994.6. Internet interconnection 3995. Pricing and pricing wars 4015.1. Regulated pricing of interconnection 4015.2. Arbitrage 4095.3. Incremental cost 4096. Interconnection around the world 4137. Interconnection and common carriage 4158. The future of regulation of interconnection 417References 419Chapter 10Universal Residential Telephone ServiceMICHAEL H. RIORDAN1. Introduction 4242. Telephone Penetration in the United States 4273. Normative Economics of Universal Service 4333.1. Price distortions 4333.2. Scale economies 4393.3. Network externalities 4443.4. Third degree price discrimination 4503.5. Second degree price discrimination 4544. Positive Economics of Universal Service 4564.1. Cross-subsidies in the price structure? 456XII Contents of Volume 14.2. Low income subsidies 4594.3. High cost subsidies 4645. Conclusions 4656. Appendix: Variable Denitions and Summary Statistics for Table 1 4676.1. Census data 4676.2. Climate data 4686.3. Cost data 4696.4. Summary statistics 470References 470SECTION III COMPETITIONChapter 11Competition Policy in TelecommunicationsDANIEL F. SPULBER1. Introduction 4782. Monopolisation 4792.1. Monopolisation 4802.2. Predatory Pricing and Raising Rivals' Costs 4832.3. Natural Monopoly 4862.4. Market Power 4893. Leveraging 4913.1. Essential Facilities 4923.2. Barriers to Entry 4943.3. Tying 4973.4. Cross Subsidisation 5004. Mergers 5025. Conclusion 505References 506Chapter 12Competition in the Long Distance MarketDAVID L. KASERMAN and JOHN W. MAYO1. Introduction 5102. Structure Conduct-Performance 5122.1. Structure 5122.2. Conduct 5222.3. Performance 5262.4. Summary 5283. Empirical Analysis of Relaxed Regulation 5284. The Access Charge Pass Through Debate 5334.1. Theoretical Issues 534Contents of Volume 1 XIII4.2. Empirical Analyses 5384.3. Summary 5435. NEIO Studies 5445.1. Residual Demand Studies 5445.2. Conjectural Variation Studies 5475.3. An Alternative Test for Tacit Collusion 5496. Competition for International Calling 5537. Conclusion 558References 559Chapter 13Mobile TelephoneJERRY HAUSMAN1. Introduction 5642. Description of the Mobile Industry 5672.1. Technology 5672.2. Competing Firms and Countries 5712.3. Government Frequency Allocation 5723. International Comparisons 5753.1. Performance Across Countries 5753.2. Pricing Within the U.S. 5793.3. Pricing in Europe 5824. Increased Consumer Welfare from Mobile Telephone 5834.1. Economic Theory to Measure Increased Consumer Welfare 5834.2. Estimation of the Amount of Increased Consumer Welfare 5854.3. Estimation of a Corrected Telecommunications Service CPI 5865. Government Regulation of Mobile Telephone 5885.1. Estimation of the Cost of Regulatory Delay in the U.S. 5895.2. Regulation of Mobile Prices 5915.3. Regulation of Mobile Access and Mandated Roaming 5945.4. Regulation of Wireline Call Termination on Mobile 5956. Taxation of Mobile Services 5966.1. Estimation of Economic Efciency Losses 5976.2. Comparison with Alternative Taxes 6017. Conclusion 602References 603Chapter 14Spectrum AuctionsPETER CRAMTON1. Introduction 6062. Why Auction the Spectrum? 6073. Auction Design 608XIV Contents of Volume 13.1. Open Bidding is Better than a Single Sealed Bid 6093.2. Simultaneous Open Bidding is Better than Sequential Auctions 6103.3. Package Bids Are Too Complex 6113.4. Other Issues 6124. Simultaneous Ascending Auction 6135. Demand Reduction and Collusive Bidding 6176. Lessons Learned and Auction Enhancements 6217. Package Bidding 6238. UMTS Auctions in Europe and Asia 6269. Advice to Governments 6309.1. Allocating the Spectrum is Just as Important as its Assignment 6319.2. Use Care When Modifying Successful Rules 6319.3. Allow Discretion in Setting Auction Parameters 6319.4. Reduce the Effectiveness of Bidders' Revenue-Reducing Strategies 6329.5. Use Spectrum Caps to Limit Anticompetitive Concentration 6339.6. Implement Special Treatment for Designated Entities with Care 6339.7. Implementing an Effective Auction: Time and Difcult Tradeoffs 6359.8. Facilitate Efcient Clearing When Auctioning Encumbered Spectrum 6369.9. Promote Market-Based Tests in Spectrum Management 63610. Conclusion 637References 637Chapter 15Local Network CompetitionGLENN A. WOROCH1. Introduction 6421.1. Scope and objectives of this chapter 6421.2. Patterns and themes 6432. Local Network Competition in Historical Perspective 6442.1. Local competition in one city, a century apart 6442.2. U.S. experience with local competition and monopoly 6472.3. The experience abroad 6563. Economic Conditions of Local Network Competition 6593.1. Dening local services and markets 6593.2. Demand for local network services 6633.3. Cost of local service and technical change 6684. The Structure and Regulation of the Local Network Industry 6764.1. Structure of the U.S. local exchange industry 6764.2. Regulation of local network competition 6805. Strategic Modelling of Local Network Competition 6845.1. Causes and consequences of local network competition 6845.2. Strategic choices of local network entrants 6895.3. Strategic models of local network competition 692Contents of Volume 1 XV5.4. Entry barriers 6936. Empirical Evidence on Local Network Competition 6977. Wireless Local Competition 6987.1. Wireless communications technologies 6997.2. Wireless services as wireline competitors 7027.3. Structure of the wireless industry 7057.4. An assessment of the wireless threat 7068. The Future of Local Competition 708References 711Subject Index 717XVI Contents of Volume 1Chapter 1STRUCTURE, REGULATION AND COMPETITIONIN THE TELECOMMUNICATIONS INDUSTRY*MARTIN E. CAVEUniversity of WarwickSUMIT K. MAJUMDARImperial College of Science, Technology and Medicine,University of LondonINGO VOGELSANGBoston UniversityContents1. Introduction 32. Economic Characteristics 52.1. Historical Overview 52.2. Network Effects 92.3. Customer Demand Analysis 112.4. Econometric Cost Functions 132.5. Representation of Technology and Production 163. Regulation 193.1. Price Regulation 213.2. Theory of Access Pricing and Interconnection 223.3. Interconnection Practices 243.4. Universal Service 253.5. Interaction Among Regulatory Institutions 27* This Handbook would not have seen the light of day but for the untiring efforts of our adminis-trative colleagues Ms. Dannie Carr at Imperial College and Ms. Marion Cherrie at Brunel University.Their professional approach and good cheer has been an extraordinary input towards the completionof the project.Handbook of Telecommunications Economics, Volume 1, Edited by M.E. Cave et al.# 2002 Elsevier Science B.V. All rights reserved4. Competition 294.1. Competition Policy 294.2. Long Distance Competition 324.3. Mobile Telephone 344.4. Economics of Spectrum Auctions 354.5. Local Network Competition 365. Conclusion 38References 392 M. Cave, S. Majumdar and I. Vogelsang1. IntroductionThe last two decades have seen exceptionally fast rates of change in every aspectof the telecommunications industry. These include technology changes espe-cially the effects of digitalization; introduction of new products including internet-based services, and convergence associated with the coming together of thebroadcasting, information technology and telecommunications industries. Si-multaneously, market structure has changed through the replacement of theformer monopolistic, vertically integrated telephone companies by a variety ofcompeting rms. These developments have been accompanied by major legislativeand regulatory developments, including the passage in the United States of the1996 Telecommunications Act and the introduction of a large number of new lawsand regulations in Europe and elsewhere. The same changes have seen a massiveexpansion of independent regulatory agencies. Their major role has been todetermine the terms of competitive interactions between the existing incumbentoperators and new entrants.This volume provides an economic review of and commentary on these changesin the landscape of what is arguably a very major industry. It attempts to providea comprehensive account of the major applications of economic analysis to oureld of interest, which is essentially two-way communications between parties thatare not in physical contact with each other by means of xed line telephony,mobile telephony, satellite telephony, cable TV telephony and Internet telephony.To