Competition, Regulation and the New Economy

COMPETITION, REGULATION AND

THE NEW ECONOMY

In addition to being the principal medium for communication, educationand entertainment, the new economy is now a leading provider of goodsand services through electronic channels. The new economy rides on thecrest of technological developments in computers, telecommunicationsand satellites creating new interactive mediums and on the deregulation and privatisation of state-owned enterprises in the telecommunicationsand broadcasting sectors. Whilst the economic viability of the dotcoms isquestioned, the existence of a new economy with novel methods of pro-duction, distribution and exchange is here to stay. Evidence of this is thefact that there are 300 million active computers in the world, with 350million people who use the world wide web (expected to grow to one bil-lion in four years), and the speed of microprocessors continuouslyincreases, facilitating the use of IT.

The question which is pursued in the series of essays in this book iswhether the conceptual underpinnings of competition law and interna-tional regulatory mechanisms are adequate or appropriate to deal with thedevelopments raised by the new economy.

Competition, Regulation and the New Economy

Edited byCosmo Graham and Fiona Smith

The University of Leicester

OXFORD AND PORTLAND OREGON2004

Published in North America (US and Canada) byHart Publishing

c/o International Specialized Book Services5804 NE Hassalo Street

Portland, Oregon97213-3644

USA

© The editors and contributors severally 2004

The Editors and Contributors have asserted their right under the Copyright, Designs and Patents Act 1988,

to be identified as the authors of this work.

Hart Publishing is a specialist legal publisher based in Oxford, England.To order further copies of this book or to request a list of

other publications please write to:

Hart Publishing, Salters Boatyard, Folly Bridge, Abingdon Rd, Oxford, OX1 4LB

Telephone: +44 (0)1865 245533 Fax: +44 (0)1865 794882email: [email protected]

WEBSITE: http//:www.hartpub.co.uk

British Library Cataloguing in Publication DataData Available

ISBN 1-84113-384-1 (hardback)

Typeset by Olympus Infotech Pvt Ltd, India, in Sabon 10/12 ptPrinted and bound in Great Britain by

Lightning Source UK Ltd

Contents

1. Introduction 1COSMO GRAHAM

1. The New Economy and its Characteristics 22. Potential Problems for Competition Policy 4

A. Predation 6B. Tying and Bundling 7C. Intellectual Property Rights 8D. Collective Conduct 8E. Mergers 9

3. Conclusions on Competition Policy 94. Regulatory Questions 105. Subsequent Chapters 12

2. Article 82 EC and New Economy Markets 17GIORGIO MONTI

1. Competing Perspectives of New Economy Markets 18A. Neo-structuralism 18B. Neo-Schumpeterianism 22C. The Response of the Authorities 24

2. Relevant Market Analysis 24A. On-Line Music 25B. Top-Level Internet Service Providers 28C. Structuralism Prevails 31

3. Dominance 31A. The Traditional Test: Market Shares

Plus Entry Barriers 32B. Neo-Schumpeterian Definitions of

Dominance 334. Abuse 36

A. Protecting One’s Dominant Position 36B. Obligations to Co-operate

with Competitors 41C. Extending Dominance 45

5. An Innovation Defence? 476. Remedies 507. Conclusion 53

3. Abuse of a Dominant Position and Intellectual Property Rights: A Suggestion toReconcile the Community Courts’ Case Law 55ESTELLE DERCLAYE

1. Confusion in the Case Law 56A. The Factual Background of the Case 57B. Conditions Required to Establish the Abuse 59

2. An Illustration of the Confusion: The IMS Case 61A. The Factual Background of the Case 61B. The Decision of the Commission 62C. The Orders of the President of

the Court of First Instance 64D. Comments 64

3. How to Solve the Dilemma: A Suggestion 65A. Argument 66B. A Few Other Corroborating Arguments 73

4. Conclusion 74

4. B2B E-Marketplaces: A New Challenge to Existing Competition Law Rules? 77JOACHIM LÜCKING

1. B2B E-Marketplaces: Technical and Commercial Developments 77

2. Regulatory Responses 793. The Legal Assessment of B2B E-Marketplaces

under EC Competition Law 80A. Legal Basis 80B. Market Definition 81C. Competition Issues 83D. Co-ordination Effects 83E. Market Dominance and Foreclosure 86

4. Best Practice Guidelines 895. Conclusion 90

5. Authorities, Competition and Electronic Communication: Towards Institutional Competition in the Information Society 91PLG NIHOUL

1. National Systems Replaced by a European Organisation 92A. First Stage of the Reform 92B. The 1987 Green Paper 93C. What Legal Basis? 94D. An Institutional Provision 95

vi Contents

E. Modification of the Rapport de Forces 96F. Legislative Power 97

2. The Debate Between Harmonisation and Liberalisation 99A. Second Stage of the Reform 99B. Rules Adopted by the Commission 99C. New Interpretation of Competition Law 100D. Harmonisation Process 101E. Harmonisation Directives 101F. Two Categories of Rules 103G. Contradictions 104

3. Conflict at National Level 106A. Third Stage of the Reform 106B. National Regulatory Authorities (NRAs) 107C. National Competition Authorities (NCAs) 108D. National Courts 109

4. Convergence from an Institutional Point of View 109A. Fourth Stage of Development 109B. Illustration 110

5. An Institutional Interpretation of the Reform 111A. Conflict and Competition Among Authorities 111B. The Process of Regulatory Competition 112C. Institutional Competition in

Electronic Communications 114D. Competition Among Authorities

Within the Same Legal Order 115E. Expanding the Line of Business 116

6. What Can We Learn About the Future? 119A. More Competition among Authorities 119B. First Difficulty: Regulation

Not an Ordinary Good 120C. Solution: Competition Within a

Given Framework 120D. Second Difficulty: Legal Certainty and

Predictability 121E. Place of Unity in Society and the Legal System 122F. Conformity with an Information and

Communication Based Society? 123

6. Controlling the New Media: Hybrid Responses to New Forms of Power 126ANDREW MURRAY AND COLIN SCOTT

1. New Media and the Problems of Effective Control 128A. The Regulatory Arbitrage Problem 128B. The Anonymity Problem 130

Contents vii

C. The Scarce Resources Problem 1322. Extending the ‘Modalities of Regulation’ Analysis 1363. Putting Controls to Work 143

A. Hierarchy/Community 143B. Hierarchy/Competition 146C. Hierarchy, Competition and Design 150D. Other Forms of Control 152E. Excluding Hierarchy 156

4. Conclusion 156

7. Regulating E-Commerce in the WTO: Exploring the Classification Issue 159FIONA SMITH

1. Background 1622. Defining E-Commerce 1673. Applying GATT and GATS to

E-Commerce: Is E-Commerce Trade in Goods or Services? 172A. GATT 173B. GATS 175

4. Conclusion 183

8. Public Services in the New Economy 185ERIKA SZYSZCZAK

1. The Provision of Public Services 1852. The Political Economy of State Intervention 1873. Challenges to the Provision of Public Services 1894. The Challenge to State Autonomy 1915. Hybridisation 1946. The Provision of Public Services by Non-State Actors 199

A. Economic Regulation 202B. Market Power and Special Responsibility 203

7. Conclusion 205

viii Contents

1

Introduction

COSMO GRAHAM*

IN ADDITION TO being the principal medium for communication,education and entertainment, the new economy is now a leadingprovider of goods and services through electronic channels. It rides on

the crest of new technological developments in computers, telecommunica-tions, broadcasting and satellites, thereby creating new interactive mediums.Primarily, it is the result of entrepreneurial activity in the computer sector,but the deregulation and liberalisation of state owned enterprises in thetelecommunications and broadcasting sectors have played a contributoryrole. Whilst the initial dotcom bubble has now burst, the existence of a neweconomy with novel methods of production, distribution and exchange ishere to stay: there are now 300 million active computers in the world, with350 million people using the World Wide Web (expected to grow to onebillion in four years); the speed of microprocessors continuously increases,facilitating the use of Information Technology (IT).

The new economy is seen as central to future prosperity, but regulating itis complex because it moves at great speed and produces novel forms of co-operation and competition. As a result, there has been great debate aboutwhether the existing forms and principles of competition law, in particular,are effective to address regulatory problems. This is not simply an issue forcompetition law, but also a general problem for schemes designed to controlaspects of the new economy. Indeed, this was one of the impulses behind thepassing of the Communications Act 2003. The current volume looks at aspects of both issues:1 the first three chapters examine some difficult

* Cosmo Graham is Professor of Law at the University of Leicester. He is a public lawyer whospecialises in the law relating to the regulation of public utilities. His other main specialism isin competition law and he has teaching interests in company law. He is the Director of theCentre for Utility Consumer Law. He is also a member of the UK Competition Commission.1 This collection of essays originates in a seminar on this topic held in Leicester in July 2001sponsored by the Modern Law Review. Chs 2, 4, 5, 6 and 8 began life as papers presented to

competition law issues raised by the new economy. The next three chaptersconsider a variety of regulatory issues, focusing on institutional design andthe book ends with a discussion of the implications of the European Union(EU) law concept of public services on high technology industries.

This first chapter attempts to set the scene for the contributions that followby discussing briefly what we regard as the new economy and its salientcharacteristics. The discussion then outlines some general issues for compe-tition law raised by the new economy and concludes by putting these issuesinto the context of recent developments in United Kingdom and EU compe-tition policy. Finally, the contributions to the book are summarised and putinto that context.

1. THE NEW ECONOMY AND ITS CHARACTERISTICS2

What do we mean by the phrase the ‘new economy’? On one level, it can berestricted to the use of computer software and hardware and its applicationthrough digitalisation to the communications industry, especially the use ofthe Internet to business transactions. But this focuses narrowly on theindustries themselves and neglects wider regulatory problems. To be com-prehensive therefore, when determining the scope of the ‘new economy’, itmakes more sense to discuss the general characteristics that distinguishthese industries from the ‘old economy’, which then allows a discussion ofthe challenges for the development of competition and regulatory policy.

It will become apparent that the majority of characteristics affect a widerange of industries beyond computers and communications embracing, forexample, biotechnology and agriculture. Although the notion of the neweconomy sounds vague, there is substantial agreement amongst the com-mentators regarding its distinguishing features.3

One of the most obvious features of new economy industries is that theyare characterised by rapid technical or technological change, which leads tothe alteration of the markets under consideration either through the creationof new markets or the transformation of old ones. Computers are a goodexample of this: in the 1960s the industry focused on mainframe computing,

2 Cosmo Graham

that seminar. We would like to thank the contributors to that occasion for their thoughtswhich have helped to shape the contributions to this volume.

2 See also ch 2.3 General discussions can be found in D Teece and M Coleman, ‘The Meaning of Monopoly:Antitrust Analysis in High-Technology Industries’ (1998) Antitrust Bulletin 801; Office of FairTrading, Innovation and Competition Policy (2002) Economic Discussion Paper 3—Part I;and D Evans and R Schmalensee, Some Economic Aspects of Antitrust Analysis inDynamically Competitive Industries (2001) National Bureau of Economic Research WorkingPaper 8268.

whereas the growth of personal computing in the 1980s has moved theindustry from one selling primarily to large organisations to one which sellsdirectly to individual consumers, as well as of course still to large organisa-tions. The mobile phone market in the UK is another example. The firsttwo mobile licences were issued in 1985 and the next two were only issuedin 1991. From those small beginnings, the market has grown to be a sizeablephenomenon both in terms of the services provided and the manufacture ofmobile phones. The mobile phone industry is also a good example currentlyof technological change through the current development of third genera-tion mobile phones.

The second characteristic that is usually emphasised is the importance ofintellectual property to these developments. Technological changes arefocused around intensive research and development with the intention ofcreating new intellectual property rights, in other words, creating a meansof protecting the investment that the firm makes in the new development.There is a well-known potential for tension between intellectual propertyrights and competition policy and this is something that may become a par-ticularly difficult issue in the high technology industries.

Third, the implications of network effects are emphasised in the litera-ture. It is often the case in new economy industries that a network of somekind is involved. The most obvious example is telephones: here, the effectthat is most commonly referred to is that the network becomes more valu-able, as the number of members increases: this is the so-called ‘networkexternalities’. This is obvious in relation to networks such as telephones,fax machines and picture messaging, but there may be a similar effect inother sectors that do not immediately rely on a network. For example, themore people who use compatible computer software, the more valuable itbecomes because of the ability to swap files and interchange informationeasily. The popularity of a network may lead to what Shapiro and Variancall demand-side economies of scale,4 that is, people value a productbecause it is popular. When combined with the more traditional supply-sideeconomies of scale this creates what they refer to as ‘positive feedback’,which allows strong firms to get stronger and weak firms to get weaker,potentially leading to extreme results. To be more precise, one firm and itstechnology eventually dominate the market. An example here is ‘Word’word processing software. These network effects lead to competition formarkets, rather than competition within markets. The natural outcome is amarket dominated by one firm, something traditional competition lawlooks at with some suspicion.

Finally, these industries tend to have high fixed, or sunk costs and lowmarginal costs. The costs of producing the first CD or software programme

Introduction 3

4 C Shapiro and H Varian, Information Rules (Harvard, Harvard Business School Press, 1999)179.

are relatively high, while the costs of reproducing it are very low, even negligible. Furthermore, as Shapiro and Varian point out, there may not beany capacity constraints in reproducing certain goods.5 There are capacitylimits in a factory producing automobiles and concrete blocks, but the samedoes not apply to producing additional copies of information. This charac-teristic has particular implications for the pricing policies of new economyfirms. In particular, a central pricing strategy for such firms will be pricediscrimination. Price discrimination is not a new phenomenon, but newtechnology makes it much easier for firms to obtain information aboutprice preferences and buying patterns, as well as making it easier to createdifferent versions of the product that they are selling.

A further implication of an industry with high fixed or sunk costs andlow marginal costs is that it tends to be concentrated because of the highcosts of entry. The picture that comes through is of an industry subject todramatic changes, but dominated by a few firms, or one major player. It ispossible that this firm, with a large market share, will engage in price dis-crimination, create complementary products, encourage customer loyaltyand be highly profitable. On the face of it, this is the sort of behaviour ofwhich the competition authorities have always been highly suspicious. It isargued, however, that this is to misunderstand the nature of what is hap-pening in these industries.6 In these circumstances, these high market sharesare fragile and potentially temporary because the rapid pace of technologi-cal change means that markets can change radically quite quickly.Consequently, on this hypothesis, intervention by the competition authori-ties is not needed, as the market will provide the appropriate correction.The more appropriate question for competition authorities should be whichis the best characterisation of a market in a particular industry; unfortu-nately, there is no simple answer. This can be illustrated by examining theproblems posed for the application of competition policy methods in thecontext of high technology industries.

2. POTENTIAL PROBLEMS FOR COMPETITION POLICY

The starting point for all competition inquiries is the question of marketdefinition. Once that is agreed upon, the crucial question is whether or

4 Cosmo Graham

5 Ibid, 21.6 See eg C Ahlborn, D Evans and J Padilla, ‘Competition Policy in the New Economy: IsEuropean Competition Law up to the Challenge?’ [2001] 22 European Competition LawReview 156 and C Veljanovski, ‘EC Antitrust in the New Economy: Is the EuropeanCommission’s View of the Network Economy Right? [2001] 22 European Competition LawReview, 115.

not the firm or firms under investigation have market power or are dominant in the context of European Community (EC) competition law.There is a substantial amount of literature on how to define the relevantmarket and all the competition authorities have provided guidance onhow they undertake this task.7 Much of the criticism of the competitionauthorities’ approach has emphasised the need to avoid a ‘static’approach to market definition. In other words, to take into account notonly how the market is at the moment, but also how it will develop in thefuture. The implication is that with a dynamic high technology industry,competition will be for the market and that a wide market definition isgenerally required.8 In terms of the principles that the authorities applyin Europe and the United Kingdom, this criticism is overstated. The guid-ance recognises that market definition is a tool in the inquiry and that theissue is establishing whether market power exists. In doing this, theauthorities recognise that they must look at supply-side substitutabilityand, although strictly not an issue of market definition, potential futureconstraints on the companies concerned. Indeed, the issue generally forcompetition authorities in merger cases is what will be the impact of themerger on competition in the future; therefore, almost by definition, theycannot adopt a static view of the market, but must instead try and predict future events. This is not an easy exercise and opinions will differon whether, in any one case, the authorities have adopted the correctapproach.

A more interesting point has been raised in the OFT’s paper on competitionpolicy and high technology industries. The authors argue that market power isusually defined as a firm having power over prices and hence output. They goon to argue that it is important to include exclusionary power within thecontext of market power because in dynamically competitive industriescompetition issues generally arise from exclusionary power rather thanpricing power.9 This is in the context of a general argument that the compe-tition authorities should apply a ‘first principles’ approach to their analysisin particular cases. By this they mean that:

it would focus the analysis from the beginning upon the alleged anti-competitiveconduct. It would then use the definition and analysis of markets, marketpower (both pricing and exclusionary), and competitive constraints broadly

Introduction 5

7 European Commission, Commission Notice on the Definition of the Relevant Market for thePurposes of Community Competition Law; Competition Commission, Merger References:Competition Commission Guidelines and Market Investigation References: CompetitionCommission Guidelines both in section 2; Office of Fair Trading, Mergers: SubstantiveAssessment Guidance (May 2003, OFT paper 516).8 A good example is Ahlborn, Evans and Padilla above n 6, 161–62.9 Above n 3, pt I para 4.11. Much of this is derived from work in the US, as acknowledged inthe report.

defined to determine whether the alleged anti-competitive behaviour did haveanti-competitive effects and whether these effects caused significant consumerharm. … It would … allow for explicit analysis of whether a particular firmhad the ability, power, or incentive to carry out the alleged anti-competitiveact. … Further, one could analyse explicitly what the effect would be uponinnovation, short-term competition, long-term competition, price in the short-and long-run, and upon product diversity and quality.10

A. Predation

This has been one of the most controversial areas in competition policy fora long time and there are a number of new twists brought about by thedevelopment of high technology industries. Traditionally, it is perceived asa rule about pricing which prevents dominant firms from charging pricesthat are so low that they will drive out competitors. Substantial controversyexists over whether or not predation is an economically rational strategyand, furthermore, what the test for predation should be in the context ofEC law. In particular, there has been a debate about whether or not thealleged predator must be able to recoup their losses once the other firm orfirms have been driven from the market.

Given that high technology industries are characterised by high fixedcosts and very low marginal costs, a number of problems for the analysis ofpredatory pricing are created, as the Office of Fair Trading (OFT) pointout.11 It may be possible for a firm to price above its variable costs, but thiswill still undercut a smaller competitor who may be forced to exit the mar-ket. Alternatively, with very low marginal costs, there are circumstances inwhich it would be sensible to price below variable costs, for example tobuild market share. Under current EC competition law there would be apresumption that such behaviour was predatory even though it may just bea sensible business practice, particularly in the context of ‘winner take all’markets.12

Neither of these are actually new problems in relation to predatory pric-ing. The issue of whether or not pricing above average total costs can bepredatory has recently been discussed in Compagnie Maritime Belge13

where the European Court of Justice (ECJ) concluded that this was indeedpossible. Wider discussions have also taken place over the reasons whyfirms might price below marginal costs.

6 Cosmo Graham

10 Ibid para 4.69.11 Ibid para 5.8.12 Shapiro and Varian above n 4, ch 2.13 Cases C–395/96 P and Case C–396/96 P Compagnie Maritime Belge Transports vCommission [2000] ECR I–1365 [2000] 4 CMLR 1076.

In addition, the OFT paper argues that the concept of predation shouldbe extended beyond pricing to cases where:

A firm either incurs costs or undertakes other actions which may be cost freeor cost reducing, that it otherwise would not have taken had it not been forthe anti-competitive benefits to the firm undertaking these actions.14

This would be an extension of the current concept of predation, but theidea is to catch certain practices such as product pre-announcements that, itis alleged, have been used in a way to stifle competition.

B. Tying and Bundling

This is another area of controversy where there are likely to be importantrepercussions for high technology industries. As Shapiro and Varian pointout, any evidence of such behaviour is likely to be a red flag to antitrustauthorities.15 The OFT indicates that there are a number of different prac-tices which need to be distinguished and that there can be perfectly goodreasons for businesses undertaking such practices.16 The report distin-guishes between tying, which it defines as where there is a contractual linkbetween products, and bundling, where a group of products are soldtogether. It further distinguishes between pure bundling, where only thebundle of products are sold, and mixed bundling, where the products maybe sold as a bundle or separately, albeit more expensively. It goes on toargue that there are a number of plausible reasons, not related to anticom-petitive practices that would encourage businesses to engage in tying andbundling. These range from concerns over the quality of products to beingable to sell more products through this method.

The authors of the report do argue that there are certain situations wheresuch practices may be aimed at market foreclosure. They claim that in amarket for tied goods, that is Good A and Good B are sold together, theability to reduce the sales volume of a competitor for Good B may reducethe competitor’s profits to the point where they exit the market. This is alsothe case where entry is undertaken through the provision of two goods,rather than just the one. In that case, bundling and tying practices mayagain have detrimental effect on competition. The overall conclusion is thatthere needs to be a thorough analysis in each case as to why tying andbundling is being used and what the effects are.

Introduction 7

14 Above n 3, para 5.17.15 Shapiro and Varian above n 4, 309.16 The analysis in this section is drawn from Innovation and Competition Policy above n 3,paras 5.37–5.69.

C. Intellectual Property Rights

The OFT report discusses two broad issues in relation to intellectual propertyrights (IPRs): the conditions under which such rights are licensed and theissue of access to those rights. In relation to the first issue, it is clear thatcompetition authorities must be alert to conditions in licences that restrictcompetition. Broadly speaking, these will be conditions that are not relatedto the protection of the right in question. For example, a condition limitingthe geographic area in which a licensee may use his or her rights would notbe a problem from the point of view of economic analysis. By contrast, acondition, which limited the licensee’s abilities in other markets, would be problematic as in the case where a licence to write software for a partic-ular platform prevented the software producers from writing for any otherplatforms.

The essential facilities doctrine is discussed in more detail in chapterthree, in the context of recent case law from the EC and wider literature onthe subject. As the OFT report puts it:

the core of the issue is balancing short run gains in efficiency with long runincentives to invest and compete dynamically17

In order to determine this question, the report notes a number of furtherkey questions. These include, is the market for the underlying facilitydynamic or potentially dynamic? What is the size of the benefit? Has thedominant firm invested in the facility? What is the potential impact onfuture investment? Can access be priced appropriately and what are thecosts of monitoring the access regime? Although these are difficult ques-tions, they are all questions that have commonly been asked in the contextof the so-called essential facilities cases.

D. Collective Conduct

In addition to the issues discussed above, further problems arise in relationto standard setting and joint ventures with the latter being discussed morefully in chapter four. In information and high technology industries the needfor compatible standards between different manufacturers may be critical.Shapiro and Varian18 make the point that it may be better for competitorsto co-operate over standards, rather than fight a standards war as this mayhave the effect of increasing the market for the product. Sometimes standardsetting is needed for the product to take off at all.

8 Cosmo Graham

17 Innovation and Competition Policy above n 3, para 5.101.18 Above n 4, 259.

When there is co-operation between parties, there must be a carefulassessment of the arrangement by the competition authorities. It has beensuggested that there are a number of critical questions. Do the firmstogether have market power? Is membership open or closed? Do they pos-sess blocking patents or other IPRs? What ancillary restraints are imposedon members of this group?19

E. Mergers

The most important question for mergers is to characterise the type of com-petition: is it competition in the market or competition for the market?20

The European Commission has been very concerned in the new economycases that have come before it to ensure that the future development of mar-kets is not restricted by the creation of bottlenecks over which the partieshave control.

3. CONCLUSIONS ON COMPETITION POLICY

The debate about the application of competition policy to new economy indus-tries will continue, particularly in the light of the European Commission’sinvestigations into Microsoft and its practices. Its recent statements arguethat Microsoft is leveraging its dominant position from the personal com-puter (PC) market into low-end servers and that Microsoft’s tying ofWindows Media Player to the Windows PC operating system weakenscompetition on its merits, stifles product innovation and ultimately reducesconsumer choice.21 What this very brief overview suggests is that the prob-lems facing competition policy are not necessarily conceptual ones, but willarise from the generic difficulty of applying general principles to specificfactual situations and, in particular, the need to take a view about the futuredevelopment of markets. It is now useful to remind readers of developmentsin EU and UK competition law and policy, which will form the contextwithin which such decisions are taken.

At EC level the two most striking developments are the increasingemphasis on economic analysis adopted by the European Commission(Commission) and demanded by the courts and the new procedural frame-work that will follow enlargement. The new procedural framework comesinto place from May 2004 and gives more responsibility for enforcement of

Introduction 9

19 Innovation and Competition Policy above n 3, paras 6.22–28.20 Ibid para 7.37.21 See Commission Press Release IP 03/1150, 6 August 2003.

EC competition law to national authorities and courts. The hope is that itwill free up resources in the Commission, so that it is able to concentratemore on serious European-wide cases of breach of competition law. Thismay help with the issue of economic analysis. The greater emphasis on thisissue can be seen as a Commission response in part to some long-standingcriticisms of the operation of the system under Article 81 EC, illustratedinitially in the block exemption on vertical restraints. The need for betterand more accurate analysis of the cases’ facts before it is one of the themesin three recent merger cases before the Court of First Instance (CFI), wherethe Commission’s decisions were annulled in each instance.22 The decisionscame at a time when the Merger Control Regulation was under review andprompted the Commission to undertake certain reforms in its internal pro-cedures, notably the creation of a panel to have a second look at mergercases and the position of Chief Economist. On the substantive level, theCommission appealed the Tetra Laval case to the ECJ because it involvesthe proper approach to be taken to leveraging, which is also an issue in theGE/Honeywell case, currently also before the EC courts.

At the UK level, major changes have come about through the EnterpriseAct 2002 as regards the control of mergers and market investigations.Broadly speaking, the role of politicians has been dramatically decreasedand the final decisions in these areas have been given to the CompetitionCommission which will decide, in essence, on the basis of whether or notthere has been, or will be, a substantial lessening of competition in the casein front of it. In legal terms, this represents a move away from the verybroad range of criteria that could be taken into account under the publicinterest provisions of the Fair Trading Act 1973. In terms of the new econ-omy, this should lead to a more focused investigation, with less chance ofnon-economic policy goals intruding, although there are special provisionsfor broadcasting and media mergers.

4. REGULATORY QUESTIONS

One of the issues surrounding the new economy is what alternatives arethere if a pure competition approach is deemed inadequate? One of theusual policy reactions to the failure of competitive markets to meet publicinterest goals is to set up a regulatory system for the industry concerned.As regards the new economy industries with the characteristics identifiedabove, this is a hazardous undertaking because most experience of regu-lation has been with industries that have not been characterised by such

10 Cosmo Graham

22 Case T–342/99 Airtours v Commission [2002] ECR II–2585, Case T–5/02 Tetra Laval vCommission [2002] ECR II–4381 (under appeal to the ECJ), Case T–310/01 Schneider Electric vCommission [2002] ECR II–4071.

a rate of technological change—the water industry for example. It is notimpossible to regulate a fast moving industry, or to use a statutoryframework in this context. Over the past 20 years regulation of thefinancial services industry has moved from a system dominated by prac-titioner regulation to one regulated under the Financial Services andMarkets Act 2000.

In this context, there are four broad regulatory questions that can beasked. First, is industry specific regulation necessary or would competitionlaw alone be appropriate? Second, if industry specific regulation is needed,what are the aims and objectives of such regulation? Third, what are thebest institutional arrangements and, finally, what are the best techniques orinstruments?

In dealing with the first question, in the context of new economy indus-tries, the popular answer is that only competition law is needed. This is thebasic assumption behind the EU’s new framework for communications regu-lation, although there is provision for other forms of regulation for operatorswith significant market power. In the UK, the regulator for telecommunica-tions has been given concurrent competition law powers with the OFT, whichsuggests that policy makers felt there was room for both. The argument thatyou cannot just simplistically move from industry specific regulation to com-petition regulation is explored in depth by Pierre Larouche who argues thatin addition to the areas that are not suitable, attempting to extend competi-tion law to these areas may have detrimental effects on its integrity.23

This links into the second question because, if we distinguish betweeneconomic (price control) regulation and social regulation,24 then theassumption is that the new economy will not be subject to the former.Again, this may be subject to exceptions, particularly in the case of infra-structure networks, but there are many broad social objectives that may bedesirable for new economy industries to accomplish. This is seen clearly inthe broadcasting sector and the arrangements in the UK provided by theCommunications Act 2003,25 but is also a theme underlying the EU discus-sion of public services, as further examined in chapter eight.

The aims and objectives of regulation will affect the type of institutionschosen as well as the methods of regulation that are considered most suit-able; these are both themes examined in depth in chapters five and six. In ageneral sense, the move is away from politically controlled regulatory agen-cies towards those that are considered, in some sense, independent andthere is also increasing discussion and interest in notions of self-regulation

Introduction 11

23 P Larouche, Competition Law and Regulation in European Telecommunications (Oxford,Hart Publishing, 2000).24 See A Ogus, Regulation (Oxford, Oxford University Press, 1994) for this distinction.25 For discussion prior to the act, see M Feintuck, ‘Walking the High-Wire: The UK’s DraftCommunications Bill’ (2003) European Public Law 105.

and co-regulation. As Andrew Murray and Colin Scott recognise at the endof chapter six, this raises questions about the legitimacy of such agencies’activities. In terms of competition regulation, increasingly the answer seemsto be a resort to reliance on agency expertise, on the basis of a boundedmandate, with checks provided by judicial mechanisms. This does not,however, provide an answer in terms of social regulation where, althoughexpertise can be prayed in aid, the relevant range of considerations is muchwider. All these issues suggest that the debate about new economy indus-tries’ regulation is only just beginning and that, as well as throwing up someold problems in new guises, it is also likely to bring forward some newquestions.

5. SUBSEQUENT CHAPTERS

In chapter two, Giorgio Monti conducts an overview of Article 82 EC: howit has been applied to new economy markets; and how analysis under thisprovision should be structured for future cases. He distinguishes between twobroad approaches to competition law enforcement: the ‘neo-structuralists’who favour aggressive antitrust enforcement in the appropriate circum-stances and the ‘neo-Schumpeterians’ who view dominance in new economy markets as temporary and short-lived. He concludes that the com-petition authorities in the EC and the US have preferred the neo-structuralistapproaches, which have on the whole been backed by the courts. He arguesthat neo-Schumpeterian ideas do not easily fit within the current competi-tion law culture, or the analytical framework of Article 82 EC. He notesthat although the authorities have engaged in aggressive enforcement, theyhave focused not on imposing financial sanctions but on ex post behav-ioural remedies and ex ante divestitures. The problem is that these types ofremedy turn competition law into a market regulator, a task for which it isill equipped. Since one of the aims of this approach, he argues, is to removedominance by facilitating entry, the more pressing question is whether thispolicy will eliminate incentives to innovate.

This leads into Estelle Derclaye’s discussion in chapter three of the appli-cation of Article 82 EC to the licensing of intellectual property rights, whichis the subject of some notoriously confusing case law. This is done in thecontext of a careful discussion of the recent IMS Health case26 where thecentral issue is whether or not there is an essential facilities doctrine. At issuein the case is whether the structure that IMS Health created for analysing

12 Cosmo Graham

26 Commission Decision 2002/165/CE, 3 July 2001, Case COMP D3/38.044, NDCHealth/IMS Health: interim measures [2002] OJ L 59/18; Case T–184/01 IMS Health vCommission [2001] ECR II–2349; [2001] ECR II–3193.

sales of drugs by German pharmacies must be shared with its competitors.On its analysis of the case law, the Commission argued that it should, andimposed interim measures requiring sharing. In the initial proceedings, theECJ did not agree that the imposition of such measures was appropriate.Derclaye argues that the Commission’s interpretation of the case law isincorrect and that there is a distinction in the case law between cases dealingwith intellectual property rights and those dealing with other forms of prop-erty. The result of this is that since the conditions under which a refusal tolicense are cumulative, then it will only be in rare cases that they are all ful-filled. Consequently abuse of dominance of intellectual property rights willbe unusual and IMS Health should, on this interpretation, win the case.

In chapter four Dr Joachim Lücking examines the challenges for compe-tition law rules set by business-to-business (B2B) electronic marketplaces.He distinguishes between four different types and discusses the case lawthat has grown up in the EC around such marketplaces. He identifies thatissues can potentially either arise under the Merger Control Regulation orunder Article 81 EC, although the working of the market may raise eitherArticle 81 or 82 EC concerns. He goes through the appropriate approach tomarket definition, possible co-ordination effects, the concern with foreclo-sure, before discussing best practice guidelines. His conclusion is, however,that the current competition rules are fully capable of dealing with B2Belectronic marketplaces because they do not raise issues that are uniquelyoutside the existing framework.

In the second part of the book, the focus switches to regulatory issues,beginning with Paul Nihoul’s examination of the relationship between insti-tutional competition and the information society. He begins this by an exam-ination of the reform of telecommunications regulation at EU level, ending inthe new framework for the regulation of electronic communications. Heinterprets this in part as a process of competition between various regulators,both at EU and national level, as well as between levels. Putting this into thecontext of the convergence of communications markets and drawing parallelswith the experience of international law, he asks what lessons can be learnt.He suggests that for the future, we abandon the idea of law and regulationhaving a central unifying authority. Instead, a plurality of regulators, albeitoperating within some common framework, would allow operators to have awider range of choice over which regulatory systems they deal with.

In chapter six, Andrew Murray and Colin Scott examine the issue ofwhat regulatory techniques should be applied to the new media, in particu-lar the Internet. They discuss the argument that the Internet is not capableof regulation, although they find that view flawed. They extend LawrenceLessig’s work on the modalities of regulation27 to draw up a typology of

Introduction 13

27 L Lessig, Codes and Other Laws of Cyberspace (New York, Basic Books, 1999).

four types of control system: hierarchical control, community-based control, competition based control, and design based control. Within eachof these four types, there are three techniques: standard setting, informa-tion gathering and behaviour modification. They argue that within the newmedia are existing forms of control mechanisms and that the challenge is tofind ways of stimulating or steering those mechanisms towards meeting thepublic interest objectives of regulation. Thus, for example, hierarchy can beused to steer systems that involve control based in community market ordesign. This feeds into Nihoul’s idea of a common framework, althoughMurray and Scott also point out that effective control can involve any ofthese systems, with or without hierarchical control.

In chapter seven, Fiona Smith looks at a specific issue, namely the regu-lation of e-commerce by the World Trade Organisation. This is really a spe-cific example of the need for an existing regulatory mechanism to adapt tonew ways of working by industries, in this case, new ways of trading. Shefocuses on the problems raised by the definition of e-commerce adopted bythe WTO Work Programme on Electronic Commerce. Specifically, the dis-cussion focuses on the classification of e-commerce as trade in goods and/ortrade in services and the implications of failing to clarify where that boundary lies. She explores the disagreements and arguments over how e-commerce is to be classified for the purposes of the various agreements.She argues that in marginal cases the classification decision will be made bythe member state potentially making a decision based on the needs of itsdomestic economy, rather than on global considerations. This arguablyleads to an inappropriate politicisation of the decision and she concludesthat the classification issue needs to be resolved if the WTO is to be used asan appropriate regulatory framework.

Finally, Erika Szyszczak examines the provision of public services in thenew economy, of which the most important in this context is the universalservice obligations imposed on telecommunications providers. The discus-sion begins by explaining the general context of the debate over the defini-tion of the concept of ‘public services’ in the EU and goes on to examine insome depth the recent case law on the concept of an ‘undertaking’ and thedefinition of state aid. She argues that it is important to understand andclarify where the boundaries end for the state to claim total immunity fromthe market and where the market rules begin. She maintains that under theexisting legal tools the ECJ has been unable to offer either clarity or consis-tency in addressing this question. Although the insertion of Article 16 EChas signalled a greater respect for the role of public services, this has not yetprovided the legal base for a positive set of EU values. The Commission’sGreen Paper28 takes a holistic approach to the issue and this may bringtogether the so far fragmentary discussion.

14 Cosmo Graham

28 COM (2003) 270 final.

We cannot pretend that the essays in this volume cover all aspects of therelationship between competition, regulation and the new economy indus-tries. We hope that by bringing them together in this format, the collectionraises some valuable questions and stimulates discussion on what will con-tinue to be an important policy problem for the 21st century.

Introduction 15

2

Article 82 EC and New EconomyMarkets

GIORGIO MONTI*

THE AIM OF this chapter is to provide an overview of the application of Article 82 EC, which prohibits the abuse of a domi-nant position by one or more undertakings, in new economy

markets. The ‘new economy’ is normally associated with three high technology industries: computer software, Internet-based businesses, andcommunication services.1 There is broad agreement that these industriespresent a unique set of characteristics that distinguish them from tradi-tional manufacturing industries. In particular, four features are associatedwith new economy markets: economies of scale, network effects, consumerlock-in, and high rates of innovation.2 There is much less agreement overthe implications of these differences for competition policy.

In broad terms, we can identify two divergent approaches: those whichfocus on the first three attributes and particularly on the dangers of networkeffects as a reason to justify aggressive antitrust enforcement; and thosewhich focus on the dynamic characteristic of these industries, suggestingthat any dominance is temporary as markets are contestable. According tothe latter camp, antitrust enforcement is largely unnecessary.3

These opposed views are reminiscent of a more general debate over therole of competition law: on the one hand the Chicago School argues for a

* Law Department, London School of Economics. An earlier version of this chapter was pre-sented at the Competition Law and the New Economy Conference on 12 July 2001, organisedby the University of Leicester. I am grateful for the many helpful and encouraging points raisedby the participants and for the comments of an anonymous referee. All errors and viewsexpressed are mine.1 R Posner, Antitrust Law 2nd edn (Chicago, University of Chicago Press, 2000) 245.2 Eg RE Litan, ‘Antitrust Law and the New Economy’ (2001) 62 University of Pittsburgh LawReview 429; SN Weinstein, ‘Sherman Act Violations: Monopolization: Tying’ (2002) 17Berkeley Technology Law Journal 273.3 These are ideal types; there are many positions in between these extremes.

limited role for competition law because markets tend to be contestable,whilst its opponents criticise the Chicago School for being overly optimistic about the market’s abilities to cure itself, noting that certainmarket structures are likely to lead to dominance and result in anticompet-itive practices. In my view, the current controversies over the role of compe-tition law in new economy markets are an extension of that more generaldebate, with novel economic arguments.

The chapter is structured in this manner: part one outlines the two con-flicting perspectives over new economy markets—the ‘neo-structuralist’which favours aggressive antitrust enforcement when anti-competitivestructures manifest themselves, and the ‘neo-Schumpeterian’, for whichdominance in new economy markets tends to be temporary and short-lived.To date, enforcement agencies have favoured the neo-structuralist view.This is evidenced in the next three parts of the chapter, which review mar-ket definition, the concept of dominance and the notion of abuse in neweconomy markets. This focus on neo-structuralism may be criticised forfailing to take into account the adverse effect that aggressive competitionlaw enforcement might have on innovation. Consequently the fifth sectionof this chapter suggests how the promotion of innovation might affectfindings of abuse under Article 82. The sixth section observes that thekinds of remedies that are likely to prevail in new economy markets arebehavioural and structural, designed to regulate the markets rather than topenalise the dominant firms.

While the focus is on Article 82, it will be useful to refer to comparablecase law from the United States,4 where a number of significant antitrustcases in new economy markets have already been considered and to exam-ine the approach of the EC and US authorities under the merger rulesbecause considerations of dominance and potential abuses of dominance ifthe merger is allowed to be consummated show the fears that competitionauthorities perceive in new economy markets and can serve as an indicatorof the kinds of actions likely to be taken under Article 82.

1. COMPETING PERSPECTIVES OF NEW ECONOMY MARKETS

A. Neo-structuralism

From a neo-structuralist perspective, there are three market characteristicswhich indicate the likelihood of dominance in new economy markets.