a signicant extent, voice communication is the process of interest. We pro-pose in a second volume to focus our attention on the economic analysis of datacommunications, including in particular the development and use of the Internet,and on a number of other strategic, institutional and organizational issues of theincredibly fast-moving contemporary telecommunications landscape.The three principal sections of this volume contain fourteen chapters written byexperts in the relevant areas. Each chapter is intended to be self-contained, andcan be read independently. Together, however, the chapters provide a survey ofthe bulk of recent work in the economics of the telecommunications sector. Thereader pursuing a particular subject, for example the application of engineeringcost models, or the relation between wholesale and retail price regulation, is likelyto want to consult several of them.Authorship of the chapters by different individuals carries the advantagethat readers can gain access to their special expertise on the subject in ques-tion. It also exposes differences among authors that may exist on key currentpolicy questions. For example, some authors, who have greater condence inthe potentially benevolent effects of regulation than others, would like to seethe process of deregulation taking place more quickly. This diversity of opi-nion is an advantage, because it reects genuine disagreement among econo-mists and policy makers on matters which are still of major topical concern.Kuhn (1962) has described the early stages of the evolution of a eld asCh. 1: Structure, Regulation and Competition 3consisting of phenomena reporting in the absence of a unifying theoreticalframework. Thus, a plurality of perspectives may be observed being brought tobear on issues. Even in the presence of theories that may be seemingly logical,the empirical phenomena noted might be at odds with the assumptions thatunderlie the framework. The existence of multiple views enables us, as editors,in this introductory chapter, to highlight some of these differences withoutnecessarily attempting to reconcile them. It also enables us to articulate the keyconceptual issues that we think ought to form the core of the discipline oftelecommunications economics that is evolving.Although we have sought to be comprehensive, particularly in explainingconcepts that underlie policy making and in reviewing the literature, the singlemost important bias of the current volume is its focus on highly developedcountries, the U.S. in particular. The main reasons are that the most importantwork has been on American telecommunications problems and that most of ourauthors are U.S. based. Critical policy experiments were carried out in the U.S.,and the extant literature covers U.S. experience in the main. Nevertheless, generalconsequences from U.S. policies are lessons drawn by the authors and in thisintroduction.Two features of the U.S. telecommunications sector are of particularimportance the ownership structure and the degree of vertical integration. Mostcountries in the world have a dominant telecommunications carrier that is or wasstate owned. In contrast, the U.S. Bell system was always in private hands1. Inspite of this difference, the resulting policy issues and outcomes are strikinglysimilar across countries. The U.S. telecommunication sector has been less verti-cally integrated than elsewhere, since AT&T was split up in 1984. This verticalseparation has provided an interesting policy experiment and has justied a se-parate chapter on long-distance competition.Our second bias is the omission of macro-economic consequences of tele-communications. The design of regulatory institutions was planned as a chapterbut could not be completed on time. However, this area is partially covered inseveral chapters, notably in the chapters dealing with the historical overview, ininterconnection practice, mobile telephony and local network competition. Wealso take it up in this introduction.The contents of the Handbook are set out in three main sections. The rstsection deals with the economic characteristics of the sector. These characteristicsdene the structure of the industry, and we discuss how different aspects ofstructure can interact to inuence industry development. The next section dealswith regulation and the third section deals with competition. In theory, the sectionon competition should precede that on regulation. In practice, regulatory issues1This particular volume of the Handbook does not cover the topic of privatization. Nevertheless, giventhe salience of the topic, some comments are in order and these comments are included at variouspoints in this introductory chapter.4 M. Cave, S. Majumdar and I. Vogelsanghave dominated the policy debates for quite some time. While issues related tocompetition are critical, they are of more recent origin. Of course, the regulatoryissues and competition issues continuously interact with each other, and each setof issues inuences the other. The industry evolves through the interaction of thetwo processes. We invite the reader to keep this in mind while reading the chapterswithin the three sections.2. Economic characteristicsThe objective of the rst section of the Handbook is to set out some history as wellas the crucial economic and structural characteristics of the telecommunicationssector, so that policymaking and rms' decisions are guided by the appropriateprinciples. In designing the ordering of the chapters in this section, the basicorganizing principle followed is that structural issues of demand and supply aresequentially looked at in detail. These demand and supply issues then underpinthe theoretical discussions related to all of the other topics dealt with in thevolume.Following the ``Historical Overview'' of the telecommunications industry in theUnited States by Gerald Brock (Brock), two chapters deal with the issues ofdemand. In their chapter, Stanley Liebowitz and Stephen Margolis (Liebowitzand Margolis) deal with ``Network Effects'' that gain particular importance in anindustry setting where interconnectivity is a unique characteristic. These authorstake a relatively macro-level look at the industry-level consumption character-istics and provide the backdrop for the chapter by Lester Taylor (Taylor), wholooks at micro-level ``Customer Demand Analysis'' issues in the sector.Following the two demand-side chapters, Melvyn Fuss and Leonard Waver-man (Fuss and Waverman) analyze ``Econometric Cost Functions,'' and WilliamSharkey (Sharkey) provides a ``Representation of Technology and Production.''Both these chapters deal with the supply-side of the telecommunications industryequation. The chapter by Fuss and Waverman is pitched at a relatively aggregatelevel of analysis, while the one by Sharkey is written intrinsically at a more micro-level of analysis.2.1. Historical overviewGerald Brock's ``Historical Overview'' is a broad review of the evolution of thetelecommunications industry in the United States. It is an entirely country-centricchapter. Yet, there are crucial lessons to be learnt from the United States'experience by other countries around the world for several reasons. First, ofcourse, at a basic technological level the telephone was invented in the UnitedStates. Nevertheless, the industry evolution patterns and public policy evolutionpatterns have much salience for other countries. In the United States, after theCh. 1: Structure, Regulation and Competition 5expiration of the basic patents held by the Bell system towards the end of thenineteenth century a number of new telephone companies entered the industry.These companies expanded so rapidly that prices fell sharply. As a result of entrya considerable enhancement of infrastructure took place so that not only werelarge cities such as Boston and New York supplied with telephone lines, but deeprural portions of the country had substantial penetration as well. And, this degreeof penetration had taken place at the turn of the last century! It is thus well worthtaking into account the lessons from history2.In the light of this decline in prices and increase in investment in infrastructurethat was observed almost a century ago, when monopoly power dissolved, thedeep and fundamental question remains, what should be public policy towardsreducing the monopoly of contemporary operators so that consumer welfare isenhanced? The evidence from the early institutional history of the United Statesthat Brock presents shows that when the institutional context changed away frommonopoly to a competitive market then entry of new rms was rapid and pricesfell, at least initially. Whether such benets would have existed or not if entry wasnot permitted is a counter-factual issue. However, from Brock's chapter it is clearthat the salient contemporary issues of access and interconnection raised theirheads at the turn of the century.Theodore Vail, as president of American Telephone and Telegraph Company(AT&T), led a process of industry consolidation and the creation of a tele-communications conglomerate. He bought up smaller operators after realisingthat interconnection was technologically necessary so that customers belonging toa multitude of phone companies could talk to each other and not just to sub-scribers of one particular network. Implicit in his plan was the need to exploiteconomies of scale and scope3. However, as Chandler (1977), in his seminal workon the theory of the rm, has noted, such economies came in a large measure fromthe administrative efciencies in handling through trafc that interconnectivitywould permit. Such administrative efciencies also generated system-wideresources to be invested in implementing an expansion policy for the Bell systemas a whole. The planning for such system-wide resource deployment and expan-sion could be done from the head ofce of the system. Thus, the sources ofeconomies were organisational and not just technological4.2The historical events described in this chapter are also treated in the chapters written by Eli Noamand Glenn Woroch for this Handbook.3However, the elucidation and empirical examination of scale and scope economies only began totake place much later in the 1960s.4In fact, it was suggested that the operating costs per subscriber increased with the number of sub-scribers and that there were diminishing returns to scale in the construction and operation of telephoneexchanges. Thus, cost efciencies, if any, would come through administrative efciencies that would beenjoyed by one large rm operating in the sector relative to two small rms operating in the sector. Theadministrative efciencies of one large rm would outweigh the combined operating efciencies en-joyed by two small rms. Thus, the issue of subadditivity and the recognition that economies of scopewere organizational in nature were important conceptual constructs a century ago.6 M. Cave, S. Majumdar and I. VogelsangAccess by companies to each other's networks might have solved the techno-logical interconnection problem. Whether, however, Theodore Vail was justied,given his search for administrative economies and the desire to establish a centralstrategic planning apparatus, in creating a monopoly is moot. Between the ex-piration of the Bell system patents in 1893 and 1894 and 1907 independent tele-phone companies operated 51 percent of the telephone lines in the U.S. Between1907 and 1912 the Bell system started absorbing independent companies into itsfold and the market share of independent companies dropped. The presence ofcompetition between telephone companies for franchises and between franchisesfor subscribers clearly stimulated the extension of telephone service, the reductionof telephone rates, and technological improvements in telephony. The process ofconsolidation that was started by AT&T at the turn of the century halted thesepositive outcomes.Today it is recognised that a competitive market structure is the viable option ifmarkets expand and economies of scale and scope are exhausted at small operatorsize. Other writers on the history of the U.S. telecommunications industry alsoconclude that clearly competition was better than unregulated monopoly. Butwhether the continuation of competition in the U.S. would have been better thanthe presence of a regulated monopoly in continuing to yield such superior out-comes will remain unknown5. In Brock's chapter, the United States experiencemoves from one of an early competitive market structure to that of a regulatedmonopoly, which then persisted for almost three-quarters of a century till thebreak up of the Bell system in 1984. Starting in the 1950s, a process of gradualintroduction of competition in the customer premises equipment, long distanceand local segments has taken place. Such moves are being copied all over theworld.Another crucial aspect of the United States industry is the role of the privatesector. AT&T was a regulated privately owned rm. As Brock observes, AT&Tchose to accept regulation and use it as a substitute for market forces. Yet, theregulation of the private company was instrumental to the enormous infra-structure development that took place in the United States. What were the aspectsof the regulatory regimes in place that led to the development of the tele-communications infrastructure to the degree that it did in the United States? Thisquestion becomes important as nations around the world are struggling withquestions of appropriate institutional design for their telecommunications sector.Allied to these changes is the design of regulatory instruments and organisa-tions. From the United States case it is apparent that public ownership, per se, isnot a prerequisite for effective infrastructure development. This is a theme, oncenovel, now increasingly taken as axiomatic around the world. What has perhaps5Nevertheless, a disquiet that a regulated monopoly was impeding technical progress and creatingX-inefciencies was implicit in fuelling the process of litigation and institutional changes that eventuallyled to the re-structuring of AT&T in 1984 (Shepherd, 1983).Ch. 1: Structure, Regulation and Competition 7led to the development of the U.S. telecommunications infrastructure is the de-centralisation of regulatory activities, to the extent possible, and effective andstable implementation of regulatory policies and rules under rate-of-returnregulation. As far as the former aspect is concerned, there was, and still is, aregulatory body in each of the states as well as the Federal CommunicationsCommission at the federal level. With respect to implementation of regulations,the co-ordinated application of the separations rules is one of the factors that hasled to the widespread diffusion of the basic telephone service, which most recentlywas at 95 percent. Thus, the design and implementation of rules do matter veryclearly. In the next section of the Handbook, several chapters deal with con-temporary regulatory design issues.Nevertheless, as Brock also observes, in the face of technical change, en-trenched monopolies can use the regulatory process to play entry deterrencegames. AT&T played these games so that the Federal Communications Com-mission (FCC) prohibited other rms from developing specialised video trans-mission facilities. These games may be currently occurring as telecommunicationsmarkets are being opened to technology-driven competition. Brock's chapterimplicitly brings out, from a historical perspective, some of the organisational andprocessual dimensions involved in the design and implementation of regulatoryinstitutions.A second issue emerging from Brock's chapter is the role that should beaccorded to the FCC's Computer Inquiry II decision (Computer II). Thisdecision has had far-reaching impact, and is an interesting institutionaldecision to study. First, the Computer II decision explicitly impacted on theboundaries of the telecommunications rm. By separating customer premisesequipment (CPE) provision from telecommunications services, the Computer IIdecision clearly stipulated the boundaries of telephony service provision6. Bymaking CPE provision unregulated, the Computer II decision led to the verylarge growth of the CPE segment of the industry. Thus, freeing up a segmentto competition has had enormous impact on new rm entry and the evolutionof an industry segment.Second, the Computer II decision has played a pivotal role in providing thebasis for regulating access to the Internet in the United States. The Computer IIdecision created a right for customers to provide services within their ownpremises and to connect those