18 Giorgio Monti

4 The US approach under s 2 of the Sherman Act is comparable. In US v Grinnell Corp 384 US563, 570–71 (1966) the Court defined illegal monopolisation as ‘(1) the possession of monop-oly power in the relevant market and (2) the wilful acquisition or maintenance of that poweras distinguished from growth or development as a consequence of a superior product, businessacumen, or historic accident’.

First, many products (for example computer software) are characterised bysignificant economies of scale—whilst it is very expensive to devise newsoftware, once the programme is written, it is virtually costless to repro-duce, marginal costs are almost zero, which leads to a situation where onedominant firm can capture most of the market.5 However, in itself this doesnot lead to dominance unless the high market shares can be held for sometime. This is where the other two characteristics of new economy marketssupport the risk of dominance: network effects and lock-in.

Many new economy businesses are characterised by network effects—the more people use a particular service, the more value it has to an individualuser.6 For example, the more people use a fax machine, the more valuable afax machine is to each user because it can be used to communicate withmore people. In these markets there will be a convergence to the mostpopular network standard. Much like the periodic table of elements, onceenough chemists agree to use a set of abbreviations for chemicals, then theminority who use alternative formulae will have to abandon these if theywish to participate in scientific discourse. The difference between the sym-bols in the periodic table and a network like the ‘instant messaging’ serviceoperated by an Internet Service Provider (ISP) is that the latter is owned byan undertaking, consequently those outside the network cannot communi-cate to those within, because each ISP uses its own separate and incompati-ble standard so that only its customers can communicate through it.7 Ifenough consumers migrate to one instant messaging system, then thatprovider will have a dominant position, because new consumers will wishto enter the network which has the greatest number of users; similarly,those consumers using smaller networks will migrate to the larger networkwhere they can communicate with more people. The upshot is that a pro-prietary network creates risks of dominance if there is sufficient consumerdemand for the services that can be obtained via the network and if othernetworks are not compatible. All three elements are necessary in order tocharacterise a network effect as potentially anticompetitive. 8

Instant messaging networks are a good example of incompatible propri-etary networks where there are still many players on the market because thereis not sufficient consumer demand for a single network. Similarly, the marketin video games consoles is characterised by more than one network and is

Article 82 EC and New Economy Markets 19

5 O Shy, The Economics of Network Industries (Cambridge, Cambridge University Press,2001) 5.6 ’The utility that a user derives from consumption of the good increases with the number ofother agents consuming the good’: ML Katz and C Shapiro, ‘Network Externalities,Competition and Compatibility (1985) 75 American Economic Review 424.7 See C Veljanovski, ‘EC Antitrust in the New Economy’ (2001) 22 European CompetitionLaw Review 115 who notes correctly that proprietary standards and incompatibility betweenstandards are necessary conditions for the theory of network effects to apply.8 Thus the existence of a network does not mean a fortiori that there is or will be dominance.

highly competitive. In contrast, in the early decades of the 20th century,when AT&T was developing the local telephone it refused to connect withindependent competitors and because there was strong consumer demandfor a telephone network that would allow a user to communicate with allother users, AT&T became dominant.9

20 Giorgio Monti

9 HA Shelanski and JG Sidak, ‘Antitrust Divestiture in Network Industries’ (2001) 68University of Chicago Law Review 1, 8.10 AOL/Time Warner [2001] OJ L/268/28.

Network effects can also manifest themselves in an indirect manner: theowner of a dominant, proprietary network may attract businesses wishingto sell consumers products that run on the dominant network, thusstrengthening the dominance of the network. A good example of the com-petition authorities seeing indirect network effects is the EC Commission’sanalysis of AOL’s position in the dial-up Internet access market in the UK.10

AOL has a network of users that navigate through its site. Navigating withAOL is not the same thing as surfing the whole Internet because AOL hascreated a more limited network of sites that can be reached by clicking onits pages. Thus whilst an AOL customer may feel like she is surfing theWorld Wide Web, it is a smaller web that she is surfing. Moreover, its webdesign is so attractive that many users never step outside the web pagesthat it supplies. This ‘stickiness’ allows AOL to enter into lucrative

Figure 2.1

contracts with content providers who want presence in its territory. In theview of the Commission, the more content providers contract with AOL,the more people will subscribe, and in a virtual circle, the more consumers,the more content providers want to be present on the AOL network. Thisis sometimes called a ‘snowball effect’. AOL’s dominance would increasedramatically if it were vertically integrated with a leading music contentprovider, transforming it into an ‘on-line store/essential facility’.11 Musicwould attract many new subscribers, because music is one of the mostpopular and sought after elements of Internet content.12 This approachsuggests that an internet service provider (ISP) could consolidate a domi-nant position either by vertical integration with popular internet contentor presumably, by having exclusive access to popular internet content,without ownership.

The operator of a complex proprietary network attracts both consumersand suppliers of goods that run on the network. The network is strength-ened by the number of participants on either side: the more consumers, thegreater the network’s ability to draw more consumers and more ancillarycontent, and the more ancillary content, the more consumers are drawn tothe network. In this setting a new entrant has to promise consumers a supe-rior network and superior ancillary content also, if the existing ones cannotrun on the new network. This makes entry even more difficult because twomarkets may have to be broken into by a new competitor.

By gathering enough content, AOL could become the dominant ISP,causing many more Internet subscribers and content providers to migrateto its system. This analysis assumes that there are certain ancillary productsthat add so much value to the network so as to allow the creation of a dom-inant position. This is an indirect variation of the network effects theorybecause the network effect is caused by the presence of other propertylinked to the network, but the anticompetitive risks are similar: a domi-nant network owner can behave independently of competitors, who willfind it difficult to secure sellers of ancillary products wishing to place themwith an ISP that has a low number of consumers, and independently ofcustomers wishing to supply goods that run on the existing networkbecause they need the network more than the network needs them.

While network effects create entry barriers for new competitors andinvite more consumers to the dominant network, another phenomena—consumer lock-in, affects the ability of consumers to look for substitutes.Lock-in occurs when the cost of switching to a new product is higher thanthe marginal benefits to be gained by the use of the new product. In termsof computer software, for example, a consumer will find it too costly to


11 Ibid para 82.12 Ibid para 82.

switch from one brand of software to another because of the need to learnhow to use new software and the problem of transferring files from oneprogramme to the next. Only a considerably better product will inducethose who are locked-in to switch. This makes entry more demanding, andlocked-in consumers risk exploitation by the dominant firm.

B. Neo-Schumpeterianism

In response to these concerns about the risks of dominance, some haveargued that the new economy moves on Internet time—where it is saidthat competitive advantages are temporary and markets contestable.13

The incentives to contest markets are also high: whoever finds the new‘killer application’ will take a large chunk of the market and the prize isextremely high profits. Many commentators have argued that the domi-nance achieved, even through the creation of a successful network, isalways temporary. Thus, the dynamism of the new economy can cancel outdominance.14 Joseph Schumpeter had set out this vision of competition inthe 1940s:

Competition from the new commodity, the new technology … which strikesnot at the margins of the profits of the existing firms but at their foundationsand their very lives.15

If the holder of a network is in constant fear of being outdone by a newproduct, that will diminish the dominance of the network vis-à-vis its con-sumers. In addition, the fear of an alternative network evolving will alsocreate incentives for the network owner to co-operate with producers ofancillary goods, because the provision of these to its consumers is a way ofmaking the network more competitive. According to Teece and Coleman‘antitrust authorities need to be cognizant of the self-corrective nature ofdominance’ in new economy markets.16

The consequences of the market being so dynamic also impact on thestrategies of firms: if markets change so quickly then predatory strategies toexclude competitors become much more difficult to plan because of theuncertainty that they will succeed in consolidating the predator’s dominance.

22 Giorgio Monti

13 Eg MA Cusumano and DB Yoffie, Competing on Internet Time (New York, Touchstone,2000).14 Shelanski and Sidak above n 9 relying on Joseph Schumpeter’s analysis.15 J Schumpeter, Capitalism, Socialism and Democracy 3rd edn (originally publ 1950, reprLondon, Routledge, 1992) 84.16 DJ Teece and M Coleman, ‘The Meaning of Monopoly: Antitrust Analysis in High-TechnologyIndustries’ (1998) Antitrust Bulletin 801.

Furthermore, recoupment of predatory losses is obviously threatened by aparadigm shift.17

The implication of this analysis is that competition law is largelyunnecessary: new economy monopolists are fragile and exclusionary tac-tics are unlikely to be successful. Moreover, competition law interventionmay even be counterproductive: if industry develops at such quick speeds,then it is more difficult for the competition authorities to impose remediesthat improve competition—there is an obvious regulatory lag between theanticompetitive act and the imposition of a penalty and therefore condi-tions may have changed by the time the competition authorities are able tointervene.18 A related point is that the dynamism of the market is caused, ingreat part, by the possibilities of the winner making profits. One concern isthat with heavy handed regulation the incentives to invent new productswill diminish significantly. This argument applies especially to requirementsthat a dominant firm grant access to its ‘essential facilities’ which can workas a tax on innovation.

Finally, even if one were to concede that network effects exist, theseeffects are demand led and therefore have increased consumer welfare. Ifso, where is the antitrust problem? Moreover, when considering remedies,how can one be sure that the remedy will not lead to a less efficient provi-sion of services? Take the AOL example used above—the Commission isconcerned if the owner of popular Internet content enters the market forservice provision, yet arguably could there be synergies in this vertical inte-gration? On this view, the competition authorities should not attempt torecreate perfect competition market structures, as these may not be themost efficient in promoting consumer welfare.19

In sum, under a neo-Schumpeterian model, new economy firms withlarge market shares have little incentive to exploit consumers, becausethese will quickly find a substitute, and little incentive to damage rivals,because such tactics will be counterproductive. Thus, dominance is rare innew economy markets. Moreover, even assuming there is dominance, therole of competition law remedies is questioned: the remedy may arrive toolate, and the market may have already cured itself; alternatively, the remedy may give firms less incentives to develop new products, reducingconsumer welfare.


17 Ibid at 810.18 Ibid at 803. They also add that the markets are complex and this makes intervention evenmore risky. This point cannot be accepted—every market is complex, so that cannot be anargument against intervention.19 C Ahlborn, DS Evans and AJ Padilla, ‘Competition Policy in the New Economy (2001)European Competition Law Review 156, 160.

C. The Response of the Authorities

The competition authorities so far have sided with the neo-structuralists.Mario Monti’s position is that ‘even if the pace of high technology sectorsmeans that market failures last only for a short time—and I have seriousdoubts about this—this does not mean that we should be less concerned’.20

Similarly, the former Chairman of the Federal Trade Commission, RobertPitofsky, said that it would be ‘naïve’ to conclude that there is no role forcompetition law when market power is based on intellectual property;rather competition law is flexible enough to play a role in regulating thenew economy.21 In his view, a firm with demand-side economies of scaleand high consumer switching costs due to network effects can make marketpower more durable.22

This rhetoric is matched by actions—most notably the high profile law-suit against Microsoft, but also a number of other investigations into therisks of dominance in new economy markets. In the three parts that followwe review how competition law analysis has been adapted to apply to neweconomy markets, following the structure of analysis under Article 82:market definition, dominance and abuse. In each section we reveal howthe neo-structuralist model has been applied and the criticisms of thatmodel by the neo-Schumpeterians. One theme running through these sections will be why the neo-Schumpeterian position has failed to gainacceptance.

2. RELEVANT MARKET ANALYSIS

Market definition focuses upon consumer needs and identifies what prod-ucts consumers could switch to if the products under scrutiny became moreexpensive and whether suppliers might enter the market with little timedelay. According to the Commission’s Notice on Market Definition, theemphasis is upon demand substitution, using empirical and econometricevidence to identify product markets.23 This approach should not changewhen considering new economy markets, so long as the special characteristicsof the industry are taken into account.

24 Giorgio Monti

20 M Monti, ‘Competition and Information Technologies’ <http://europa.eu.int/comm/compe-tition/index_en.html> (18 September 2000).21 R Pitofsky, ‘Antitrust and Intellectual Property: Unresolved Issues at the Heart of the NewEconomy’ (2001) 16 Berkeley Technology Law Journal 535–36.22 JD Balto and R Pitofsky, ‘Antitrust and High-Tech Industries: The New Challenge’ (1998)66 Antitrust Bulletin 583, 585; R Pitofsky, ‘Challenges of the New Economy: Issues at theIntersection of Antitrust and the New Economy’ (2001) 68 Antitrust Bulletin 913.23 EC Commission Notice on the Definition of the Relevant Market for the Purposes ofCommunity Competition Law [1997] OJ C/372/5.

In this section we examine the approach to defining markets in two decisions: AOL/Time Warner and WorldCom/Sprint. In the former, theCommission faced the difficulty of determining whether the delivery of on-line music, an emerging mechanism of distribution, could be considered arelevant product market. In the latter the Commission analysed whethertop-level Internet Service Providers (ISPs) were distinct from lower-levelISPs. From a neo-structuralist perspective the market definitions are con-vincing; however a neo-Schumpeterian perspective would criticise thesemarket definitions as unnecessarily narrow, for two reasons: first inAOL/Time Warner the Commission failed to deploy the correct economet-ric methodology favouring a set of factors that make little economic sense;secondly while the market definition in WorldCom/Sprint is more persua-sive, it fails to take into account the dynamic nature of the new economy.These critiques are not unprecedented: the EC has long been accused ofdefining certain markets too narrowly.24 In addition to these criticisms, itwill be argued that the AOL/Time Warner decision seems to adopt anunnecessarily narrow definition of the market for policy reasons—the wishto regulate the development of the on-line industry.

A. On-Line Music

In AOL/Time Warner, the Commission found a market for on-line music,distinguishing between on-line music and other sales channels. However, atfirst sight it is difficult to see why the on-line music market should be iso-lated from the market for the sale of music in general. The merged entitywould hold between 30 and 40 per cent of music publishing rights and thegeneral risk identified by the Commission was:

One entity controlling such a sizeable music catalogue could exercise substan-tial market power, by refusing to license its rights, or threatening not tolicense them, or imposing high or discriminatory prices and other unfair com-mercial conditions on its customers wishing to acquire such rights (such asInternet retailers offering music downloads and streaming).25

The approach is problematic in two respects—first because it is unclearwhether from the perspective of consumers’ on-line music is a separatemarket. On the demand side the Commission pointed out that consumershave immediate access to music. This may be so, but as most people like tolisten to music in places other than their computer chair, they will still need


24 V Korah, An Introductory Guide to EC Competition Law and Practice 7th edn (Oxford,Hart Publishing, 2000) 84–85.25 AOL/Time Warner above n 10, para 47.

to go out and buy some blank CDs to burn the music on. The fact thatthey can download single tracks is also unconvincing on its own—it pointsto a physical difference, but says nothing about whether this differencewould affect consumer choice. That special hardware needs to be obtainedto download on-line music but does not show that on-line music is a sepa-rate market.

On the supply side, it is true that the structure of music supply on-line isdifferent for the supplier, but this does not make the two markets separatefor the consumer, and it is the consumer perspective which should deter-mine the relevant product market. The fact that the prices of CDs have notdecreased as a result of on-line supply of music can be explained by the factthat the level of Internet penetration is low, and that not all consumers willuse the Internet to download music—some will simply not see it as a viable‘market’ for purchasing music.26 However, the recent losses reported byrecord companies are a sign that pirate copies downloaded on-line havehad a negative impact on sales of CDs, indicating that the two markets aresubstitutable, not separate.

Lastly, the Commission also noted that on-line supply of music wasan emergent market, but it does not follow that it should be a separatemarket.27 It is also worth noting that the Internet distribution channel isnot always seen as a separate market—for example in Telefonica/Terra/Amadeus28 the Commission thought that an on-line travel agency was inthe same product market as traditional travel agencies. However, it is diffi-cult to see why buying music on-line is so different from buying music fromshops, whereas buying a travel ticket from the web is the same as going to atravel shop.

In sum, the first criticism which may be levelled at the Commission’sMarket Definition in this case is that it eschews the methodology propounded in the Notice on Market Definition (under which the emphasisshould be upon whether consumers, on the basis of econometric evidence,find on-line music supply a separate market) in favour of a general, subjec-tive evaluation based exclusively upon product characteristics. Thisapproach is particularly unhelpful in the new economy because high tech-nology products are all highly differentiated but may still be considered assubstitutes. Thus, as Pleatsikas and Teece observe, the traditional approach

26 Giorgio Monti

26 M Haig, ‘A Battle for the Future of Digital Music’ The Independent (20 August 2001)review s 8.27 AOL/Time Warner above n 10, para 21.28 Decision of 28 April 2000 (text in Spanish available from <http://europa.eu.int>). To be fair,there was no need for the Commission to analyse market definition in depth because evenunder a narrow market definition it would have concluded that no dominance was likely toresult; moreover the Commission noted that the markets may change and a reassessment ofthe relevant market may be warranted (para 12 of the decision).

to market definition does not reflect the manner of competition in neweconomy industries.29

The second criticism is that the Commission’s principal concern doesnot seem to be that the merger would lead to higher prices for con-sumers, but that it would foreclose entry by potential competitors. Thereare two ways through which an owner of substantial music content could foreclose entry on the on-line music market. First, if a lot of on-linemusic is formatted with one proprietary standard, then the owner of thattechnology can control the development of music player software by notlicensing this technology unless the software manufacturers promise not tosupport other technologies; in addition other music publishers will find itnecessary to format their music according to the technical standard of theleading firm, which would act as a gatekeeper. Second, a large music pub-lisher could also enter the downstream market and format its music to fitits own music player, thus eliminating other music player software.30 TheCommission’s market definition therefore seems to be policy oriented,because it allows the Commission to find dominance in the market foron-line music and by allowing it to prevent the creation of such domi-nance, the Commission can protect the development of competing tech-nological alternatives and prevent the foreclosure of emerging markets.These considerations were articulated expressly when the Commissionblocked mergers in the field of digital pay TV in the German market, notingthat technological progress would be adversely affected if the dominantprogramme supplier on digital pay TV were to join forces with the domi-nant supplier of technical services for pay TV.31 Arguably, the narrowmarket definition does allow the Commission to regulate the developmentof technical standards in emerging markets, and also the development ofthis emerging industry. Whilst this is laudable, it is questionable whetherthis is something to be achieved by competition law through a narrowmarket definition in a merger case. The narrow market definition isused as a means to an end—it allows for the creation of new marketsand protects those that are developing new technologies (here softwaremanufacturers).


29 C Pleatsikas and D Teece, ‘New Indicia for Antitrust Analysis in Markets ExperiencingRapid Innovation’ in J Ellig (ed), Dynamic Competition and Public Policy (Cambridge,Cambridge University Press, 2001) 112–17.30 Note similar fears about the development of the pay TV market in Germany if the merger inDeutsche Telekom/BetaResearch [1999] OJ L/53/3 were to be allowed: paras 33–34 (concernsthat BetaResearch’s d-box technology would become the digital standard); paras 35–36 (con-cerns over the development of new technology); paras 37–39 (concerns that all new operatorswould be dependent on BetaResearch’s licensing policies). See 28th Competition Report(1998) paras 96 and 141.31 Bertlesmann/Kirch/Premeire [1999] OJ L/53/1 passim, but esp paras 119–22. Again, onecould question whether digital pay TV is in a separate product market from terrestrial TV.

If the concern is about standards and about access by new entrants, thenthese seem to be issues best left to regulation rather than to competitionenforcement. A regulatory body specifically charged with overseeing theindustry would be a more direct and effective means through which to safe-guard the interests of market participants. In the United States, similar con-cerns about the market for Internet music and the development of playersoftware have led to legislative proposals to reform the copyright laws.These proposals also stemmed from increased vertical integration betweenrecord companies and on-line music distributors and are designed to makeit easier for other on-line companies to obtain licences to sound recordings,providing further that the record companies may not dictate the use of anyparticular digital music player.32

B. Top-Level Internet Service Providers

In mergers of Internet Service Providers (ISPs), the concept of networkeffects has been deployed to aid the process of market definition. For theInternet to work, there has to be an infrastructure that allows for computersto communicate with one another and for communication lines to allow forworldwide connectivity—this infrastructure works like this:33 consumerscontract with ISPs to gain access to the Internet, and ISPs enter into con-tracts with each other to ensure that their customers can communicate withcustomers of other ISPs. Now, it would be quite expensive and time con-suming for an ISP to enter into commercial relations with all other ISPs,especially given the entry of new competitors. To facilitate the process, themarket has evolved so that there are a group of ‘top-level’ ISPs and asmaller ISP can merely contract with one of these to allow its customers toaccess all other Internet users.34 The top-level ISPs have high capacitytransmission facilities and ‘peering’ arrangements among themselveswhereby they receive traffic from other top-level ISPs free of charge. Thesmaller ISPs enter into so-called ‘transit’ arrangements with one of the top-level ISPs, for which there is a charge, in exchange for which the smallerISPs can offer their customers universal access to the Internet. Thus, theinfrastructure is hierarchical and top-level ISPs are in a separate market.This is supported by the finding that most Internet communications are carried over top-level networks.35

28 Giorgio Monti

32 Music Online Competition Act 14 HR 2724 107th Cong § 4(b). For comment see A Davieand C Soares, ‘The Music Online Competition Act of 2001: Moderate Change or RadicalReform?’ (2001) Duke Law and Technology Review 0031.33 This analysis is taken from WorldCom/MCI [1999] OJ L/116/1 and US v WorldCom andSprint complaint by the Department of Justice (26 June 2000) <http://www.usdoj.gov/atr/index.html>. The analysis in the two jurisdictions is very similar.34 DOJ WorldCom/Sprint ibid para 22.35 DOJ WorldCom/Sprint ibid para 27.


36 WorldCom/MCI above n 33, para 126 (the DOJ did not use this theory in their complaint).

Figure 2.2 The analysis by the EC and US Authorities sees a market where theconsumer wishing to connect to an ISP can either connect directly to a Tier 1 ISP orto a Tier 2. If he chooses to connect with a Tier 2 ISP, it is necessary for the Tier 2ISP to be connected to a Tier 1 ISP so that the customer can connect to all otherinternet users.

With this perspective, top-level Internet service provision is a separatemarket, so if one firm owned all or a substantial part of the top-level, itwould hold a dominant position. This conclusion is sustained by the theoryof network effects—an Internet consumer will prefer to join an ISP whichwill guarantee universal connection and if universal connection is onlyavailable if the consumer can contact directly or indirectly, the top-levelISPs, then the first tier is a compulsory partner for any ISP and indirectly,for any Internet user. In fact, the EC Commission concluded that a domi-nant top-level firm would own a network which could be characterised as an‘essential facility.’36

Crucial to this analysis was the finding that the Internet only works hierarchically. As the Department of Justice’s (DOJ) analysis makes clear,the market has evolved so that it works in a hierarchical manner because itis convenient. But the real question to be asked is whether, if the top tierfirms raise prices, the shape of Internet communications would change—for

example, could not all communications be re-routed to lower tiers? TheDOJ answered this question in the negative. It stated that top-level ISPshave two advantages: a cost advantage because they have peering relation-ships with other top-level ISPs and therefore do not need to buy transit inorder to provide universal service connection. In addition lower level ISPsexchange traffic at inferior public interconnection facilities, whereas thelarger ISPs own private interconnection facilities meaning data transmissionis more secure. Consequently, the fewer the number of network connec-tions connecting to a large top-level network, the fewer the cross-networkconnections.37 Moreover, as the EC Commission pointed out, the cost oflower level routing of all Internet communications would entail a lot ofconnections, each carrying very low levels of traffic, so the connectionswould not be cost effective. In addition, even if the lower level ISPs man-aged to create a sub-network, this would not provide universal accessbecause the customers of the top-level ISP could not be connected unless atransit arrangement was entered into with the top-level ISP.38 Thus, for universal connectivity, contact with the top-level is needed and no supply substitution is readily available. This is a laudable application of economicevidence to the market and stands in stark contrast to the less rigorousapproach in AOL/Time Warner.

This carefully constructed analysis by EC Commission and DOJ showsthat network effects can be deployed to define the relevant product mar-ket. However, the rejoinder from the advocates of dynamic markets is thatthe above market analysis is already out of date. There have been at leasttwo developments that challenge the identification of top-tier ISPs as aseparate market. First, the cost of alternative routing has become cheaper,consequently peering at lower levels of the Internet hierarchy is now economical.39 Second, corporate users now resort to multi-homing: pur-chasing connections from more than one provider. The impact of this isthat top-tier ISPs are no longer a compulsory partner for much Internettraffic and an economic study has noted that these developments havehad the effect of reducing the bargaining advantage of the tier 1 ISPs.40

These observations cast some doubts over the accuracy of using networkeffects to decide upon market definition. The difficulty is that theapproach is unduly static, which some have said is unsuitable in marketswhere there is rapid innovation, because it will lead to unnecessarily narrowmarkets.41

30 Giorgio Monti

37 DOJ WorldCom/Sprint above n 33, paras 27–30.38 WorldCom/MCI above n 33, paras 75–76.39 The new routing standard is known as BGP4.40 Besen et al, ‘Advances in Routing Technologies & Internet Peering Agreements’ (2001) 91(2)American Economic Review 292.41 Pleatsikas and Teece above n 29, 131.

C. Structuralism Prevails

The two decisions considered above illustrate that the Commissionfavours a structural approach to defining markets. Why have the neo-Schumpeterian criticisms not found a place in the competition law analy-sis of the Commission? There seem to be two reasons. First, it is difficult totranslate the commission’s economic insights into rules of market definition—how can one take rapid innovation (which necessarily occurs in thefuture) into account when determining the relevant product market?Moreover, this kind of consideration can be taken into account at thenext stage, when assessing dominance, for long-term supply substi-tutability is considered fully at that stage in Article 82 cases.42 The second reason is that if the definition of the market is policy-driven, thenthe analysis will shun all evidence that contradicts the narrow marketdefinition which aids the Commission in targeting a market it is con-cerned about.

3. DOMINANCE

A firm is dominant if it has a high market share which is protected by bar-riers to entry. This is a structural approach which allows the court to inferdominance.43 Direct proof of dominance would require evidence of monop-oly prices maintained over time, but this would be hard to obtain and maynot be the way in which dominance manifests itself, specifically becausedominance need not be absolute, it is merely ‘a position of economicstrength enjoyed by an undertaking which enables it to prevent effectivecompetition from being maintained on the relevant market by giving it thepower to behave to an appreciable extent independently of its competitors,customers and ultimately of its consumers’.44

This is significant because it does not tally with an economist’s conceptionof monopoly—in EC and US antitrust law there can be some competition in amarket dominated by one firm and dominance is the power to defeat that competition, not to ignore it.45 Therefore, to prove dominance, thecompetition authorities have to show that the firm is likely to be able todefeat the challenges posed by potential competitors. Dominance in most


42 EC Commission Notice on the Definition of the Relevant Market above n 23, paras 23–24.43 US v Microsoft 253 F 3d 34 (DC Cir 2001) (en banc).44 Case 27/76 United Brands [1978] ECR 207 para 65.45 Similarly, Faull and Nikpay (eds), The EC Law of Competition (Oxford, Oxford UniversityPress, 1999) paras 3.25–3.30; Judge Wyzanski in US v United Shoe Machinery 110 F Supp 295(DMA 1953) also thought that monopolisation for the purposes of s 2 of the Sherman Actmerely required the power to exclude competition.

cases is thus based upon the combination of high market shares and highbarriers to entry.46

A. The Traditional Test: Market Shares Plus Entry Barriers

For most products, market shares can easily be calculated by looking at thelevel of sales.47 However in other markets the calculation of market sharesmay require different means of measurement. In markets for Internet serv-ices, for example, the Commission used revenue and traffic flows to deter-mine market shares.48

Given the speed of market developments, the relevance of market sharescan be questioned—if dominance is ‘fragile’ as the neo-Schumpeteriansclaim, then the relevance of market shares diminishes considerably. AsAhlborn, Evans and Padilla put it: ‘imagine what Nokia’s market sharewould be two years from now if it left innovation to Ericsson, Siemens,Motorola or any of a dozen of firms with the technical capacity to createmobile phones with a greater number of functionalities: it would plungeirremediably’.49 However, it must be recalled that market shares are notthe exclusive measure of dominance.50 Barriers to entry, as will be shownbelow, are as important; consequently this criticism on its own is nottenable.

In the Microsoft case, the barrier to entry which the court found waslabelled an ‘applications barrier to entry’.51 Consumers want to buy anoperating system for which a large number of applications have alreadybeen written and developers of applications prefer to write for an operatingsystem that has a substantial consumer base—the upshot is that onceMicrosoft has such a large market share for its operating system, then appli-cations writers will prefer to write applications for Microsoft, further consolidating its dominance. This conclusion rests on the theory of indirectnetwork effects, although it is not a real network where consumers speak toeach other, but a virtual network, created by consumer demand.52 It is anindirect network effect because the utility of each user increases as the number of users increases because the more users, the more applications

32 Giorgio Monti

46 Although other factors will also come into play. These are set out in Faull and Nikpay aboven 45, paras 3.73–3.84.47 Eg Microsoft’s 95% share of the market for Intel-compatible operating systems. US vMicrosoft 84 F Supp 2d 9 (DDC 1999) (Findings of Fact).48 WorldCom/MCI above n 33, paras 104 ff.49 Ahlborn et al above n 19, 162.50 See eg Case 85/76 Hoffmann La Roche [1979] ECR 461; Case C–62/86 AKZO [1991] ECRI–3359.51 US v Microsoft above n 47, 19–24.52 Ibid, 20.

will be written for the operating system, and the greater the number ofapplications, the greater the utility of consumers. It leads to dominancebecause an application written for one operating system is not compatiblewith other systems.

The fact that Microsoft had created this network effect through superiormarketing and excellent product design does not remove the fact that thenetwork effect placed Microsoft in a dominant position. The Microsoftcourt defined entry barriers as ‘factors … that prevent new rivals fromtimely responding to an increase in price above the competitive level’.53 Asin all competition cases, barriers to entry were looked at using a short-runanalysis.54 This short-run analysis, coupled with a static approach to mar-ket definition has been criticised as inappropriate in new economy marketswhere it is said that long-run fears of technological paradigm shifts thatrender contemporary technology obsolete loom large in the minds of fragilemonopolists. As a result, neo-Schumpeterians have suggested several alter-native approaches to analysing dominance, which are reviewed below.

B. Neo-Schumpeterian Definitions of Dominance

Two alternative approaches have been suggested to modify the test of dom-inance to better reflect the realities of new economy markets. One authorhas suggested three points which should be borne in mind when decidingwhether a firm is dominant: (1) the existence of competing networks asopposed to simply competing products; (2) the history of innovation in theindustry, with special attention to the rate of innovation; (3) barriers toinnovation rather than barriers to entry.55 Competing networks can provide the same uses to consumers and should be included in the productmarket, a history of innovation in the market can serve to undermine the probative value of a large market share, and low barriers to innovation canallow new products to supplant existing standards, like CDs did for vinyldiscs or the VHS standard did for Betamax. A barrier to innovation couldbe defined as a factor which makes innovation more costly for the new


53 Above n 43, referring to S Pac Communications Co v AT&T 740 F 2d 980, 1001–2 (DC Cir 1984). The court expressly refused to decide whether a barrier to entry should beidentified when rivals face higher costs than the incumbent, or whether it exists when rivalsface the same costs as the incumbent. The court simply found that even adopting the narrowerdefinition of entry barriers, Microsoft did not face the same costs as new entrants because:(1) Microsoft did not face an incumbent dominant firm when it introduced its operatingsystems (although it did face a highly uncertain market and it did manage to create substantialconsumer demand for its product) and (2) the cost to Microsoft of an upgrade from sayWindows 95 to 98 was lower than the introduction by a new firm of an operating system.54 Approved by the Court of Appeal, above n 43.55 Note ‘Antitrust and the Information Age’ (2001) 114 Harvard Law Review 1623.

entrant than for the incumbent, for example if all research facilities areowned by the incumbent, or the unavailability of a distribution channel fornew products.56

The examination of competing networks is clearly of value, but applyingthis to the Microsoft case, it can be seen that the court did consider poten-tially competing networks, but rejected these as unlikely to pose a threat toMicrosoft. The argument that such potential networks should be includedrests upon the assumption that entry is speedy, but the court found that itwould only be in the long run that competing networks could enter.57 Thehistory of innovation, for example, past displacement of market leaders,58

is not all that useful, because the rate of innovation does not follow any setpattern in any industry. Barriers to innovation are useful only insofar asone is able to predict that the new product will supplant the existing incum-bent. This is a very speculative exercise. To find simply that there are anumber of products that could challenge the pre-eminence of the ‘PC plusoperating system’ paradigm, that innovation here is occurring at freneticspeeds, and that there is current investment in alternatives to Microsoft’soperating system, does not allow a judge to conclude that Microsoft willnot be dominant in the near future. Barriers to entry analysis is much lessspeculative—the court already knows that there is consumer demand forthe product and if another person can offer it, then the new entrant will besuccessful. The main weakness of this proposal is that it requires the courtsto speculate upon future events in the market. This need not make the sug-gestion worthless, but the facts show that Microsoft has been dominant forseveral product generations and no new product has been able to eat awayat its dominance for so many years. Even if we were to concede that itsdominance is now waning, this could affect the remedy that is chosen bythe courts, but not the conclusion as to dominance. Of course, the sugges-tion above might be more applicable in other markets where it is easier toprove that market innovations are regularly threatening the incumbent’sposition, but such regular threats can already be covered by the short-runbarrier to entry analysis.

Another way of measuring dominance in new economy markets was sug-gested by Richard Schmalensee at the Microsoft trial.59 In his view, thestructural approach is inappropriate in dynamic markets where marketboundaries are not well defined. In these markets, what is important is toassess whether a firm is fragile, hence market definition and market shares

34 Giorgio Monti

56 Ibid 1638.57 Above n 47, p 14 ff.58 R Schmalensee, ‘Antitrust Issues in Schumpeterian Industries’ (2000) 90 AmericanEconomic Review (Papers and Proceedings) 192, 194.59 Direct Testimony of RL Schmalensee, Direct Testimony of 3 January 1999 available at:<http://www.neramicrosoft.com> (visited 4 February 2002). See also Schmalensee above n 58.

are not as important. Instead, a behavioural approach should be adoptedwhich identifies: (1) constraints that limit the ability of firms to behave anti-competitively; (2) whether the accused firm acts like a firm with monopolypower.

The examination of constraints is similar but more expansive than thebarriers to entry analysis, considering past, present and future constraintsof market power. Applying this standard, Microsoft faced competition firstfrom its past behaviour (in that consumers who had bought earlier versionsof its operating system would have to be persuaded to buy the upgrade); inthe present, higher prices would increase the risks of piracy and lead tosome consumers to buy other goods; and as for the future, Schmalenseeidentified a number of firms that were Microsoft’s potential competitors forthe long run. On this basis, the structure of the market does not onlyinclude Intel compatible operating systems and on either the structural orbehavioural approach, Microsoft lacks monopoly power. In consideringMicrosoft’s behaviour, Schamlensee said that its prices were much lowerthan the monopoly prices that Microsoft could have charged if it were amonopolist. Moreover, in the context of Microsoft’s tactics to excludeNetscape, the relevant question is whether Microsoft would be able torecoup the costs of its predatory strategy in the long run, a possibility pre-cluded by the existence of many other potential competitors.60

This approach is also speculative in that by focusing on long-run possi-bilities defendants can always claim that there will be threats to dominance.Moreover, it would require a considerable change of direction for EC com-petition law. The Commission would have to examine the market in thelong run, whilst at present only short-run constraints are examined, and itwould have to amend the concept of abuse to require a finding that theabuse is likely to create a monopoly. This explicitly is not the case in preda-tory pricing in the EC,61 and is not a requirement for other types of abuseeither (for example refusal to supply). Furthermore, the approach does notfit within the methodological approach taken in dominance cases, eitherunder the Sherman Act or the EC Treaty, for it confuses the concepts ofabuse and dominance.

However attractive these neo-Schumpeterian alternatives may seem, theyshare two features which makes their adoption unlikely by competitionauthorities: they require great faith in time curing many anticompetitiverisks and consider that what is anticompetitive about dominance is only the ability to exploit consumers. These two perspectives contrast with theapproach to competition that is taken today: the principal concern is the


60 This is supported by Brooke Group v Williamson 509 US 209 (1993) which requires proofof the possibility of recoupment following predatory pricing.61 Case C–333/94P Tetra Pak 2 [1996] ECR I–5951 expressly refuses to include recoupment.

unlawful suppression of competitors and because this is the main concern,a short-run analysis of the market is more suitable. The fact that dominancemay be eroded five years from today by market forces will not stop antitrustaction if the authorities perceive that with some regulation dominance canbe eroded more quickly; the authorities will also take the view that by pre-venting a dominant firm from prolonging its dominance, they are promot-ing innovation. Consequently, the above suggestions are problematic in tworespects: they are difficult to put into practice (because of speculation aboutlong-term effects) and are inimical to the type of antitrust culture that existsat present.

4. ABUSE

The abusive behaviour which most worries the competition authorities innew economy markets is that exercised against potential competitors. Inthe EC context this has traditionally been the case, witness the formulationof abuse in one frequently cited passage:

[B]ehaviour … [which] has the effect of hindering the maintenance of the degreeof competition still existing in the market or the growth of that competition.62

The concern here is not with higher prices, but with exclusionary conductby the dominant firm. Below we review three types of exclusionary strate-gies which might be checked by Article 82: protecting dominance; refusalto co-operate with competitors; extension of dominance to new markets.63 Neo-Schumpeterians find it hard to work out why these exclusionary tactics are a problem: so long as these tactics exclude lessefficient competitors, consumers do not suffer and more efficient com-petitors will be able to enter the market because a paradigm shift willtransfer power to the holder of the new technology. As the three examplesof abuse below demonstrate, the neo-Schumpeterian view of the aim ofcompetition law is not accepted by the courts keen to protect marketaccess for competitors.

A. Protecting One’s Dominant Position

The most well known case representing this kind of abuse is Microsoft. TheUS courts found the company guilty of exclusionary conduct designed to

36 Giorgio Monti

62 Hoffmann La Roche above n 50, para 91.63 Space precludes a comprehensive review, but it is hoped that the examples selected offer arepresentative survey of the types of legal and economic issues that arise.

maintain its monopoly by preventing the distribution of potential competitors’products. This has been achieved inter alia by attempting to prevent theNetscape browser from gaining a significant foothold in the browser mar-ket. Microsoft’s reason for preventing the establishment of a competingbrowser was not so much so that it could dominate the browser market withInternet Explorer, but because a browser is another means of access to appli-cations. If applications are written to work with a particular browser, then itdoes not matter which operating system the consumer is using, so long asshe has the browser in question. If Netscape were able to circulate enoughcopies of its browser to attract applications designers to write applicationsfor it, then the Netscape browser could provide a competitive threat to theMicrosoft operating system monopoly, because it would not matter whichoperating system people bought so long as they could run Netscape.


Figure 2.3

Before the entry of browsers, Microsoft’s applications barrier to entrymade it hard for a competing operating system to enter the operating systems market. If Netscape Browser gains a significant foothold, then itdoes not matter what operating system consumers have, they will accessapplications through the browser. This would make the operating systemsbrowser more competitive.

Microsoft achieved its aim by foreclosing the most cost effective ways bywhich Netscape might have been able to place its browser onto many con-sumers’ PCs and promoting its own browser, Internet Explorer. The twomost cost effective ways of promoting a browser are to have it pre-installedon a PC or to have Internet access providers promote a browser. Microsoftused contractual and technological means to ensure that OriginalEquipment Manufacturers (OEMs) pre-installed Internet Explorer to theexclusion of other browsers and entered into agreements with Internetaccess providers whereby many promoted Internet Explorer on an exclu-sive basis. The upshot is that Netscape was prevented from entering thebrowser market using the most cost effective ways because of Microsoft’sactions,64 thereby foreclosing any possibility for Netscape to challenge theOperating System monopoly. On these facts the same conclusion wouldhave been reached under EC competition law.65

One additional circumstance in this kind of scenario, noted by the ECCommission in Tetra Pak 2 and the court in Microsoft, was that attemptsby a dominant firm to bind purchasers to its products were aggravated ina market characterised by rapid technological development: this slows theevolution of competing technologies, both by existing competitors and bypotential investors in new technologies.66 However, it might be arguedthat if Netscape’s strategy had been successful, then everyone would wishto have the Netscape browser in order to access the largest range of appli-cations. On this basis, it might be legitimately argued that the welfareeffects of shifting from ‘operating system dominance’ to ‘browser domi-nance’ are non-existent because one is simply replacing one dominant firmwith another. Moreover, it may be argued that it is more efficient to havean operating system with an integrated browser. However, recall that com-petition in network industries is territorial—the undertaking is competingfor the market, not in the market. Consequently, if Netscape is allowed tocompete for the network, this will give incentives to other firms to com-pete Netscape’s future dominance away, or conceivably a new operatingsystem could be devised that runs applications more efficiently than a

38 Giorgio Monti

64 Microsoft also carried out similar actions to exclude Java but the legal reasoning is analo-gous to that deployed when considering Netscape and will not be reviewed here.65 See Tetra Pak 2 above n 61.66 Tetra Pak 2 [1992] OJ L/72/1; Microsoft above n 47, 111–12.

browser and the market might then shift back to operating systems. Theanticompetitive nature of the acts by Microsoft his not only in the factthat by preventing the emergence of a successful browser it protects itsdominance in the operating system market, but also that such tacticsthwart innovation by other competitors. One possible market outcome ifNetscape had been allowed to enter the market unobstructed could havebeen the creation of two competing networks—like in video games wheremultiple standards exist.

Microsoft’s licensing practices in the EC during the early 1990s can alsobe seen as attempts to safeguard dominance: Microsoft forced OEMs topay royalties for Windows on every PC sold, regardless of whether or notMicrosoft software was loaded. As many consumers wanted Windowsloaded on the machine, OEMs were forced to deal with Microsoft, but itslicensing practices discouraged OEMs to install competing software pack-ages. Microsoft subsequently undertook to amend its licensing practices.67

It has been argued that it is important to prevent these tactics because theyalso foreclose innovation.68 Thus, a neo-structuralist approach does notdeny the dynamism of markets, but is concerned about maintainingdynamism in the market.

The neo-structuralism that prevailed in Microsoft is also evidenced bythe Court of Appeal’s approach to the question of causation. Here, theplaintiff was unable to prove that but for Microsoft’s conduct the operatingsystem market would have been competitive, but held that the causationinquiry should simply address whether (1) the exclusion of new threats is atype of conduct reasonably capable of contributing to monopoly and (2) whether Navigator and Java constituted such a nascent threat.Significantly for the purposes of this discussion, the court said: ‘it would beinimical to the purpose of the Sherman Act to allow monopolists free reignto squash nascent, albeit unproven, competitors at will—particularly inindustries marked by rapid technological advance and frequent paradigmshifts’.69 On this view, the court believes that the dynamic nature of indus-try is a reason for higher, not lower antitrust scrutiny because the more dan-gerous innovations are for incumbent, the greater the incentive to eliminatecompetitors. Thus the suggestion that the courts should instead requireproof that the practices of the dominant firm are likely to be successful inconsolidating dominance in order to establish liability (as Schmalenseewould suggest) is inappropriate because it would not allow competitionlaw to safeguard innovation.


67 Undertaking of 15 July 1994 (see Commission’s Press Release of 17 July 1994).68 M Dolmans, ‘Restrictions on Innovation: The EU Antitrust Approach’ (1997–98) 66Antitrust Law Journal 455.69 Microsoft above n 43. The same low standard of causation is applied in EC Law: see Faulland Nikpay above n 45, paras 3.131–134.

In sum, the neo-structuralist approach adopted by the authorities isdesigned to ensure participation by new entrants and to safeguard thepotential for future innovation rather than relying on innovation to demolishthe dominant firm. The Microsoft decision has affected Microsoft’s stancein its negotiations with Kodak. Microsoft’s new operating system,Microsoft XP, contains digital photography software and Kodak (who alsodeveloped such software) complained that Microsoft would give unfairpreference to its software at the expense of Kodak. Microsoft has nowagreed that when consumers plug in their digital cameras they will be ableto select a variety of services (including Kodak’s) and will not be led tochoose Microsoft’s software.70

Critics of this analysis posit that the line distinguishing anticompetitiveconduct and aggressive competition has been drawn inaccurately. Klein forexample suggests that the better way to test whether Microsoft’s practiceswere abusive is to examine their effect on consumers and whether the prac-tices were efficient.71 On this analysis he suggests that both Netscape andMicrosoft were able to compete for distribution of their browsers throughInternet access providers but that Netscape failed to keep up withMicrosoft’s efforts. Moreover he notes that Microsoft’s practices merelyraised Netscape’s distribution costs but did not make entry prohibitivelyexpensive. Combined with the fact that consumers benefited from a moreadvanced product manufactured by Microsoft, the tactics used should notbe declared anticompetitive. In this light, competition law seems to havebeen used to protect a competitor. This again illustrates the differencesbetween a neo-structuralist and a neo-Schumpeterian view of new economymarkets: according to the former, there is competition and innovation ifmany firms are able to participate actively in the marketplace, according tothe latter, there is competition so long as the incumbent feels sufficientlythreatened by potential competition that it takes steps to innovate andincrease the quality of its product so as to keep competitors out. The factthat innovation is carried out by Microsoft and not by other firms is not ofconcern to a neo-Schumpeterian, but is of concern to those adopting a neo-structural approach.72 A neo-structuralist interpretation of the competitionlaws seems therefore to embrace considerations of distributive justice.

40 Giorgio Monti

70 M Pesola, ‘Kodak Persuades Microsoft on XP Photo Changes’ <www.ft.com> (13 August2001). Although there are still controversies. See P Spiegel, ‘Microsoft gives in to Kodak’<www.ft.com> (14 August 2001).71 B Klein, ‘Did Microsoft Engage in Anticompetitive Exclusionary Behaviour?’ (2001) 46Antitrust Bulletin 71, 108.72 Eg FM Fisher,’Innovation and Monopoly Leveraging’ in J Ellig (ed) Dynamic Competitionand Public Policy (Cambridge, Cambridge University Press, 2000) 1156–57. A related criticismof the neo-structuralist analysis of Microsoft is the emphasis they place on Microsoft’s intentionto damage competitors; however intention is generally irrelevant as a means to establish aninfringement of competition law.

B. Obligations to Co-operate with Competitors

The presence of network effects in many new economy industries can leadto claims that the holder of a dominant network has an ‘essential facility’which gives rise to obligations to make this available to other market play-ers. For example, the Commission hypothesised that a single firm domi-nating the top level of Internet connectivity would control an essentialfacility to which all lower level ISPs would have to connect;73 a duty tocollaborate with lower level ISPs can be imposed by analogy with the caselaw on refusals to supply,74 and is comparable to the undertaking offeredby IBM to the Commission in 1984 where it agreed to provide advancetechnical information to manufacturers of computer peripheral productsto allow them to manufacture competing peripheral equipment.75 TheCommission has identified a number of other bottlenecks where the dom-inant incumbent must co-operate.76 Similarly, in an indirect network scenario, the dominant supplier of on-line music which combines thedistributional might of AOL’s ‘on-line outlet’ will become a compulsorypartner for other music companies. Moreover, if AOL develops a techno-logical standard for downloading music it could impose this standard onothers, and ‘because of its control over the relevant technology the newentity would be in a position to control downloadable music and stream-ing over the Internet’.77 Here too, liability may be imposed for a refusal toco-operate with other firms wishing to participate in markets that requireaccess to the AOL network.78

The duty of dominant firms to co-operate with other market partici-pants has been set out in a controversial line of cases under the headings‘essential facilities’ or ‘refusal to supply’. Even before the advent of thenew economy, this jurisprudence was criticised from three perspectives:as an intrusion into the commercial freedom of undertakings; as a tax on


73 WorldCom/MCI above n 33.74 The refusal to deal and the ‘essential facilities’ case law address the same point and no dis-tinction need be drawn between them for the purposes of this discussion. The differencebetween the two concepts was identified by AG Jacobs: in essential facilities cases a duty tosupply arises because of the facility held and is not related to any other action by the dominantfirm (eg discontinuation of supplies to existing customers). Case C–7/97 Oscar Bronner [1998]ECR I–7791 para 50. For a recent review see B Doherty, ‘Just What Are Essential Facilities?’(2001) 38 Common Market Law Review 397.75 Fourteenth Report on Competition Policy (1984) paras 94–95.76 Eg telecoms markets in Germany and France (Atlas [1996] OJ L/239/23); transatlantic cables(BT/MCI II [1997] OJ L/336/1); technical and administrative services in the pay TV market(MSG Media Services [1994] OJ L/364/1); digital pay TV markets (Bertelsmann/Kirch/Premiere above n 31).77 AOL/Time Warner above n 10, para 56.78 On analogy with US v Associated Press 326 US 1 (1945). The same principle underpins thecase law on Art 86 where the ECJ condemns the grant of two conflicting rights that create aconflict of interest (eg Case C–260/89 ERT [1991] ECR I–2925).

firms that innovate (for their incentive to innovate is dented by the factthat they are subsequently forced to share their discoveries); and as oftenbeing concerned about protecting competitors rather than competition.79

Consequently, attempts have been made to limit its scope. Three pointsarise: first, which market players have a right to demand access? Second,what property is so essential that it must be shared? Third, can a duty toshare be imposed when the property in question is protected by intellectualproperty rights?

First, in terms of who has the right to seek access, EC competition lawplaces no restriction on who can demand access so long as they can estab-lish that the access to the dominant firm’s property is essential for them toenter the market. This is justified by reliance on the principle that theabuse of a dominant position does not have to be in a market where thedominant firm is present.80 In fact, Article 82(b) that identifies as an abuse‘limiting production, markets or technical development to the prejudice ofconsumers’ makes no reference as to whether the dominant firm needs tobe present on the market where the anti competitive effects occur.81 Incontrast, in the Intel case the US courts have placed a limit on the personsentitled to seek access to essential facilities. In this case Intergraph, a man-ufacturer of workstations for Intel-compatible computers, required accessto technical information about new Intel chips so that its workstationcould operate jointly with the Intel chip. Any refusal by Intel to communi-cate technical information in advance of the release of the new chip couldseriously undermine Intergraph’s position because it would not be able todevelop compatible software until after the release of the Intel chip. Intelarguably holds a dominant position in the market for microprocessors: onthe FTC’s analysis it has an 80 per cent share of the market and high barri-ers to entry (eg large sunk costs of design and manufacture, economies ofscale, the need of a new entrant to attract the support of software developersand reputation).82 When Intel refused to give advance technical informationto Intergraph, the Court of Appeals found that the essential facilitiesdoctrine could not apply because

no court has taken it beyond the situation of competition with the controllerof the facility, whether the competition is in the field of the facility itself or ina vertically related market that is controlled by the facility. That is, there must

42 Giorgio Monti

79 See AG Jacobs in Bronner above n 74, paras 53–62.80 ‘[T]here is no doubt … that an abuse of a dominant position on one market may be censuredbecause of effects which it produces on another market’. Case T–128/98 Aéroports de Paris vCommission [2000] ECR II–3929 para 164 (affirmed by the ECJ: Case C–82/01, judgment of24 October 2002, not yet reported).81 It has been suggested that the same result will not occur under section 2 because it is onlyconcerned with monopolisation and not with abuses of monopoly power.82 FTC Analysis of Proposed Consent Order to Aid Pubic Comment 17 April 2003 (availableat <www.ftc.gov>).

be a market in which plaintiff and defendant compete, such that the monopolistextends its monopoly to the downstream market by refusing access to thefacility it controls.83

On the facts, Intel was not competing with Intergraph and no duty to supplywas found. However, it is submitted that competition between plaintiff anddefendant should not be a sine qua non for liability: the same anti-competitiveeffects can arise for example, if the dominant firm only supplies to oneundertaking downstream, then that one undertaking will hold a dominantposition and a refusal to supply another creates dominance downstream.Another type of anticompetitive effect could arise if some companiesdevelop a totally new type of software that Intel has objections to—if Intelmakes it impossible for the market for this new software to be created, thenthe market does not evolve. In this context there may be a loss of consumerwelfare and possibly a diminution of innovation. In addition, if the refusalto supply one firm is as a means to another end,84 then again, the courtsshould be ready to see an anticompetitive effect.85

Second, the duty to supply is only required when the facility held is trulyessential. This requires adequate market definition and an appropriate defi-nition of when facilities are essential. We have noted above that sometimesthe market definition appears to be too narrow, which may overstatewhether a facility is truly essential by excluding potential substitutes, and indefining ‘essential’ the court in Bronner held that a refusal to supply maybe unlawful only if (1) it is likely to eliminate all competition in the relevantmarket on the part of the undertaking requesting the service, and (2) theservice is ‘indispensable for the requesting undertaking’s business’.86 Thesecond limb is appropriately narrow, indeed the ECJ went on to say that tocharacterise a facility as indispensable one would have at least to establishthat it is not economically viable to duplicate it, either alone or in collabo-ration with others.87 However, the first limb is worrying because it doesnot require that competition in the market be eliminated—only that compe-tition from the requesting undertaking be eliminated.88 However, the


83 Intergraph v Intel 195 F 3d 1346 (Fed Cir 1999).84 Intel had a longstanding intellectual property dispute with Intergraph that led to Intel’s lackof co-operation. The FTC forbade Intel from using the IP dispute as a justification for refusingadvance access to relevant information (see FTC above n 82).85 Analogous with the refusal to resume supplies to disobedient distributors in United Brandsabove n 44.86 Bronner above n 74, para 41. But against AG Jacobs at para 65 holding that the dominantfirm must also have dominance on the related market.87 Above n 74, para 46.88 Against Doherty above n 74, 427, who says that if there are other competitors who are noteliminated by the refusal, this just means that the facility held by the dominant firm is notessential because there are alternative suppliers. But if this were so then the two requirementsare unnecessary, while para 41 of Bronner clearly envisages two conditions.

elimination of a competitor does not mean that the market is thenmonopolised; what one should look for is the effect on competition in general. The Advocate-General’s reasoning in Bronner (comparable to theCFI’s definition in European Night Services89) is preferable, requiring proofthat it is ‘extremely difficult not merely for the undertaking demandingaccess but for any other undertaking to compete’.90 Confined to such limits,and combined with an appropriate market definition, refusals to co-operatewill only breach Article 82 when there is a tangible effect on competition.This limitation would also bring EC law more in line with US antitrust.

The third issue, which is of particular importance in new economy mar-kets, is whether the duty to co-operate can extend to situations when thedefendant’s ‘essential facility’ is protected by intellectual property rights.Insofar as EC competition law is concerned, this is not a major issue: thejurisprudence is such that the exercise of intellectual property rights fallsunder the scope of EC law and the court in Magill had no qualms in find-ing that when the circumstances warrant it, an order to license IP rightscould be made.91 However the case law is not particularly clear as towhether the criteria for imposing a duty to share IP rights are stricter thanthose for other property rights. The recent decision in Bronner failed toresolve this point conclusively. The argument is still open that for a licenceof IP rights, the claimant has to fulfil the Bronner requirements and alsoestablish that the refusal prevents the appearance of a new product forwhich there is consumer demand and that the refusal would exclude allcompetition in a secondary market.92 It seems difficult to justify a differ-ent treatment for IP rights. The policy of the ‘essential facilities’ doctrine isto prevent foreclosure of markets, and the ECJ does allow for a defence of‘objective justification’ which, as will be explored in the following section,can be used by defendants to justify the non-imposition of a duty to sharewhen this would have the effect of stifling innovation. A similar contro-versy has affected the application of the essential facilities doctrine to IPrights in the United States. While some writers take the view that the doc-trine should apply to IP rights without any differentiations, some courts

44 Giorgio Monti

89 ‘[A] product or service cannot be considered necessary or essential unless there is no real orpotential substitute’ Joined Cases T–374/94, T–375/94, T–384/94 and T–388/94, ENS [1998]ECR II–3141 para 208.90 Above n 74, para 66 (emphasis added).91 Joined Cases C 241 and 242/91 P RTE and ITP v Commission [1995] ECR I–743.92 I justify this reading by reference to para 41 of Bronner where the court considered whether‘the case law on the exercise of an intellectual property right [is] applicable to the exercise ofany property right’. This indicates that there may be two separate tests. Case T–504/93 TiercéLadbroke SA v Commission [1997] ECR II–923 also suggests two tests at para 131: either theproduct is essential for the exercise of the activity in question or it is a new product whoseintroduction might be prevented despite consumer demand. For a view that there is a single,unified test set out in Bronner see: F Fine, ‘NDC/IMS: A Logical Application of the EssentialFacilities Doctrine’ (2002) 23 European Competition Law Review 457.

have restricted the application of the doctrine to cases where the patentholder is seeking to use his rights to extend his monopoly into another market.93

All this suggests that while the policy arguments in favour of imposingan obligation to co-operate with competitors are based upon the desire toprevent market foreclosure, the case law is still incoherent—at times thecase law goes too far and protects individual competitors and not theprocess of competition (eg the first limb of the Bronner test); while at timesthe law is too restrictive, showing a tendency to limit the scope of obliga-tion only to cases where the holder seeks to extend the dominance into newmarkets. The reason for the unsettled state of the law is probably due toconcerns that by imposing too extensive an obligation to share, innovationwould be stifled. Here, some of the neo-Schumpeterian concerns appear tohave influenced the decisions of the court. However it is submitted thatthese concerns are best addressed more openly by seeking a mechanism tobalance the competing demands of market access and innovation. Section 5charts a possible approach.

C. Extending Dominance

A related strategy by a dominant firm is to use ‘its monopoly power in onemarket to gain a competitive advantage in another’.94 Three examples ofleveraging in the new economy case law have been canvassed. First, such apossibility was noted in the Internet markets by the Department of Justice:a dominant top-tier ISP would be able to enter the market for new internetservices and exclude other ISPs—for example by developing a proprietaryprotocol for voice over the Internet, which only its subscribers can use. Thistactic has anti-competitive effects in two markets: first it will further under-mine the position of other ISPs and will drive more consumers to join thedominant firm;95 second, it will also have an anti-competitive effect in themarket for voice transmission over the Internet because there will only beone firm supplying it.

Second, the Commission’s current investigation of Microsoft also restsupon the theory of leveraging. Microsoft is suspected of refusing to disclose


93 Contrast LA Sullivan and WS Grimes, The Law of Antitrust (St Paul, MN, WestPublishing, 2000) 848–53 with the ruling in CSU and Others v Xerox 203 F 3d 1322 (Fed Cir2000), criticised for failing to strike a balance between IP and competition law by R Pitofsky‘Challenges of the New Economy: Issues at the Intersection of Antitrust and IntellectualProperty’ 68 Antitrust Law Journal 913, 2001.94 Berkey Photo v Eastman Kodak 603 F 2d 263, 276 (2nd Cir 1979)—no requirement ofdominance in second market, whilst some circuits require acquisition of monopoly in secondmarket (eg Alaska Airlines v United Airlines 948 F 2d 536 (9th Cir 1991) ).95 DOJ WorldCom/Sprint above n 33, para 46.

information about its operating system (where it has a dominant position)to developers of server software, preventing the possibility of developingsoftware that works on a computer with a Windows operating system. Theaccusation is that Microsoft is attempting to extend its dominance in themarket for servers as well. The press release by the Commission notes thatMicrosoft did not refuse to disclose such information to everyone, but thatit disclosed it on a discriminatory basis, according to a ‘friend–enemy’scheme.96

Third, using the concept of ‘tipping’, the DOJ considered that a domi-nant top-level ISP could try to drive the customers of competing top-levelISPs firms away from them by lowering the quality of the services offeredto competing ISPs.97 This would make users of lower level ISPs migrate tothe top-level, where connectivity is better. Top-level ISPs may also raise theprice of transit agreements, costs which lower level ISPs would pass on toconsumers who would then switch to the top-level ISP who would be offer-ing lower rates.98 As the Commission pointed out, the top-level ISP couldpick off competitors one by one and take over the market gradually.99 Thistheory of liability might apply by analogy with Aspen Skiing—the owner ofcertain ski slopes who refused to continue to co-operate with the neigh-bouring owner of fewer ski slopes in the expectation that skiers would pre-fer to buy its ski pass for a greater number of slopes.100 This strategy madeit more difficult for the smaller operator to attract skiers.

None of these scenarios may merit the imposition of liability. As for thefirst, if a company creates a new service; which consumers want, it wouldbe inefficient to penalise the provision of that service; furthermore, theexclusion of others from that market should be seen as the reward thatthe firm deserves for creating the new service. On this basis the analysisof the DOJ is flawed for failing to take into account innovation. The reasonfor heightened scrutiny is the theory of network effects, which might sug-gest that a dominant network holder can very easily become dominant in arelated market: in the example of voice over the internet, if two ISPsdevelop proprietary technology but one has a larger network, that firm’stechnology will prevail. This prevents competition over technological standards. However, if a dominant firm wins the race to produce the nextgeneration of products because of network effects and with a somewhatinferior product, this is not necessarily an anticompetitive outcome.101

In fact, the dominant firm has not done anything to harm competition.

46 Giorgio Monti

96 EC Commission Press Release, 3 August 2000.97 DOJ WorldCom/Sprint above n 33, para 35.98 Ibid paras 40–41.99 MCI/WorldCom above n 33, paras 120–22.

100 Aspen Skiing v Aspen Highland Skiing 472 US 585 (1985).101 Teece and Coleman above n 16, 816.

This makes the DOJ’s analysis hard to sustain.102 Moreover, it is not clearwhat competition law can do to remedy this situation.

Similarly, the Commission’s evidence against Microsoft does not (yet)make it clear whether there has been harm to competition in the server mar-ket or whether there has only been harm to Sun Microsystems (one of thefirms which did not receive information about the operating system soft-ware). In the US for example, the Federal Court of Appeals takes the viewthat leveraging requires an anticompetitive effect in the second market.103

This requirement is necessary—the sheer fact that a dominant firm canenter another segment of the market should not in itself be objectionableunless consumers suffer. The third scenario is also one which seems to pro-tect competitors rather than competition—if the dominant ISP wanted toincrease its profits it would be able to raise the transit charges—thus verti-cal integration would not raise prices or reduce output any more than if thedominant firm raised transit charges.104

In these cases, the neo-structuralist approach leads to the protection ofcompetitors without any apparent benefit to competition in general as theanticompetitive effects are assumed to result from the exclusion of rivals.105

It may be doubted whether there will be many leveraging cases which falloutside the scope of the essential facilities doctrine. To date the antitrustauthorities prefer to prevent the possibilities of leveraging through mergercontrol, by ensuring that the market structure does not facilitate the exten-sion of monopoly power.

5. AN INNOVATION DEFENCE?

The review of how the concept of abuse might be deployed in the new econ-omy demonstrates the pre-eminence of neo-structural analysis: dominant


102 In California Computer Products v IBM 613 F 2d 727 (9th Cir 1979) the court refused toimpose liability on IBM when it introduced a new line of computers that integrated the centralprocessing and memory units, causing disk drives losses to manufacturers.103 Intel above n 83.104 The analysis borrows from Hovenkamp’s sceptical treatment of leveraging claims inFederal Antitrust Policy (St Paul, MN, West Publishing, 1994) 284. The decision in Aspen(above n 100) may be justified because the refusal to co-operate reduced consumer welfare byremoving an existing product (a four mountain ski pass) from the market. However, my analy-sis of the third scenario might be challenged by considering the facts of the Kodak case(Eastman Kodak Co v Image Technical Services 112 S Ct 2072 (1992) ) where after denyingindependent repairers access to its spare parts, Kodak raised the price of the spare parts it sup-plied directly to consumers.105 In the US this follows from the reasoning in Aspen Skiing (above n 100) and Kodak (aboven 104) as summarised in Data General v Grunman Sys Support Corp 36 F 3d 1147, 1183(1st Cir 1994): ‘a monopolist’s unilateral refusal to deal with its competitors (as long as therefusal harms the competitive process) may constitute prima facie evidence of exclusionaryconduct in the context of a Section 2 claim’.

firms are not allowed to safeguard their positions by unlawfully excludingother market players, nor may they extend dominance to other markets andmust even co-operate to support other market participants.106 It has evenbeen suggested that in new economy industries the duty to co-operate willencourage innovation by fringe firms who will be confident of being able tointerconnect with the goods or services of the dominant firm.107

However, as noted above in some cases the authorities at times are protecting competitors and not competition, resulting in an unjustifiablyaggressive policy. More fundamentally, from a neo-Schumpeterian per-spective, all impositions of duties to co-operate, and all restrictions onbusiness expansion are suspect, for two reasons: first, even if it appearsthat competition is being suffocated, dominance may be overthrownquickly (the fragile monopoly point again); second, the imposition of co-operative obligations dents the incentive of undertakings (dominant andnon-dominant) to develop new products knowing that they will be forcedto share the fruits of their investment. To a certain extent, the restrictiveinterpretation of the essential facilities doctrine in Intel may be a responseto these concerns.

The first objection can be countered by setting a strict definition ofwhether co-operation with the dominant firm is necessary.108 If technicalinnovations can quickly surmount the entry barrier created by the networkeffects, then co-operation is unnecessary. The second objection, about elim-inating the incentives to innovate is an important one: too expansive a dutyto co-operate and a too far-reaching restriction on extending one’s marketto new products can act as a tax on innovation and retard future develop-ments. To date this concern has not been addressed fully—while intellectualproperty rights are offered as a reward for innovative activity by the incum-bent, there is no rule of law that balances the risk that a dominant firm mayforeclose market access with the potential benefits which might accrue tosociety if the dominant firm is left unregulated.

One possible way of taking into account the pro and anticompetitiveeffects of practices carried out by dominant firms in the new economy is bydrawing on Teece and Coleman’s suggestion that before imposing liabilityfor breach of competition law, there should be proof that liability will notimpede aggregate innovation in the industry.109 This would ensure dynamicefficiency is not killed off by the enforcement of the antitrust laws. In what

48 Giorgio Monti

106 In certain contexts access to essential facilities is regulated by statute (eg Art 9 of Dir 97/33[1997] OJ L/199/32; s 251(c) of the Telecommunications Act 1996).107 JB Baker, ‘Promoting Innovation Competition Through the Aspen/Kodak Rule’ (1999) 7(3)George Mason Law Review 495.108 As noted above the ECJ’s definition is still too wide.109 Teece and Coleman above n 16, 838.

follows, I consider two questions: first, how an ‘efficiency defence’ mightwork; and second, whether it would be juridically acceptable within theframework of Article 82.110

The way the innovation defence would operate is to require the decisionmaker to consider whether the imposition of liability or an obligation toco-operate with competitors would have the effect of stifling innovationand whether this effect would outweigh the foreclosure effects of theanticompetitive activity. In considering whether innovation would be sti-fled, one would have to consider both innovation on the part of thedefendant, and also more generally how other firms might be affected bythe judgment—law after all is an incentive mechanism and the potentialrepercussions on imposing liability in one instance will be felt by others ina similar position to the defendant.111 There are some easy cases wherethe innovation defence would fail, for example Magill. As AG Jacobsobserved, the copyright protection afforded to the information in Magillwas hard to justify ‘in terms of rewarding or providing an incentive forcreative effort’. Here it would be impossible for the defendant to justify arefusal to co-operate with competitors on the basis that it would stifleinnovation. Marginal cases will be more difficult to resolve as currenteconomic theory has no suggestion for how to balance the incentives toinvest with the costs of abuse of dominance.112 However, a precise balanceis unnecessary and a qualitative analysis will suffice—for example, apply-ing the innovation defence to the Intel scenario, Intel’s non participationin Intergraph’s market means that granting the latter access would notdiminish Intel’s economic incentives to develop new products; in fact, byallowing others to develop complementary products, demands for Intelgoods should rise.

This kind of approach will be criticised as being too imprecise to beworkable, but it does address the need to ensure that competition and inno-vation are both safeguarded. Moreover, this method can be justified onthree grounds. First, the lack of a precise cost benefit analysis is not a fatalweakness (for example we are quite happy to tolerate this in the context ofthe duration of an IP right: while IP rights are designed as an incentive toinnovation,113 the level of protection is not set on a case by case basis (as itshould be to provide perfect incentives) but there is a standard level available


110 For a formal model along these lines which considers the balance between static anddynamic efficiency at the remedy stage, see Shelanski and Sidak above n 9.111 This point is one which is recognised in English common law. See the discussion of themajority and the dissenting Law Lords in Donoghue v Stevenson [1932] AC 562 as to thepotential impact of imposing negligence liability on manufacturers.112 For a demonstration of how malleable the arguments are, see RM Brunell, ‘Appropriabilityin Antitrust: How Much is Enough?’ (2001) 69 Antitrust Law Journal 1.113 R Posner, Economic Analysis of Law 5th edn (New York, Aspen, 1998) 43.

whether you patent a new safety mechanism that saves thousands of livesor a new toy for toddlers). Second, it is more legitimate to try to ensure thatcompetition law is applied so as to maximise competition and innovationby trying to measure the potential effects of imposing liability than to drawarbitrary lines to distinguish the lawful from the unlawful, as is currentlythe case in the field of essential facilities. Third, the type of imprecise bal-ancing process recommended above is not uncommon in legal reasoning: itoccurs also in applying the exemption in Article 81(3) EC and the efficiencydefence in merger cases;114 and it is, like common law reasoning, partlysubjective and impressionistic.115 Through pragmatic decision making areliable set of precedents will allow for the evolution of imperfect but work-able criteria for assessing the consequences of imposing a duty to share ondynamic efficiency.

Juridically, the innovation defence could be used as an additional objec-tive justification under Article 82. This places the burden of proof on thedefendant, who must demonstrate that his abusive conduct is necessary andproportionate to preserve incentives to innovate, and that this outweighsthe potential reduction in competition.116 The value of innovation is recog-nised in Article 157 EC and Article 82 should be interpreted purposively soas to ensure that the aims of the Treaty are met. This defence will be able tocounter the aggressive use of the essential facilities doctrine as well ascounter leveraging claims.

6. REMEDIES

Remedies must ‘unfetter a market from anticompetitive conduct’.117

Achieving this in new economy markets is complex. Usually theCommission issues a cease and desist order plus a fine.118 The order mayput at an end the enjoined practice, but does not prevent the deployment ofother, equally anticompetitive practices119 and the level of fines set by the

50 Giorgio Monti

114 On the basis of Art 2(1)(b) of Reg 4064/89 as amended by Reg 1310/97. Consolidated textat [1997] 5 CMLR 387. See eg Bertelsmann/Kirch/Premiere [1999] OJ L/53/1 paras 119–22.115 It is not unlike consequentialist reasoning by the English courts. See N McCormick, LegalReasoning and Legal Theory rev edn (Oxford, Oxford University Press, 1994) chs 5 and 6, esp101–16. Space precludes further discussion at this time, but it is submitted that competitionlaw rules cannot be derived purely from economic tests which are absolutely true or accurate—consequently at the margins where the controversy is significant, the decision makerhas to rely upon a subjective perspective and rest the decision upon settled legal principles.116 This is contrary to the suggestion by Teece and Coleman (above n 16) for whom the bur-den of proof should be on the competition authority.117 Ford Motor Co v US 405 US 562, 577 (1972).118 See Arts 3(1) and 15 of Reg 17/62 OJ Sp Ed [1962], no 204/62.119 Eg the legal wranglings over Microsoft’s 1995 consent decree: US v Microsoft 147 F 3d 935(DC Cir 1998).

EC may not deter many undertakings,120 although the possibility of trebledamages in the US may offer adequate deterrence. Moreover, a fine willoften not be the most direct way of remedying the anticompetitive situation.

Consequently, the Commission will often require that the dominant firmact in a specific way (eg licensing IP rights or entering into collaborativerelations with undertakings). Such orders, like the recent interim measuresin IMS Health,121 or the decision in Aéroports de Paris,122 require the dom-inant firm to co-operate with other undertakings in a way which is non-discriminatory and commercially reasonable. The parties fix all terms byagreement, but in cases of disagreement (eg as to the level of royalties) anindependent expert will be appointed to fix the terms of the agreement.These positive obligations, provided they can be implemented in a timelyfashion,123 are a more effective and direct way of preventing future anti-competitive conduct.

Perhaps the most eagerly anticipated remedy requiring co-operation isthe recently concluded Microsoft consent decree. In brief, this obligesMicrosoft to ensure that competitors have access to the market (for exam-ple preventing Microsoft from forcing Original Equipment Manufacturers(OEM) to exclude competing products), and to disclose information to aidinteroperability between its products and those of competitors.124

According to the judge the latter was in the public interest because ‘suchdisclosure has the potential to increase the ability of competing middlewareto threaten Microsoft’s Operating Systems monopoly’.125 In addition theenforcement order provides for the setting up of a committee to enforce thefinal judgment, aided by a technical committee. The judge praised this orderfor being ‘forward looking’ and designed to eliminate the effects ofMicrosoft’s illegal conduct. However, even though the decree will only lastfor five years, it seems as if this kind of forward looking enforcement mightbe best left to a permanently appointed regulator who can intervene whennecessary. Competition law remedies seem ill suited to guide the market ina forward looking manner.

In complex network markets, one issue which has been of concernto the Commission in merger cases is the separation of complementaryfunctions—the owners of content are not allowed to own the means of


120 Department of Trade and Industry, A World Class Competition Regime, July 2001, Cm 5233 Box 7.3.121 Press Release IP/01/1232, 3 July 2001.122 Above n 80.123 The risk of delay in new economy markets is particularly problematic.124 US v Microsoft, final judgment of 12 November 2002 (available at <www.usdoj.gov>). 125 US v Microsoft, US District Court for District Columbia, 1 November 2002 (Kollar-Kotelly J)(available at <www.dcd.uscourts.gov>).

distribution of that content. The reasoning to justify this concern iseither: (1) if a firm owns two segments of the market, this raises barriersto entry for newcomers who would also have to own both segments; (2)one of the segments may be an essential facility, in which case the holderhas an inherent interest in favouring its content and will discriminateagainst others.126 In merger cases, divestiture is not too controversial aremedy because the parties will agree to this or refuse to merge. TheCommission has devoted considerable efforts to prevent vertical integrationin network sectors like media and telecommunications using the mergerregulation and Article 81(3) EC127 and has also imposed obligations onundertakings that are vertically integrated to ensure non-discriminatoryaccess to other market participants.128 In the merger cases in media andtelecommunications, the Commission often states that preventing suchvertical consolidation is intended to allow the new markets to grow on acompetitive basis.129

More problematic is divestiture ex post, when an undertaking is accusedof abusing its dominant position. In US law divestitures have been orderedinfrequently,130 while the EC has only recently obtained the power todivest.131 In contemplating divestiture, two matters should be grasped:when this remedy should be imposed and how the firm should be divested.Only the first question will be addressed here.132 As a matter of lawdivestiture must be necessary to restore an acceptable level of competition.Consequently, the remedy should be imposed only if there is no less restric-tive remedy that would achieve broadly similar results. If a less restrictiveremedy offers as good a competitive effect as divestiture, it is submittedthat the less restrictive remedy should be selected, because the risks of anunsuccessful divestiture are so high (viz the new companies may not be successful, the market as a whole may not function as effectively, innova-tion is retarded, loss of economies of scale) that antitrust authorities shouldbe satisfied with an outcome that reintroduces some competition, even if itis a second best. At a practical level, the authority must consider the time lag

52 Giorgio Monti

126 On analogy with ERT above n 78.127 See generally MJ Reynolds, ‘Article 81 and the Merger Regulation: Breaking the NetworkAccess Bottleneck’ in CD Ehlermann and L Gosling (eds), European Competition Law Annual1998: Regulating Communications Markets (Oxford, Hart Publishing, 2000), 387.128 The type of regulatory effort here is analogous to that in the telecoms field. See for similaranalysis in the context of broadband services: P Kremmyda, ‘Walled Gardens? Cable ProvidedBroadband Internet and Competition Law’ (2001) 24 World Competition 181.129 See eg Bertelsmann/Kirch/Premiere above n 31.130 E Gellhorn, Antitrust Law and Economics 4th edn (Minnesota, West Publishing, 1994),134–36.131 Art 7 Council Reg no 1/23 of 16 December 2002, [2003] OJ L/1/1, 4 January 2003.132 The second is mostly a matter of economic analysis of specific factual situations. For analyses supporting the divestiture of Microsoft into two firms, see Romer’s and Shapiro’ssubmissions at <http://www.usdoj.gov/atr/cases/f4600/4643.htm> Romer, 27 April 2000, and4642.htm Shapiro, 28 April 2000, respectively.

between the proposal of this remedy and its final imposition: the firm is likelyto raise several legal challenges that are likely to delay the implementation mak-ing it uncertain whether the proposed divestiture will still be effective when itcan finally go ahead. This practical consideration probably influenced theDOJ’s decision not to seek a divestiture of Microsoft, considering that aprompt but effective behavioural remedy was preferable.133 These consid-erations suggest that it will only be in exceptional circumstances thatdivestiture should be called for.

However, it is submitted that in those rare cases where an undertaking isso powerful that no fine or behavioural order is likely to restrain its powersor avoid anticompetitive effects, it may be more effective to establish a sys-tem of public regulation—this would avoid any loss of synergies that abreak-up may engender and would also allow for a more coherent and flex-ible means of ensuring competitiveness as well as other social goals. Thiswas recognised by the DOJ in the WorldCom/Sprint merger where it notedthat if one firm became the dominant top-tier ISP, ‘restoring the market to acompetitive state often requires extraordinary means, including some formof government regulation’.134 To a certain extent some of the remediesimposed in merger cases, in cases of essential facilities, and in the Microsoftcase, are already regulatory in nature, suggesting that competition law principles are being over-stretched.135

7. CONCLUSION

The new economy creates markets that require different economic tools toexamine the behaviour of undertakings. Economists agree as to the aims:maintaining competition and promoting innovation, but disagree on howthese should be obtained: non-enforcement according to neo-Schumpeterians,more aggressive enforcement according to the neo-structuralists. In litigation,this debate continues as defendants deploy neo-Schumpeterian argumentsand plaintiffs follow the dictates of the neo-structuralists. The competitionauthorities in the EC and US have preferred the neo-structuralist approach,which explains both their preference for narrower market definitions,which seek to ensure the proliferation of participants in new economy markets, and its preference for a structural analysis of dominance. Thecourts on the whole have backed this approach.


133 On 6 September 2001: <http://www.usdoj.gov/atr/public/press_releases/2001/8981.htm>.The DOJ was already embarrassed when after 13 years of trying to divest IBM, it gave up real-ising that market forces had made the divestiture unnecessary.134 Above n 33, para 41.135 For further elaboration of this theme see P Larouche, Competition Law and Regulation inEuropean Telecommunications (Oxford, Hart Publishing, 2000).

Neo-Schumpeterian concepts are unlikely to gain favour, for two reasons: first, a lot of optimistic speculation about future market develop-ments is required; second, neo-Schumpeterians take a narrower view aboutthe aims of competition law—damage that dominant firms can do to otherundertakings is irrelevant unless it can be shown that there will be loss toconsumer welfare in the long run as a result of the acts of the dominantfirm. However, current competition law is concerned with the eliminationof competitors per se, observing that this carries the risk of the dominantfirm being alone on the market or deterring new entrants.136 Thus, neo-Schumpeterian ideas do not fit easily within the current competitionlaw culture, and their theories do not fit within the analytical frameworkof Article 82 EC. The task for neo-Schumpeterians is to translate their economic theories into workable legal models. One suggestion which thischapter has made which might facilitate the integration of neo-Schumpeterian concepts in competition law analysis is to develop an inno-vation defence which would force the courts to consider the impact ofArticle 82 decisions on the incentives to innovate more openly than is doneat present.