services to the public switched telephone network(PSTN). The principles of Computer II are currently applied to regulation of theInternet through recognition that intelligence is generated and retained by theInternet service providers (ISPs) at the edges of the network. This happens withCPE within a telephone network. ISPs gain access to the PSTN without having6Of course, empirical evaluations of boundary separations are far from ever being perfect. The needfor empirical analysis to ascertain where boundaries of different sectors might lie, howsoever fuzzy themeasurement, remains crucial.8 M. Cave, S. Majumdar and I. Vogelsangto pay the often-high access charges that other telephone companies have to paythe incumbents for access to their PSTNs. This implication of Computer II,permitting ISPs almost free access to the PSTN, has resulted in the mostextraordinary diffusion of the Internet in the United States. This had con-sequences for global consumers as well, since a very large proportion of theInternet content is generated in the United States. The availability of the largeUnited States customer base has generated a large number of content-devel-oping enterprises whose output is circulated globally. The Computer II decisionhas had global consequences.2.2. Network effectsThe chapter by Stan Liebowitz and Steve Margolis (Liebowitz and Margolis) on``Network Effects'' deals with the most fundamental economic characteristic of acommunications system. The importance of connectivity and the value placed byconsumers on that feature is well recognised. In their chapter, Liebowitz andMargolis deal broadly with issues of classication of network effects, features ofnetwork effects, and the somewhat limited empirical literature7.Liebowitz and Margolis highlight the importance of the role of network sizeand network choice. From their articulation of these issues one can draw im-portant implications to current regulatory controversies. First, the ability tocapture and internalize externalities via ownership or market position is an in-ducement for the development of large networks. The network effects generatedin such networks provide prot opportunities for the owner. In this context thereare close intellectual parallels between the network externalities and the naturalmonopoly arguments: each sustains the position of the regulated incumbentmonopolist. Certainly, both consumption and production can exhibit increasingreturns to network size or scope. Such conditions would militate against someforms of unbundling if the efciencies enjoyed by the large network operatorswere to be lost as a result. On the other hand, with mixed bundling, specicelements of the local loop are leased out. Such lease arrangements exploit thespare capacities available on local loop networks. In such a case, any loss ofnetwork efciencies may well be small and may be offset by the leasing revenuesreceived. In view of the ramications of public policy toward networks, precisemeasurement and quantication of scale and scope economies should be a highpriority.Second, several small networks might interconnect. This has happened with theInternet. In that case, the consumption scale economies may be insufcientlysmall for any network operator to make meaningful gains because of limited7Starting from early work (Rohlfs, 1974), authors in the telecommunications eld have shown a greatdeal of interest in consumption externalities on the demand side. Of more recent origin is the notion ofproduction externalities, which are the supply side effects arising from connectivity between networks.Ch. 1: Structure, Regulation and Competition 9network size8. Concomitantly, congestion effects may arise and create negativeexternalities. Therefore, a proliferation of networks may create a situation akin tothe `tragedy of the commons.' The presence of additional participants will causetrafc jams and decrease the value of network membership to the other partici-pants who are members of the network9.With respect to the creation of congestion effects, there is no doubt that thegrowth in data trafc may potentially overwhelm existing xed line networks.This concern, however, has to be considered in the light of the facts presented byWilliam Sharkey in his chapter ``Representation of Technology and Production.''Sharkey suggests that advances in optical networking technologies will allowoptical signals to bypass the constraints at switching nodes, where optical signalshave to be converted to electrical signals. Optical devices may lessen the impact ofthis constraint, which is responsible for the congestion effect. Hence, in future, if anumber of small new entrants join the network and seek interconnection, theproliferation of rms, and networks, may not necessarily create a `tragedy of thecommons.'A further issue examined by Liebowitz and Margolis is whether path de-pendencies play any important role in fostering inefciencies, as is alleged in thenow famous case of the QWERTY and Dvorak keyboards. Liebowitz andMargolis, however, have shown that there is not historical support for the com-mon claim that the keyboard is an example of path dependent inefciency. Fortelecommunications networks, if diseconomies on the supply side outweigh thescale economies on the demand side, then the unbundling of incumbent localmonopoly telephone networks is likely to be economically justied. If, in the faceof potential unbundling, local telephone monopolies are able to protect theirentrenched positions, then customers on the incumbent monopolist's networkmight get locked into an inferior service, even though over time newer networksmay provide better quality services and enhance consumer efciencies. Efforts ofeither the incumbent monopolists or the regulatory authorities may frustrateefcient turnover. Here again, however, the competition from alternativenetworks wireless networks, for example limit the inefciency that can beimposed by the `locking in' of a self-isolating network. Thus, there are clear ex-trapolations from the existing literature on network effects to the contemporarytelecommunications situation.Brock's chapter usefully highlights the role that the Computer II has played, inthe form of dampening prices for ISPs to connect to the PSTN, and in fosteringInternet diffusion in the United States. Thus, the Internet has evolved as aseparate network that operates in conjunction with the PSTN. If, however, this8Even if interconnection is achieved at cost-based prices, the scale of entry matters for a networkoperator to benet from the consumption externalities that are generated by entry of that rm into thenetwork.9Therefore, the issue is should several small networks be allowed to interconnect or a few largenetworks that can benet from scale effects be allowed to do so.10 M. Cave, S. Majumdar and I. Vogelsanghigh level of Internet diffusion is accompanied by congestion on the PSTN as aconsequence of the Internet trafc ows to the nal customer, then the benets ofComputer II are outweighed by the costs of congestion. There are, however, in-direct network externalities that can be generated. The existence of both thePSTN and the diffusion of the Internet create a derived demand for new services.The customer connected to both the Internet and the PSTN can consume theseservices. The need to consider such indirect network effects then is also important.2.3. Customer demand analysisThe chapter by Lester Taylor on ``Customer Demand Analysis'' provides achange in focus, away from the aggregate industry-level perspective of the rsttwo chapters to the level of the customer in the telecommunications sector. At theconsumer level, Taylor takes the presence of a network externality as axiomatic.Factors generating the demand for calling are exogenous. There are, however,endogenous effects that lead to the creation of dynamic network externalities,since once the exchange of information has taken place it creates the need forfurther exchange of information. A network externality can be construed as ademand-side economy of scale. Taylor brings out the need to think of the dynamicincreasing returns process that propels these scale economies.The distribution between static and dynamic network externalities is vital, butsubject to considerable measurement problems, which have led to the