In spite of the authorities’ aggressive stance, firms have escaped with lit-tle financial sanctions. Microsoft for example escaped liability in spite ofcommitting flagrant breaches of the Sherman Act, but may yet face privatelawsuits.137 Instead, the authorities have focused on ex post behaviouralremedies and ex ante divestitures. As some authorities have recognised,these types of remedies turn competition law into a market regulator, a taskfor which it is not well equipped.

These developments suggest that the focus on neo-structuralism by theauthorities aims to regulate the new economy to afford market access forall participants, rather than to deter dominant firms with large fines.Competition law here is forward looking in seeking remedies that removedominance by facilitating the entry of others. Whether or not this goalshould be achieved through Article 82, a more pressing question is whetherthis policy will eliminate incentives to innovate, frustrating the dynamismof new economy markets.

54 Giorgio Monti

136 This concern may also be grounded in a political concern about excessive power—R Pitofsky, ‘The Political Content of Antitrust’ (1979) 127 University of Pennsylvania LawReview 1051 and G Amato, Antitrust and the Bounds of Power (Oxford, Hart Publishing,1997).137 AOL is reported to be seeking damages. Financial Times (24 January 2002).

3

Abuse of a Dominant Position and Intellectual Property Rights:

A Suggestion to Reconcile The Community Courts’ Case Law

ESTELLE DERCLAYE*

THIS CHAPTER CONCERNS the application of Article 82 of theEC Treaty to intellectual property rights (IPR). The conditions1

when a refusal to license an IPR amounts to an abuse are obscure.As a matter of fact, the actual rulings by the European Court of Justice(ECJ) and the Court of First Instance (CFI) on the application of Article 82to IPR are difficult to reconcile. Is there abuse if only one of the conditionsidentified by the ECJ are met, or do several have to be met? If the latter, arethe conditions cumulative or alternative? Do they apply indistinctly to allforms of property? The recent IMS2 case clarifies the confusion that cur-rently exists as regards the way the relationship between Article 82 and IPRmust be regulated and provides an excellent illustration of the problem.The purpose of this chapter is to give a suggested way of reconciling thecase law and thus clarify under which circumstances the holder of an IPRabuses its dominant position under Article 82.

* LLM, DES, Lecturer in Intellectual Property Law, Queen Mary Intellectual PropertyResearch Institute, University of London. The author is most grateful to friends who carefullyread and commented on a previous draft. She also wishes to thank the anonymous referees fortheir comments. All potential errors remain the author’s.1 The words ‘conditions’ and ‘circumstances’ will be used as synonyms. The European Courtof Justice in the Magill case originally used the term ‘circumstances’. However, the headings ofthe cases generally use the term ‘conditions of abuse’ and the literature refers to conditionsand circumstances interchangeably: see eg V Korah, ‘The Interface Between IntellectualProperty and Antitrust: The European Experience,’ (2002a) Antitrust Law Journal 801, 814;H Schmidt, ‘Article 82’s “Exceptional Circumstances” That Restrict Intellectual PropertyRights’ (2002) 22 European Competition Law Review 210, 214.2 Abbreviation of Intercontinental Marketing Services Health Inc. The shorter version (IMS)will be used throughout the chapter.

The chapter is divided into three sections. After a short review of theconditions under which Article 82 applies, section one addresses the vari-ous decisions of the ECJ and CFI dealing with refusals to license by holdersof IPRs preceding the IMS case.3 The facts will be first described, then therulings. The analysis will show the existing confusion as to the circum-stances in which there is abuse by an IPR holder. Section two discusses theIMS case. The case serves to illustrate further the confusion identified insection one. First, the facts of the case will be summarised. Then the rulingsof the Commission and the President of the CFI will be explained, com-mented upon and contrasted. Section two concludes with remarks on theconsequences that the conflict between the decisions entails. Section threesuggests a way of reconciling the cases. It first develops the argument andthen draws some consequences, including a preferred way of deciding theIMS case. A few arguments corroborating the suggested interpretation ofthe case law are finally given. In conclusion, it is hoped that the IMS casewill be a catalyst for the court to clarify the conditions at which an IPRholder should be forced to grant a license.

1. CONFUSION IN THE CASE LAW

The five conditions which trigger the application of Article 82 are as fol-lows: there must be (1) any type of abuse (2) by one or more undertakings(3) in a dominant position (4) within the common market or a substantialpart of it and (5) the abuse must actually or potentially affect trade betweenMember States. The area where most of the confusion lies in the case law isthe first condition (abuse). The only requirement on which the discussionfocuses is thus the notion of abuse. The other requirements (dominant posi-tion etc) will not be reviewed here.

Article 82 lists a few examples of abuses. These examples do not form anexhaustive list.4 Abuses such as refusals to deal with a competitor are not

56 Estelle Derclaye

3 Constraints of space prevent us making a detailed analysis of those cases. Therefore basicknowledge of the decisions will be assumed. For such an account, see eg E Derclaye, ‘Abus deposition dominante et droits de propriété intellectuelle dans la jurisprudence de la Communautéeuropéenne: IMS survivra-t-elle au monstre du Dr. Frankenstein ?’ (2002) 15 Les Cahiers dePropriété Intellectuelle 21, 26 ff and the comparative table, 54–55. For similar reasons, this dis-cussion will not review the essential facilities doctrine nor criticise it. The substance of the doc-trine can be summarised as follows: ‘An essential facility means one to which undertakingsrequire access in order to produce goods or provide services for their customers, but which theycould not reproduce themselves without prohibitive expense or inordinate delay’ (F Wooldridge,‘The Essential Facilities Doctrine and Magill II: The Decision of the ECJ in Oscar Bronner’(1999) Intellectual Property Quarterly 256). As a consequence, the dominant firm which refusesto supply an essential good or grant access to an essential facility to its downstream competi-tor(s) will be forced to do so by the competition authorities at a non-discriminatory price. 4 Case C–6/72 Europemballage Corp and Continental Can Co Inc v Commission [1973] CMLR 199, para 26. See also eg R Greaves, ‘The Herchel Smith Lecture 1998: Article 86 of the

listed, but the ECJ has since the 1970s included them within the scope ofArticle 82.5 There are at least two types of refusal to deal: refusals to sup-ply and refusals to license.6 Refusals to supply seem to be connected to tan-gible property while refusals to license, to intellectual property. As refusalsto deal are not listed in Article 82, the conditions under which such refusalswill be abusive similarly are not listed. Thus a fortiori the conditions underwhich a refusal to license by the holder of an IPR is abusive are not men-tioned either. However, the Community courts have decided that Article82’s scope encompasses refusals to license and have, in a number of deci-sions, laid down conditions at which such a refusal is abusive. The remedyfor an abusive refusal to license an IPR is to impose a compulsory licenseonto the holder of the IPR.

A. Factual Background to the Case

There are four cases in which the ECJ and the CFI have elabo-rated conditions or circumstances in which an IPR holder will be forced, under Article 82, to grant a license to its competitors. In chronological order, these are: Renault,7 Volvo,8 Magill9 and

Abuse of a Dominant Position and IP Rights 57

EC Treaty and Intellectual Property Rights’ (1998) European Intellectual Property Review 380;S Weatherill and P Beaumont, EU Law 3rd edn (London, Penguin Books, 1999) 845; R Whish, Competition law 4th edn (London, Butterworths, 2001) 150.

5 Cases C–6 and 7/73 Istituto Chemioterapico Italiano SpA & Commercial SolventsCorporation v Commission [1974] ECR 223; Case C–27/76 United Brands Continental BV vCommission [1978] ECR 207.6 P Roth QC (ed), Bellamy & Child, European Community Law of Competition 5th edn(London, Sweet & Maxwell, 2001) n 9–092–110 do not make the distinction exactly as suchbut make an attempt to distinguish refusals to supply and refusals to license. The structure inwhich they discuss this area of the law is as follows: (v) refusals to supply; including (a) refusalsto supply existing customers; (b) access to essential facilities; (c) refusals to supply in other cir-cumstances and (vi) abuses of IPR. Whish above n 4, 611 ff only speaks of refusals to supplyand classifies them in several headings similar to those of Bellamy & Child and also discussesabuses of IPR within a separate category (698).7 Case C–53/87 Consorzio italiano della componentistica di ricambio per autoveicoli &Maxicar v Régie nationale des usines Renault [1988] ECR 6039. 8Case C–238/87 AB Volvo v Erik Veng (UK) Ltd [1988] ECR 6211. On the Renault and Volvodecisions, see eg I Govaere, The Use and Abuse of Intellectual Property Rights in EC Law:Including a Case Study of the EC Spare Parts Debate (London, Sweet & Maxwell, 1996) 195 ff;V Korah, ‘No Duty to License Independent Repairers to Make Spare Parts: The Renault, Volvoand Bayer & Hennecke Cases’ (1988) European Intellectual Property Review 381; P Groves,‘The Use of Registered Designs to Protect Car Body Panels’ (1989) Business Law Review 117.9 Case T–69/89 Radio Telefis Eireann v Commission [1991] ECR II–485 and Case T–76/89ITP v Commission [1991] ECR II–575 (both CFI’s judgments carry the same substance); CasesC–241/91 P and C–242/91 P Radio Telefis Eireann (RTE) & Independent TelevisionPublications Ltd (ITP) v Commission [1995] ECR I–743. The literature abounds in commentson the decisions of the Commission, the CFI and the court, of which only a few will be men-tioned here. B Sufrin, ‘Comment on The Magill Case’ (1992) Entertainment Law Review 67;A Reindl, ‘The Magic of Magill: TV Program Guides as a Limit to Copyright Law?’ (1993)

Ladbroke.10 In addition, even if it does not deal with an IPR, theBronner11 case is relevant to the analysis for at least two reasons. First, thecourt, to reach its decision in Bronner relied upon the above cited cases.Second, the Commission has heavily relied on that ruling to decide theIMS case.12

The facts of the cases can be summarised briefly as follows. In theRenault and Volvo cases,13 Renault and Volvo had design rights on theirmodels for car body panels. They refused to grant a license of their designrights to independent repairers thereby preventing them from supplyingspare parts. In the Magill case, the Irish and British broadcasters BBC, RTEand ITP,14 each respectively holders of copyright on their weekly televisionlistings, denied Magill, an Irish publisher of comprehensive weekly televi-sion guides, a license to reproduce them. In Ladbroke, the French sociétésde courses, held copyright in the pictures and sound of horse races. Theyrefused to grant to Ladbroke, a Belgian bookmaker, a license to rebroad-cast French horse races live. Finally, in Bronner, Mediaprint, an Austriannewspaper publisher, refused to distribute the daily newspaper of another

58 Estelle Derclaye

International Review of Industrial Property and Copyright Law 60; C Miller, ‘Magill: Time toAbandon the “Specific Subject Matter” Concept’ (1994) European Intellectual PropertyReview 417; R Thompson, ‘Magill: ECJ Upholds Use of Article 86 to Control Conduct ofCopyright Holders on Ancillary Markets’ (1995) Entertainment Law Review 143; R Greaves,‘Magill est arrivé’ (1995) European Competition Law Review 244; S Taylor, ‘Copyright VersusRight to Compete: The Judgment of the ECJ in Magill’ (1995) Computer andTelecommunications Law Review 99; T Vinje, ‘The Final Word on Magill: The Judgment ofthe ECJ’ (1995) European Intellectual Property Review 297; H Calvet and T Desurmont,‘L’arrêt Magill: une décision d’espèce ?’ (1996) 167 Revue Internationale de Droit Economique3; Govaere above n 8, 135–50, n 5.47–5.67; S Anderman, EC Competition Law andIntellectual Property Rights: The Regulation of Innovation (Oxford, Oxford University Press,1998) 204 ff.

10 Case T–504/93 Tiercé Ladbroke SA v Commission of the European Communities [1997]ECR II–923. See eg D Fitzgerald, ‘Magill Revisited’ (1998) European Intellectual PropertyReview 154; V Korah, ‘The Ladbroke Saga’ (1998) European Competition Law Review 169.11 Case C–7/97 Oscar Bronner GmbH & Co KG v Mediaprint Zeitungs- und ZeitschriftenverlagGmbH & Co KG, Mediaprint Zeitungsvertriebsgesellschaft mbH & Co KG & MediaprintAnzeigengesellschaft mbH & Co KG [1998] ECR 7791. For comments on the Bronner case,see P Treacy, ‘Essential Facilities—Is the Tide Turning,’ (1998) European Competition LawReview 501; F Wooldridge, ‘The Essential Facilities Doctrine and Magill II: The Decision ofthe ECJ in Oscar Bronner’ (1999) Intellectual Property Quarterly 256; M Bergman, ‘The BronnerCase: A Turning Point for the Essential Facilities Doctrine’ (2000) European Competition LawReview 59.12 See Commission’s decision, analysed below and also C Stothers, ‘The End of Exclusivity?Abuse of Intellectual Property Rights in the EU’ (2002) European Intellectual PropertyReview 88.13 The facts and rulings of the two cases are substantially the same; thus the cases will be dis-cussed together.14 ITP was not strictly speaking a broadcaster but ITV and Channel 4 had granted it theircopyright in their television programmes. NB: copyright will be used throughout this articlein its broad meaning, ie encompassing both civil law author’s right and common law copyright.

smaller Austrian publisher, Oscar Bronner, through its national newspaperhome delivery scheme. Only the broadcasters were compelled to give alicense. In all the other cases, no compulsory license was forced onto theright holders. All cases have been decided by the ECJ except the Ladbrokecase which was decided by the CFI.

B. Conditions Required to Establish the Abuse

When is a refusal to license by the holder of an IPR an abuse? In otherwords, in what circumstances can a compulsory license be imposed onto anIPR holder?

First of all, it is well established that the mere ownership and mere exer-cise of an intellectual property right (here, the mere refusal to grant alicense) cannot in itself confer a dominant position nor consist in an abuseof such a position.15 Thus there will only be an abuse when the IPR is exer-cised in certain—exceptional—circumstances. These circumstances inwhich an abuse may be found have been laid down in the five cases.

In the Renault and Volvo cases, the court did not set out circumstancesin which a refusal to license is abusive, but merely gave non-exhaustiveexamples of abusive conduct resulting from the exercise of an IPR, namely:the arbitrary refusal to supply spare parts to independent repairers, the fix-ing of prices for spare parts at an unfair level and the decision no longer toproduce spare parts for a particular model even though many cars of thatmodel are still in circulation.16 As can be seen, the refusal to license wasnot even listed, but only the refusal to supply,17 and it was only cited as oneexample. No conditions were given at which such refusal was abusive.Thus, Renault and Volvo only set the scene for the more detailed ruling inMagill.

Magill constitutes the first case in which the Court addressed in detail theconditions under which a refusal to license18 will be abusive. It stated that


15 See eg Parke Davis & Co v Probel, Case 24/67 [1968] ECR 55. The ECJ recalls this princi-ple in each of the above mentioned decisions whereas in Ladbroke, the CFI omits to do so.16 Paras 9 of the Volvo decision and 15 of the Renault decision.17 F Siiriainen, ‘ “Droit d’auteur” contra “droit de la concurrence”: versus “droit de la regula-tion” ’ (2001) Revue Internationale de Droit Economique 422 thinks that Volvo and Renaultwere not really concerned with a refusal to license but rather with a refusal to supply. However,it can be said that in effect, the car manufacturers’ refusal to supply amounted to a refusal tolicense, as it was not possible to make spare parts without having a proper license.18 Some think that Magill is actually a case of refusal to supply raw materials (ie here infor-mation), not a refusal to license the reproduction of works protected by copyright, probablydue to the fact that the copyright held by the broadcasters was on basic information. See M Van Kerckhove, ‘Magill: A Refusal to License or a Refusal to Supply?’ (1995) 51Copyright World 26. However, it is to be noted that despite that fact, one cannot deny thatthere was copyright in the television listings and that strictly speaking it was a refusal tolicense (ie not to allow a third party to reproduce a copyright work).

these conditions must remain exceptional. These are: (1) the prevention ofthe appearance of a new product which the IPR holder did not offer and forwhich there was a potential consumer demand, (2) the refusal is not justi-fied and/or (3) the IPR holder reserves to himself a secondary market byexcluding all competition on that market.19

In Ladbroke the CFI refined one of the conditions in Magill: a refusal tolicense will infringe Article 82 if it is a new product whose introductionmight be prevented, despite specific, constant and regular potential demandon the part of consumers.20 However, it added a new alternative condition:a refusal to license will infringe Article 82 if it concerns a product or servicewhich is essential for the exercise of the activity in question, in that therewas no real or potential substitute.21

Finally the Court in Bronner set a tripartite test which combines condi-tions from Magill and Ladbroke: (1) the refusal of the service comprised inthe home delivery must be likely to eliminate all competition in the dailynewspaper market on the part of the person requesting the service; (2) suchrefusal cannot be objectively justified;22 and (3) the service in itself must beindispensable to carrying on that person’s business, in as much as there isno actual or potential substitute in existence for that home deliveryscheme.23

As it transpires from this brief summary, it is difficult to know exactlyunder which conditions a refusal will be abusive and which conditionsshould be applied in subsequent cases.24 First, the conditions set down inthe cases differ: some are restated in the same terms, others with a changein the wording, some are withdrawn, some are added. Second, there isuncertainty as regards whether the conditions should be applied in a cumu-lative or alternative way (and if so which ones). Indeed from the wordingof the Magill ruling, it is impossible to know whether the Court intendedthe circumstances to be alternative or concurrent.25 In Ladbroke, the CFI

60 Estelle Derclaye

19 See paras 54–56 of the Magill decision.20 Para 131 of the Ladbroke decision (emphasis added). This wording already appeared inits judgments RTE (para 62) and ITP (para 48) which were appealed to the ECJ in theMagill case.21 Para 131 of the Ladbroke decision.22 Emphasis added. The Court here adds to the Magill ruling by further qualifying the refusal.23 Para 41 of the Bronner decision.24 See eg P Landolt and J Ysewyn, ‘Intellectual Property Rights and EC Competition Law’(2001) 111 Copyright World 19, 19 note that the cases decided by the ECJ and the CFI dis-play a lack of finality.25 The court’s wording in paras 54–56 is unclear. Upon reading para 54, it seems that the con-dition set out there is sufficient to establish abuse. However, the court goes on to enumerateother conditions in paras 55 and 56 without using the term ‘and’ to connect them. Finally, itconcludes, in para 57: ‘in light of all those circumstances, the Court of First Instance did noterr in law in holding that the appellants’ conduct was an abuse of a dominant position withinthe meaning of Article 86 of the Treaty’. This final statement seems to infer that the conditionsare cumulative but this is only speculation. The Commission sees the conditions as alternative,

sets two alternative conditions. Ladbroke thus broadens the Magillruling.26 Finally, in Bronner, the court takes the opposite stance and optsfor a cumulative test.

What happened in the IMS case? Not too surprisingly, the Commissionon the one hand and the Presidents of the CFI and of the Court, on theother, made two different, and conflicting, interpretations of the case law.

2. AN ILLUSTRATION OF THE CONFUSION: THE IMS CASE

A. The Factual Background of the Case

In Germany, IMS27 provides reports informing pharmaceutical compa-nies of regional sales of their pharmaceutical products. To provide thisdata, any provider must comply with the German data protection law andthus ensure that individual sales by any given pharmacy cannot be identi-fied (at least three pharmacies must be grouped in a geographical zone,also called ‘brick’ or ‘module’, for which the data on the sales is given). Inorder to abide by German law, IMS has created a structure which dividesGermany into 1860 modules. On the basis of its structure, IMS wrotereports on sales of pharmaceutical products and provided them to phar-maceutical companies. IMS claimed it had copyright on its structure as a


the President of the CFI, as cumulative. V Korah (2002a), above n 2, 810, 811 and 814, seesthe Magill conditions as cumulative even if subsequent case law has suggested that they arealternative. Anderman above n 9, 209–10 and H Lugard, ‘ECJ Upholds Magill: It Sounds Nicein Theory, But How Does it Work in Practice?’ (1995) European Business Law Review 233,also wonder whether some of the Magill conditions are cumulative. F Fine, ‘NDC/IMS: InResponse to Professor Korah’ (2002a) 70 Antitrust Law Journal 247, 251 believes that theconditions are alternative: ‘The President therefore read the factors listed in Magill as beingcumulative, contrary to what subsequent case law—as even Korah confirms—tells us’.

26 V Korah (2002a) above n 2, 814.27 On the IMS case see eg P Landolt and J Ysewyn above n 24; D Paemen and C Norall, ‘IMSHealth Reveals EC’s Steel Resolve’ (2001) Managing Intellectual Property, September, 60;Stothers above n 12; D Hull, J Atwood and J Perrine, ‘Intellectual Property and CompulsoryLicensing’ (2002) European Antitrust Review 97; J Ratliff, ‘Major Events and Policy Issues inEC Competition Law, 2001: Part 2’ (2002) International Company and Commercial LawReview 64; R O’Donoghue and D Ilan, ‘Court Halts Commission Shake-up of CompulsoryLicensing Rules for Intellectual Property’, (2002) 118 Copyright World 8–9; V Korah (2002a)above n 2; F Fine (2002a) above n 25; V Korah, ‘Essential Facilities and a Duty to License—IMS’ paper given at the Fordham Intellectual Property Conference, New York, 4–5 April 2002;J Temple Lang, ‘Comment on Professor Korah’s Paper: “Essential Facilities and a Duty toLicense—IMS” ’ (2002) paper given at the Fordham Intellectual Property Conference, 4–5April 2002; F Fine, ‘NDC/IMS: What is the Real Subject Matter?’ (2002) 121 Copyright World19; S Anderman, ‘Microsoft in Europe’ (2002) paper presented at the 2002 FordhamConference; Derclaye above n 3, 21–55 (detailed account of the facts and of the several rulingsin the case, 36 ff).

database. In 1999, National Data Corporation Health InformationServices (NDC) and Azyx Deutschland GmbH Geopharma InformationServices (Azyx)28 entered the German market by creating a structurewhich was compatible with that of IMSs. IMS sued NDC for infringe-ment of copyright in its structure in Germany and won.29 NDC thenbrought a complaint to the European Commission against IMS for abuseof dominant position because IMS denied NDC a license to use its struc-ture. The Commission decided to oblige IMS to grant a license to allundertakings already present in the market.30 On IMS’s appeal (applica-tion for interim relief), the President of the CFI suspended the decision31

and the President of the ECJ confirmed the suspension.32

B. The Decision of the Commission

First, the Commission established the relevant product and geographic mar-ket as being the market for services of data provision on regional sales ofpharmaceutical products in Germany alone.33 Second, the Commissionestablished that IMS has a dominant position.34 IMS is in a situation ofquasi-monopoly because before NDC and Azyx’s entry on the market, itwas the sole player on the market. The respective positions of NDC andAzyx were negligible. In addition, there was an effect on the commercebetween Member States.35

As to the question of abuse, the Commission briefly summarised the caselaw relating to refusals to deal (namely Commercial Solvents,36 UnitedBrands,37 Volvo, Magill, Ladbroke, Bronner).38 The way in which the

62 Estelle Derclaye

28 As the situations of NDC and Azyx are very similar, reference will be made exclusively tothose of NDCs.29 The Commission was bound by the German decision in this regard as it cannot decide uponthe validity of national copyrights. This issue is left to the national courts.30 Commission’s Decision 2002/165/CE of 3 July 2001, Case COMP D3/38.044, NDCHealth/IMS Health: interim measures [2002] OJ L/59/18. 31 First order (provisional stay) of 10 August 2001, Case T–184/01 R, IMS Health Inc vCommission [2001] ECR II–2349. Second order of 26 October 2001, Case T–184/01 R, IMSHealth Inc v Commission [2001] ECR II–3193. 32 IMS Health Inc v NDC Health Corp, Case C–481/01, 11 April 2002 [2002] 5 CMLR 1(ECJ). As this order merely confirms the two orders of the President of the CFI and does notcontain a further analysis of the substance of the issues, it will not be discussed.33 Paras 45 & ff of the Commission’s decision.34 Ibid paras 57 ff.35 Ibid paras 175–78. It is generally quite easy to find such an effect because it has been inter-preted broadly, see eg R Greaves, ‘The Herchel Smith Lecture 1998: Article 86 of the ECTreaty and Intellectual Property Rights’ (1998) European Intellectual Property Review 380;Paemen and Norall above n 27, 63. 36 Above n 5.37 Ibid.38 Paras 63–74 of the decision of the Commission.

Commission reviews the decisions indicates that it reads them as reflectinga trend towards the adoption of the essential facilities doctrine by theCommunity courts.39 In the Commission’s view, the case law is to be seenin an evolutionary way where Commercial Solvents reflects the implicitadoption of the essential facilities doctrine40 and Bronner confirms it. TheCommission puts all the cases together in a straight line as constitutingsome sort of proof of the same, single trend. Thus for the Commission, theapplicable conditions for judging any abuse of a dominant position underArticle 82 (including abuses of IPR) are the three cumulative ones now laiddown in Bronner.

To justify this interpretation, the Commission reasoned (albeit perhapsimplicitly) in the following three-stage way. First, it set out Magill’s excep-tional circumstances. Second, it recalled that in the Ladbroke case, the CFIimplied that it is not necessary for a refusal to prevent the appearance of anew product in order for it to be considered an abuse.41 In other words,Ladbroke allows Magill’s conditions to be read in an alternative fashion.Third, the Commission found that the Bronner ruling confirms theLadbroke approach. The reasoning is as follows. The condition of the pre-vention of a new product, first set out in Magill and restated but in a clearalternative fashion in Ladbroke, is omitted in Bronner. Thus, for theCommission, it is not to be tested anymore.

In sum, the Commission’s interpretation of the cases seems to be thatBronner is the definitive test to apply in cases involving refusals to deal,including refusals to license. In other words, for the Commission, Bronnerlays down the definitive exceptional circumstances or contains the decisivereinterpretation of the notion of exceptional circumstances as first set inMagill. This is why the Commission based its decision in the IMS caseexclusively on the Bronner tripartite test.

Applying the Bronner test to IMS’s facts, the Commission came to theconclusion that IMS had abused its dominant position since the three cumu-lative conditions were fulfilled:42 first, IMS’s refusal was unjustified; sec-ond, it eliminated all competition in the market of the provision of regionaldata because the structure was indispensable for NDC and Azyx to con-tinue their activities, and third, there was no actual or potential substituteto this structure. As a result, the Commission imposed a compulsory licenseon IMS.


39 Even if the Commission acknowledges that neither the CFI nor the ECJ has ever referredexpressly to the essential facilities doctrine in their judgments (para 64). The doctrine is brieflydefined and explained at above n 3.40 Whish above n 4, at 611, 613 and 615, shares this opinion as well.41 Paras 68 and 180 of the decision of the Commission.42 ‘These exceptional circumstances meet the test set out in Bronner for a refusal to supply tobe considered an abuse of a dominant position’. (ibid para 181). For a criticism of this findingby the Commission, see Derclaye above n 3, 47 and 52.

C. The Orders of the President of the Court of First Instance

The President of the CFI was not convinced that the Commission’s viewwas the only interpretation that can be given to the case law. In his firstorder, the President put emphasis on the Magill case. It seems implicitfrom the rather short reasoning that what should be stressed is the factsof the cases. In substance, his reasoning can be put in the form of a syllo-gism: (1) the approach underlying the Commission’s decision seems todepend on the notion of exceptional circumstances as elaborated inMagill; (2) there are potentially important differences between the factsof Magill and those of IMS; (3) the facts of IMS and Magill are different,but nevertheless the Commission imposes a compulsory license on IMSdespite the difference in their facts, IMS has made a prima facie case thatthe Commission misapplies Magill.43

In his second order, the President recalled that there are differencesbetween the facts of the Magill and IMS cases and added that theCommission’s interpretation of the Magill ruling (ie as setting alternativeconditions) ‘constitutes at first sight an extensive interpretation of the Magillcase’.44 The President concluded that while the Commission’s reading of thecase law might be correct, there were reasonable grounds to conclude thatthere may be another interpretation of the case law, such as seeing the Magillconditions as cumulative.45 Thus the order seems to suggest that there maybe two interpretations of the Magill case: the conditions for imposing alicense are cumulative, or, on the contrary, they are alternative. Given theexistence of a serious dispute as to the legal reasoning that must be appliedto decide the IMS case, the President, having found that all other conditionsto grant interim relief (urgency, balance of interests, etc) were fulfilled, suspended the decision of the Commission. In conclusion, he found thatthere was clearly a problem of interpretation and only a final decision onthe merits could clarify it.

D. Comments

As the IMS case illustrates, there are at least two possible ways of interpret-ing the case law. None of them is clearly untenable. It is true that theCommission in its interpretation of the case law failed to apply the firstcondition set in Magill. But can one criticise this approach in view of therather confusing state of the case law? The Magill decision is unclear as to

64 Estelle Derclaye

43 President’s order of 10 August 2001, para 24.44 Para 100 and 102 of the President’s second order, referring to para 67 of Commission’sdecision.45 Para 104 of the second order.

the nature of the circumstances it sets forth (alternative or cumulative). TheLadbroke ruling has set two alternative conditions. The Bronner case, whilebringing together conditions deriving from the previous cases, has omitted(or at least considered as non-applicable) the condition of the prevention ofthe appearance of a new product despite potential consumer demand. Thusit seems difficult at first sight to criticise the Commission’s chronologicaland in the end, rather logical construction of the case law. In addition, sucha construction is hardly surprising. The Commission is naturally bound toprefer such a view, as its prime role in competition matters is to be theguardian of any potential abuses of dominant firms.

The different interpretations of the case law reached by the Commissionand the President of the CFI lead to different results. If, on the one hand,the conditions in Magill are applied in a cumulative fashion, the number ofcases where an abuse will be found will probably be small. If, on the otherhand, the Bronner test is solely applicable, it should lead to the oppositeresult (as the important condition of the prevention of the appearance of anew product is not required). Thus the choice as to which of those interpre-tations should prevail will have extremely important consequences on thedetermination of future cases relating to refusals to license.

In conclusion, the IMS case raises once again the fundamental questionunderlying the application of Article 82 to the exercise of an IPR: whatare the exceptional circumstances in which there will be abuse of an IPR?And, more generally, what should they be? While this second question isbeyond the scope of the discussion as it involves a deeper legal and philo-sophical analysis of the relationship between IPR and competition law,the following section will attempt to answer the first, namely make senseof the case law as it stands. It will include a discussion of why theCommunity courts should make a distinction in their treatment of intel-lectual property compared to other forms of property.

3. HOW TO SOLVE THE DILEMMA: A SUGGESTION

As illustrated by the abundant literature on the topic and by the IMS case,it appears to be difficult to reconcile the rulings of Magill, Ladbroke andBronner.46 The root of the problem seems to be the uncertainty overwhether the conditions in Magill are cumulative or not. The question is:are the cases irreconcilable? Is there a problem of interpretation of the caselaw and if so, how can it be solved? If there is no problem, how can thecases be explained? The short answer to this is that there is no dilemma. Inother words, the case law is reconcilable. The argument is as follows.


46 For the literature, see above n 25.

A. Argument

The answer lies at the heart of the Bronner ruling. As Bronner is the mostrecent case on refusals to supply a downstream competitor decided by theECJ, not surprisingly, the Court reassesses its previous case law. If one readsthe decision carefully, the ECJ seems to give the answer to the apparentproblem. In paragraph 40, the ECJ restates the Magill conditions, butclearly reinterprets them as cumulative.47 It even reinterprets the third con-dition in stating that the refusal must be objectively justified.48 Whetherthis is intentional or not is left to the appreciation of the reader. However,in view of the confusion the Magill decision created as far as the interpreta-tion of the exceptional circumstances is concerned and of the amount of lit-erature it has generated commenting on this imprecise language, one mightwell take the view that this (re)wording is not innocent. This is reinforcedby the language of paragraph 41 which reads:

Therefore, even if that case-law on the exercise of an intellectual propertyright were applicable to the exercise of any property right whatever, it wouldstill be necessary, for the Magill judgment to be effectively relied upon inorder to plead the existence of an abuse within the meaning of Article 86 ofthe Treaty in a situation such as that which forms the subject-matter of thefirst question, not only that…49

In this paragraph, the ECJ suggests one thing: as regards the application ofArticle 82, or at least in the case of one type of abuse, refusals to deal, intel-lectual property rights have to be treated differently from other propertyrights. In paragraph 41, the Court’s opinion is that it is far from clear thatthe case law relating to abuses of IPR applies to abuses of other forms ofproperty.

The finding that IPR should be treated differently from other types ofproperty has two consequences. First, it is certain that in similar situations

66 Estelle Derclaye

47 ‘In Magill, the Court found such exceptional circumstances in the fact that the refusal inquestion concerned a product (information on the weekly schedules of certain television chan-nels) the supply of which was indispensable for carrying on the business in question (the pub-lishing of a general television guide), in that, without that information, the person wishing toproduce such a guide would find it impossible to publish it and offer it for sale (para 53), thefact that such refusal prevented the appearance of a new product for which there was a poten-tial consumer demand (paragraph 54), the fact that it was not justified by objective considera-tions (para 55), and that it was likely to exclude all competition in the secondary market oftelevision guides (para 56) ‘ (emphasis added). Schmidt above n 2, 215 suggests that ‘Bronnerclearly emphasises the ruling in Magill that all the “exceptional circumstances” should bepresent before Article 82 can apply to confine the ambit of an IPR’, however without givingthe reasons or making a demonstration of his argument.48 Para 55 of the Magill ruling only states: ‘Second, there was no justification for such refusaleither in the activity of television broadcasting or in that of publishing television magazines… .’49 Emphasis added.

to the facts of Bronner—or if it can be extrapolated further, in casesinvolving other property rights—other conditions must be met (ie the threeconditions the court sets forth in paragraph 41 as further characterized inparagraphs 44 and 46). At the risk of stretching the Court’s wording orintention too much, Magill is the authority dealing with IPR and does notin itself have direct application as it stands in other property fields.

Second, this suggests therefore, implicitly and a contrario, that in situa-tions dissimilar to those of Bronner, ie in cases involving refusals to licenseIPR, it is Magill which is to be relied upon. At least it is far from clear thatthe Bronner conditions should apply. In other words, the ECJ seems towarn that it may not be self-evident that the same conditions should applyto cases involving refusals to license (involving intellectual property) andrefusals to supply (involving other types of property (mainly tangible) ).Thus the interpretation of this important part of the Bronner judgment,which is passed over in silence by the Commission, but interestingly pointedto by the President of the CFI in his second order,50 seems to strongly sug-gest that only Magill should apply to intellectual property rights and notBronner. It could be argued therefore that the Bronner judgment makes adistinction between cases involving refusals to supply and refusals tolicense. In the first set of situations, only the tripartite test set out inBronner applies whereas in the second set, only the cumulative conditionsof Magill (as reformulated in paragraph 40 of Bronner) apply. In conclu-sion, there is no conflict between the several rulings.

The important consequence that can be drawn from this interpretation isthat the circumstances under which a refusal to license will be abusive beingcumulative, it will be rare that they will all be fulfilled. As a result, abuse ofIPRs should remain scarce. The IMS case readily illustrates this point. As aresult of the suggested interpretation of the case law, it is clear now thatIMS should win its case. The first condition set forth in Magill is met by theprevention, by the holder of the IPR, from the appearance on the market ofa new product for which there is potential consumer demand. IMS’s refusaldoes not prevent the appearance of a new product. On the contrary, NDCand Azyx do not wish to either change or improve the structure, nor createnew types of report on regional sales of pharmaceutical products; their onlydesire is to use IMS’s structure in order to provide similar or identical services to those of IMS. There is no potential demand from consumers (iethe pharmaceutical companies) for another new hypothetical product.Therefore the condition is not fulfilled, IMS’s refusal is not abusive and itwill not be forced to grant a license. As suggested by the President of theCFI, the facts of Magill are different from those of IMS.


50 In its para 104, the President seems to hint at the solution by recalling paras 40 and 41 ofBronner. Perhaps this is further evidence of the approach advocated in this article.

The paragraphs above sought to bring out a coherent legal interpretationof the reasoning of the Community courts in order to understand their caselaw. The analysis was confined to interpreting the wording of the courts’judgments. It has been seen that the courts seemingly wish to make a dis-tinction between intellectual property and other forms of property (tangible(such as land, living creatures and goods) and intangible (such as stocks,shares, capital, money)51). However, they do not state the reasons why. Thequestion is then whether a distinction should indeed be made, as a matterof policy, in the treatment of abuse of a dominant position related to intellectual property and other forms of property.

This issue is no doubt extremely important but it would require a longand elaborate discussion, which could constitute a whole new article initself. Within the framework of this chapter, the author will only concen-trate on the specific features of copyright (since most of the cases whichcame in front of the Community courts involved copyright). The discussionwill list the differences between copyright and other forms of property,52

and highlight those which point towards justifying a difference in the treat-ment by competition law of abuses of dominant positions through refusalsby copyright holders and holders of other forms of property. This analysisis meant to attract the courts’ attention to the special character of copyrightand advise them to reflect carefully before applying a common test, identi-cal to both copyright and other forms of property. For reasons of space, theadequate condition(s)—ie conforming to policy—under which a compul-sory license should be imposed to a copyright holder will not be discussed.The author puts this to the agenda for a future article.

Copyright is generally categorised, as regards its nature, as a type ofintangible property. However, this categorisation is far from being obvi-ous. A long running controversy exists over the exact legal nature ofcopyright in both common law and civil law jurisdictions.53 And it is still raging.54 This debate is due to the difficulty of classifying copyrightas a type of property since there are important differences between

68 Estelle Derclaye

51 For a classification of types of property, see F Lawson and B Rudden, The Law of Property3rd edn (Oxford, Oxford University Press, 2002) 19 ff. In this chapter, these categories will bejointly called under the terms ‘other forms of property’. 52 The accent will be on the UK law of property. The comparison will be made especiallybetween copyright and other forms of personal property because copyright’s characteristicsare closer to it than to land law. Indeed s 90(1) of the UK Copyright Act classifies it as per-sonal or moveable property.53 Although it has been more of an issue in civil law countries.54 For a detailed account, see A Strowel, Copyright et droit d’auteur, divergences et conver-gences (Bruxelles, Bruylant, 1993) 80–171 or see C Colombet, Propriété littéraire et artistiqueet droits voisins 9th edn (Paris, Dalloz, 1999) paras. 16 and ff, 13–14 for a shorter overview.French modern authors generally think that it is vain to try and find a unique nature toauthor’s rights. It is a combination of property and personality rights. Its nature is hybrid. SeeColombet, ibid and A Bertrand, Le droit d’auteur et les droits voisins 2nd edn (Paris, Dalloz,1999) n 1.54, 76.

copyright and other forms of property. The result is that it is still uncertainwhether copyright can really be said to be a type of property. The author’saim is not to determine once and for all that copyright is property or thatit is not. Rather, even if it is accepted that copyright is (a type of) prop-erty, it is submitted that it is so dissimilar from other forms of propertythat refusals to license copyright should be treated differently fromrefusals to supply other forms of property.

There seems to be broad agreement among scholars that property rightshave two essential characteristics. They are rights against the world (in otherwords, the owner has the right to prevent anyone from interfering with theuse and enjoyment of the property)55 and they are transmissible (inter vivosand upon death).56 In addition, property rights are generally perpetual (ordurable)57 and absolute.58 It will be seen that copyright lacks all of thesecrucial characteristics (at least in their full extent). In addition, at least threespecific features of copyright can be identified that other forms of propertydo not have (copyright includes moral rights, the same copyright can belicensed to several persons simultaneously and copyright has an ethical element) which further estrange it from the remainder of property.