emergenceof only a small empirical literature on the subject, as compared to the empiricalliterature on the measurement of supply-side scale economies which is the topic ofthe next two chapters. As the types of available networks increase, with theaddition of wireless and Internet connections to the customers` choice sets, thedynamic externalities increase. Therefore, the measurement problems affectingthe concept of network externalities increase as well. Taylor's detailed analysis ofthe large literature on consumers' demand for access and usage, in conjunctionwith a consideration of some of the developments that have taken place inestimation methodologies, can help sort out the inherent difculties in measure-ment.Taylor's review of the demand modelling literature highlights the role thatadvances in discrete choice modelling have played in shedding empirical light onkey questions. Allied with those are also the advances in simultaneous equationestimation techniques. Hence, as data on demand for access and usage of differenttypes of network services become available, simultaneous estimates of xed-line,wireless and Internet access and usage can capture the interdependencies betweenthe demand characteristics for different networks. This type of modelling remainsan agenda item that ought to shed light on consumption dynamics, as networksproliferate and converge. There are clear strategic implications of such modellingexercises for rms, as they begin to understand what motivates demand for dif-ferent types of networks.Ch. 1: Structure, Regulation and Competition 11Taylor also highlights the related question of option demand. The benet oftelecommunications connectivity arose not only from the completed commu-nications but also from the ability to do so. This issue of option demand gets morerelevant as diffusion of the Internet and the mobile networks continue apace.Connectivity to the xed-line network not only entitles the customer to make andreceive voice calls, but also to the option of access to the Internet for the contentcontained therein. Simultaneous connectivity to mobile networks allows con-sumers enhanced option sets for accessing voice functionalities. Once wirelessaccess to the Internet becomes widespread, the options available to the consumerwill increase with the availability of alternative voice and content choices. AsInternet based e-commerce becomes mobile networks based m-commerce, theoptions to both business and residential customers increase substantially. Then,basic telecommunications connectivity is expanded to include not just commu-nicability, but also commercial capability. The choice sets for customers are in-creased still further.How does the option demand interact with the dynamics of information ex-change to alter consumer behaviour patterns? This important question requiresresearch, at the theoretical and the empirical level, though the latter is more dif-cult since the data requirements are large. If consumer behaviour patterns dochange substantially, then there are major implications for rms as well as policymakers. Take the issue of pricing. The way that prices are regulated for access tothe Internet, and the way that prices are regulated for calls from xed networks toterminate on mobile networks and vice-versa, vary across countries. These var-iations in prices are key drivers of consumers `consumption behaviour given optiondemand and the dynamics of information exchange. If prices are set sufcientlylow so as to encourage access, then customers' option sets increase substantiallysince there are more opportunities to make connections10. The availability of anincreased number of potential connections can impact on the increasing returnsprocess that underlies the dynamics of information exchange and increase thevolume of digital signals owing through networks. This enhanced volume canpotentially cause congestion on the xed-line network. If the supply-side econo-mies of scale are robust, then the increased trafc volumes may be accommodatedwithin one network11. Melvyn Fuss and Leonard Waverman, and William Sharkeydeal with the consideration of supply side issues in their chapters.10On this issue, we have remarked that the almost non-existent access charges for ISPs to connect tothe local telephone networks in the U.S. have led to a substantial diffusion of the Internet in the U.S.Consumers in the U.S., thus, have considerably more choices than consumers elsewhere.11An issue that the Handbook does not cover is what motivates rms in the telecommunications sectorto make investments in the new technologies. A rich tradition in the literature, going back toSchmookler (1966), highlights the role of demand factors in inuencing technology adoption. In theevolving telecommunications sector, how factors such as option demand and the dynamics of in-formation exchange inuence customer demand and thereby the potential deployment of new tech-nologies by carriers remains an exciting topic awaiting research and investigation.12 M. Cave, S. Majumdar and I. Vogelsang2.4. Econometric cost functionsThe issue of whether the telecommunications industry is a natural monopoly ornot has a long history. The fundamental institutional outcome, if a naturalmonopoly situation exists, has been the perceived need for regulation. In thatcontext, the need for understanding what is an optimal market structure hasgenerated a substantial literature, based primarily on Canadian and United Statesdata, on scale, scope and efciency issues. Melvyn Fuss and Leonard Waverman(Fuss and Waverman), judicially survey this literature, and the accompanyingconcepts, in their chapter ``Econometric Cost Functions.''The rst issue of concern is the distinction between three concepts: mono-polisation, actual monopoly and natural monopoly. This distinction is highlightedin the chapter on competition policy by Daniel Spulber but relevant for this onetoo. The rst concept is a process of behaviour by rms; the second concept is anoutcome of that behavioural process in terms of market conditions; the thirdconcept deals with the structural and technological conditions determining thecost-minimising number of rms in a particular market or industry. The lastconcept is what the current chapter deals with.While in the decades from the 1960s to the early 1990s issues of naturalmonopoly and cross-subsidies were relevant, Fuss and Waverman identifyaccess pricing and access conditions as the concerns of today. This observationimplicitly goes to the heart of what is one of the crucial questions in thetelecommunications eld today: how many competitors should be allowed toenter the industry and on what terms and conditions? To understand the issueof how many requires detailed analysis of production conditions, which scale,scope and efciency studies are meant to achieve. Thus, it is useful not tothink of natural monopoly questions per se, but to ask: what is the structureof production like?Having reviewed the available literature, and the techniques used in the studies,Fuss and Waverman reach the sobering conclusion that the existing literaturedoes not really tell the policy maker what the permissible structure ought to be fora given market situation. There is a clear reason for this conclusion. Historically,the telecommunications sector has consisted of a monopoly operator that hasoperated both long-distance as well as local networks. The question of naturalmonopoly, at a system level, has arisen because it is assumed that an operator hasboth economies of scale in providing calls in general and that there are economiesof scope, or multi-product economies, for that operator to produce two or moretypes of calls. Additionally, for producing any one type of call the costs of thatproducer are less than the total costs of two or more producers generating thesame output. That is, at the system level costs are sub-additive. The empiricalliterature has never been able to test whether a stand-alone toll producer's costfunction is different from that of a producer of both toll calls and local calls.Undertaking this test remains a challenge for the profession.Ch. 1: Structure, Regulation and Competition 13The Fuss and Waverman analysis raises the issue: what is the appropriatedenition of the telecommunications industry for the purpose of such