Copyright is a bundle of economic and moral prerogatives.59 Whileeconomic rights can be alienated,60 moral rights cannot61 and are thus


55 In other words, property rights are universal or erga omnes. A Bell, Modern Law of PersonalProperty in England and Ireland (London, Butterworths, 1989) 6 (who defines this character-istic in terms of property being both a defensive and assertive right (7–9): a property right is adefensive right in the sense that no one may interfere with my ownership of a thing withoutmy permission. A property right is also ‘against the world’ in the sense that it can be assertedagainst anyone into whose hands the property comes; W Murphy and S Roberts,Understanding Property Law 2nd edn (London, Harper Collins Publishers, 1994) 46–47; M Bridge, Personal Property Law (London, Blackstone Press, 1996) 8. 56 In other words, property rights are alienable, assignable or transferable. Bell above n 55, 6;Murphy and Roberts above n 55, 46–47; F Lawson and B Rudden above n 51, 39–40.57 Murphy and Roberts, above n 55, 47.58 See eg French and Belgian Civil Code, Art 544: ‘Property is the right to enjoy and dispose ofthings in the most absolute manner, as long as one does not use it in a way that is prohibitedby statute and regulations’ (translation by the author). See also the discussion below.59 The economic aspects of copyright cannot be severed from their moral counterparts. TheCourt of Justice has itself declared that the specific subject matter of copyright is to protect theeconomic and moral rights of right holders (Collins (Phil) v Imtrat Handelsgesellschaft MbH[1993] ECR 545) and that the essential function of copyright law is to protect the moral rightsin the work and to ensure a reward for the creative efforts, while respecting the aims of, inparticular Art 86 (CFI rulings in Magill, Cases T–69/89 Radio Telefis Eireann v Commissionand T–76/89 ITP v Commission above n 9. One cannot thus see copyright amputated from apart of its true nature (ie its moral rights).60 In some countries, notably Germany, copyright (Urheberrecht) is seen as a unified, uniqueconcept, where moral and economic prerogatives are inseparable. Hence copyright (thus as awhole, comprising both its economic or moral prerogatives) cannot be alienated but onlylicensed. In this sense, the whole German bundle of copyright rights lack the important prop-erty characteristic of transmissibility.61 Like other national copyright acts, the UK Copyright Act expressly states that copyright istransferable (s 90(1) ). However s 94 states that moral rights are not assignable. JAL Sterling,

not property.62 Moral rights protect the personality (name, honour, reputation, integrity) of the author and are therefore classified as person-ality rights. Unlike other forms of property, the copyright holder canlicense his/her rights to several people, ie s/he can choose between grant-ing one exclusive license or several non-exclusive licenses.63 Copyright ismoral. Copyright has an ethical component that other forms of propertydo not have: courts will not enforce immoral copyrights, because if theydid, they would legitimise immorality.64 Copyright is also territorial. Itonly forbids unauthorised acts done within the borders of the state inquestion.65 Even if international conventions66 tend to make copyright‘truly international’, there are still countries not party to them, so there isstill a difference between copyright and other forms of property. In otherwords, copyright is not strictly a universal right. As soon as the BerneConvention (or the TRIPs Agreement which incorporates it) is applicablein all countries of the world, copyright will be truly a universal right.

In addition, copyright is not absolute: intrinsically, by nature, it is lim-ited. Copyright statutes contain in themselves a generally well-thoughtthrough and carefully tailored internal balance. Three limitations to copy-right’s scope readily exist in the statutes.67 Copyright law is first limited byprotection requirements that need to be fulfilled in order to attract protec-tion (ideas are not protected but only their original expression). Once therights are acquired, they are further limited by exceptions which exist to thebenefit of users. Finally, copyright does not last perpetually.68 The limited

70 Estelle Derclaye

World Copyright Law (London, Sweet & Maxwell, 1998) 43 n 2.12; Colombet, above n 54,paras 16 and ff pp 13–4.

62 Lawson and Rudden, above n 51, 40. Already at the end of the 19th century, commentatorscriticised the classification of copyright into property because, among other things, it disre-gards the moral rights aspects of copyright. Thus some theories emerged arguing that copy-right is either a personality right or a sui generis right. See Strowel above n 54, 97. Murphyand Roberts above n 55, 47, wrongly generalise that since intellectual property rights aretransmissible, they are property. But they omit to see that moral rights are not transferable.63 Lawson and Rudden above n 51, 123. See also P Roubier, Le droit de la propriété intel-lectuelle, (Paris, Sirey, 1952) tome 1, 97–8 cited by Strowel above n 54, 104. Indeed, there can only be one lessee of a car, an animal or a house. This is due to the fact that unlike otherforms of property, copyright is a public good: it is essentially information and thus infinitelyreproducible. 64 J Phillips and A Firth, Introduction to Intellectual Property 4th edn (London, Butterworths,2001) n 10.13, 136: ‘Neither real nor personal property depends for their continued existenceupon criteria of inherent or functional morality, nor is there any reason why they should’.65 In other words, the work of author X, national of country X, or who has published hiswork in country X, will not automatically be protected in country Y.66 The major one being the Berne Convention for the Protection of Literary and Artistic Worksof 1886, last amended in Paris in 1971. Lawson and Rudden above n 51, 40.67 An additional limitation exists in UK copyright law. Copyright is firstly delimited by the factthat only a certain number of subject matter can be protected (the UK Copyright Act uses thesystem of categorisation of works, see ss 1–8).68 Phillips and Firth above n 64, n 10.6, 130; Sterling above n 61, n 2.12, 41–42.

duration, characteristic of copyright, is totally contrary to the durable quality of other forms of property. It does not make sense that a car, land ormoney ever ‘dies’,69 ie that it is not be owned by someone. It makes sensethat someone always owns the object, even if it is not bought by a subse-quent owner but is appropriated because it has been abandoned by its orig-inal owner. On the contrary, this finding does not make sense in relation tooriginal expressions of thought (be they literary, musical, artistic or othercopyright protected works). As a matter of policy, the public should be ableto use them freely once the author has recouped his/her investment. Hencethe rationale for the limited duration of copyright. This entails an impor-tant consequence: since copyright is not perpetual, the monopoly or domi-nant position it can create will never last for ever, contrary to other formsof property.

There are no such limitations on other forms of property law. To becomethe object of property, a thing does not have to fit in a category nor fulfilcertain conditions. As regards rights, copyright is much more limited thanother property rights, for instance rights in land.70 Landowners’ rights areindefinable, comprising every possible use of land: a complete list of themcannot be made. The same can be said of personal property. In contrast,copyright rights are well defined in the statute.71 Unlike copyright law,which provides for exceptions to rights, the law relating to other forms ofproperty does not further oblige the owner to let anyone use it once certainconditions are fulfilled.72

In sum, unlike other forms of property, copyright can be adequately envi-sioned as (primarily) creators’ rights but also as users’ rights. Thus,although there may be rules against abuses or misuses of other forms ofproperty, overall these forms are much more absolute, or at least less lim-ited, than copyright.

The reasons for these limitations to copyright are briefly as follows.Copyright works are information goods. The nature of information goodsis that they are public goods. Like other public goods, they possess a


69 It is true that living things such as animals will die but they remain different from copyrightin the sense that the duration of their owner’s property is not limited by consideration of pub-lic policy at the core of copyright law.70 J Penner, The Idea of Property in Law (Oxford, Oxford University Press, 1997) 119–20.71 See also ibid who takes patents as an example: ‘The patent is an exclusive right to a particu-lar use of the invention or idea, that is working it to produce goods for sale in the market. Butthis is only one of a limitless number of ways in which an idea may be ‘used’; one can study it,use it to illustrate scientific principles, use it as the basis for further inventive endeavours, andso on. That the market use of the idea is often the most valuable use in economic terms (…)that does not alter the fact that it is one use only’.72 This is especially true of personal property. As regards land, there can be easements on realestates such as grazing rights for instance, but unlike exceptions in copyright law, they are notautomatic (ie not every piece of land has them) and their extent is generally less far-reachingthan exceptions (or ‘users rights’) in copyright law.

‘non-rival’ quality. Non-rivalry means that the consumption possibilities ofone individual do not depend on the quantities consumed by others. Inother words, if one person consumes the good, s/he will not reduce the con-sumption opportunities of others in the good. Thus there is a potential forcollective consumption of the same good (eg the showing of a film). Purepublic goods are also non-excludable.73 This means that it is very costly toexclude anyone from enjoying them. In other words, one person cannotexclude another person from consuming the good in question. Copyrightworks are also generally non-excludable. This non-excludability arisesbecause reproduction costs are generally very low for anybody other thanthe creator of the good. Because creation costs (for the creator) are highand reproduction costs (for the free rider) are low, if the creator does nothave a means to stop the free riders, he will not have the chance to recouphis investment. Therefore a market failure is generated (no one will createas it is not profitable). To remedy this market failure and encourage cre-ation, the state enacts a copyright law which grants an exclusive right tothe creator (s/he is the only one to prevent or authorise the use of the work).This right enables him/her to have a chance to recoup his/her investment.

At least two out of the above discussed features of copyright plead fora specific treatment by competition law. First, contrary to other forms ofproperty, copyright arises from an individual’s intellectual creation. Theauthor has moral rights in his/her creation. A compulsory license onsomething so personal should be imposed with extra care. One mightobject that many creations are now made collectively or under the super-vision of a legal entity and that moral rights are less important for certainfunctional creations.74 Nonetheless, there is a second very important dis-tinctive feature of copyright, which other forms of property do not share:its ‘limitedness’.

Copyright is not a complete property right because its scope isrestricted in many ways. Copyright law is a legal remedy to the marketfailure which is the consequence of the non-excludability of works. Partsof these public goods (ie not ideas, but only original expressions of them)are thus artificially privatised for a limited amount of time. The rationalefor this is that the privatisation of these public goods is economically effi-cient (if copyright did not exist, no works or too few would be created,hence the public would be worse off). Copyright is thus in essence fragileand is inherently balanced:75 it is a compromise between the natural free

72 Estelle Derclaye

73 On all these notions see, eg R Cooter and T Ulen, Law and Economics 2nd edn (London,Harper Collins Publishers, 1997) 100 ff.74 As a matter of fact, in some countries, notably the UK, eg computer programmers do nothave moral rights (ss 79(2) and 81(2) ).75 On this balance see eg T Cotter, ‘Intellectual Property and the Essential Facilities Doctrine’(1999) Antitrust Bulletin 211.

use of original expressions by everyone76 and the just (and economicallyefficient) reward for the creator’s labour. Consequently, more cautionmust be exercised when further limiting these rights. In other words, thereare reasons to be more prudent in imposing compulsory licenses on copy-right holders than on owners of other types of property. A difference oftreatment, most probably in the direction of a lower incursion of compe-tition law into copyright’s scope than into the scope of other forms ofproperty, is justified.

B. A Few Other Corroborating Arguments

The first argument which confirms the suggested construction of the caselaw is the following. It is true that strictly speaking, there is no rule ofprecedent within the Community courts. The ECJ, and the CFI, are notbound by their own previous decisions and thus in theory, the ECJ is notbound to continue following its Magill ruling. But it is also true that theECJ, and the CFI, in practice do not often depart from their previous decisions.77 The Bronner case could be said to be an example of this prac-tice. The Court, in Bronner, does not depart from its case law, but refines it.It simply does not overrule Magill but distinguishes between two decisions.Magill and Bronner apply in two different types of situations. Both decisions are equally good law.

Second, it can hardly be denied that the Court, if it is consistent, willinterpret its ruling in Bronner as being one which already warned litigantsthat it will be cautious in applying or extending the Bronner ruling (or assome call it, the essential facilities doctrine) to intellectual property situa-tions such as Magill-type situations. This view is at least confirmed by thePresident of the ECJ in the IMS case.

Finally, it is worth noting that this interpretation is in line with otherarguments and general policy reasons which plead against too active anincursion of this part of competition law into the realm of intellectual prop-erty. It is beyond the scope of this discussion to develop them all here.Instead, the most important one will be mentioned. The Bronner case is seen


76 For maximum efficiency, public goods should normally be owned publicly, see Cooter andUlen above n 73, 102.77 A Arnull, The European Union and its Court of Justice, (Oxford, Oxford University Press,1999) 529; Weatherill and Beaumont above n 4, 202. In addition, even if the CFI is not boundby decisions of the ECJ (except in a few cases: res judicata and Art 54 of the Statute of theCourt, see eg Case T–162/94 NMB France & others v Commission [1996] ECR II–427, para36), when the precedents of the ECJ are clear, the CFI should follow the ECJ’s decisions. SeeArnull, 533 citing Brown and Kennedy, Brown and Jacobs’ the Court of Justice of theEuropean Communities 4th edn (London, Sweet and Maxwell, 1994), 351 and Weatherill andBeaumont above n 4, 207.

by the Commission and some commentators as the confirmation of theincorporation of the doctrine of essential facilities in European law.However, besides the fact that the court has never referred to the doctrine inany of its judgments,78 legal journals and reviews in Europe and across theAtlantic are literally flooded with academic literature79 heavily criticisingthe doctrine. In addition, the judiciary has expressed strong reservationabout its adoption in the European legal order.80 The argument is that theessential facilities doctrine should not be applicable in competition law or atleast should be applied with great care because of its many negative effects.These considerations favour an interpretation of the Bronner decision aseither not confirming the doctrine at all or—at most—as adopting it butwith extreme prudence and only as regards property rights other than IPR.

4. CONCLUSION

The interpretation suggested in this chapter is one way to solve the appar-ent conflict in the case law. This interpretation is based on the wording usedin the cases themselves, with some policy reflections on the justification todifferentiate between different types of property in their interface with com-petition law. Perhaps the Community courts are clear on the constructionto be given to their cases. However, as shown by the divergent opinions ofcommentators over the interpretation to give to the cases, confusion stillremains in the legal community as regards the exact conditions under whichthe holder of an IPR abuses it.

The IMS case offers a unique opportunity for the European courts toclarify their case law and provide a transparent test to help IPR holders in a

74 Estelle Derclaye

78 Case C–7/97 Oscar Bronner v Mediaprint [1998] ECR I–7791, per Advocate-GeneralJacobs, para 35; Commission’s Decision 2002/165/CE of 3 July 2001, para 64, above n 30. 79 Among others, see P Areeda, ‘Essential Facilities: An Epithet in Need of Limiting Principles’(1989) Antitrust Law Journal 841; J Temple Lang, ‘Defining Legitimate Competition:Companies’ Duties to Supply Competitors and Access to Essential Facilities’ (1994) 18Fordham International Law Journal 439; D Ridyard, ‘Essential Facilities and the Obligationto Supply Competitors under UK and EC Competition Law’ (1996) European CompetitionLaw Review 438; A Overd and B Bishop, ‘Essential Facilities: The Rising Tide’ (1998)European Competition Law Review 183. See also for strong criticism, H Hovenkamp, FederalAntitrust Policy 2nd edn (St Paul, MN, West, 1999) 305–11. For other references on criticismof the essential facilities doctrine, see E Sheehan, ‘Unilateral Refusals to Deal and the Role ofthe Essential Facility Doctrine, A US/EC Comparative Analysis’ (1999) World Competition73, fnn 26 and 28.80 See Opinion of Advocate-General Jacobs in Bronner above n 78, at paras 56–58 (the doc-trine goes against freedom of contract, it stimulates competition on the short-term stimulationbut not on the long-term; para 62 (if the doctrine is to be applied at all, it should be appliedwith even more caution in cases dealing with IPR) and para 69 (difficulty to decide on thecompensation to be paid for access); see also Whish above n 4, at 617 and 700; Treacy aboven 11, 502.

dominant position know when they are likely to breach competition law. Itis hoped that the Court will, one way or the other, clarify the circumstancesin which the refusal to license by a holder of an IPR is abusive. It would beinteresting if the reading of the case law suggested in this discussion were tobe adopted by the court when deciding the IMS case on the merits. Thusthe Court would clarify or confirm what it meant in paragraphs 40 and 41of the Bronner case. It could also go further and state whether the condi-tions under which a refusal is abusive are similar for all property rights orwhether they should be different for intellectual property rights. If it decidesthat the conditions are different for the two types of rights, only the Magillcase should be relied upon for abuses by holders of IPRs and its conditionsshould be applied concurrently.

Even if the suggested construction is not embraced, adopting the condi-tions set out in Magill and applying them in a cumulative manner wouldensure, better than an application of the Bronner tripartite test or of theessential facilities doctrine, that the cases in which there will be an abuse ofan IPR remain truly exceptional. The case law should preferably evolveaccordingly in order to maintain the fragile balance between competitionand intellectual property laws.


4

B2B E-Marketplaces: A NewChallenge to Existing Competition

Law Rules?

JOACHIM LÜCKING*

BUSINESS-TO-BUSINESS (B2B) e-marketplaces constitute one ofthe most important developments of the ‘new economy’. They arealso an interesting case study for the attempts of competition

authorities to find a response to contractual agreements that, in the viewof industry itself, offer the potential to revolutionise purchasing and supply-chain management. This chapter briefly reviews the commercialdevelopments in the field of e-marketplaces and the history of the regula-tory response to them. It then proceeds to an in-depth discussion of thecompetition issues raised by e-marketplaces and the analysis theEuropean Commission conducts to examine them.

1. B2B E-MARKETPLACES: TECHNICAL AND COMMERCIAL DEVELOPMENTS

Business-to-Business (B2B) e-marketplaces are specialised Internet sites thatallow buyers and suppliers to meet each other virtually and to trade. Theirdevelopment in Europe can be traced back to 1996 when British Telecomestablished a Private Digital Exchange known as BT Trading Places.1 It didnot work successfully however and was withdrawn. E-marketplaces weremore widely developed from 1998 onwards.

* Dr Joachim Lücking is an Administrator in the European Commission, CompetitionDirectorate-General. All views expressed are personal and do not necessarily reflect those ofthe European Commission.1 European Commission, E-Marketplaces: New Challenges for Enterprise Policy, Competitionand Standardisation (Brussels, Workshop Report, 2001) 2.

Since then, the development of e-marketplaces has been marked by achange from euphoria to scepticism. Quantitatively, the number of suchmarkets exploded between 1998 and 2001. In 2000 around 200 market-places were active in Europe,2 in 2001 this number had grown to more than500.3 After the dot.com bubble burst, a number of ventures failed andfewer new marketplaces were created. In 2002, between 1000 and 1200marketplaces are estimated to be active worldwide, in Europe this numberis supposed to lie between 380 and 540.4

The number of all business-to-business transactions that are currentlyconducted through these exchanges is also comparatively low. Less thantwo per cent of all business-to-business transactions currently take place onthese exchanges. This number is however expected to increase sharply.5

There are four general market types under which there are many variations:

1. Buyer-Managed Exchanges: these are markets that are set up bylarge buyers, often in conjunction with technology partners. Anexample would be Covisint, a marketplace for car components,set up by General Motors, Ford, DaimlerChrysler, Renault andNissan as buyers, together with Oracle and i2 as technology partners.

2. Supplier-Managed Exchanges: these markets are being set up bysuppliers, such as the aircraft component producers UTC andHoneywell, who together with their technology partner i2, devel-oped a marketplace for aerospace equipment (MyAircraft.com).6

3. Market Makers: these are independent exchanges not controlledby buyers or sellers. They tend to be backed by venture capital andoften were early innovators. An example would be the internetauction house Freemarkets.

4. Content Aggregators: content aggregators are sites that build andmaintain multi-vendor catalogues which allow customers to accessthe offerings of several suppliers using a common search structure.

78 Joachim Lücking

2 UBS Warburg, Europe: The A-Z of B2B (London, UBS Warburg, 2000) 7.3 Estimate by Jupiter MMXI <http://uk.jupitermmxi.com/press/releases/20010212.jsp> (lastvisited on 20 July 2001).4 Estimates by Berlecon Research <http://www.berlecon.de/services/b2bbd> and eMarketService<http://www.emarketservices.com> (last visited on 26 April 2002).5 In 2000, UBS Warburg predicted that in 2005 16 % of all B2B transactions in Europe would be conducted through exchanges, above n 2, 3. Goldman Sachs expected this numberto be 13 %, see Goldman Sachs, The Old World Meets the New Economy: B2B in Europe—Book 2 (London, Goldman Sachs, 2000) 52. In 2002, Forrester Research estimated that nettrade in Europe will surge from 2% of total business trade in 2002 to 10% in 2004<http://www.forrester.com/>, (last visited on 26 August 2002).6 This exchange has since merged with the buyer-managed exchange AirNewCo to formCordiem.com.

Another useful categorisation can be built upon the scope of anexchange and its position in the value chain.7 Horizontal exchanges offeruniform, non-strategic products across multiple industries. Exampleswould be the maintenance, repair, and operations (MRO) exchangeGrainger.com or the office supplies provider Staples.com. Vertical mar-ketplaces, on the other hand, concentrate typically on one industry andtry to offer all the inputs, strategic and non-strategic, that are required tooperate in this industry. Examples would be Covisint for the automotiveindustry or Rubbernetwork for the tyre industry. In addition, there aremulti-industry vertical marketplaces which seek to create trading commu-nities in several industries. An example would be Freemarkets.

B2B electronic markets are generally considered as a potential sourceof significant efficiency gains. They generate efficiencies in three ways.First, they put a downward pressure on purchasing prices by increasingmarket transparency. Second, they decrease informational costs andexpand everyone’s market reach by removing the geographic barriers tobuyers and sellers efficiently discovering each other. Third, they allow areduction in transaction costs and an improvement of inventory manage-ment. Transaction costs may not only be reduced through the direct inter-action between buyers and sellers via the Internet, but also by allowingbuyers to form a more direct relationship with upstream manufacturerscutting out intermediaries (disintermediation). Estimates about the size ofthese gains vary: UBS Warburg, for example, predicted cost savings fromB2B electronic markets of seven per cent of turnover on average.8

2. REGULATORY RESPONSES

Regulators took note of the existence of e-marketplaces fairly early. InJune 2000 the US Federal Trade Commission (FTC) conducted an explana-tory workshop to examine competition issues in B2B electronic market-places. As a result, the FTC published a staff report in October 2000.9

The UK Office of Fair Trading (OFT) ordered a study on ‘E-Commerceand Its Implications for Competition Policy,’ which it published in August 2000.10 The OECD held a ‘Mini-Roundtable on ElectronicCommerce’ in October 2000 with written contributions from competition

B2B E-Marketplaces 79

7 DA Devine, CB Dugan, ND Semaca and KJ Speicher, ‘Building Enduring Consortia’ (2001) 2The McKinsey Quarterly 28.8 UBS Warburg above n 2, 4.9 Federal Trade Commission, Entering the 21st Century: Competition Policy in the World of

B2B Electronic Marketplaces (Washington, 2000).10 Office of Fair Trading, E-Commerce and Its Implications for Competition Policy, preparedby Frontier Economics (2000, OFT Report 308).

authorities of 10 jurisdictions.11 The EuropeanCommission, finally, followed a twin-track approach. On the one hand, it began to developand publicise the principles for the assessment of e-marketplaces in aseries of conferences.12 On the other hand, it gained practical experiencethrough a number of notifications, both under Article 81 and theEuropean Community Merger Regulation (ECMR).

As a result of this public debate, one can identify a common ground ofissues which competition authorities across the world consider as impor-tant when analysing B2B e-marketplaces. These issues also feed into someform of implicit ‘guidelines’ which industry seems to follow when settingup exchanges in order to pre-empt possible competition concerns and ofteneven the need for a notification.

3. THE LEGAL ASSESSMENT OF B2B E-MARKETPLACES UNDER EC COMPETITION LAW

A. Legal Basis

B2B e-marketplaces may be subject to European Community (EC) competi-tion law at different stages in their life cycle. First, the creation of anexchange may constitute a concentration under the ECMR.13 This is thecase if the conditions of Articles 1 and 3 ECMR are fulfilled. Article 1 setsthe turnover thresholds which are used to define the concept of ‘communitydimension’. This concept is employed to divide competence betweenMember States and the Commission in the field of merger control. Accountis taken not of the turnover of the marketplace but of the turnover of theparent companies. In general, their combined aggregate worldwide turnovermust exceed e5000 million and the aggregate Community-wide turnover ofeach of at least two parent companies must exceed e250 million.14

Article 3 defines the notion of a ‘concentration’. In the case of the cre-ation of a joint venture (JV) in particular, it requires that the JV perform ona lasting basis all the functions of an autonomous economic entity. This isnormally the case for B2B e-marketplaces which are generally endowedwith all the functions necessary to operate in an industry and which do not

80 Joachim Lücking

11 OECD, Competition Issues in Electronic Commerce (DAFFEE/CLP(2000)32).12 Conference ‘The E-Economy in Europe: Its Potential Impact on EU Enterprises and Policies’on 1–2 March 2001 in Brussels; Conference ‘E-Marketplaces: New Challenges for EnterprisePolicy, Competition and Standardisation’ on 23–24 April 2001 in Brussels.13 Council Reg (EC) No 4064/89 [1989] OJ L/395/1, 30 December 1989, last amended byCouncil Reg (EC) No 1310/97 [1997] OJ L/180/1, 9 July 1997.14 Art 1(2) ECMR. In the cases set out in Art 1(3) ECMR, lower turnover figures may besufficient to indicate a ‘Community dimension’.

only provide services to the parent companies. Another important criterionis that the JV must be controlled by one or more companies. This is oftenthe case if a B2B e-marketplace is set up by a small number of large indus-trial companies which jointly control the electronic market. However, if thenumber of shareholders gets too large, no individual shareholder and noclearly identifiable coalition of shareholders may exercise control. In thiscase, the agreement to create an exchange does not constitute a merger butmay be restrictive of competition in the sense of Article 81 EC.

Secondly, the operation of an exchange must comply with Articles 81and 82 EC. Article 81 covering restrictive agreements could for instancebe applicable to agreements between several buyers on an exchange whowant to bundle their purchasing volumes. Article 82, which deals withthe abuse of a dominant position, could be relevant as regards questionsof access to an exchange which has developed into the dominant electronic market.

B. Market Definition

The definition of the relevant antitrust market is the starting point foralmost all antitrust examinations. In many cases, the outcome of the mar-ket definition process also determines the outcome of the assessment. TheCommission has described the principles of market definition in a Notice.15

These principles remain valid in the context of B2B electronic markets.Their application may become more difficult however due to the lack ofreliable sales and price data and the current speed of change in e-commercemarkets generally.16 In particular, the definition of the relevant productmarket will raise two sets of questions:

First, whether electronic marketplaces compete with ‘normal’ bilateralsales or whether they constitute a separate, narrower product market. Theformer would be likely if the parties used electronic marketplaces only asan additional sales channel; the latter if the exchange offered additionalservices which clearly differentiated them from other sales forms.

These questions were first discussed in the Commission’s clearancedecision of MyAircraft, a B2B exchange for aircraft parts and services.17

In this case, the Commission investigated the question whether this on-line exchange was part of the wider market for airline equipment orwhether it constituted part of a narrower market for exchanges(exchanges for airline equipment).


15 Commission Notice on the Definition of the Relevant Market for the Purposes ofCommunity Competition Law [1997] OJ C/372/5, 9 December 1997.16 See OFT above n 10, 37.17 Case M–1969 —UTC/Honeywell/i2/MyAircraft.com, Art 6(1)b—decision of 4 August 2000.

The Commission’s market investigation revealed that industry participantsin general considered the B2B e-marketplace as one segment among themany modalities by which companies transact business. The exchangewould increase the efficiency of communications between aerospace indus-try participants without changing the way transactions were conducted inthe aerospace industry.

The Commission also considered whether specific services offered by theexchange constituted product markets in their own right. In this case, theservices offered by MyAircraft to its customers included supply chain man-agement tools and e-procurement. To a large extent these services were con-sidered to be an integral part of the services offered by MyAircraft in orderto enable customers of the site to use MyAircraft as a purchasing or sellingtool. However, some elements of the supply chain management service wereregarded as going beyond what is normally required by a user of MyAircraftin order to use this site to make business. This would be the case for theinventory planning tools and forecasting tools in particular. The marketinvestigation revealed, however, that a majority of third parties consideredthat these services would be distinct components that may be offered separately.

The results of the investigation seem to suggest that the B2B electronicmarketplace constitutes part of a wider market. It should be noted, how-ever, that in this case the precise relevant product market definition was leftopen since, irrespective of the market definition chosen, the proposed con-centration did not give rise to the creation or strengthening of a dominantposition. All decisions since have followed this line, without arriving at anyformal conclusion.

A second question that might arise when defining the relevant productmarket is whether or not distinctions can be drawn between different B2Be-marketplaces based on their industry focus. A currently open questionconcerns the degree of substitution between vertical exchanges that are setup to cater to a given industry, but which may be open to outside buyersand horizontal exchanges which cut across industries but only offer certaingoods or services. An example would be a car producer that needed MROgoods: it could turn either to a vertical exchange, such as Covisint, a hori-zontal exchange, such as Grainger, or to an auction house, such asFreemarkets. For other, more car-specific goods, its choice might be morelimited because there may not be a horizontal exchange or the good maynot lend itself to an auction. This suggests that the demand side might befaced with a continuum of choices, depending on the number and nature ofB2B e-marketplaces that allow trading in a specific good. Again, one shouldnote that the Commission’s decisional practice has so far not addressed thisissue, as the marketplaces analysed so far did not create competition problems irrespective of the market definition adopted.

82 Joachim Lücking

When looking at the definition of the relevant geographic market, therelevant question is likely to be whether the geographic market will bewidened, as location becomes less important for the interaction betweenbuyers and sellers. One can expect that such a widening of the geographicmarket will indeed be brought about by many B2B electronic marketplaces.However, the Commission did not need to pronounce on this issue, as eitherthe markets concerned were already global (such as the market for aircraftparts), or the creation of the marketplace did not raise any competition concerns irrespective of the delineation of the geographic market.

C. Competition Issues

Once the relevant market is defined, one can turn to the competitionissues which B2B e-marketplaces potentially raise. Here, a distinction canbe made between issues relating to co-ordination effects which could bebrought about by an exchange and issues relating to market dominanceand foreclosure.

D. Co-ordination Effects

The likelihood of co-ordination effects has been one of the most discussedfeatures of B2B exchanges. The problem arises as B2B electronic market-places not only increase the transparency in the market, but also facilitatethe exchange of sensitive information between competitors. For competi-tion policy, such an information exchange is of concern as communicationis central to collusion.18

A problem of information sharing is created if some or all buyers or sell-ers can use an electronic market in order to discover or to exchange sensi-tive information on prices and quantities. This concern is linked to thedesign of the system, in particular its openness in terms of individual dataoriginating with other parties. When analysing this concern, a distinctioncan be made on the basis of whether or not information is exchangedbetween all market participants or only between a sub-group of the market,in particular the owners.

Whether or not information sharing between all market participantscreates a competition problem depends very much on the nature of themarket. In EC competition law the exchange of sensitive and detailed


18 K-U Kühn, ‘Fighting Collusion by Regulating Communication Between Firms’ (2001)Economic Policy 197.

information might as such be caught by Article 81(1) EC if it takes placeon an oligopolistic market. The judgments of the European Courts in theTractor cases19 and the Steel Beams cases20 provide useful clarification inthis respect.

It emanates from these judgments that in addition to the market struc-ture, the following elements need to be taken into account when assessingthe potential impact of information sharing: the type of informationexchanged, especially the level of aggregation, the age of the informationexchanged and the frequency of information exchange.

B2B electronic marketplaces are thus less likely to raise concerns if theyonly provide summary statistics (eg on trade volumes) to all market partic-ipants while individual data is only accessible to the owner of such data.Similarly, competition concerns are reduced when the system only allowsaccess to historical data. In the example of an auction market, this couldfor instance be done by only showing the leading bid without revealing theidentity of the bidder and by not specifying the order of other bids.

Enforcement practice shows that marketplace operators are well awareof the need to ensure data protection and to impede improper informationexchanges. They are therefore setting up ‘firewalls’ and use other technicalmeans which ensure that data flows are limited to necessary informationand that they can be controlled. Antitrust agencies will in turn need to pro-vide or acquire the technical expertise that enables them to judge whetheror not these safeguards are sufficient.

A different concern is raised if only a few market participants haveaccess to certain information. This problem may exist where an on-linemarketplace is controlled by a number of market participants. Theseowner-participants could then receive privileged information about trans-actions in the market which would create competition problems relatingto both information sharing and discrimination.

This issue was addressed in the Volbroker case, the first B2B exchangecleared under Article 81. In this case, six major banks set up a joint ventureoffering an electronic brokerage service for trading foreign currencyoptions. The case raised concerns regarding the access to confidential infor-mation by the parent companies. To deal with this concern, the owners ofthe Volbroker.com exchange gave the following assurances to theCommission:

None of Volbroker.com’s staff or management will have any contractual orother obligation towards any of the Parents and vice versa. Volbroker.com’s

84 Joachim Lücking

19 Case C–8/95/P New Holland Ford v Commission [1998] ECR I–565 and Case C–7/95/PJohn Deere v Commission [1998] ECR I–3111.20 Cases T–134/94, T–136/94, T–137/94, T–138/94, T–141/94, T–145/94, T–147/94,T–148/94, T–151/94, T–156/94 and T–157/94 (‘Steel Beams’ Cases) [1999] ECR II–0347.

staff and management will be in a geographically distinct location from thatof the Parents. The representatives of the Parents on Volbroker.com’s Boardof Directors will not have access to commercially sensitive informationrelating to each other or to third parties. The Parents will not have access tothe information technology and communication systems of Volbroker.com.The Parents will also ensure that the staff and management of all the partiesunderstand and appreciate the importance of maintaining the confidential-ity of sensitive commercial information and that sanctions for breach arespelled out.21

To alleviate further concerns related to market access, the parent companiesalso agreed to allow so-called voice brokers to participate in Volbroker.comwhere they are acting as principals.

The undertaking thus aims at building ‘Chinese walls’ between the jointventure operating the exchange and the parent companies which are activeas market participants. While such a separation can certainly be regardedas a necessary condition to ensure that the marketplace is operated withsufficient independence from the parents, it remains to be seen whether it isalso a sufficient condition. Undertakings involving ‘Chinese walls’ are diffi-cult to monitor for a competition authority. While they might be appropri-ate for financial services industries which are used to such provisions andwhich have established a certain ‘compliance culture’ based on the financialmarkets rules, such undertakings may not always be sufficient for otherindustries or market situations.

A second problem regarding possible co-ordination in electronic marketsrelates to the question whether market participants can effectively bundlepurchasing or selling volumes. This question is in principle not differentfrom ‘normal’ joint purchasing or joint commercialisation. Therefore, thediscussion of these questions in the Commission’s ‘Guidelines on theApplicability of Article 81 EC to Horizontal Co-operation Agreements’constitutes a good starting point for the assessment under EC law. It isnotable in this respect that these Guidelines propose a safe haven of 15 per cent market share below which a purchasing or commercialisationagreement would be assumed either not to restrict competition or to fulfilthe conditions for an exemption.22 These safe havens would also apply tohorizontal agreements involving e-commerce. Thus, joint purchasing in anexchange by companies whose combined market share is below 15 per centon both the purchasing and the selling market would not breach Article 81(1).


21 ‘Commission Approves the Volbroker.com Electronic Brokerage Joint Venture between SixMajor Banks’, Commission Press Release IP/00/896 of 31 July 2000.22 Guidelines on the Applicability of Article 81 of the Treaty to Horizontal Co-operationAgreements [2001] OJ C/3/2, 6 January 2001, paras 130 and 149.

Many B2B electronic markets are, however, addressing the possiblecompetition problems of horizontal co-operation not by reference to mar-ket shares but by only allowing joint purchasing or commercialisation ofaccessory or MRO products. For example, an exchange for the widgetindustry would only provide for joint purchasing of office supplies whilethe widgets themselves would be bought individually by the member ofthe exchanges.

This would not necessarily bring the joint buying activities within theboundaries of the safe haven provided by the Guidelines. Collectively, theproducers of widgets are unlikely to have a market share larger than 15 per cent in the buying market for office supplies. However, their mar-ket share on the selling market for widgets may be larger than 15 per cent.They would therefore not fulfil the conditions of the safe haven.

It must nevertheless be acknowledged that such a set-up would reducethe risk of collusion significantly. An individual assessment would thereforeprobably lead to the result that Article 81(1) is not infringed. This becomesvery clear from the discussion in the horizontal guidelines.

The Guidelines make clear that joint purchasing can be a problem fortwo reasons: (1) because it creates buyer power; and (2) because it can leadto co-ordination on the downstream market.23 If we assume that the mar-ket share of the widget producers in the buying market for office suppliesis lower than 15 per cent, we can already conclude that buyer power isprobably not a problem.

Competition problems on the downstream market could exist if the par-ties achieve a high communality of costs through joint purchasing.24 Officesupplies, however, are unlikely to constitute a high proportion of the totalcosts of a widget producer. Therefore, the joint purchasing of these prod-ucts is also unlikely to create competition problems on this front. The jointpurchasing of office supplies is unlikely to pose a competition problem.

E. Market Dominance and Foreclosure

A second set of possible competition problems relates to issues of marketdominance and foreclosure. They are created through the network charac-ter which is inherent in B2B electronic marketplaces. Positive networkexternalities and potential problems of network dominance are presentwhen the value of a system to the individual user increases with the num-ber of users.25 They can lead to market ‘tipping’ and the creation of a

86 Joachim Lücking

23 Ibid paras 127–29.24 Ibid para 128.25 M Katz and C Shapiro, ‘Network Externalities, Competition, and Compatibility’ (1985)75(3) American Economic Review 424.

dominant position if the network effects are strong enough to induce allmarket participants to use the same network. This problem could poten-tially arise in the context of B2B electronic marketplaces, as the benefitswill often increase with the number of buyers and suppliers which arelinked to the same system.

The possible prevalence of network effects in B2B electronic markets cre-ates a dilemma for competition policy. On the one hand, it needs to takeaccount of the fact that a larger network can be the source of substantialefficiencies. The fact that exchanges try to sign up as many industry playersas possible should therefore not be considered as a competition problem initself. On the other hand, competition policy needs to acknowledge thatnetwork effects can lead to a ‘tipping’ effect which will substantially raisebarriers to entry and expansion and which could create substantial marketpower for the owner-operator of the largest exchange.

The likelihood of such ‘tipping’ effects will depend on the value thatbuyers and sellers put on liquidity. Liquidity is important for B2B elec-tronic marketplaces which operate as true exchanges. Such exchanges arecharacterised by the interaction of many buyers and sellers and thedynamic setting of a market-clearing price. They therefore require a suffi-cient amount of liquidity to operate.26

Most B2B electronic marketplaces, however, do not seem to constitute‘exchanges’ in the sense of a commodity or stock exchange, as there is notrade in standardised products at a market price in an anonymous transac-tion using intermediaries. Most B2B electronic marketplaces are ratherfacilitating devices, which allow buyers and suppliers to engage directly inindividual transactions. These transactions could take the form of an auc-tion, a reverse auction or of a vendor catalogue. In all these cases, the criti-cal success factor is not so much the volume of actual transactions, butrather the number of buyers and sellers connected to the system thatactively monitor it and which could thus potentially make an offer.

Network dominance will be harder to achieve in such a context. Anyoperator trying to build a dominant position would need to base thisattempt not only on a ‘tipping effect’, but also on the need to create otherlock-in mechanism such as exclusivity provisions. Exclusivity provisionscan interact with network effects and create substantial barriers to entry.For users willing to switch B2B exchanges the combination of networkexternalities and exclusivity provisions creates ‘what can be a prohibitiveopportunity cost of joining the new network: cutting themselves off fromthe larger, established network’.27 Lock-in could also be achieved where


26 L Benzoni, ‘La Place de Marché est-elle proconcurrentielle?’ (2001) 121 Revue de laConcurrence et de la Consommation 7, 8.27 C Shapiro, ‘Exclusivity in Network Industries’ (1999) 7 George Mason Law Review 673.

market participants are tied into the market via proprietary supply chainmanagement systems.

Competition policy should therefore carefully assess the design of themarket and any ancillary provisions to ensure that the owner–operatordoes not try to enhance any existing network effects by contractual orother means. Competition authorities should in particular generally notaccept provisions which require the users of an electronic marketplace topurchase all its requirements of a certain good through an exchange.Minimum purchase requirements expressed in absolute or percentageterms, or other contractual terms which indirectly seek to achieve the samegoal should similarly be seen with suspicion.

Nevertheless, such requirements may be acceptable during the start-upphase of a marketplace in order to provide a minimum level of liquidity. Inthe emaro case, where a B2B electronic market for office furniture andequipment was set up by Deutsche Bank and SAP, the Commission acceptedthat Deutsche Bank would contribute a minimum of 68 per cent to the jointpurchases of both parent companies through the exchange. This clause wasconsidered as directly related to and necessary for the setting-up of the mar-ket, provided its duration did not exceed three years.28

So far, the Commission’s enforcement experience with B2B electronicmarkets has not revealed problems related to network effects and marketdominance. This is due to the fact that in these cases several e-marketplacescompeted heavily even in a narrowly defined market.29

In the future, a shakeout phase leading to a reduction in the number ofmarketplaces seems, however, unavoidable.30 This raises the question ofthe appropriate regulatory response as regards access to the remaining mar-ketplaces. Problems of foreclosure could arise if marketplaces, especiallythose owned by industry participants, exclude certain participants from themost efficient trading platform, consequently putting them at a competitivedisadvantage.

These problems could be addressed under Article 82 EC, provided thatan exchange occupies a dominant position and that it can be considered asan essential facility.31 In this case, an exchange would abuse its dominant

88 Joachim Lücking

28 Case M–2067—Deutsche Bank/SAP/JV, Art 6(1)b—decision of 13 July 2000.29 See eg the decisions in cases M–1969—UTC/Honeywell/i2/MyAircraft.com, Art 6(1)b—decision of 4 April 2000; M–2270—Babcock Borsig/mg technologies/SAP Markets/DeutscheBank/VA Tech/ec4ec, Art 6(1)b—decision of 22 January 2001; M–2398—Linde/Jungheinrich/JV, Art 6(1)b—decision of 25 April 2001.30 Jupiter MMXI, a research company, predicts, for instance, that ‘fewer than 100 NetMarkets will survive out of the 500 existing today’: see <http://uk.jupitermmxi.com/press/releases/ 20010212.jsp> (last visited on 20 July 2001).31 On essential facilities in general, see C Esteva Mosso and S Ryan, ‘Article 82—Abuse of aDominant Position’ in J Faull and A Nikpay (eds), The EC Law of Competition (Oxford,Oxford University Press, 1999) 159–62.

position if it did not grant access, unless such refusal is objectively justified.The test set out by the European Court of Justice in the Oscar Bronnerjudgment is fairly strict. It requires that the refusal is likely to eliminate allcompetition in neighbouring markets; that access must be indispensable forthe competitor to carry on its business in that there is no actual or potentialsubstitute in existence to the facility; and that the refusal is incapable ofbeing objectively justified.32 Application of the essential facilities doctrinewould thus require not only that the exchange was dominant in relation toother exchanges but also the proof that conventional means of distributionwere no longer a substitute to trading through an exchange.

While application of the essential facility doctrine to B2B electronicmarketplaces remains an uncertain proposition,33 operators of these mar-ketplaces may nevertheless find it in their interest to create a marketplacethat is as open as possible. In most circumstances, this will make com-mercial sense and it will also help to refute any allegations relating to thepossible abuse of a dominant position. However, openness does notrequire that an operator admit every interested party to a marketplace.Earlier Commission decisions concerning a number of commodityexchanges34 suggest that admission standards would generally seemacceptable, provided that the standards are objectively necessary and areapplied on a non-discriminatory basis.

4. BEST PRACTICE GUIDELINES

From the discussion above, one can deduce a number of rules that compa-nies setting up e-marketplaces may want to follow to ensure that EC com-petition rules are not infringed. Elements of such ‘guidelines’ are:

1. open, non-discriminatory access from all interested buyers andsellers;

2. no provisions which directly or indirectly try to impose the exclu-sive use of the exchange by its participants;

3. joint purchasing or commercialisation only within the boundariesset by the Commission’s horizontal guidelines;


32 Case C–7/97 Oscar Bronner GmbH & Co v Mediaprint Zeitungs-und ZeitschriftenverlagGmbH & Co KG [1998] ECR I–7791, 41.33 Other commentators are, however, of the opinion that B2B e-marketplaces may quickly turninto essential facilities: see F Stroud, ‘B2B E-Marketplaces—The Emerging Competition LawIssues’ (2001) 24(1) World Competition 125, 134.34 London Sugar Futures Market [1985] OJ L/369/25, London Cocoa Terminal MarketsAssociation [1985] OJ L/369/28, Coffee Terminal Market Association of London [1985] OJ L/369/31.

4. structural separation between the exchange and its parents whichis supported by ‘Chinese walls’;

5. credible data protection and safeguards against the exchange ofinformation.

EC enforcement experience shows that many e-marketplaces seem to fol-low these principles. The press release issued on the occasion of theCovisint clearance35 states for instance that ‘the agreements show thatCovisint is open to all firms in the industry on a non-discriminatory basis,is based on open standards, allows both shareholders and other users toparticipate in other B2B exchanges, does not allow joint purchasingbetween car manufacturers or for automotive-specific products, and pro-vides for adequate data protection, including firewalls and securityrules’.36

Adherence to these ‘guidelines’ may also explain the relatively smallnumber of notifications under Regulation 17/62. Most companies settingup e-marketplaces seem to assess themselves whether or not Article 81(1) isinfringed and arrive at the conclusion that their agreement is not restrictiveof competition. The Commission should welcome this development as itcorresponds to its proposals for the future application of Article 81.37 Itobviously has the option of examining any such non-notified agreement,either on its own initiative or following a complaint, if an e-marketplacethreatens to create competition problems.

5. CONCLUSION

B2B e-marketplaces have generated a large degree of interest both in thebusiness world and the antitrust community. The theoretical debate anddecisional practice has however shown that the current competition rulesare fully capable of dealing with this new phenomenon because most ema-nations of the ‘new economy’ B2B e-marketplaces are not so unique thatthey cannot be analysed within the existing framework.

90 Joachim Lücking

35 This joint venture between General Motors, Ford, DaimlerChrysler, Renault, Nissan, Oracleand i2 was found to be not restrictive of competition. Accordingly, the Commission issued anegative clearance ‘comfort letter’.36 ‘Commission clears the creation of the Covisint Automotive Internet Marketplace’,Commission Press Release IP/01/1155 of 31 July 2001.37 Proposal for a Council Regulation on the implementation of the rules on competition laid down in Art 81 and 82 of the Treaty and amending Reg (EEC) No 1017/68, (EEC) No 2988/74, (EEC) No 4056/86 and (EEC) No 3975/87 (‘Regulation ImplementingArticles 81 and 82 of the Treaty’ [2000] OJ C/Sol.365, 19 December 2000).

5

Authorities, Competition andElectronic Communication: Towards

Institutional Competition in theInformation Society

PLG NIHOUL*

THIS CHAPTER EXPLORES the links existing between authorities1

and competition, in the context of measures adopted in electroniccommunications within the European Union over the last 15 years.

As we know, these measures have given rise to numerous commentariesmost of them concentrating on economic matters.2 These economic

* Prof Dr PLG Nihoul is Professor of Law, Jean Monnet Chair on the European InformationSociety, Director of the Centre on Consumer Law (Consumer Choice), University of Louvain,Belgium; and Professor of Law, University of Groningen, The Netherlands. The paper on whichthis chapter is based was presented at a workshop held at the University of Leicester on the rela-tionship between competition law and the information society. It was also delivered as the inau-gural speech when the author was appointed as a Professor of Law at the RijksuniversiteitGroningen in May 2002. A first version has been published under the same title in (2001) TheJournal of Policy, Regulation and Strategy for Telecommunications, Information and Media.This is a slightly different version, where more emphasis is placed on the relationship betweeninstitutional competition and the information society. I would like to thank the editor of theaforementioned journal for allowing republication and the editor of this book for taking inter-est in the manuscript although already published in an earlier version.1 This latter term is used in the chapter without any technical meaning. It generally refers toorgans established by, or under the control of, the European Union, the Member States or acomponent thereof. It thus designates a public entity the task of which is to carry out missionsentrusted to it by the above mentioned levels of power. I have avoided using ‘institution’ asthis has a special meaning in European law where it refers to specific bodies: the EuropeanParliament, the Court of Justice of the European Communities, the Council of Ministers, theEuropean Commission and the Court of Auditors. See LWG Gormley, Introduction to theLaw of the European Communities (Deventer, Kluwer Law International, 1998) 181.2 Thus, some commentators have glorified the introduction of competition on markets thatwere formerly organised on other bases. They have underlined, in substance, that competitionleads to greater efficiency and hence produces better economic results. Others have by contrastlamented that competition would probably imply the disappearance of a ‘public service’

considerations will not be addressed in this discussion, which insteadfocuses on institutional issues. For many analysts, rules are relevant in sofar as they modify behaviour: norms deserve study because they influencesociety. That opinion is certainly worthwhile, but I am inclined to thinkthat reasons explaining the adoption of rules should also be taken intoaccount. Among these reasons, those with an institutional character oftenplay a significant role.

The chapter is constructed as follows. First, I will introduce the majordevelopments that have shaped, in my opinion, the institutional side of thereform that has recently taken place in European electronic communications.Second, I will try to analyse what role considerations of an institutionalnature may have played in the process. Third, I will embark on a prospectiveventure and consider what we can learn from these developments for thefuture of public intervention.

I. NATIONAL SYSTEMS REPLACED BY A EUROPEANORGANISATION

A. First Stage of the Reform

The first development is related to the adoption of the instruments whichofficially launched reform within the Community.3 Prior to this, electroniccommunications were characterised by national monopolies within theEuropean territory. Member States entrusted one undertaking—often adepartment within their administration—with the task of installing andoperating a network in their national territory. This undertaking was alsoresponsible for providing services through infrastructure, limited to voicetelephony in many instances. The monopolistic undertaking had an exclu-sive right to commercialise terminal equipment and to establish the techni-cal requirements to be satisfied for connection.

That situation was deemed unsatisfactory by the European Commission.4

Traditionally it has been argued that this reaction was motivated by a

92 PLG Nihoul

mentality. They argue a profit making attitude would further develop in society as a result ofthe necessity of seeing all activities on the basis of their cost.

3 For a general presentation on EU law on electronic communications, see L Garzaniti,Telecommunications, Broadcasting and the Internet (London, Sweet & Maxwell, 2000); P Larouche, Competition Law and Regulation in European Telecommunications (Oxford,Hart Publishing, 2000); P Nihoul, Droit européen des télécommunications—L’organisationdes marchés (Brussels, Larcier, 1999); Simmons & Simmons, Telecommunications: The EULaw (London, Palladian Law Publishing, 1999); I Walden and J Angel, TelecommunicationsLaw (London, Blackstone Press, 2001).4 European Commission, Towards a Dynamic European Economy—Green Paper on theDevelopment of the Common Market for Telecommunications Services and Equipment, COM

comparison with the state of affairs in the United States: (1) American customers seemed to have access to better communications at lower tariffs;(2) American companies appeared to have a technological advantageEurope would not be able to compensate without immediate action;5

(3) regulatory action was undertaken in America to encourage furtherdevelopment in the communications sector in that country; it was fearedthat Europe would definitely be left aside if no similar move was taken onthis side of the Atlantic.6

B. The 1987 Green Paper

On the basis of these findings, the Commission published a Green Paper7

where it outlined the measures it thought should be taken in order to sup-port and develop electronic communications in Europe. Among other meas-ures, this document suggested the introduction of competition in thetelecommunications markets (liberalisation).

This proposal did not raise much enthusiasm among the Member States.It was indeed running counter to the system that had governed electroniccommunications for decades within the national territories. These activitieshad been consistently reserved to one undertaking by the Member States.The European Commission was proposing to set that system aside and tolet all interested firms enter the markets. Member States voiced their oppo-sition, but the Commission decided to go ahead.8 It successively adoptedtwo Directives which converted the liberalisation goals envisaged in theGreen Paper into law. Competition was thus successively introduced in the

Authorities, Competition and Electronic Communication 93

(87) 290 final (30 June 1987); H Ungerer, Télécommunications en Europe: le libre choix pourl’utilisateur sur le grand marché européen de 1992 et l’enjeu pour la Communauté européenne(Brussels, Office des Publications Officielles des Communautés Européennes, Luxembourg,1988).

5 Innovations made in communications are essential to sustain economic growth. Firstly, com-munications per se constitute an important sector in the economy. Growth in that sector hasa substantial effect on economic activity. Secondly, innovations in communications have animpact on other activities. Economic activities generally imply and presuppose informationand data exchange. Innovation in that field acts as leverage to increase dynamism in othersectors.6 Y Benkler, Rules of the Road for the Information Superhighway: Electronic Communicationsand the Law (St Paul, MN, West Publishing, 1996); HJ Brands and ET Leo, The Law andRegulation of Telecommunications Carriers (Norwoord, Artech House, 1999); CH Kennedy,An Introduction to US Telecommunications (Norwoord, Artech House, 2001).7 Towards a Dynamic European Economy—Green Paper on the Development of theCommon Market for Telecommunications Services and Equipment, COM (87) 290 final (30 June 1987). 8 For an excellent account see P Larouche, Competition Law and Regulation in EuropeanTelecommunications (Oxford, Hart Publishing, 2000) 1–17.

markets for terminals,9 services10 and infrastructure.11 Through theseinstruments, the Member States were instructed to suppress the special andexclusive rights they had granted to national operators.12

C. What Legal Basis?

As a legal basis for these instruments, the Commission invoked a treatyprovision that had not been used earlier—Article 90 EC renumbered toArticle 86 EC after the Treaty of Amsterdam.13 In its first paragraph, thatprovision makes it clear that Member States must respect treaty provisionsas regards undertakings they have close relationships with. These relation-ships may be in the form of participation in the capital of these entities.They may also derive from regulatory advantages granted to these under-takings in order to facilitate their activities (special or exclusive rights).Pursuant to Article 86(1) EC, Member States may not take any measurecontrary to the treaty in respect of these undertakings. All rules containedin the treaty are concerned, but Article 86 EC specifically refers to thoserelating to competition.

The second paragraph of Article 86 EC concerns undertakings entrustedby Member States with the operation of services of a general economicinterest. According to that provision, these undertakings may be granted anexemption with respect to the rules of the treaty, including those on compe-tition. As a result of such an exemption, the provision, an undertaking con-cerned is exempt from the relevant rules. An exemption may only begranted where a legitimate objective is pursued. Furthermore, it onlyapplies in so far as it is useful and necessary to allow firms to perform theparticular tasks assigned to them in connection with such an objective.

94 PLG Nihoul

9 Dir 88/301 of 16 May 1988 on competition in the markets in telecommunications terminalequipment [1988] OJ L/131/73. 10 Dir 90/388 of 28 June 1990 on competition in the markets for telecommunications services[1990] OJ L/192/10.11 Dir 96/19 of 13 March 1996 amending Dir 90/388 with regard to the implementation offull competition in telecommunications markets [1996] OJ L/174/13. All directives adopted bythe Commission in electronic communications may be found in Simmons & Simmons,Telecommunications: The EU Law (London, Palladian Law Publishing, 1999). They can alsobe found on the internet at �http://europa.eu.int/information_society/topics/telecoms/regula-tory/index_en.htm�.12 The vocabulary then used by the Commission echoes words used during the FrenchRevolution. The Commission announced that special and exclusive rights would be abolis, inthe same way that the sans-culottes required in 1789 the abolition of the Monarchy, the AncienRegime and the privileges that were attached to these institutions. Dir 88/301, Art 2 and Dir 90/388, Art 2—both in the French version.13 All articles mentioned in this paper are numbered after the changes introduced in the Treatyof Amsterdam. See mainly JL Buendia Sierra, Exclusive Rights and State Monopolies underEC Law—Article 86 (Former Article 90) of the EC Treaty (Oxford, Oxford University Press,2000).

A relationship can be established between the first and the secondparagraphs of Article 86 EC. As they are undertakings with an exclusiveright, monopolies fall under the rules of competition as provided by thefirst paragraph. That situation does not appear, however, to conformentirely14 to the requirements of competition. Competition normallyrequires the presence of several undertakings, all struggling to obtainresources from potential partners. In order to justify the apparent contra-diction, Member States referred to the second paragraph of Article 86 EC.Their argument was that exclusive rights were necessary to fulfil objectivesof a public nature. According to them, the monopoly holders had to be pro-tected against rivals. That was a condition for these monopolists to achievethe public objectives assigned to them. These objectives could not bereached in a competitive environment alone.15 An exemption was thus war-ranted, along the lines laid down in Article 86(2) EC.

D. An Institutional Provision

In addition to regulating monopolies and the general interest, Article 86 ECcontains an institutional measure (paragraph 3). Pursuant to it, theCommission has power to adopt directives or decisions as deemed necessaryto implement the first and second paragraphs. Thereby, the Commission isgranted power to ensure the application of treaty provisions to firms withspecial or exclusive rights (paragraph 1).16 The Commission also has the


14 The Court of Justice has repeatedly insisted that exclusive rights should not as such be considered contrary to the rules of competition. However, it has ruled that the exercise of suchrights could be abusive in violation of Art 82 EC (abuse of dominant position). Later, itadopted the line that such exercise was necessarily abusive. According to the Court, the mereexercise of a right of that nature will lead the firm to abuse its dominant position. Thereby,the Court’s interpretation endorses the general economic literature in which a firm will tendto increase prices and diminish output where it is in a monopolistic situation. P Nihoul, Laconcurrence et le droit (Paris, EMS, 2001) 236–38.15 A typical example is the natural monopoly. Previously, the Member States justified exclusiverights granted to their telecom operator by arguing that these rights merely stated in law astructure that was imposed by the markets. According to them, it was less costly for society tofinance one infrastructure and concentrate demand on it. The alternative—letting competitorsinstall their own network—would lead to redundancy, losses and bankruptcy. Granting exclu-sive rights further allowed the national authorities to impose public policy objectives on firmsholding the monopoly. For instance, these firms were ordered in various Member States towire the whole national territory irrespective of the cost such operation would imply for someregions that could not be accessed easily. On that point, profit making or even financial bal-ance were considered less important than the establishment of a national infrastructure. FM Scherer, Industrial Market Structure and Economic Performance (Boston, HoughtonMifflin Company, 1980); W Sharkey, The Theory of Natural Monopoly (Cambridge,Cambridge University Press, 1982); R Schmalensee, The Control of Natural Monopolies(Lexington, Lexington Books, 1979).16 As well as to public undertakings.

authority to verify that exemptions are conferred in accordance with theconditions laid down in the treaty (paragraph 2).

On the basis of that provision, the Commission claimed a legal mandateto organise the telecom markets in Europe on a competitive basis. It consid-ered that competition rules had to be applied to that sector (paragraph 1).According to the Commission, competition implied that markets should inprinciple be opened to new entrants. Exemptions could admittedly beaccepted (paragraph 2). They would however have to be assessed with careand precaution.

(1) As to the scope, these exemptions should be limited to specificservices—those recognised as being of general economic interestin national legislation on the basis of criteria laid down atEuropean level (universal service obligations).

(2) As for modalities, exemptions must be made available to all firmswithout discrimination. One should thus not grant any firm anautomatic right to a derogatory status in the name of the generalinterest. A procedure should be organised whereby all interestedfirms would be given an opportunity to apply, if interested, forthe provision of these services and selection would be on the mer-its. The selected firm would be allowed to perform these services,if need be, under derogatory conditions (monopoly).

E. Modification of the Rapport de Forces

The initiative taken by the Commission was not accepted by all MemberStates. An application was submitted to the Court of Justice of theEuropean Communities (hereinafter ECJ) to obtain the annulment of thedirectives involved. The ECJ issued two judgments—one per directive. Inboth decisions, it confirmed the Commission’s power to act.17

The position adopted by the ECJ will not be discussed here as it hasalready been the subject of excellent academic commentaries. What isimportant for us to note is that in its judgments, the ECJ substantially mod-ified the institutional rapport de forces that existed before between theCommission and the Member States. The ECJ in fact ruled that legislativepower vested in the Commission for the application of treaty provisions to

96 PLG Nihoul

17 Cases C–202/88 France v Commission (1991( ECR I–1223 and 271/90, Spain v Commission(1992) ECR I–5833. Through these judgments, the Commission was given a green light to goahead with its projects for the telecom sector. As a result, it adopted other directives that com-pleted and even went beyond the ideas submitted in the Green Paper. Competition wasextended to infrastructure whereas that step had not earlier been envisaged. For a list of all direc-tives adopted by the European Commission in electronic communications, see the referencesmentioned above n 12.

public monopolies. According to the ECJ, Article 86 EC entrusts theCommission with the power to specify, by way of general instruments, theconsequences deriving from the treaty for the organisation of economicactivities carried out by this kind of undertaking. The judgments confirmedthe Commission’s willingness to act as a legislative authority to organiseservices of a general economic interest and/or services provided underderogatory conditions (special or exclusive rights).

Yet, the legislative power was undisputedly vested with the Councilbefore these judgments were issued. Until the European Parliament wasgranted more legislative power, the Council of Ministers was considered tobe the legislative organ of the European Communities. As we know, theCouncil is made up of the representatives of the Member States. It couldthus be said that the Member States exercised legislative power throughtheir representation in the Council, pursuant to the rules provided thereforein the treaties. That competence was not limited to a particular given field.On the contrary, it applied to the majority of activities directly or indirectlycovered by the Treaty.18

F. Legislative Power

That division of power applied, amongst others, to the making and imple-mentation of European competition policy. In that area, the legislativepower is clearly, explicitly and without any doubt granted to the Councilby several treaty provisions. These provisions are located in chapter one:‘Rules of competition’, within title five: ‘Common rules on competition,taxation and approximation of laws’, of the treaty establishing theEuropean Community. They roughly contain four ideas with respect to theinstitutional allocation of power in that context.

1. The Council is granted power to adopt the instruments that are nec-essary in order to give effect to the competition provisions contained in thetreaty: Article 83(1) EC. A distinction is established depending on theperiod when these instruments were adopted. Where the enactment tookplace within three years after the entry into force of the treaty, unanimity isrequired. Since that date, measures only have to be adopted with a qualifiedmajority.

Within three years of the entry into force of this Treaty the Council shall, actingunanimously on a proposal from the Commission and after consulting the


18 That situation was modified with the adoption and ratification of new treaties. The principalmodifications brought by these treaties did not affect, however, the institutional balancebetween the Council and the Commission. They rather promoted the position of theParliament vis-à-vis both institutions: the Parliament was progressively given a greater say inthe legislative process. LWG Gormley, Introduction to the Law of the European Communities(Deventer, Kluwer Law International, 1998) 209.

European Parliament, adopt any appropriate regulations or directives to giveeffect to the principles set out in Articles 81 and 82. If such provisions havenot been adopted within the period mentioned, they shall be laid down by theCouncil, acting by a qualified majority on a proposal from the Commissionand after consulting the European Parliament.

(Article 83(1) EC)

2. The Council also has power to determine how the treaty provisionson competition apply in the various branches of the economy: Article 83(2)(c)EC. That article implies that the Council and not the Commission is theauthority in charge of determining what the consequences of competitionshould be in a given sector such as electronic communications.

The regulations or directives … shall be designed in particular: … (c) to define,if need be, in the various branches of the economy, the scope of the provisionsof Articles 81 and 82.

(Article 83(2)(c) EC)

3. The Council also has authority to define the role other institutionsmay be allowed to play in the implementation of the rules of competitioncontained in the treaty: Article 83(2)(d) EC. In that context, the Councilhas the right to determine what role is granted to the Commission and howthe latter is permitted to act in the framework of competition.

The regulations or directives … shall be designed in particular: … (d) to definethe … functions of the Commission … in applying the provisions laid down inthis paragraph.

(Article 83(2)(d) EC)

4. As appears from the first quotation above, the Commission is notabsent from the procedure in cases where the Council adopts competi-tion law instruments in application of the treaty. However, the treatysees the Commission enjoying only a limited role. On the one hand, ithas to draft the documents on which the Council adopts its directives ordecisions (Article 83(1) EC quoted above). On the other hand, theCommission is granted power to apply the rules of competition in concretecases: Article 85 EC. As seen above, the rules to be applied by this institu-tion are embodied in the treaty. They also encompass the instrumentsadopted by the Council pursuant to the treaty provisions, as seen above.

1. [T]he Commission shall … ensure the application of the principles laiddown in Articles 81 and 82 … [T]he Commission shall investigate cases ofsuspected infringement of these principles. If it finds that there has been aninfringement, it shall propose appropriate measures to bring it to an end.

98 PLG Nihoul

2. If the infringement is not brought to an end, the Commission shall recordsuch infringement of the principles in a reasoned decision. The Commissionmay publish its decision and authorise Member States to take the measures,the conditions and details of which it shall determine, needed to remedy thesituation.

(Article 85 EC)19

2. THE DEBATE BETWEEN HARMONISATION ANDLIBERALISATION

A. Second Stage of the Reform

A second institutional development can be found in the oppositiondebate to the Commission and the Council in the course of the reform.We have analysed how the reform was unleashed by an initiative takenby the Commission (adoption of liberalisation directives). We have also commented on the reaction of the Member States (judicial application).When they saw they could not win the judicial debate, the MemberStates decided to adopt their own measures. Their hope was therebyto reach a balance which reflected their own choices. To that effect,they used the power to adopt harmonisation directives originallygranted by the treaty to the Council before its extension to the EuropeanParliament.20

B. Rules Adopted by the Commission

In the liberalisation directives, the Commission went far beyond orderingthe disappearance of special and exclusive rights. It specified what theMember States would still be allowed to do if they wanted to organise activ-ities in the sector. It thus distinctly traced borders Member States would notbe allowed to cross. Measures of that nature were also introduced in deci-sions adopted on the basis of Article 81 EC (anticompetitive agreements),Article 82 EC (abuse of a dominant position) as well as the merger control


19 That provision may be interpreted as giving the Commission a certain margin for manoeuvre.That margin is however limited. The means it may use are restricted: investigation, possibilityto propose measures, possibility to record the infringement in a decision. In drafting that pro-vision (Art 85 EC), Member States have obviously exercised extreme caution to avoid givingthe Commission too much room in the application of competition rules. It is difficult to find insuch a provision a basis to claim that the Commission has authority to set aside an organisa-tion set in place and maintained for decades by Member States. 20 Above n 19.

(Regulation 4064/89, as amended). Two examples are given hereinafter.Distinct publications may be looked at for other examples:21

Example 1. Member States often want to control firms before lettingthem carry out their economic activities. That control gives authorities anopportunity to impose conditions or obligations on these undertakings.In several instruments, the Commission strictly regulated the matterbecause it feared that burdensome intervention might deter undertakingsfrom entering markets, which would contradict the objective it pursued forthe sector—introducing and stimulating competition (presence of severalparticipants).22

Example 2. The Commission also determined how the universal serviceobligation should be financed. Several Member States remained committedto maintaining a system whereby some fundamental services would bemade available to the population at equal and reasonable conditions. Thequestion was to determine how these services would be financed.23 TheCommission did not want new entrants to have any obligation to con-tribute imposed upon them. An obligation of that kind, it felt again, wouldincrease the cost of entry and have a deterrent effect to entrance on the mar-kets contrary to what was being sought in the reform.24

C. New Interpretation of Competition Law

The measures adopted by the Commission were greeted with surprise.Competition rules had long been understood to concern principally under-takings (Articles 81 to 85 EC).25 Admittedly, some treaty provisions areaddressed to the Member States in connection with competition. Forinstance, Article 87 EC prohibits national authorities from granting financial

100 PLG Nihoul

21 P Larouche, Competition Law and Regulation in European Telecommunications (Oxford,Hart Publishing, 2000) 37; P Nihoul, ‘European Telecommunications: A Real Departure fromRegulation?’ in G Haibach (ed), Services of General Interest in the EU: ReconcilingCompetition and Social Responsibility (Maastricht, European Institute of Public Administration,1999) 127–66; P Nihoul, ‘Norme, régulation et réforme des télécommunications’ (1998) 4Annales de Droit de Louvain 389; P Nihoul, ‘Convergence in European Telecommunica-tions—A Case Study on the Relationship between Regulation and Competition (Law)’ (1998)2 International Journal of Communications Law and Policy �http://www.digital-law.net/IJCLP�.22 Dir 97/13 of 10 April 1997 on a common framework for general authorisations and indi-vidual licenses in the field of telecommunications services [1997] OJ L/117/15.23 The said services are provided at derogatory conditions with respect to those provided bythe markets. The tariffs, for instance, are not based on costs. They are set by an authoritybelow cost. The loss must then be compensated by funds coming from another source. 24 Dir 97/33 of 30 June 1997 on interconnection in telecommunications with regard to ensur-ing universal services and interoperability through application of the principles of OpenNetwork Provision (ONP) [1997] OJ L/199/32, Arts 5 and ff; for vocal telefony, Dir 98/10 of26 February 1998 on the application of ONP to voice telefony and on universal service fortelecommunications in a competitive environment [1998] OJ L/101/24, Arts 3 and ff.25 See title V, ch 1, s 1, with the heading: ‘Rules Applying to Undertakings’.

advantages that would improve the competitive position of one or severalfirms to the detriment of others. Article 86(1) EC further prohibits MemberStates from enacting or maintaining in force any measure contrary to therules of the treaty, including the rules of competition.

Yet, the position of these provisions has generally been considered pecu-liar in the treaty. Competition by its nature concerns undertakings. Publicintervention is warranted in that context where the equilibrium is chal-lenged by one or several firms holding market power. Public bodies are notexpected to damage that equilibrium by granting advantages of a financialor regulatory nature. That prohibition had however not been interpreted asimplying that rules would be adopted at European level in order to system-atically determine what Member States would still be allowed to do andwhat initiative taken by them would be analysed as unacceptable given their(sometimes remote) effect on the market.

D. Harmonisation Process

As we have seen, Member States did not succeed in their judicial challengeagainst the liberalisation Directives. They thus resorted to legislative actionin a bid to assert their presence and influence market organisation in adirection closer to their own choices. To that effect, they used an instru-ment provided for by the treaty—harmonisation.

Harmonisation (officially called ‘approximation of laws’) is organ-ised by several treaty provisions. Among them, Article 95 EC (formerly Article 100(a) EC) is generally considered the standard basis. Pursuant tothat provision, approximation may be carried out where national rules hin-der by reason of their disparity the establishment and/or the functioning ofthe internal market. The provision is thus meant to realise the internal mar-ket. Rules generally differ among Member States. They express choices thathave long been made in separate contexts. Approximation provides theCouncil and the Parliament with a mandate to establish a common fieldwhere undertakings will carry out activities on a similar basis. In theprocess of approximating the rules, these institutions have the opportunityto decide what norms will apply to undertakings throughout the Union.They are thus in a position to regulate entire sectors of the economy.

E. Harmonisation Directives

In the harmonisation framework, the Council and the Parliament adoptedseveral electronic communications directives.26


26 They may be found at �http://europa.eu.int/information_society/topics/telecoms/index_en.htm� (visited 11 January 2002).

Terminal equipment Some of them regulate the conditions under whichterminal equipment can be commercialised and connected to infrastructurein the various Member States. The aim underlying these measures is toensure that manufacturers will not have to adapt their products to specifictechnical requirements in force in each Member State. Technical standardsare thus harmonised throughout the Union. The terminals manufactured inconformity with these standards have to be accepted in the other countriesof the Union. The directives further harmonise the formalities that can beimposed at national level in order to determine whether these technicalrequirements are complied with.27

Provision of services Similar measures were taken with regard to serv-ices. In the absence of harmonisation, firms are submitted to more or lessstringent rules depending on the territory where they are established.Suppose they want to perform services in several European countries, theywill normally have to comply with the rules applicable in each national ter-ritory. To avoid these situations, the choice was made to determine atEuropean level the obligations that could be imposed on undertakings pro-viding services in the electronic communications. A rapprochement alsotook place with respect to the formalities firms may have to comply withbefore starting their activities in the different Member States.28

Open Network Provision (ONP) A third and wide set of measures wasadopted with a view to opening networks throughout the Community. Thegoal was to ensure networks would interconnect easily across MemberStates. A technical harmonisation was adopted to achieve that end. Thatprocess facilitates transmissions implying several Member States. Measureswere added to regulate the behaviour of former national operators vis-à-visoperators or service providers from other European countries.29

Behaviour of operators In the same context (Open Network Provision),the Council and the Parliament adopted measures to a lesser extent related

102 PLG Nihoul

27 Last instrument: Dir 1999/5/EC of the European Parliament and of the Council of 9 March 1999on radio equipment and telecommunications terminal equipment and the mutual recognitionof their conformity [1999] OJ L/091/10.28Dir 97/13 of 10 April 1997 on a common framework for general authorisations and individuallicences in the field of telecommunications services [1997] OJ L/117/15. Despite these meas-ures, it remains that firms active on several territories will still be subject to the formalities ineach Member State. The formalities undertaken in one territory will not automatically berecognised for services to be performed by the same firm in the other countries of the Union.29 Dir 90/387 of 28 June 1990 on the establishment of the internal market for telecommunica-tions services through the implementation of open network provision [1990] OJ L/192/1; Dir 92/44 of 5 June 1992 on the application of open network provision to leased lines [1992]OJ L/165/27; Dir 97/33 of 30 June 1997 on interconnection in telecommunications withregard to ensuring universal services and interoperability through application of the principlesof Open Network Provision (ONP) [1997] OJ L/199/32; Dir 97/51 of 6 October 1997 amend-ing Dirs 90/387 and 92/44 for the purpose of adaptation to a competitive environment intelecommunications [1997] OJ L/295/23; Dir 98/10 of 26 February 1998 on the application ofONP to voice telephony and on universal service for telecommunications in a competitiveenvironment [1998] OJ L/101/24.

to the market. In some directives, these institutions harmonised the conditionsat which network operators have to grant access to service providers. Suchan intervention was certainly useful to establish common conditionsthroughout the European market—and thus ensure all undertakings wouldbe placed on equal terms. It was not specifically related, however, to the sit-uation of network operators or service providers wishing to carry out crossborder activities and being hindered in their project by disparity betweennational rules.

Thus, the Council and the Parliament regulated via the harmonisationdirectives the behaviour network operators could, or could not, adopt vis-à-vis service providers. They determined for instance under what condi-tions interconnection should be given or could be refused. They imposed anobligation on operators to provide some kind of service or infrastructure.They also established accountancy constraints, in the hope that activitieswould be easier to control.30

Social measures The Council and the Parliament further adopted measuresto protect certain categories of people. These measures are not specificallyrelated to the internal market either, with the reserve that they establishcommon rules across the European territory (influence on the competitiveposition of companies). Thus, provisions relating to a universal service obli-gation were introduced: Member States were granted the right to set up asystem whereby some fundamental services could be made available to thewhole population at reasonable and equal conditions. These provisions alsodetermine how the system can be financed, what costs can be taken intoaccount and who may be asked to pay for them. Other provisions wereadopted in consideration of handicapped persons or other groups deservingspecial protection.31

F. Two Categories of Rules

As a result of harmonisation and the application of competition rules, twocategories of rules were adopted in European electronic communications bydifferent authorities or groups of authorities. They are based on diverginglegal bases (harmonisation: Article 95 EC; competition rules: Articles 81and 82 and 86 EC as well as the Merger Regulation). They also rest on dif-ferent concepts (achieve the internal market in one case; introduce andmaintain competition in the other one). Despite these differences, both categories were used to regulate similar issues.

Liberalisation and competition were interpreted widely by the Commission.Thus, that institution considered it its duty to regulate anything that may


30 Above n 30.31 In particular Dir 98/10 as cited above in n 29.

directly or indirectly affect competition. Yet, all rules have an effect of thatnature. As they force firms to act in given directions or prohibit them toadopt actions, rules necessarily influence behaviour and positions on themarkets. Hence, the Commission undertook to regulate the sector of elec-tronic communications entirely.

Parallel to that, a similar stance was adopted by the Council and theEuropean Parliament through harmonisation directives. These instrumentssought to eliminate disparities that may affect activities. Yet, rules necessar-ily differ among Member States as they result from diverging contexts. TheCouncil and the Parliament thus sought to establish common grounds foractivities across the European territory. Consequently, they regulated justabout any behaviour in the electronic communications sector.

G. Contradictions

As both categories of rules governed similar issues, contradictions wereinevitable. Some indeed occurred in various instances. One example isexamined below, with respect to the determination of undertakings thatmay have obligation to contribute to the cost of the universal serviceimposed upon them.32

Example As we have seen, the Commission, the Council and theParliament agreed that the Member States could organise the provision offundamental services at derogatory conditions. These services would beperformed at prices below the cost. The loss would however have to becompensated. The compensation could either come through public inter-vention or through contributions paid by a fund set up to that effect. Inthis latter case, the fund would collect payments by undertakings. Thequestion was to determine what obligation to pay could be imposed onundertakings.

At the same time, the Commission sought to prevent such obligationsbeing imposed on new entrants, as this might deter these new undertakingsfrom entering the markets. This would signal a failure for the purposesought by the Commission—to introduce competition by allowing new par-ticipants on the markets. To avoid that result, the Commission stated thatthe fund should ideally be financed by former national operators. Theseoperators would thus face an extra burden. By contrast, new entrantswould bear no obligation of that kind. They would thus be at an advantagethat would allow them to compete more effectively with former operators.

104 PLG Nihoul

32 Other examples are available in distinct publications. A full account may be found in P Larouche, Competition Law and Regulation in European Telecommunications (Oxford,Hart Publishing, 2000) 70 and in P Nihoul, publications cited above n 22.

The still enormous market share retained by the latter would expectedlyshrink to the benefit of competition. In order to achieve that result, theCommission specified in liberalisation Directive 96/19 that universal serv-ice obligations could only be financed by operators of public networks—in fact former national monopolies—in states where the cost would befinanced by a fund.

A different attitude was taken by the Member States via the Council.The Member States appeared to accept liberalisation introduced by theCommission after they realised they could not oppose it. They were alsoin favour of fundamental services being offered to the population on rea-sonable conditions, but they thought that the cost of providing such serv-ices should be borne by all undertakings present on the telecom markets.Liberalisation would provide opportunities for new entrants. It would belegitimate, according to them, that part of the benefits acquired by theseundertakings be transferred to the population in the form of universalservice obligations. Member States also feared that placing an extra bur-den on former national operators might adversely impact on the finan-cial situation of the latter. Such an impact would further complicate any possible negotiation with third parties interested in buying part orall of the shares of these operators (privatisation). It would also forceoperators to drastically reduce costs, predominantly at the expense ofworkers.

For these reasons, the Council sought to enlarge the circle of undertak-ings that could be called to contribute to the cost of universal service.These endeavours resulted in several provisions being inserted in ONPDirectives 97/33 and 98/10. Pursuant to these instruments, universal servicenow principally entails the provision of vocal telephony to the public. TheCouncil found it legitimate to provide that a contribution could be requiredfrom any firm active in the provision of public telecommunications networkor in the provision of public vocal telephony.

A contradiction thus emerged between the provisions adopted by theCommission and by the Council (with the Parliament). It was lateraddressed by a communication adopted by the Commission. Pursuant tothat communication, Member States may impose financial obligations inthe context of universal service on voice telephony providers (other than former national operators). These obligations however have to bedetermined in proportion to their usage of public telecommunicationsnetworks.