economicstudies? Does it comprise all of the producers, collectively connected via theirindividual networks, in one large meta-network? Is it the long-distanceoperators who operate the trunk networks that may connect various parts of acountry? Or is it the local exchange networks that operate in specic jur-isdictions? Depending on what denition of industry is chosen, the econometricstudies are carried out at that particular level. However, results obtained forone level of analysis, say the long-distance sector or segment, need not beconsistent with the results obtained at another level of analysis, say that of thelocal exchange sector.The appropriate denition of the industry is critical for several reasons. Forthe regulator, it is important to know what segment to regulate and whatsegment to keep free from regulation. For the prospective operator,industry denition is important because the efcient boundaries of a rm can bedened12. From an academic perspective, the denition of an industry is criticalbecause it helps dene the level of aggregation at which studies are carried out.The extrapolations to policy made from these studies are a function of the levelof aggregation associated with each industry denition. If a study is carried outfor one sector and the results are extrapolated to another, this can lead towrong policy conclusions.Fuss and Waverman's rather negative conclusions of the existing literaturearise because academics have principally used aggregate-level data to answerquestions that are inherently micro-level in nature. Take the controversies basedon the U.S. Bell system data and captured in the works of Evans and Heckman(1984) and Charnes, Cooper and Sueyoshi (1988). Based on the use of a Bellsystem data-set, Evans and Heckman came to the conclusion that AT&T'snation-wide telephone system, consisting of the Long Lines long distance op-erations plus the local exchange operations, did not display subadditivity13.Therefore, a case could be made for breaking up the Bell system. Based on theuse of exactly the same data set, Charnes, Cooper and Sueyoshi came to theopposite conclusion14. A problem in interpreting such diametrically opposingresults is the level of aggregation at which the studies have been carried out. It isa awed approach to club together data for long-distance and local operationsinto one data set and then to conclude whether the local or the long-distancesector is competitive or not. Such aggregated analysis can lead to potentiallywrong policy decisions. Therefore, it is very important to use the right type ofdata for analysis.12A related issue is the question of market denition, concerned with the set of competitors and theiroutputs.13Shin and Ying (1992) use local operator level data for operators only in the state of California toreach similar conclusions.14Ro ller (1990) comes to broadly the same conclusion as Charnes, Cooper and Sueyoshi.14 M. Cave, S. Majumdar and I. VogelsangAllied to the appropriate choice of industry segment to be analysed is the questionof technique. In the core, the primary approach has been to use regression-basedparametric efciency estimation techniques. Such techniques, especially the trans-log variant, are exible but can create problems because of underlying functionalformspecication oddities. Nevertheless, suchtechniques are extensively used in therecent literature. Another approach is to use operations research based techniquesthat do not, a priori, specify any functional forms. Not only is it feasible to estimateoverall scale economies, but the precise optimal size of each unit of observation canalso be estimated. Thus, data on the number of cost-minimising rms that can co-exist within a given industry can be ascertained. The use of this approach, involvingthe computation of rm or operator level scale economy parameters, augments theestimation of system-wide scale economy statistics.The above property is useful for policy makers and for rm-level strategicdecision-making. If the broad total size of an industry segment is known, givenhistorical data on consumption and connections, these techniques permit deri-vation of optimal market structure since the size of the most efcient rm oroperator can be estimated. Thus the number of rms that can co-exist, withoutany rms becoming inefcient, can be gauged. If the most efcient size is relativelysmall, there can be many operators within a given industry segment. For theprospective operator, such data reveal the level of investment necessary to be aviable player within a given industry segment. If the most efcient size of a rm isrelatively small, this calls for proportionately less investment than otherwisewould be the case. Given the potential policy and managerial prescriptions thatcan follow from using these new techniques, we hope that a signicant body ofwork, exploiting new sources of data and new techniques, becomes available toshed continuing light on the dynamic yet vexatious problem of optimal marketstructure.The question of what the supply-side economies are has to be considered inconjunction with the chapter on ``Network Effects'' by Liebowitz and Margolis.Network externalities are a demand-side scale economy characteristic. Such scaleeconomies are a function of network size, but network size is also endogenous.Supply-side scale economies will only be generated if there is a large volume ofcalls being made on the system. The generation of demand-side and supply-sidescale economies is a simultaneous process. From an academic research perspec-tive, there are enormous benets to combining the empirical literature of networkeffects with the empirical literature on scale and scope economies so as to un-derstand the dynamics of the underlying increasing returns process.Nevertheless, policy decisions have to be made in stages and sequentially. Thesupply-side scale economies can be such that they initially peter out at very smalloutput levels. Then, given the generally burgeoning size of the total market,several entrants may be permitted to enter the market, as the market can ac-commodate many players. In fact, it is this assumption that led to the opening upthe local exchange segment to competition in the U.S. 1996 TelecommunicationsCh. 1: Structure, Regulation and Competition 15Act. Interconnection should permit a multitude of small entrants to eventuallyenjoy the system-wide demand-side scale economies that are feasible within thetelecommunications sector. If some small entrants thereafter become more ef-cient because of rm-level managerial competencies, then the process of naturalselection will mean that a consolidation of the sector might occur and large lea-ders emerge. The non-parametric approaches permit an ex-ante assessment ofpotential scale, or size, at entry. How, then, the dynamics of the increasing returnsprocess play out in generating further economies is an empirical question.2.5. Representation of technology and productionWilliam Sharkey (Sharkey) in his chapter titled ``Representation of Technologyand Production'' takes a deep micro-micro look at the underlying technologyof telecommunications production. It is the last chapter of this section forseveral key conceptual and technical reasons. First, the micro-level issues thatSharkey deals with are now in the forefront of policy-making. Understandingwhat the supply-side economies are is useful in designing market structures.But the setting of prices requires an understanding of what it actually costs toproduce a unit of output. In the erstwhile multi-product system based litera-ture there have been considerable problems with the top-down allocation ofjoint and common costs, as well as estimating the cost functions. Therefore,the bottom-up cost literature has evolved so as to provide detailed estimates ofwhat it might cost to provide the various services on an incremental basis15.This generation of ex-ante incremental cost structure estimates is veryuseful from a regulatory point of view. These estimates provide the basis forcreating regulatory yardsticks against which subsequent performance may bemeasured.Second, the bottom-up cost literature has evolved substantially in the U.S.because of the belief that the local exchange network, once believed to be thelast bastion of natural monopoly, is prospectively