Commission: Initiative

[A]ny national scheme which is necessary to share the net cost of the provi-sion of universal service obligations entrusted to the telecommunicationsorganisations, with other organisations whether it consists of a system of


supplementary charges or a universal service fund, shall: (a) apply only toundertakings providing public telecommunications networks.33

Parliament and Council: Reaction

Where a Member State determines … that universal service obligations rep-resent an unfair burden on an organisation, it shall establish a mechanismfor sharing the net cost of the universal service obligations with other organ-isations operating public telecommunications networks and/or publiclyavailable voice telephony services.34

Commission: Conciliation

The Commission will … interpret both Article 4C of the Commission direc-tive and 5(1) … as allowing contributions only to be imposed on voicetelephony providers in proportion to their usage of public telecommunica-tions networks.35

In line with the Full Competition Directive and the InterconnectionDirective, only organisations providing public telecommunications networksand/or public voice telephony services may be required under NationalSchemes to contribute to a Universal Service Fund or to any system of supple-mentary charges … [C]ontributions may only be imposed on voice telephonyproviders in proportion to their usage of public telecommunications net-works … [T]he Commission will, in the case of an application (extension) ofobligations to new entrants and/or mobile operators, assess in particular ifthe burden is allocated according to objective and non-discriminatory criteriaand in accordance with the principle of proportionality.36

3. CONFLICT AT NATIONAL LEVEL

A. Third Stage of the Reform

Measures have thus been adopted at European level by several authori-ties. Some of them produce direct effect in the internal orders of the

106 PLG Nihoul

33 Commission Dir 90/388 of 28 June 1990 on competition in the markets for telecommuni-cations services [1990] OJ L/192/10, Art 4(c) introduced by Commission Dir 96/19 of 13 March 1996 amending Dir 90/388 with regard to the implementation of full competitionin telecommunications markets [1996] OJ L/74/13.34 Council and Parliament Dir 97/33 of 30 June 1997 on interconnection in telecommunica-tions with regard to ensuring universal services and interoperability through application of theprinciples of Open Network Provision (ONP) [1997] OJ L/199/32, Art 5(1).35 Commission Communication on Assessment Criteria for National Schemes for theCosting and Financing of Universal Service in Telecommunications and Guidelines for theMember States on Operation of Such Schemes, COM (96) 608, 27 November 1996, Annex C:Commission Statement to the Minutes of the 1910th Meeting of Council(Telecommunications), on 27 March 1996 on Who Contributes to Universal Service.36 Commission Communication above n 36, Guidelines for National Regulatory Authority,point III.

Member States.37 Most of them, however, have to be implemented in thenational orders. The legislators and the government play an important rolein that regard, as they adopt the national rules necessary for implementa-tion. Other more specialised authorities are called on to intervene as well.As we will see, several authorities or public bodies are entrusted with thetask of ensuring a proper application at national level of the European rulesthat have been adopted in the telecom sector. The co-existence of theseauthorities partly mirrors the division into two categories that we havementioned while studying liberalisation (‘competition law’) and harmonisa-tion (‘sector specific regulation’).

B. National Regulatory Authorities (NRAs)

As part of the measures they have adopted in electronic communications,Member States have created National Regulatory Authorities. Theseauthorities are in charge of implementing the European regulation on electronic communications in their territory. The establishment of theseauthorities is the result of an obligation imposed on the Member Statesby the harmonisation rules due to the Council and the Parliament. Suchauthorities were to be created as early as 198538 in order to ensureproper treatment in other European countries of requests for recognitionof terminal equipment legally manufactured in a Member State. Otherduties were assigned to these bodies in the course of the directivesadopted by the Council and the Parliament. The Commission also playedits part, namely by imposing on the Member States an obligation to sep-arate operational and regulatory tasks, as well as insisting that the lattermission had to be entrusted to independent bodies.39 NRAs must fulfil avariety of tasks. Among them is the obligation to ensure that sector spe-cific regulation is properly implemented in the national orders. Theseauthorities also intervene in the rule making process, among others by determining under what conditions dominant operators provide interconnection.40


37 Decisions adopted by the Commission on the basis of Art 81 EC, Art 82 EC and the MergerReg; regulation adopted by the Council on specific aspects (eg on unbundling the local loop).Some provisions contained in directives may have direct effect after the deadline set for theirimplementation by the Member States.38 See Council Dir 86/361/EEC of 24 July 1986 on the initial stage of the mutual recognitionof type approval for telecommunications terminal equipment, [1986] OJ L/217/21.39 Formerly, the national operators were entrusted with regulatory tasks in addition to theiroperational activities. In its liberalisation directives, the Commission insisted that these tasksshould be assigned to different bodies.40 P Nihoul, Droit européen des télécommunications—L’organisation des marchés (Brussels,Larcier, 1999) 239.

C. National Competition Authorities (NCAs)

In addition, competition authorities have been created at national level.There does not appear to exist, at the present time, any explicit obligationfor the Member States to set up such bodies in their internal order. Onecan, however, consider that a constraint of that sort derives from the gen-eral provisions of the treaty. In particular, Article 10 EC provides thatMember States are to take all measures that are appropriate to ensure theimplementation of treaty objectives. Yet, one of these objectives is to estab-lish a market where competition is undistorted.

In accordance with this obligation, Member States have established com-petition authorities. At the present time, these authorities are principallyresponsible for the application of national rules relating to competition.41

European rules have indeed largely been applied so far in that field by theEuropean Commission.42 That situation is now changing. The Commissionhas prepared a project aimed at decentralising the application of Europeancompetition rules. Should the project be adopted, the responsibility forapplying these rules will mainly be assigned to the NCAs.43

As a result of decentralisation, national authorities will hold responsibil-ity for applying all rules related to competition within the European legalorder. Among these rules are the competition principles laid down in thetreaty (Article 81(s) EC). To these rules we have to add the instrumentsadopted by European institutions on the basis of these principles, including:the directives and regulations adopted by the Council (see Article 83 ECdiscussed above); as well as the directives adopted by the Commission inthe framework of telecom liberalisation. In connection with that latter cate-gory of instruments, one can thus say that NCAs will have authority toapply the rules introduced by the European Commission on the basis ofArticle 86 EC within the electronic communications sector.

108 PLG Nihoul

41 They already have a certain authority to apply European competition rules. As publicorgans, they are under a general obligation to ensure proper application of European rules.They however have to respect the power granted to the European Commission, includingthe exclusive competence of that authority to grant individual exemption pursuant to Art 81(3) EC.42 National authorities always had a role, but the Commission incontestably remained theleading figure. The reasons for these are probably (1) the absence of competition authorities insome countries during a long period and (2) the clear desire, on the part of the Commission, tomaintain some sort of leadership in the area.43 These authorities would have full responsibility for applying the prohibitions contained inArts 81 (anticompetitive agreements) and 82 EC (abuse of a dominant position). TheCommission would retain authority to withhold cases in the Community interest. It wouldalso retain exclusive competence under the Merger Reg, for operations falling in the scope of suchregulation. The White Paper is available at �http://www.europa.eu.int/comm/competition/antitrust/wb_modernisation_en.pdf� (visited 11 January 2002). On the basis of the WhitePaper, the Commission has proposed a Regulation that is available at �http://www.europa.eu.int/comm/competition/antitrust/others/modernisation/comm_2000_582/en.pdf�.

D. National Courts

We have mentioned the power vested in national specialised authorities forthe application of competition law (NCAs) as well as sector specific regula-tion (NRAs). One also has to take into account the competence nationalcourts have in that same area.

On the one hand, national courts have a general jurisdiction to applyEuropean rules. That jurisdiction includes the possibility of applying rulesadopted at European level in the field of competition. Courts may forinstance rule over applications introduced by undertakings or customersaiming at obtaining the nullity of an agreement, or at getting payments inreparation of damages caused by violation of European competition rules.That power is not limited to a given sector. It thus applies to electronic com-munications as well, in so far as competition rules apply to this sector.

On the other hand, national courts normally hold power to applynational competition rules. That competence is mainly regulated bynational law. It is generally accepted in the Member States that courtsshould handle civil aspects of claims based on competition rules. They areindeed in national orders the bodies that have charge of the protection ofcivil rights.44 National competition rules are normally similar in substanceto those in force at European level.45

National courts further have jurisdiction to apply telecom sector-specificregulation. Applications may be submitted to them, on the basis of nationallaw, in order to solve disputes involving national rules implementing theEuropean telecom regulation. As these national rules normally reflect thoseadopted by the European institutions, one may thus consider that nationalcourts apply, in substance, the European rules. Where implementation hasnot been timely or correctly made, national courts may give precedence toEuropean rules. Some conditions however have to be fulfilled for Europeanrules to be given direct effect in the internal legal order.

4. CONVERGENCE FROM AN INSTITUTIONAL POINT OF VIEW

A. Fourth Stage of Development

The three developments we have examined above do correspond to various stages in the reform that has taken place in the European electronic


44 C Jones, Private Enforcement of Antitrust Law in the EU, UK and USA (Oxford, Oxford University Press, 1999).45 They could not be different, as they would otherwise put into danger the effet utile ofEuropean law. They would also affect legal certainty by creating separate standards for theapplication of a common concept—that of competition. See P Nihoul, La concurrence et ledroit (Paris, Editions Management et Société, 2001) 275.

communications: the introduction of competition; the negotiation that hasfollowed between the Commission and the Council (with the Parliament),which has given rise to two separate categories of rules; the consequence ofthis duality in national law where two authorities are entrusted with spe-cific tasks in addition to ordinary public bodies (legislator, government,judiciary). The fourth development is of a different nature. It is not reallyassociated with any given stage in the reform. Therefore it is mentioned atthe end of this discussion. As a matter of fact, it appears independent of anyinstitutional dimension, although it has consequences on that point.

As we have mentioned, convergence is currently taking place betweenthe technologies that are used in the broadcasting, computing and elec-tronic communications industries. These moves are accompanied byalliances between undertakings within these sectors. They have an effect onregulation, particularly on authorities in charge of these sectors. Prior tothis, the said sectors were subject to different rules. From an institutionalpoint of view, they were administered by separate regulators. As a result ofconvergence, activities now involve hybrid technologies. Thus, they cannotbe associated with a particular sector any longer. Given their hybrid charac-ter, they fall under the ambit of several regulations and regulators.46

B. Illustration

As mentioned above, that situation has institutional consequences. Supposea broadcasting company sends video advertisements via the Internet to allpeople with an Internet address in a given country. The activity will proba-bly fall under several regulations. As a result, it will call for an interventionfrom more than one regulator. (1) The broadcasting regulator will probablyintervene, as the message is sent by a broadcasting authority and the mes-sage is in the form of a video.47 (2) The activity will further attract theattention of the authority entrusted with privacy protection in mostMember States.48 (3) The authority in charge of customer protection willalso be willing to intervene, should the message contain information thatmay mislead the addressees. (4) One can additionally envisage the interven-tion of a competition authority (an NCA or the European Commission asthe case may be). Such an intervention may be mandated, for instance, as aresult of the power held by a firm that reserves to itself the possibility of

110 PLG Nihoul

46 C Blackman and P Nihoul (eds), The Convergence Between Telecommunications and OtherMedia: How Should Regulation Adapt? (Amsterdam, Elsevier, 1998).47 In our example, the advertisements are made in the form of a video. They are sent to anyonewith an electronic address in a given country. They are meant for the public. 48 In this example, the advertisement was sent to all people with an address in a given countryin the absence of a request from these customers. Addresses may have been communicated tothe broadcasting company without prior authorisation by the addressee.

sending messages of that nature via its infrastructure. (5) The NRA mayfeel involved as well, as telecom sector specific regulation governs relationsbetween the service provider and the network operator.

5. AN INSTITUTIONAL INTERPRETATION OF THE REFORM

A. Conflict and Competition among Authorities

The examination that has been carried out above points to the plurality ofauthorities with competence to intervene in respect of transactions takingplace in electronic communications. That situation is remarkable as itreflects in the institutional structure a transformation that has taken placeon the telecom markets. As we have seen, markets have been opened to newundertakings after being reserved for decades to a single organisation. Theliberalisation that has taken place may also be interpreted in terms of theintroduction of a plurality of participants, where singularity was previouslythe rule with the former monopolists.

It now appears, with the observations made above, that this develop-ment has not been limited to economic activities. A similar phenomenonhas occurred on the institutional side. In the monopolistic era, telecomactivities were governed by a single body. Regulatory functions were thenentrusted to the undertaking in charge of managing the network and pro-viding the services. This organisation has often been labelled as peculiar, asit allowed a regulatory task to be combined with economic activities.49 Wecan now emphasise a second feature, as we see that regulation and publiccontrol were entrusted to a single authority where several now intervene inthe same sector.

Not only are several authorities claiming competence in order to regu-late the sector, or at least adopt rules that will apply to activities carried outin that sector, in many respects, these authorities also seem to be in conflictwith each other. Let us take a look back at the first conflicts we have exam-ined in the first part of this chapter.

First conflict. In launching the liberalisation process, the Commissionwas willing to organise the sector according to the principles it consideredwere appropriate. Consequently, it wanted to replace an organisation thatreflected political and economic conceptions shared by the Member Statesbefore the reform was enacted.50


49 This created a situation that is now perceived as dangerous because governments try toorganise authorities in an independent manner. Such independence is considered necessary toallow authorities to reach decisions irrespective of the peculiar interest of any particular under-taking involved. The issue was discussed in Case 41/83 Italy v Commission (BritishTelecommunications I) [1985] ECR–873.50 The telecom sector was organised along the same lines in most European States. TheUnited Kingdom was the only country to provide an exception. Still, it should be noted that

Second conflict. Subsequently, the Commission and the Council wereclearly opposed in attempting to define the rules that would govern elec-tronic communications in the European Union.

Third conflict. Practitioners underline the difficulties emerging from theexistence of several authorities as a result of convergence. Due to that phe-nomenon, these authorities are in charge of applying European rules innational legal orders. Practitioners fear these authorities may not co-operate,but rather conflict with each other to attract attention and funding.

Fourth conflict. As a result of convergence, regulators formerly responsi-ble for separate activities are bound to intervene simultaneously. They willadmittedly remain responsible for different aspects of a same activity, butthese aspects cannot be totally separated: the intervention of one will nor-mally have consequences on aspects normally regulated by another.51

B. The Process of Regulatory Competition

As we see from the discussion above, the four developments that have takenplace at institutional level during the European electronic communicationsreform have the form of conflicts. These conflicts may be analysed in termsof competition. In all these cases, two or more authorities are competingfor the power to regulate and control. How can we understand this phe-nomenon? The best way to analyse it is probably to refer to other kinds ofconflicts that appear in legal systems.

As we talk about regulatory conflicts and competition, one naturallyturns to international private and public law. Conflicts have long opposedstates that wanted to simultaneously assert their jurisdiction on a givenissue. We have experienced several times in the last years this kind of multi-jurisdictional conflict in the application of competition rules, which is thetopic for our discussion. In that field, American and European authoritiesoften tend to apply their rules as soon as their territory, or even merely their

112 PLG Nihoul

the British national operator was originally granted the same position as that enjoyed by itscounterparts in the other Member States. The change only occurred a few years before it wasintroduced by the Commission in the European Union. See M Thatcher, The Politics ofTelecommunications, National Institutions, Convergence and Change in Britain and France(Oxford, Oxford University Press, 1999).

51 Thus, content and transmission were regulated beforehand by separate regulators. Yet, deci-sion taken on one of these aspects will have consequences for others. Suppose an authoritywants to ensure cultural pluralism in content production. In order to implement that goal, itmust allocate resources in a way that will stop one form of production taking precedence overanother. Absent any measure, alternatives will progressively be eliminated. Yet, networkcapacity constitutes one of these resources. If capacity is limited, the authority will have toallocate that resource to make it possible for producers of alternative programs to reach customers. Transmissions must be regulated therefore to ensure proper allocation in view ofpluralism in content.

interest is affected by a transaction (effects doctrine).52 This situation is noexception. In general, one can say that markets have an increasingly inter-national nature. The result is that authorities assert ever more their jurisdic-tion simultaneously on transactions, thereby creating numerous potentialconflicts that are not always easily solved.

Conflicting claims of states have contributed to developing amongundertakings the feeling that they should choose, wherever they can, thelegal order to which they would like to be subject. Suppose a firm wants toprovide Internet services around the globe. As we know, some countries aremore stringent than others as regards content that may be published on theInternet. For instance, France does not allow the publication of any mate-rial that may encourage racism. By contrast, information of that sort is notprohibited in the United States.53 In the business world, these two regimeswill be compared by potential Internet Service Providers (ISPs). On the basisof that comparison, these providers will determine how to structure theiroperations with respect to these countries. The decisions will presumablybe based on one main concern: alleviate costs and thus choose the legal sys-tem that will impose the most lenient obligations.

In international law, one has traditionally called this behaviour ‘forumshopping’. Firms shop around to determine what regime will be most bene-ficial. As a matter of fact, the pattern could also be described in terms of‘competition’. Through their behaviour, undertakings set competitionamong the various states and legal orders concerned. In that context, thecompetition process may be analysed a follows.54 Firms must establish theiroperations in a given territory. To that effect, they still need some sort ofterritorial resource.55 Territory however does not come alone. It encompasses


52 See Y Akbar, ‘The Extraterritorial Dimension of US and EU Competition Law: A Threat tothe Multilateral System?’ (1999) 53 Australian Journal of International Affairs 113; KM Meessen, Extraterritorial Jurisdiction in Theory and Practice (Deventer, Kluwer LawInternational, 1996); E Nerep, Extraterritorial Control of Competition under InternationalLaw, with Special Regard to US Antitrust Law (Stockholm, Norstedt, 1983).53 For a discussion, see the judgment issued by the French judiciary in the Yahoo! case. Acivil rights association introduced an application against the American ISPs, on the groundsthat the latter hosted a site containing racist information and statements in contradictionwith French law. The case may be found at �http://www.juriscom.net/txt/jurisfr/cti/tgi-paris20001120.htm� (visited 11 January 2002) with a discussion in French and English.54 P Nihoul, La concurrence et le droit—entreprises, consommateurs et autorités (Paris,Editions Management et Société, 2001) 113.55 It is uncertain that that necessity will persist in the future, due to the development of virtualworlds. Consider the example of the example of the American on-line music firm Napster.This firm proposed software allowing users to download copyright protected material free ofcharge. This activity had to be terminated at the request of copyright holders. The requestcould succeed because the undertaking was located in a given territory, where an action couldbe brought against it. To circumvent the prohibition, new programs are designed that do notrequire any physical location on a given territory. These programs are downloaded by users ontheir computers. They establish a network whereby users exchange their data, including copyrightprotected material. No intervention of a centrally located server is requested, contrary to what

political, social and economic conditions. These conditions include the regulatory environment (the type of authority and rule). Where they haveto take business decisions, firms analyse the offers that are available in thatregard. Among these offers, they choose the option which best correspondsto their needs.

As a consequence of ‘forum shopping’, one can observe a type of ‘regulatory’ or—more appropriately—’deregulatory’ competition amongstates. Countries propose special treatment in the hope of attracting for-eign investment or companies. This type of competition is common withrespect to taxation. Some countries tax non-resident firms or individualsat very low rates, whereas others grant subsidies to the firms they want toattract. In these situations, firms basically introduce a type of competitionamong states. Through the possibility they have to select where they estab-lish their presence, they force the states to improve their offer. They choosea territory and a legal order as they would do a supplier of a distributionchannel.56

C. Institutional Competition in Electronic Communications

One may wonder whether these international law situations help us under-stand the institutional developments associated with European electroniccommunications. These developments do not appear to involve any inter-national dimension. In conflict one, the Commission was claiming prece-dence over the Member States. It wanted to organise the telecom sectoralong the lines of competition, whereas the Member States wanted to maintain their own organisation. In conflict two, the Member States soughtto challenge the Commission by adopting harmonisation measures withinthe Council and in association with the Parliament. That opposition waschannelled at national level in conflict three, where NCAs and NRAs mayconflict with each other when implementing European and/or national rulesrelating to electronic communications.

None of these three conflicts feature any international law situation sim-ilar to those encountered in classical international (private or public) lawissues. In short, we do not see in these conflicts states fighting with eachother in order to win the right to apply their rules to a given situation. Theabsence of an international reference is still more obvious in conflict four.

114 PLG Nihoul

happened with Napster. The new system makes it impossible to efficiently initiate a judicial pro-cedure. An action should be brought simultaneously against all users in all countries—which isnot materially possible.

56 The competitive process that then takes place is not necessarily positive for all people insociety. As they try to improve the offer made to undertakings, states will often lower socialprotection as well as the level of service that is generally made available to their citizens.

As we have seen, this conflict is related to convergence. Before convergenceemerged, regulators had jurisdiction over distinct activities. For instance, aBroadcasting Authority would be responsible for broadcasting activities,whereas a Telecom Authority would prepare and implement regulationdealing with telecommunications. These tasks were carried out within agiven state and legal order (for instance the British system), outside anyinternational situation.

D. Competition Among Authorities Within the Same Legal Order

The observations made in electronic communications show that institu-tional competition is not limited to states. Lawyers have long been aware ofinter-state competition.57 It now appears that conflicts are not limited torelations at an international and supranational level. Institutional competi-tion is not based on national boundaries. It opposes authorities attached todifferent countries, as well as authorities pertaining to the same legal order.Legal orders have traditionally been said to form coherent and consistentsystems where authorities would be allocated distinct and specific tasks tobe exercised for the common good. A different picture emerges when onelooks at the institutional divisions we are experiencing in European elec-tronic communications. From these divisions, one can infer that the idea ofa harmonious division of power among authorities has to be abandoned.The division takes place amid rivalry and competition, where each author-ity uses available instruments (including the power to adopt regulation) tomaintain and if possible improve its own situation as well as that of itsmembers. As a result, the conflicts we observe in international law situa-tions should not be analysed as exceptions. They form the visible face of areality present at all levels of institutional organisation.

Institutional competition may be analysed with tools elaborated in inter-national law, if one gives the concept of legal order another definition aswell as another scope. Let us consider an analysis of conflict four (result ofconvergence). Prior to convergence, authorities normally administeredactivities that were distinct from one another. These regulatory tasks werecarried out within the same legal order. For instance, one authority con-trolled broadcasting in the UK, whereas another one administered telecomsin that same country. As we have seen, convergence brings these activitiescloser to one another. Authorities remain responsible for distinct aspects ofeach activity, but conflicts are bound to occur, as one intervention will havean effect on others.


57 That form of competition was not limited to an regulatory one: it extended in other fields,including economy.

One can probably compare to territories these various aspects of thesame activity. The conflicts emerging among authorities responsible for dis-tinct aspects or areas may be assimilated to oppositions amongst bodies incharge of various territories. The difference between both contexts is thatterritories refer in one case to material fields (convergence), whereas theyare traditionally associated with geographic boundaries (states) in interna-tional situations.

Regulatory conflicts may thus be described, whatever their level(national or international), in terms of competition among rules andauthorities that form different legal orders. The concept of legal order hastraditionally been reserved to a system proper to a state. However, it mayalso be used to describe the sort of competition that appears in conflictfour. We propose to define it as ‘any system made of rules that are to beconsistent with one another and the object of which is to govern a specificobject’. Each body of rules may thus be described as a legal order. In thatperspective, one might represent national legal orders as bodies encom-passing in themselves a multitude of systems: each of them forming a microlegal order insofar as it aims at governing specific issues in a somewhatcoherent fashion.

That sort of situation does not appear as an exception. It rather seems toflow from a general trend that may be observed in all economic and socialsectors. Throughout the ages, authorities have created rules to governhuman activity. Rules have been adopted to administer just about anybehaviour, in all its aspects. It can thus be expected that the same behaviourcan be scrutinised under several rules corresponding to various aspects thathave been regulated through the years. This plurality in rules has beenmatched by a trend to create bodies, each entrusted with a mission to ensurethe proper application of a given set of rules.58 The two trends (creation ofrules and authorities) have mutually reinforced themselves. The enactmentof rules made it necessary to create bodies to implement them. Conversely,authorities were granted the power to further regulate the areas placedunder their command. In that capacity, they issued new regulations. Amovement was thus developed whereby activities were being increasinglysubmitted to various authorities, all eager to apply their rules to the situationsthey encounter.

E. Expanding the Line of Business

The lesson to be drawn from the reform on an institutional level probablylies here—authorities compete in the same way as undertakings. The four

116 PLG Nihoul

58 As we noted above, several authorities were thus created which now hold a certain powerrelating to electronic communications: national regulatory authorities, broadcasting authorities,authorities competent to monitor the respect of privacy, consumer protection, etc.

conflicts observed in electronic communications may be interpreted in competitive terms. As competition does for undertakings, these conflictsprobably play an important role in the initiatives that are taken by authori-ties. One should not infer from that statement that competition explains allmoves adopted by public bodies. For instance, the European telecomreform can certainly be explained by economic reasons: the willingness toorganise European activities in line with modern economic thinking. Tobelieve that the reform can only be explained by considerations of thatnature would however be short sighted. One would lose sight of the rolecompetition and conflict play in institutional development.

In order to understand that process, one may refer to methods used byfirms in the business world. An undertaking normally manufactures prod-ucts or performs services. A good manner for it to expand activities consistsof using one of several methods it has developed in the course of its business in new markets. Take the Internet firm Amazon. That firm wasoriginally engaged in the business of selling books. It thereby developed aspecific ability—a ‘business function’ in management terms. That functionwas to serve as an intermediary between producers (editors) and clients(readers or in some instances other intermediaries closer to the final cus-tomers). In a hope to improve its situation, Amazon decided to expandactivities. To that end, it entered markets where it could apply that businessfunction. It thus started distributing other products—in the first instanceobjects similar to books (CDs, etc) and then products that were moreremote (travel, etc) but were still offered by a producer for which it couldact as an intermediary.

A similar pattern of interpretation may be used in order to analyse ininstitutional terms the reform that has taken place in European electroniccommunications. The interpretation starts with the position held by theEuropean Commission. That institution considers itself as having the powerto design and implement competition policy within the European Union.59

For the Commission, that task constitutes an activity, like selling goods andservices constitutes an activity for the American intermediary Amazon. TheCommission drafts and implement rules in situations where competitionissues emerge.60


59 See the following statement on the Internet site of DG IV: ‘the mission of CompetitionDirectorate General is to establish and implement a coherent competition policy for theEuropean Union’: �http://www.europa.eu.int/comm/dgs/competition/index_en.htm� (visited11 January 2002). That position is already the result of expansion. The role of the Commissionis indeed confined by the treaty to the implementation of the rules of competition that areembodied in the treaty as well as those that are adopted by the Council, among others thosewhereby the Council determines what role other institutions—including the Commission—will play in the realm of competition.60 The European Commission has other tasks as well. We are here focusing on that which isassociated with DG IV—the division responsible for competition matters within theCommission.

Underlying that activity is the ‘business function’ characterising the intervention of the Commission in that context. Through using the expres-sion ‘business function’, we designate a sort of archetypal activity. Wethereby determine what the firm or the institution basically does, independ-ent of the accurate and concrete context (identity of the product or service)where it operates. In this case, the business function may be expressed inregulatory terms. The Commission basically ‘talks’ competition rules, what-ever the context where it wants to apply these rules and expand the policyit is devising under that name. The regulatory aspects of competition forma legal order, which grants the Commission tools to express its views andintervene in a variety of situations. The legal or regulatory order may beanalysed as the equivalent in institutional terms of what we have labelled‘business function’ using management vocabulary.

In a system where power is attributed, institutions may only act on thebasis of competence entrusted to them. To expand their activities they havelittle choice but to receive new powers or to extend those which they have previously received. Given the difficulty of receiving new powers inthe course of treaty revisions, their best chance is to apply that power tonew contexts. A new equilibrium will thus progressively be reached,between the authorities that could pretend to power in that new context.Conflicts may in some instances be brought to a superior organ, which willthen play the role of an arbitrator between these authorities.61

To come back to electronic communications, we can interpret the behav-iour of the Commission as that of an authority willing to enter new activities.The rules of competition could be applied to vast sectors of the economy.However, it had never been considered that these rules should also apply tosectors that had so far been regulated by states. We are alluding here to allsectors where states had attributed special or exclusive rights to certainundertakings, as well as to sectors where states had a particularly strongpresence through public undertakings. These sectors formed a significantpart of the economy. They also provided states with an opportunity still toinfluence economic structure and strategy within the European Union, at atime when it was felt a European approach might be useful. Electronic com-munications provided a good opportunity given the reasons mentioned atthe outset of this discussion—principally the importance acquired by thatsector in the economy as a whole as well as the technological advantagesthat were developed by American firms in that field. In order to enter these

118 PLG Nihoul

61 The ECJ played a role of that nature in the conflict opposing the European Commission andseveral Member States in the cases relating to the liberalisation directives adopted by the for-mer in the telecommunications sector. The ECJ acted in this role with respect to the conflictopposing the European Commission and several Member States in the cases relating to the lib-eralisation directives adopted by the former in the telecommunications sector. On the role ofarbitrator of the ECJ, see K Lenaerts, Le juge et la constitution aux Etats-Unis d’Amérique etdans l’ordre juridique européen (Brussels, Bruylant, 1988).

areas, the Commission applied the function it had developed in other sectors.It thus adopted competition rules meant to organise the sector. Using com-petition concepts, it designed regulation tailored to the needs of the telecomindustry as these needs were perceived by the Commission.62

6. WHAT CAN WE LEARN ABOUT THE FUTURE?

A. More Competition among Authorities

As these elements have been presented for an institutional interpretation ofthe reform, the time is ripe to dare a prospective question. In our analysis,we have seen that authorities compete with each other and that such com-petition may sometimes play an essential role in the determination of therules that are adopted in a given sector. Can we envisage a situation wherethat sort of competition—one which concerns regulation—would be organ-ised further? Competition is structured by law and intervention where itcomes to firms and markets. Is that way of organising relations limited tothese entities per se—or can it be adopted with respect to institutions andregulatory processes?

The question should be further refined as follows. The debate is not set upin terms of: ‘Should we organise authorities in a more competitive manner?’Such a formulation would entail political choices and considerations thatdo not have their place in a scientific discussion like the one which is takingplace in the context of this seminar. The question should rather be under-stood in the form of: ‘Are we engaged in a process where relations areincreasingly regulated in competition terms, to such an extent that thisorganisation will not be limited any longer to undertakings but will also beused to tailor institutional relations in the future?’

In order to understand the question better, let us return to conflict three.As we have seen, it is often feared by undertakings that NRAs may competein the future with NCAs to obtain attention and funds. Does the increasingimportance acquired by competition—an importance that is attested by theintroduction of rivalry in previously monopolistic markets and the conflictswhich have taken place at an institutional level in the European electroniccommunications—announce an era in which firms will be allowed tochoose the authority they see as best to intervene in their situation and solvetheir legal disputes? On the basis of what we have observed can we predicta time when firms will be allowed to choose ‘their’ authority and ‘their’legal system as they choose a provider or a distribution channel? Will sucha situation occur independent of any international context?


62 As mentioned above, these rules were enacted in the liberalisation directives as well as in thedecisions adopted on the basis of Art 81 EC, Art 82 EC and the Merger Regulation.

B. First Difficulty: Regulation Not an Ordinary Good

When we try to envisage such a possibility, a first difficulty readily emergesto the extent that undertakings will probably choose the legal system (rule,authority) that is the most favourable to them. Firms normally sell goodsor/and services. A choice is made by customers among these goods and serv-ices. In the process, firms adapt their offer to the needs of the customers, inthe hope of being selected by the latter. Only on those conditions will thesefirms obtain the resources that are necessary in order for them to continuetheir activities and carry out their projects.

Can we envisage a similar pattern with authorities? In the representationproposed above, authorities are regarded as bodies offering ‘regulatory’goods or services. Among the goods and services of that nature, firms andindividuals will normally choose the offer they see as best corresponding totheir needs. In that process, authorities would probably try to adapt theiroffer as we have seen undertakings do. There is a risk that authorities willdilute their rules and requirements in order to please their customers.Regulation would then stop fulfilling its mission, which is to provide par-ticipants with a platform where all can interact in a manner conforming topublic objectives and values.

C. Solution: Competition Within a Given Framework

The difficulty can probably be overcome by resorting to the mechanismused in auctions. Consider the approach that has been taken by theEuropean authorities with respect to universal services obligations.63

Pursuant to the European regulation, the Member States may decide thatsome fundamental services will be provided at derogatory conditions ontheir territory. That decision is based on the perception that these servicesmay be deemed essential to the population in these national legal orders. Inthe framework established by the European rules, the Member States deter-mine what services may be concerned as well as the conditions under whichthe latter will have to be provided. Candidate firms are asked to bid a priceat which they would be ready to offer all or part of the services.

In such a mechanism, firms are called to apply in order to fulfil goals ofan objective nature. The application occurs within a framework that is setby authorities and corresponds to the requirements justified by the publicinterest. That framework is indeed made up of decisions where authoritiesset the objectives to be reached as well as the criteria that will be used toevaluate performance. Within that framework, competition occurs betweenthe candidates. The best offer will win the auction. The firms thus carry out

120 PLG Nihoul

63 Above n 23.

their tasks in an environment made up of competition within a frameworkwhere public goals are to be achieved.

A similar system may be imagined for authorities. Objectives may befixed which authorities will have to implement. These objectives would befixed by some superior organ—for instance the European Parliament, asolution that appears legitimate in a democratically oriented legal order.Within these objectives, authorities would have to compete. They wouldneed to carry out their tasks so that they are chosen by customers (firms,individuals) without endangering the realisation of the public objectiveswhich have been assigned to them.

D. Second Difficulty: Legal Certainty and Predictability

A second difficulty may come from the impossibility to determine with cer-tainty and accuracy the solution that might be applied to a legal dispute inan environment where participants (firms, individuals) may choose theirregulation as well as their regulatory partner. Economic agents need toknow in advance what rules will apply to their behaviour. That is possible,although not always easy, where one solution only may be admitted.64 It isprobably more difficult, if not impossible, in cases where two or severalsolutions may eventually be applied, depending on what legal order is cho-sen by one or the other party.

To understand the problem, let us go back to our example where firmsand individuals are allowed to choose between competition law or sectorspecific regulation (conflict three). Suppose one firm chooses the competi-tion legal order. It enters into conflict with a rival that has opted for thesector specific legal order. What rules and what authority will eventuallyintervene to solve the dispute? The choice in favour of one or the othermay influence the outcome. Yet, the probable outcome of a dispute formsa decisive factor where firms envisage engaging in certain activities in thefuture.65


64 The observation of the case law shows that economic agents are not always in a position todetermine what solution will apply to their behaviour. That is probably due sometimes todefects in their reasoning but at times to ambiguity and obscurity of the legal system as well.65 The difficulty is illustrated by the fact that the European system does not allow a choice tobe made between competition law or sector specific regulation. Pursuant to legislation, bothare to apply simultaneously. Mechanisms are set in place in order to solve possible contradic-tions. See eg Arts and ff of the Amended Proposal for a Directive of the European Parliamentand of the Council on a Common Regulatory Framework for Electronic CommunicationsNetworks and Services (taking into account the amendments put forward by the Parliamentat its first reading) COM (2001) 380, 4 July 2001, available at �http://www.europa.eu.int/information_society/topics/telecoms/regulatory/new_rf/com2001-380en.pdf� (visited 11 January 2002).

E. Place of Unity in Society and the Legal System

This comment refers us back to the importance of unity in the legal system.Unity plays an essential part, because without unity, legal solutions may nolonger be identified. Predictability therefore disappears, with the negativeconsequences such a development may have on business decisions.66 Weare touching here on a very delicate issue in terms of how legal systems arenow designed. In order to address this issue, let us return to electronic com-munications. We have seen that telecommunications have long been organ-ised in the form of a monopoly. In that system, one undertaking wasgranted the right to perform activities. As a result, one infrastructure wasestablished in each national territory. One offer was thus presented to thecustomers. The demand was even unified, as customers were supposed toform one body interested by standard goods and services.

That organisation was not limited to communications. It was appliedto other sectors where authorities had a leading role. All public utilitieswere in fact concerned—water, electricity, gas and transportation. Besidesthese sectors, non-economic activities were also affected. This unity basedorganisation was also used to structure legal and administrative arrange-ments. For instance, each country was supposed to have one legal systemwhere one function was assigned to each category of institutions (legisla-tive, executive, judiciary). The case is particularly clear for judicial com-petence, as courts normally have distinct jurisdictions, so overlappingpowers are considered a major issue that requires the intervention of asuperior authority.67

In short, unity appears symptomatic of the way authorities organisetasks and institutions placed under their control whatever the sector andthe nature (economic or not) of the activity. That unity is associated withhierarchy. Things are organised in a hierarchical manner. Each is assigned adomain where it will hold exclusive competence. At a higher level, severalentities will depend on a superior body that will solve potential contradictionsor difficulties.

122 PLG Nihoul

66 Unpredictability of legal outcomes is an important factor explaining the difficulties faced bycountries in transition, where they endeavour to attract foreign investment and companies.67 Illustrations about the importance of certainty, predictability and unity are pervasive in thecase law of the ECJ. A good example is provided by Case 314/85 Fotofrost [1987] ECR 4225,para 11. In that case, the question for the court was whether national courts could invalidateacts adopted by European institutions. The ECJ claimed exclusive jurisdiction for decisions ofthat kind. That solution was justified by the necessity to have identical solutions throughoutthe Community. (National courts may decide differently where asked to evaluate the validityof Community acts.) It was also based on the necessity of achieving consistency among severalprocedures. As the ECJ has exclusive jurisdiction to rule on direct applications for annulment,the same solution had to apply with respect to preliminary rulings with a similar object. Withoutsuch a solution, decisions may have differed on the same act depending on the procedureinitiated by the applicant and thus the court that would have been called to intervene.

F. Conformity with an Information and Communication Based Society?

Does this unity based system still conform with society as it develops? Wehave had unity for a long time in our lives. Our ancestors had one religion,partner, family, political party, employer and place of residence. Their sta-tus as well as their activities was organised around that semi-consciousprinciple of necessary unity. Yet, we see that unity is progressively beingreplaced by plurality. Monopolies are set aside to welcome participants onthe markets—not just in electronic communications. Syncretism is now thenorm in religion, where everyone picks up pieces wherever he or she can inorder to constitute his/her own religious environment. Families are increas-ingly built on plurality, with people choosing partners in various periods oftheir lives and having different families as a result. Political parties cannothope to attract supporters for their whole program any longer: citizens turnto one for some ideas and to other ones for other inspiration.

This evolution poses a challenge to the legal system. Is it going to remainattached to principles such as unity and hierarchy, which have marked theorganisation of public power as well as institutional relations for the pastcenturies—or will it be organised along these new lines being suggested bythe appearance of plurality in all sectors of our lives?

This challenge may be addressed with the help of technological progress.Thanks to information technology, we are now able to treat larger amountof data and information more quickly. Communications technology hasalso significantly increased our capacity to transmit information. With thistechnology, we may envisage regulatory arrangements that are more com-plex than the ones we have known before. Thanks to it, we may be in aposition to introduce plurality in the legal system.

Think of the information based mechanisms that are used for productsand services, where it comes to giving customers data about the items theyare likely to purchase or order. For a more concrete example, take the datathat must now be placed on meat packages as a result of decisions recentlytaken by European authorities. With this data, it is now possible for cus-tomers to know the place were the animal was born, where it was raised,where it was slaughtered as well where the packaging took place. Choicesare made by economic agents on the basis of the information which is provided to them.68

Can we envisage a similar pattern about legal regimes that may beapplied to given behaviour in a society? One can imagine goods and serv-ices being accompanied with detail informing customers about the author-ity and the law that would apply to possible disputes. The system is already


68 Here again, a superior authority has to intervene in order to decide what kind of informationhas to be provided by the firms.

used in international transactions, where firms specify the national legalorder that will be deemed competent to solve disputes.

In our prospective search, a similar mechanism could probably be usedwith respect to competing authorities and rules dealing with similar issueswithin a single nation. To illustrate that possibility, let us go back to ourideal world where firms and individuals would be allowed to choosebetween one legal order based on competition law and another one basedon sector specific regulation. In that world, firms could specify the regimethey want to be associated with. They should inform possible customersabout their choice. Customers would then be in a position to make theirdecisions based on information of that nature.

The situation would be similar to those encountered in international law.In such situations, parties structure their operations in anticipation of therules that would apply and the authorities that would intervene, should adispute occur. For instance, they choose to sign an agreement in Francebecause they know that French rules will then be applicable to them. In ourprospective thinking, a similar system would apply with different rules andauthorities being organised in a competitive manner independent of anynational or international consideration. Choosing to deal with a given firmwould mean that disputes would be submitted to the rules and authoritiesselected by that firm. One would thus encounter typical international lawrelated multi-jurisdictional issues, with the difference that the choice wouldnot be attached to a territory. Instead of choosing between, say, French orBritish law, the protagonists would choose between different legal ordersthat may co-exist on a given territory and which may offer variable solu-tions within a framework of public goals and values.