competitive16. As Sharkeynotes, the local competition provisions of the 1996 Telecommunications Actprovide fundamental rights to prospective entrants. These rights are to inter-connect with the incumbent operator at rates based on cost, to obtain un-bundled network elements, and to obtain retail services at wholesale discounts.From an incumbent's point of view the knowledge of what costs actually areso that interconnect prices are set has been the raison d'etre for this approach.As countries around the world open up their monopoly networks to inter-15The bottom-up approach is based on engineering models and primarily uses optimization techni-ques. We have remarked earlier on the use of non-parametric operations research based techniques forefciency and optimal size measurement. These techniques provide a bridge to link the top-down andbottom-up approaches.16This assumption is remarkably consistent with the position adopted a century ago!16 M. Cave, S. Majumdar and I. Vogelsangconnection by new entrants, the bottom-up engineering cost methodology willdiffuse further.Third, the practice of generating ex-ante cost estimates permits a forward-looking view to be adopted as to what a communications network ought to be like.While the adoption of a forward-looking view is enshrined in the 1996 Tele-communications Act as an economic principle, the implications of a forward-looking approach are more far-reaching. In taking a forward-looking view theengineering viewpoint converges with the economic viewpoint to create efciencies.Because of the technological change in progress, various alternate networkstructures are possible. Sharkey goes into details of the various network topolo-gies. Each possible network topology can be created using alternate technologies.Thinking in terms of optimizing the trafc ows and using the appropriatetechnologies to then build the network prima-facie creates the least-cost trafcow solutions. Networks are constructed to take into account local trafc anddemographic idiosyncrasies, and this forward-looking approach helps operatorsplan for the best feasible solution.Fourth, given the continuous technological changes taking place within theindustry, the notion of a telecommunications rm is evolving. The evolving notionof what is a telecommunications rm leads to the re-drawing of market bound-aries, a fact that Sharkey clearly notes. If the rm and market boundaries change,then the optimal size of a rm is also subject to change since the technologicalbasis of production is in a state of ux. The change in the optimal size of a rmhas implications for the overall structure of the industry. If the optimal size of therm is small, then it is feasible to allow several rms to interconnect within a givengeography.Permitting a number of rms to enter and interconnect is an option to beconsidered in the light of Liebowitz and Margolis' comments on network ex-ternalities. They comment on optimal network size from a consumption point ofview. If a prospective operator nds that an inability to enter at a large scale ofoperations leads to a denial of the consumption externalities then entry might nothappen. Nevertheless, even smaller scale entry can be economically benecial ifoptimal scale can be attained in the process. The volume of calls that terminate onthe operator's network, however small, can generate access revenues that morethan compensate for the loss of direct call revenues to be earned from having alarge number of customers making calls from a larger size network. Thus, com-bining the ideas of Liebowitz and Margolis and Sharkey is important, from acompetition policy point of view, in designing market structures and in decidingappropriate scale at entry.There are further issues that emerge from Sharkey's chapter. As Sharkey notes,new technologies radically change the economics of transmission versus switch-ing, thus expanding or contracting the scope of operations for a new entrant or anexisting player. For example, a migration of scale economies to the switchingfunction has occurred because of the use of common control features.Ch. 1: Structure, Regulation and Competition 17The above facts can have implications for the dynamic evolution of rms'capabilities within the telecommunications sector. In particular, the U.S. 1996Telecommunications Act gives facilities-based entrants the right to interconnectwith the incumbent operator on rates based on cost, for a lesser entrant toobtain unbundled network elements, and for an entrant which is a reseller toobtain retail services at wholesale discounts. It is thus feasible for a rm toenter the market as a reseller of calls that are purchased wholesale. The nextstage can be to operate as a switchless service provider, using leased switchesto operate the business. In the nal stage, the entrant can acquire switches andprovide switching capacity that can be the precursor to complete facilities-based competition where switching and transmission facilities are both pro-vided17. Small-scale participation as a reseller provides the initial entry into thesector, and as efciencies are achieved then growth to an eventual fully fa-cilities-based player may be feasible. For policy makers, an understanding ofbottom-up costs, and how such costs change, provides an indication of whattypes of rms might enter and in what number. Thus, bottom-up cost mod-eling provides a foundation for the analysis of market structures in a tech-nologically dynamic context.Finally, Sharkey provides examples of how bottom-up cost modelling has beenapplied in Argentina, Peru and Portugal. While it has been developed in the West,the methodology is robust enough to travel across borders and be applied in othercontexts. It is not necessary to have advanced telecommunications networkdesigns analysed with the help of this model. Even the establishment of basicnetworks, using simple and standard technologies, can be analysed using theapproach. This way, the limited funds available to operators in developingcountries can be wisely spent in designing the lowest cost network. Therefore, theSharkey chapter is useful for policy makers from developing countries as well asdeveloped countries.17Sharkey, in a personal communication, further suggests that resale might not appear to be aviable long-run strategy for entry at least in U.S. local markets at the present time, though in theU.K. several service providers have entered the market for local and national long-distance calling.These service providers have taken away a large chunk of the incumbent monopoly rm's marketshare in the U.K. AT&T has complained about resale perhaps as an excuse for not entering U.S.local markets more rapidly. Nevertheless, AT&T's relative lack of success in entering local marketsmight be a function of perhaps aggressive entry deterrence strategies adopted by the incumbentlocal exchange carriers (ILECs) in the U.S. Resale nevertheless remains a valuable, and perhapsessential, transition strategy for new entrants. Switching assets are relatively scalable. Smallswitches are now available and adding line card modules can easily increase their capacity. Fur-thermore, digital loop carrier (DLC) technologies allow a single switch to serve a large geographicarea at less than 100 percent penetration rates. Hence, competition in the switching element can beexpected to occur in the reasonably near term future. The main problem with facilities based entryare the potentially large sunk costs in loop investment for wireline networks. Interofce transport isby far the most competitive element, since entry by competitive access providers (CAPs) in urbanmarkets is very well established.18 M. Cave, S. Majumdar and I. Vogelsang3. RegulationThe second section of the Handbook contains a set of four chapters dealing withregulatory issues. Telecommunication has been and remains a highly regulatedindustry. However, the form of regulation has changed signicantly over the past30 years. The industry has progressively opened up to competition in all parts ofthe world, as new technology such as mobile telephony has been introduced andas the incumbent operator in many countries outside North America has switchedfrom a public rm to mixed or wholly private ownership.It is useful to identify three stages in the recent interactive processes involvingregulation and the development of competition. For this purpose, we omit theprecursors to the rst stage of private or public monopoly. In the U.S., this periodended with the historic `Kingsbury Commitment' in 1913, under which a series ofde facto local monopoly franchises emerged under the leadership of AT&T. InEuropean countries, such as the United Kingdom, the precursor to a publiclyowned telecommunications monopoly was the development of a series of privatelyor municipally owned local companies. These were unied in the early part of thetwentieth century into a monolithic organization at the national level, typically adepartment of government.The regulatory arrangements during this rst stage, which characterized theindustry for the bulk of the twentieth century, can be dichotomized into a `NorthAmerican' model, based upon the independent regulation of private monopoly,and a `European' model based upon direct regulation by governments of a publicmonopoly.In the United States, the major instrument of regulation until the passage ofthe Telecommunications Act of 1996 was the 1934 Communications Act. Thisestablished the Federal Communications Commission (FCC). The FCC, incombination with state-level regulatory bodies and under the occasional super-vision of the Courts, regulated an industry dominated by the vertically integratedBell system. Until the 1950s, competition was virtually non-existent. The structureof retail prices, subject to break-even constraints, could thus be chosen with a highdegree of freedom, without running the risk of competitive distortion. Thisstructure took the form of relatively low monthly rental and local call charges,usually combined in a at rate, and relatively high long-distance charges. Basedon a combination of call types, the telephone companies were able to earn returnsat least equal to their cost of capital. Low line rental charges encouraged highpenetration rates, and hence the achievement of universal service objectives. Amore explicit re gime of cross subsidization, based on a detailed cost separationsmethodology, was developed. This was directed at mitigating cost differencesbetween companies providing services in high cost and low cost parts of theUnited States.The absence of competition, combined with public ownership of the industry,allowed governments operating the European model even greater discretion inCh. 1: Structure, Regulation and Competition 19running the industry. The same government department was typically responsibleboth for the production and the regulation of telecommunication services. Publicownership mitigated the break-even constraint that characterized the Americansystem. The European model typically embodied a structure of tariffs similar tothat observed in North America, in the sense that line rentals are low (relative tocost) and long-distance call charges high. Within the relevant governmentdepartment, the universal service objective of uniform and affordable tariffs couldbe achieved by direct government control of retail prices.Brock, Kaserman and Mayo, and Woroch describe the development of theregulatory framework, within which competition in the United States was pro-gressively introduced, in the chapters. It is useful, however briey, to characterizeby way of contrast the operation of the same processes in Europe and elsewhere.Member states of the European Union (EU) operate in a federal framework.Under this framework the European institutions (the Parliament, the Council ofMinisters and the Commission) enact legislation. This is then transposed intonational laws and implemented by national regulatory agencies. After a numberof member states opened up their markets to the competition in the 1980s, theEuropean Commission promoted a package of legislation in the 1990s to establisha framework within which competition could develop. Crucially, a politicalcommitment to liberalize telecommunications markets in the EU from 1998 wassecured. Detailed implementation would rest with the national regulatory agen-cies. The regulatory framework did not, however, supersede European competi-tion law. The Commission in its role of Competition Authority addedconsiderable impetus to the process.This set of regulatory arrangements can be broken down into provisionsrelating to licensing, interconnection and universal service. In addition, nationalregulatory authorities (NRAs) have typically imposed restrictions on the retailprices charged by the dominant operator, and also controlled the structure ofprices. Examples are the limits on increases in line rentals and restrictions on thediscounts which the incumbent might offer to the major customers. Thesecustomers were frequently the targets of entrants' marketing plans. However, theCommission has encouraged re-balancing of price structures away from crosssubsidization.This framework of regulation has been designed to meet the needs of a period inwhich the transition to competition was beginning. It has developed in a piece-meal fashion, and the instruments employed were often blunt. In some countries,notably the United Kingdom, the promotion of entry in the form of facilities-based competition was an explicit objective of policy.Regulation in Europe is now about to enter a new phase corresponding to aperiod in which the market power of the incumbent is diminished. Regulation can,therefore, be less intrusive. The new re gime also acknowledges the widening ofmarkets associated with convergence. It is described as a framework forregulating not telecommunications but electronic communications. While main-20 M. Cave, S. Majumdar and I. Vogelsangtaining regulatory provisions aimed at the achievement of social objectives, suchas universal service, the new framework is intended to bring regulation into closerconvergence with competition law, focussing in particular upon rms in a positionof dominance.The regulatory framework elsewhere in the world has generally exhibited asimilar progression, from regulation of a monopolist via a phase of early tran-sition to competition, followed, where appropriate, by a relaxation of constraintson the historic operator. In Australia, this process has been accompanied by areallocation of responsibility for the regulation of telecommunications to thecompetition authority (the Australian Competition and Consumer Commission).This body uses newly enacted provisions in competition law to control access tobottleneck facilities in network industries.New Zealand represents the sole exception to the otherwise standard procedureof regulating the transition to competition to a regulatory agency operating underlegislation that is additional to standard competition law. When the New Zealandgovernment introduced full competition into telecommunications in 1989, iteschewed specic regulation and chose to rely instead upon its general competi-tion law, the Commerce Act 1986. Extensive litigation ensued under the Act withrespect to the interconnection arrangements between the incumbent monopolist,Telecom New Zealand, and a competitor, Clear. This nally culminated in ajudgement delivered by the Privy Council in London in 199518. This judgementdid not, however, still the controversy. In 2001 the Government in New Zealandwas giving consideration to the possibility of introducing telecommunications-specic regulation.The previous paragraphs have provided a schematic view of the process bywhich regulation has developed to match the changing competitive circumstances.It is now appropriate to discuss in more detail the three key instruments reviewedin the regulatory chapters of the Handbook.3.1. Price regulationDavid Sappington (Sappington) in the chapter entitled ``Price Regulation'' dis-cusses price regulation and incentives, with applications to telecommunications.The prime exemplar of such schemes are price caps, which are now widelyemployed by regulators throughout the world to control the retail prices chargedby incumbent operators. A more recent, and interesting, development has beenthe extension of price caps to control the prices at which the incumbent sellsnetwork services both externally to competitors and internally through transferprices to its afliates.Price caps are a member of a broader class of instruments of incentive regu-lation mechanisms which provide operators