In this pattern, the law would stop being considered as a necessarily cen-trally organised system guiding all participants on the markets or in societyin general. It would rather become a part of goods or services, as much asthe guarantee possibly granted by the manufacturer is nowadays consid-ered by customers as integral to a product they want to purchase. Asobserved several times during this chapter, an organisation of the typeenvisaged here does not eliminate altogether the necessity of a superiorauthority which would design the framework of public goals and valueswithin which institutional and regulatory competition would flourish.Thus, the expectation should not be that the law as we know it will totallybe set aside. At the present stage, one should rather expect an increasingpart of it to be organised on a competitive basis—law becoming more andmore a commodity within a framework set by classical authorities.

124 PLG Nihoul

6

Controlling the New Media: HybridResponses to New Forms of Power

ANDREW MURRAY AND COLIN SCOTT*

THE EMERGENCE AND identification of the new media, premisedupon the development and application of digital technologies, hascreated new sources and locations of power, many not fully docu-

mented or understood. Those new configurations of power which havebeen identified have stimulated distinctive literatures about the most appro-priate mechanisms of control. With much of the literature, classical or‘command and control’ regulation is held either to be undesirable or unfea-sible in the face of the new policy challenges. For one school of thought, thechanging market structures associated with the new media indicate a reducedrole for classical regulation and its virtually total displacement by competi-tion law.1 For another school, the emergence of the Internet presents insuper-able problems for classical regulation and alternative mechanisms of controlbased on self-regulation and architecture are more likely to be effective.

In this chapter we draw together some of the regulatory problems presented by the new media and apply a developed and modified version of

* Andrew Murray is a Lecturer in Law at the London School of Economics. Colin Scott isSenior Fellow in Public Law, Research School of Social Sciences, Australian NationalUniversity and a Reader in Law, Centre for the Analysis of Risk and Regulation, LondonSchool of Economics. This chapter was originally presented at a Workshop on CompetitionLaw and the New Economy sponsored by the Modern Law Review and was first published in (2002) 65 Modern Law Review 491–516. Though we accept full responsibility for errorsand infelicities we gratefully acknowledge the comments on an earlier draft from Workshopparticipants and from Julia Black, John Braithwaite, Neil Duxbury, Peter Grabosky, MathiasKlang, Robin Mansell and David Post.1 The Chicago School of Law and Economics supports market control where markets are com-petitive. If the market is uncompetitive, competition law provides an adequate remedy. Thispremise has been attacked in relation to layered communications networks. See L Lessig, TheFuture of Ideas: The Fate of the Commons in a Connected World (New York, Random House,2001) 110; C Salop and RC Romaine, ‘Preserving Monopoly: Economic Analysis, LegalStandards and Microsoft’ (1999) 7 George Mason Law Review 617.

Lawrence Lessig’s ‘modalities of regulation’2 analysis to thinking about therange of mechanisms which have been developed to address these problems.Accordingly we first provide a description of some of the key problems identified with controlling the new media. Our modified version of Lessig’sanalysis claims that there are four bases of regulation—hierarchy, competi-tion, community and design. We set the analysis to work demonstratingthat these four bases of regulation are observable as means of addressingthe range of regulatory problems of the new media. The tendency to privi-lege one basis for regulation over others appears to us to be consistent neither with empirical observation nor with the normative considerationsof institutional design for good regulation. What we observe is the prevalenceof hybrid forms of control which, when better understood, could provide thebasis for a better informed policy debate about the control of the newmedia.

Differences in approach may partly be explained by reference to the cul-tures and preoccupations within different jurisdictions. The UK and manyEuropean Union states have a strong tradition of self-regulation in the mediagenerally and the legitimacy of this form of governance is widely accepted.3

Private governance forms are generally less well recognised and accepted inthe United States and have been the subject matter of fierce debate over theirlegitimacy.4 A related bias in the US literature is the very high value placedon the constitutional ideal of freedom of speech which feeds into a stronglibertarian underpinning to much discussion of regulation of new media.5

Though freedom of speech may have some constitutional protection in EUstates, the extent to which such a right is qualified by other collective consid-erations is quite pronounced. American scholarship on new media issues isdominated by the ‘legal centralist’ perspective of law and economics whichaccords less recognition to the potential for pluralism in the generation ofnorms than is true of some European scholarship.6

126 Andrew Murray and Colin Scott

2 L Lessig, Code and Other Laws of Cyberspace (New York, Basic Books, 1999) 88.3 J Black, ‘Constitutionalising Self-Regulation’ (1996) 59 Modern Law Review 24.4 The suspicion of private governance institutions in American legal scholarship is forcefullyrepresented by Michael Froomkin’s critique of the Internet Corporation for Assigned Namesand Numbers (ICANN): M Froomkin, ‘Wrong Turn in Cyberspace: Using ICANN to RouteAround the APA and the Constitution’ (2000) 50 Duke Law Journal 17; cf the (European)views of W Kleinwächter, ‘The Silent Subversive: ICANN and the New Global Governance’(2001) 3 Info 259 which are largely approving of the innovation in governance created byICANN.5 M Castells, The Internet Galaxy (Oxford, Oxford University Press, 2001) 33. S Venturelli,‘Inventing E-Regulation in the US, EU and East Asia: Conflicting Social Visions of the Internetand the Information Society’ paper presented to 29th Research Conference on Communi-cation, Information and Internet Policy, October 2001, Alexandria, Virginia available at�http://www.arxiv.org/ftp/cs/papers/0110/0110002.pdf� (visited 19 December 2001).6 The analysis is succinctly made by R Ellickson, ‘The Aim of Order Without Law’ (1994) 150Journal of Institutional and Theoretical Economics 97, which provides a summary of the fullertreatment in R Ellickson, Order Without Law (Cambridge, MA, Harvard University Press,

Whatever the effects of intellectual bias, we suggest that research andthinking on control of new media sectors has generated novel insights onregulation which are of wider application. In particular, the current debateon how forces of control may be used to shape the future development ofnetworks is of wider interest to researchers in the fields of law, economicsand social policy. The debate is centred upon the role of the commons inthe fledgling third generation Internet. Sunstein’s claim that ‘there is noavoiding “regulation” of the communications market’7 has been met by anequally forceful counterclaim by Lessig that

[t]he issue for us will not be which system of exclusive control—the govern-ment or the market—should govern a given resource. The question for uscomes before: not whether the market or the state but, for any given resource,whether that resource should be controlled or free.’8

Lessig’s call for a debate on this issue provides a powerful rallying call tothose lobbying for the deregulation of, in the sense of making free, all lay-ers of the Internet infrastructure.

The debate called for by Lessig is not new. The open source movementled by Richard Stallman and the Free Software Foundation has lobbiedfor deregulation of the code level since the mid-1980s.9 Deregulation atthe content level was built into the original Internet infrastructure by net-work designers such as Paul Baran, Jerome Saltzer, David Clark andDavid Reed.10 This has since been substantially eroded by the develop-ment of intelligent networks such as Resource Reservation Protocol(RSVP).11 There is no doubt many in the new media sector will respondto Lessig’s analysis and over the next five years the wider dialogue of therole of regulation within political science, media and economics will bestrongly influenced by this currently narrow legal debate. Thus, while wemake extensive use of examples drawn from new media, we suggest thatthe developed models of control which we discuss are of interest to policymakers and researchers with interests in governance and regulation generally.

Controlling the New Media 127

1991). See also R Cooter, ‘Against Legal Centrism’ (1993) 81 California Law Review 417; L Lessig, ‘The Regulation of Social Meaning’ (1995) 62 University of Chicago Law Review943.

7 C Sunstein, Republic.Com (Princeton, Princeton University Press, 2001) 128. 8 Lessig, n 1 above n 12. 9 Ibid 52–61.

10 Ibid 34–44; J Saltzer, D Reed and D Clark, ‘End-to-End Arguments in System Design’ (1984)2 ACM Transactions in Computer Systems 277. On-line version at �http://web.mit.edu/Saltzer/www/publications/endtoend/endtoend.pdf� (visited 4 January 2002).11 Discussed below.

1. NEW MEDIA AND THE PROBLEMS OF EFFECTIVE CONTROL

Processes of digitalisation associated with the development of new mediahave brought about important reconfigurations of power. The Internet, forexample, provides widespread access to technology based on a network ofnetworks and addressing systems which connect computers globally.12 It issaid to create a space where users can engage in a variety of activities with asubstantial autonomy from state power which does not exist in non-digitalmedia.13 Digitalisation of broadcasting and mobile telecommunications cre-ate niches for new forms of service provider, shifting power away both fromthose who own the physical infrastructure of networks and from those whoown content. We identify in this section three general problems of new media(that is problems which apply generally or to more than one medium) whicharise from shifts in power. None of these problems is exclusive to the newmedia, though each emerges with interesting new features in this context.They are the problems of regulatory arbitrage, anonymity and scarcity ofresources. In each case once prevalent governance forms based on publicownership are no longer fashionable (and for some no longer feasible) theurgency of investigating other forms of control is enhanced. We should beclear that these are not the only problems associated with the new media.Among the other pressing policy problems are the issues relating to accessi-bility of digital broadcasting and communications services to less advan-taged consumers (which can be defined both in economic and social terms)14

and the extent to which content of digital broadcasting should be controlled(in the manner that both negative and positive content controls apply to ana-logue broadcasting).15 Discussion of these issues is precluded for reasons ofspace and in the belief that the theoretical frame developed is sufficientlyaddressed by the policy problems which we do discuss.

A. The Regulatory Arbitrage Problem

The problem of regulatory arbitrage emerges wherever subjects of regulationhave sufficient mobility in their operations or activities that they can choose


12 B Leiner et al, ‘A Brief History of the Internet’ Internet Society available at �http://www.isoc.org/internet/history/brief.shtml� (visited 7 January 2002); Castells n 5 above, ch 1. 13 S Sassen, ‘Digital Networks and the State: Some Governance Questions’ (2000) 17 Theory,Culture and Society 19, 20. According to Lessig much of this autonomy is hard-wired into thenetwork by its end-to-end architecture, Lessig above n 1, 26–41.14 M Lemley and D McGowan, ‘Legal Implications of Network Economic Effects’ (1998) 86California Law Review 479; P David, ‘The Evolving Accidental Information Super-Highway’ (2001) 17 Oxford Review of Economic Policy 159; M Cave and R Mason, ‘TheEconomics of the Internet: Infrastructure and Regulation’ (2001) 17 Oxford Review ofEconomic Policy 188.15 See C Sunstein, ‘Television and the Public Interest’ (2000) 88 California Law Review 499; D Goldberg, T Prosser and S Verhulst, Regulating the Changing Media: A Comparative Study(Oxford, Oxford University Press, 1998).

to be regulated by one regime rather than another. The effect is to create aform of market for regulation within which dissatisfied subjects can ‘exit’one regime in favour of another. Regulatory arbitrage, seen as a problemfor authorities attempting to capture activities within their web, can also beseen as a solution to problems of excessive or inappropriate regulation as itlimits the capacities of authorities.16 The problem has an interesting double-edged character in the new media, since options to relocate to avoidparticular regulatory regimes may be available both to service providersand consumers. Thus broadcasters can relocate their operations to differentjurisdictions to evade national regulation (and this predates digitalisation)while listeners and viewers can relocate from the more controllable formsof delivery to satellite and Internet. One of the problems raised by regula-tory arbitrage is the risk that competing standards for the new digitalbroadcasting transmission services might develop. This is squarelyaddressed with harmonised rules requiring all member states to legislate forcommon standards in the EU, notably in respect of consumer equipmentfor conditional access to services.17 Under the terms of European Unionlegislation the EU rules on broadcasting regulation apply only to broadcast-ers established in a state to which the applicable directive applies.18 Thedirective’s requirements that Member States apply their domestic broad-casting rules to all broadcasters established within the state has been inter-preted so as to require Member States to apply their rules as intensely tobroadcasters directing their programming at other Member States.19 Thisinterpretation is intended to preclude countries like the UK from settingthemselves up as attractive locations for establishment of overseas broad-casters through the application of a more liberal regime than would applyto domestic broadcasters.20 This is a particular issue with broadcastersseeking to evade what they regard as overly restrictive domestic rules, forexample on advertising to children or transmission of pornography.

Regulatory arbitrage in Cyberspace (that is applying to the Internet) is afocal point for two opposing schools of thought, the Cyberlibertarians andthe Cyberpaternalists. The primary argument of the Cyberlibertarians isthat Cyberspace is unregulable due to its design. Cyberspace is a uniquejurisdiction, as it has no physicality or real-world existence. It is possible toconceive of Internet users simultaneously in Cyberspace and in a grounded,real-world jurisdiction.21 It is this duality and the non-physicality of


16 See W Bratton, J McCahery, S Picciotto and C Scott (eds), International RegulatoryCompetition and Coordination (Oxford, Oxford University Press, 1996).17 Dir 95/47/EC [1995] OJ L/281/51, 23 November 1995 Art 4; Broadcasting Act 1996.18 Council Directive 89/552/EEC [1989] OJ L/298/23 as amended by Dir 97/36/EC [1997] OJL/202/ 60, Art 2(1); B Drijber ‘The Revised Television Without Frontiers Directive: Is it Fit forthe Next Century’ (1999) 36 Common Market Law Review 87, 92.19 Commission v United Kingdom Case C–222/94 [1996] ECR I–4025.20 Broadcasting Act 1990, s 43.21 Lessig above n 2 190.

Cyberspace which allows for regulatory arbitrage. In the physical worldsovereignty is exercised by governments over defined physical territories. Auser who wishes to be regulated by a different regulatory structure maytake steps to relocate either themselves or their activities. In Cyberspaceusers may transcend physical borders with ease and may choose to take onany guise or form desired (see below ‘The Anonymity Problem’). Users whoprefer a regulated environment where there are structured discussions oncarefully selected topics and where content is closely monitored and censored may choose to join a regulated and monitored Cybercommunitysuch as America Online (AOL). Users seeking uncensored discussion andcomplete freedom of speech may make use of a virtual chat room on theUSENET system or may use an Internet Service Provider (ISP) to enterunmonitored discussion boards on the Web. These freedoms allow users tochoose freely the regulatory structure they wish to follow while in Cyberspace.Thus a citizen of Germany can enter a USENET discussion group on theHolocaust and post denial messages, something he or she would be unableto do freely in their home state. Similarly a UK citizen may post informationwhich is in breach of the Official Secrets Acts. Although strictly speakingthese citizens are still committing offences within their physical jurisdiction,they can do so without fear of prosecution as in Cyberspace they have takenon a different personality and therefore are unlikely to be traced and prosecuted.22 These citizens have effectively removed themselves from theregulatory control of their sovereign government and have chosen to beregulated by another set of regulatory values and norms. This is because, asdramatically put by David Post,

Cyberspace … does not merely weaken the significance of physical location itdestroys it … . they do not cross geographical boundaries (in the way that sayenvironmental pollution crosses geographical boundaries), they ignore theexistence of boundaries altogether.23

B. The Anonymity Problem

The non-physicality of Cyberspace allows Internet users to choose to adopta different persona from their real-world personality (pseudonymity) or tohide all details of their personality (anonymity). Pseudonymity andanonymity provide a further set of problems for regulators. As well as facil-itating regulatory arbitrage by allowing citizens to conceal their identity,


22 With a degree of computer literacy they can ensure that it would be almost impossible forlaw enforcement agencies in the physical world to track them down and prosecute. This is dis-cussed further below. 23 D Post, ‘Governing Cyberspace’ (1997) 43 Wayne Law Review 155. On-line version at�http://www.temple.edu/lawschool/dpost/Governing.html� (visited 4 January 2002).

thereby inhibiting the application of civil, administrative and criminalregimes while in Cyberspace, pseudonymity and anonymity also allowNetizens to carry out transactions in an unregulated manner.24 Two exam-ples which may be given are the distribution of hate or defamatory speechand access to regulated content.

To begin with the latter, there are certain areas in our physical societieswhere we regulate access to certain persons. Children are not permitted accessto public bars or licensed sex shops. In addition there are activities that arerestricted to certain persons. Only those with driving licences may legallydrive and only those who are members of the appropriate professional soci-ety may practise as a lawyer. A lack of physical persona makes the regulationof such simple activities much more complex in Cyberspace. A child may takeon an adult personality and gain access to pornographic content.25 In thephysical world a child entering a licensed sex shop would be removed by themanager, whereas in Cyberspace the elements of physicality are lost and theability to regulate is impaired. This is not to say the anonymity problem ren-ders regulation of access impossible. Community-based control structures,supported by design-based elements have met with a high degree of success.26

More worryingly, the access control problem allows for the potentially moreharmful conduct of adults passing themselves off as children. In the same waythat children are prevented from accessing certain adult areas of the physicalworld, there are areas where unauthorised adults are kept out to protect chil-dren.27 Children nowadays are educated to keep away from strangers and tobe wary of any unusual adult contact. Again the lack of physicality inCyberspace raises problems. Users cannot discern the age of others in the chatroom intended for children. As it is at the user’s discretion how much infor-mation he wishes to reveal about himself there is no practical methodology toensure adults do not pose as minors for as long as Cyberspace supports ananonymous culture. Given that any attempt to remove the currently availableculture of pseudonymity/anonymity would probably lead to a high level ofregulatory arbitrage, there is no apparent means to deal with such problems.

Further, the easy availability of anonymous messaging allows individualsto take part in activities without being required to meet usual societalnorms. Individuals may make antisocial comments without fear of beingostracised by society at large. The technology of anonymous remailers whencoupled with encryption technology can ensure an untraceable messagesource.28 This may be used to distribute comments about an individual or


24 Netizen is the universally accepted term for a ‘citizen of the Internet’.25 Lessig above n 2, 174.26 See below 3.D. ‘Other Forms of Control’.27 Examples would be schools, children’s playgrounds, nurseries and other controlled environments.28 The technology is described in some detail by Michael Froomkin in ‘The Internet as a Sourceof Regulatory Arbitrage’ in B Kahin and C Nesson (eds), Borders in Cyberspace (Cambridge,

organisation without fear of prosecution or social exclusion.29 Anonymityin Cyberspace creates a unique culture where expression free from the nor-mal constraints of legal and social control is common. Even the UnitedStates with its particular emphasis on the right to free speech cannot allowcompletely unfettered or unrestricted freedom of expression.30 Cyberspaceuniquely offers a forum for unfettered free expression.31 Although it maybe argued that ISPs or other moderators of discussion groups may removeoffending messages, they may be reposted somewhere else in Cyberspacealmost immediately. Also Netizens may directly address others via e-mail.Again although this practice, known as spamming, is regulated in Europeby the Distance Selling and E-commerce Directives32 and by other enact-ments worldwide, the availability of anonymous communications renderssuch enactments impotent within Cyberspace. He who cannot be caughtcannot be punished. Anonymity therefore allows for perfect freedom ofexpression, which in the physical world has been tempered by even the mostliberal of regimes.

C. The Scarce Resources Problem

Regulators in the new media are called upon to oversee systems of alloca-tion of scarce resources. All new media sectors draw heavily on limited


MA, MIT Press, 1997). On-line version available at �http://www.law.miami.edu/~froomkin/articles/arbitr.htm� (visited 4 January 2002)

29 Following the enactment of the Communications Decency Act 1996 a US-based ISP has nothird party liability for any libellous messages carried on their system (s 230). In the UK andthe European Union ISPs may have third party liability if they fail to act once the nature of alibellous message is drawn to their attention. See Godfrey v Demon, Internet [1999] 4 All ER342 and the E-Commerce Directive (Dir 2000/31/EC [2000] OJ L/178/1–16), 17 July 2000Art 12.30 Sunstein n 7 above, 151–53. For a fuller account of the philosophical foundations uponwhich restrictions on the first amendment are justified see F Schauer, ‘The Aim and Target inFree Speech Methodology’ (1989) 83 Northwestern University Law Review 562; RKGreenawalt, ‘Free Speech Justification’ (1989) 89 Columbia Law Review 119. 31 Several commentators cited the success of the complainers in UEJF (L’Association Uniondes Etudiats Juifs de France) et Licra (La Ligue Contre le Racisme et l’Antisémitisme) v Yahoo!Inc., L’Ordonnance du Tribunal de Grande Instance, 20 November 2000 as evidence of theability of courts to regulate expression in Cyberspace. This confidence has been substantiallyeroded following the finding of Judge Fogel in Yahoo! Inc v La Ligue Contre Le Racisme etL’Antisemitisme (LICRA), 169 F Supp 2d 1181 (ND Cal 2001), that the French order is notenforceable in the United States as ‘[it] chills Yahoo’s First Amendment Rights … and that thethreat to its constitutional rights is real and immediate’ (at 1194).32 Distance Selling Directive, Dir 97/7/EC [1997] OJ L/144/19–27, 4 June 1997; E-CommerceDir, Directive 2000/31/EC [2000] OJ L/178/1–16, 17 July 2000. The Distance Selling Directivewas implemented in the UK through the Consumer Protection (Distance Selling) Regs 2000 SI 2334. At the date of writing the E-Commerce Directive awaits implementation.

resources, whether these be natural resources such as spectrum for thetelecommunications or broadcasting sectors, or man-made resources suchas domain names in relation to Cyberspace. Digital developments do, insome respects, reduce existing scarcity problems. Thus digital broadcastinguses spectrum more efficiently and thus enhances capacity.33 This may inturn create a problem for regulators seeking to maintain controls designedto ensure pluralism in the broadcasting sector.34

The spectrum scarcity problem is exemplified by the emergent marketfor third generation (3G) mobile communications.35 3G mobile will makemultimedia services available to mobile phone users anywhere in the world,combining satellite and terrestrial digital capacities. This development hasthe potential both substantially to displace a number of current communi-cations technologies, notably second generation mobile and fixed linktelephony and to grow new markets in mobile communications. Most EUMember States have concluded that spectrum scarcity permits them tolicense between 4 and 6 network operators for 3G mobile.36 The objectivesof the licence allocation processes have been to promote the developmentof competitive markets, to allocate the spectrum to those best placed to useit and in many cases to secure windfall fee-income to the finance ministry.Further policy making will be necessary to determine the terms on whichservice providers who do not have network operators licences can haveaccess to the networks for the provision of services.

Scarce resources are also a problem in Cyberspace. The Internet is oftenseen as a network without resource constraints. If more resources areneeded more computers can be added to the network. This though onlyincreases the available processing power of the net; there are other key areaswhere resources remain scarce. One area is bandwidth.37 Modern telecom-munications networks rely on the ability to transmit data from one sourceto another and in this respect the Internet is no different from mobiletelecommunications networks. Network content is increasingly sophisti-cated. Consumers are demanding faster and more stable access to the net-work, to allow them to listen to real time audio transmissions and to viewstreaming video transmissions. These additional network demands are put-ting the current network protocols under strain and commercial providersof such services are calling for the current protocols to be substantiallyoverhauled to provide for the flow of such services free from the current


33 R Collins, ‘Back to the Future: Digital Television and Convergence in the United Kingdom’(1998) 22 Telecommunications Policy 383, 384–85.34M Cave, ‘Regulating Digital Television in a Convergent World’ (1997) 21 TelecommunicationsPolicy 575, 590.35 Sometimes known as Universal Mobile Telecommunications System (UMTS)36 P Curwen, ‘Next Generation Mobile: 2.5G or 3G?’ (2000) 2 Info 455, 461.37 See J Glasner, ‘Move Over, Pork Bellies’ Wired News, 20 May 1999 (available at�http://www.wired.com/news/business/0,1367,19796,00.html� (visited 4 January 2002);Lessig n 1 above, 47.

problems of latency (delays in transmission) and jitter (variations indelays).38 These problems are caused by the current network protocol,Internet Protocol version 4 (IPv4) which employs a ‘best effort’ quality ofservice.39 The best effort service is simply an onward transmission servicewhich routes packets of information based upon information on congestiongiven to the sender from the next point or node in the network. This meanspackets of information relating to a single transmission can become sepa-rated and can arrive with delay variation causing jitter. Simple Internetapplications such as e-mail or web browsing can tolerate these delays anddifferentials, but streaming audio and video cannot: Internet telephony forexample cannot tolerate a delay of more than 250 milliseconds.40

To deal with these problems network designers have suggested the cre-ation of an intelligent network which would allow for quality of service(QoS) solutions.41 The implementation of QoS systems involve either theimplementation of a complex virtual overlay network (VON) which wouldallow traffic from a single network flow to pass through routers withoutcompeting with traffic from other network flows42 or, as seems more likely,the implementation of a new network protocol, Internet Protocol version 6(IPv6).43 IPv6 offers many advances over IPv4. It allows for better homo-geneity of transmission. In the event of network queuing it allows forstreaming transmissions to be packaged together. This means time criticaltransmissions such as streaming audio and video may be prioritised overless time sensitive transmissions such as e-mails. Also it crucially supportsthe Resource Reservation Protocol (RSVP) developed by Cisco Systems andMCI WorldCom which allows service providers to sell bandwidth to usersallowing them to prioritise their transmissions over other traffic using the


38 See eg C Huitema (Microsoft Corporation), ‘How Will IPv6 Change the World?’ paper presented to IPv6 2000, 19–20 October 2000, Washington DC available at �http://www.ipv6forum.com/navbar/events/xiwt00/presentations/html/huitema/� (visited 20 December2001); Y Pouffary (Compaq), ‘The IPv6 Advantage’ paper presented to IPv6 2000, 19–20 October 2000, Washington DC available at �http://www.ipv6forum.com/navbar/events/xiwt00/presentations/html/pouffary/� (visited 20 December 2001). 39 David above n 14, 173. 40 Lessig above n 1, 46. 41 Lessig above n 1, 46–47. Generally Lessig is wary of such solutions as adding intelligence tothe network allows for control in the content layer. 42 David above n 14, 173; Computer Science and Telecommunications Board, NationalResearch Council, The Internet’s Coming of Age (Washington DC, National Academy Press,2001) 102–3. Available at �http://bob.nap.edu/html/coming_of_age/� (visited 20 December2001)43 IPv6 also solved the problem of a scarcity of Internet Protocol (IP) addresses. Currentlythere are just under 4 billion available IPv4 addresses. Although this seems a healthy figure,large organisations such as AT&T and MIT hold up to 16 million addresses each. Currently ifyou use dial-up access you will be allocated a temporary IP address while connected. Thisallows several users to share the same IP address. With new networked tools some analystssuggested the total fund of IP addresses would be exhausted by 2004. IPv6 allows for 1038

IP addresses—more than enough for the foreseeable future.

same routers.44 This functionality comes at a cost. These developments willalmost certainly lead to the development of fragmented proprietary networkswithin the wider network structure and an end to the current end-to-endinfrastructure of the Internet.45

Although bandwidth scarcity is not unique to Cyberspace, the scarcity ofdomain names is.46 It may seem bizarre to claim domain names are a scarceresource. The permutation of domain names seems almost limitless. Theymay be made up of a string of up to 63 characters47 in any permutation anda top-level domain of which there are more than 250.48 Despite this there isa scarcity of usable domain names. Usable domain names reside almostexclusively in the .com top-level domain and are made up of recognisableterms in major languages.49 There is a paucity of such names as usabledomain names are of a one mark one owner architecture, whereas previoustrade mark systems had been of a one mark many owners architecture.50

Competing demands for usable domain names quickly arose and the bodiescharged with overseeing the domain name system (initially the InternetAssigned Numbers Authority (IANA) and Network Solutions Inc., andmore latterly the Internet Corporation for Assigned Names and Numbers(ICANN) )51 were required to develop a policy to deal with these competingclaims. This policy, the Uniform Domain-Name Dispute-Resolution Policy,attempts to balance the rights of trade mark holders against the first-user


44 ‘Traffic to Take the High Road—for a Price’ Wired News (27 March 1997). Available at�http://search.hotwired.com/search97/s97.vts?Action�FilterSearch&Filter�docs_filter.hts&ResultTemplate�news.hts&Collection�news&QueryMode�Internet&Query�IPv6�(visited 20 December 2001). 45 David n 14 above, 174–78. A Odlyzko, The Economics of the Internet: Utility, Utilization,Pricing, and Quality of Service (AT&T Labs-Research, 1998) 27–28. Available at �http://www.dtc.umn.edu/~odlyzko/doc/internet.economics.pdf� (visited 20 December 2001).46 M Mueller, ‘Competing DNS Roots: Creative Destruction or Just Plain Destruction?’ paperpresented to 29th Research Conference on Communication, Information and the Internet,October 2001, Alexandria, Virginia, available at �http://www.arxiv.org/ftp/cs/papers/0109/0109021.pdf� (visited 19 December 2001).47 A domain name may currently occupy a maximum of 67 characters. See J Jones, ‘InternetDomain Registrars Increases Length of Internet Domain Names’ InfoWorld available at: �http://www.infoworld.com/articles/ic/xml/99/12/16/991216iclonger.xml� (visited 27 March 2002). As the top-level domain uses three characters and the root (dot) takes onecharacter, this leaves 63 characters free for use as a SLD. 48 Currently there are 239 Country Code top-level domains (ccTLDs) detailed in ISO-3166, 12generic top-level domains (gTLDs) and two US Federal TLDs (.gov & .mil). 49 The latest Network Wizards Domain Name Survey (July 2001) records 37,502,747 .comdomains. By comparison the Oxford English Dictionary only contains ‘over half a millionwords’ (source: ‘About the Oxford English Dictionary’ �http://www.oed.com/public/inside/�(visited 4 January 2002).50 See for example A Brunel and M Laing, ‘Trademark Troubles with Internet Domain Namesand Commercial Online Service Screen Names’ (1997) 5 International Journal of Law andInformation Technology 1; A Murray, ‘Internet Domain Names: The Trade Mark Challenge’(1998) 6 International Journal of Law and Information Technology 285. 51 IANA and ICANN are non-governmental not-for-profit agencies. Network Solutions Inc isa subsidiary of Verisign Inc, a for-profit publicly listed company.

policy previously applied. It is an extremely controversial policy and will beexamined in depth below when we analyse the effectiveness of controlmechanisms in the new media.

2. EXTENDING THE ‘MODALITIES OF REGULATION’ ANALYSIS

Lawrence Lessig’s Code and Other Laws of Cyberspace is widely regardedas one of the most complete analytical attempts to capture the variety offorms which regulation of new media does or may take.52 Lessig contendsthat there are four distinct modalities of regulation. He attaches to these thelabels law, markets, norms and architecture. He thinks of these in terms ofconstraints on action.53 Thus law constrains through the threat of punish-ment, social norms constrain through the application of societal sanctionssuch as criticism or ostracism, the market constrains through price andprice-related signals, and architecture physically constrains (examplesinclude the locked door and the concrete parking bollard).

Lessig’s work is of great value for reminding us of the importance ofarchitecture as a basis for regulation. The potential for controls to be builtinto architecture have long been recognised, as exemplified by JeremyBentham’s design for a prison in the form of a panopticon (within whichthe architecture permitted the guards to monitor all the prisoners) and themore recent observations of the way in which visitors to Disney World arecontrolled by an architecture in which nearly every aspect of the design hasa disciplinary function.54

Lessig observed the various constraints that are built into software bytheir designers. Such architectural constraints in software code are chieflyused for commercial purposes (such as restricting the user’s use to whatthey have paid for or segmenting the market so as to charge higher prices in


52 n 2 above. Cf the five way analysis of controllers in ordering society put forward byEllickson, The Aim of Order Without Law above n 6, 131. Ellickson sees order as a productof first party (or self-control) second party (or contractual) control, third party control (basedon social forces and norms), organisation (with associated institutional apparatus) and, lastly(significantly) government with the laws. There is a substantial political science literature onalternative instruments of governance: see J Kooiman (ed), Modern Governance (London,Sage, 1993) and C Hood, The Tools of Government (London, Macmillan, 1983).53 Lessig above n 2, 235–39.54 M Foucault’s research on the history of the prison has been responsible for generating newinterest in surveillance generally and Bentham’s panopticon in particular: See Discipline andPunish: The Birth of the Prison (Harmondsworth, Penguin, 1977, trans A Sheridan), ch 3. C Shearing and P Stenning, ‘From the Panopticon to Disney World: The Development ofDiscipline’ in A Doob and E Greenspan (eds), Perspectives in Criminal Law (Aurora, CanadaLaw Book Co, 1984). Crime control through design is exemplified by A Lester, CrimeReduction through Product Design (Australian Institute of Criminology, Trends and Issues inCriminal Justice no 206, 2001); N Kumar Kaytal, ‘Architecture as Crime Control’ (2002) 111Yale Law Journal 1039.

some segments without the risk of arbitrage) but may also be used for otherregulatory purposes (as with the controls placed on users by Filterware).55

Lessig suggests that as a means of regulation architecture is self-executingand thus different at least from norms and law.56 This claim appears cor-rect up to a point. However the analysis which separates the functions of acontrol system shows that the standard-setting element of architecture isnot self-executing but is, by definition, designed by human hands. Somearchitecture-based regimes may be self-executing as to monitoring andbehaviour modification. A parking bollard, for example, requires no fur-ther agency on the part of a regulator to control parking. Other architec-tural controls do rely on actions by the controller. For example, Bentham’spanopticon requires that prison guards actively monitor prisoners andintervene to control deviance. The panopticon can thus be seen as a hybridof hierarchy and architecture.

The importance of Lessig’s analysis is to draw attention to the variety ofbases for control which can be deployed in the face of anxiety that techno-logical change (such as the Internet) and economic change (such as globali-sation) tends to make a variety of different forms of conduct unregulatable.The argument that variety in forms of activity requires an equal or greatervariety of bases for control if regulation is to be effective has found formalexpression in the cybernetics ‘law of requisite variety’. It is expressed inother terms as the principle that ‘only variety can destroy variety’.57 Thesceptical position which Lessig challenges is premised in part upon a myththat social and economic activity has traditionally been highly amenable toregulation, conventionally defined. Recent scholarship on the limits to con-trol has emphasised the problems of trying to regulate social and economicactivity.58 This work has highlighted the importance of developing regula-tory regimes which seek to steer or stimulate activities within the target sys-tem indirectly as an alternative to external command and control.59 Lessig’s


55 Filterware is discussed further below. See 3.D. ‘Other Forms of Control’.56Lessig above n 2, 236–37. Lessig claims that markets have in common with norms and law thefact that they require human agency and are not self-executing. This claim is contentious (thoughLessig does not recognise this) as the control exerted by a market does not operate at the level ofthe individual seller and buyer, but rather in an aggregate. In the perfectly competitive marketmodel the decisions of an individual buyer or seller cannot affect the operation of the market. 57 S Beer, Decision and Control (London, Wiley, 1966) 279–80.58 P Grabosky, R Smith and G Dempsey, Electronic Theft: Unlawful Acquisition in Cyberspace(Cambridge, Cambridge University Press, 2001) 5–11; P Nonet and P Selznick, Law andSociety in Transition (New York, Harper Row, 1978); I Ayres and J Braithwaite, ResponsiveRegulation: Transcending the Deregulation Debate (Oxford, Oxford University Press, 1992);N Gunningham and P Grabosky, Smart Regulation: Designing Environmental Policy (Oxford,Oxford University Press, 1998).59 This prescription has its origins in systems theory: Beer above n 57; and is found in both thepolitical science literature: A Dunsire, ‘Tipping the Balance: Autopoiesis and Governance’(1996) 28 Administration and Society 299, and legal literature: G Teubner, ‘Juridification:Concepts, Aspects, Limits, Solutions’ in G Teubner (ed), Juridification of Social Spheres(Berlin, De Gruyter, 1987).

work has the potential to support efforts to reconceive regulation in a sensethat is both more modest in its claims and ambitions and more useful inproviding mechanisms not only, or perhaps mainly, of direct control butalso of indirect control. A key method of this new approach, which wedeploy in this discussion, is to identify effective regulation in whatever formit takes and to seek to support it, develop it or extend it by analogy to otherdomains in which there are problems of regulation.

The concept of regulation deployed in Lessig’s analysis is a broad one,extending beyond the narrowly defined ‘systematic oversight by referenceto rules’ to encompass four ‘modalities of regulation’ which have the objector effect of holding behaviour within one state among all the possible stateswhich the behaviour might take. Lessig refers to the ‘ “net regulation” ofany particular policy’ domain as the ‘sum of the regulatory effects of thefour modalities together’.60 Regulation in this expansive sense is conceptu-ally closer to the usage of biologists and sociologists than to that oflawyers.61 It refers to any control system. To be viable, within the terms ofcontrol theory, a control system must have some standard-setting element,some means by which information about the operation of the system can begathered and some provision for modifying behaviour to bring it backwithin the acceptable limits of the system’s standards.62 With regulation,information gathering is usually achieved through monitoring by an agency,department or self-regulatory body and deviations addressed by applicationof formal and informal sanctions (See Figure 6.1 below).

When locating Lessig’s description within the stricter analysis of controltheory some problems emerge both with the labels and the concepts whichthey describe. Put simply the conceptual schema, drawn from Lessig’s workin law and economics, needs enriching if it is to capture the institutionalvariety in control. Our earlier discussion of control theory suggests that theappropriate schema involves not only a four way division between differentbases of control, but also a further fine grained analysis of the three differ-ent elements necessary to generate a control system (standard-setting, infor-mation gathering and behaviour modification). This development of theanalysis provides a clearer descriptive framework for understanding howcontrol is or can be achieved and opens up the possibility for identifying thewide range of control systems which appear as hybrids of two or moremodalities of regulation. To develop this analysis we draw not only onLessig’s work, but also on attempts to deploy cultural theory to identify


60 L Lessig, ‘The Law of the Horse: What Cyberlaw Might Teach’ (1999) 113 Harvard LawReview 501, 508. 61 R Baldwin, C Scott and C Hood (eds), ‘Introduction’ in Socio-Legal Reader on Regulation(Oxford, Oxford University Press, 1998); M Clarke, Regulation (London, Macmillan, 2000).62 C Hood, H Rothstein and R Baldwin, The Government of Risk: Understanding RiskRegulation Regimes (Oxford, Oxford University Press, 2001) 21–27.

variety in control systems.63 This analytical frame has recently been put towork in analysing variety in risk regulation regimes.64 The term ‘regime’ isapt to capture variety not only in standards and standard-setting (whichrepresents the bias in Lessig’s analysis), but also in the institutional dimen-sions of information gathering and behaviour modification. The regimeanalysis makes it transparent that the various functions which contribute toviable control systems can be widely dispersed among state and non-stateactors, even within a single regime, and can be assembled in mixed orhybrid forms.

Lessig’s conceptualisation of ‘law as command’65 suffers from a weak-ness in that it fails to capture all of the control systems which are within theset of command based or, as we label it, hierarchical control. Law, in thisconception, refers only to state law (whether made by judges, or, more com-monly in this context, legislatures)66 and neglects the plurality of formswhich hierarchical control structures may take. The richer conception ofhierarchy looks to the form of control rather than its source. Thus theregime for developing Internet domain names has important elementswhich are non-state in character and yet which are distinctly hierarchical(and are discussed further below in Section 3(a)). The term law also suffersfrom the difficulty that it is often deployed in a way which infers only stan-dards and not the institutional elements of a control system (viz informationgathering and behaviour modification). Law in Lessig’s terms is merely theconstraint placed upon the individual. Accordingly hierarchical control provides both a better label and a substantively enriched conception of thismodality of regulation.

The concept of norms as it is deployed in Lessig’s analysis follows ausage developed in the social psychological literature—referring to sharedpatterns of behaviour—but which is unconventional and unhelpful in thestudy of law. Even in its psychological usage, the term norm does notdescribe the institutional dimensions of a control system, but rather a set ofstandards which exist between a particular social group for the time being.We argue that the preferred meaning of the word norm is as the genericterm for standards, guidelines and legal and non-legal rules.67 The controlform which involves societal or group standards, peer-based informationgathering and behaviour modification based on social sanctions such asostracisation or disapproval, we refer to as community-based control.


63 C Hood, ‘Control Over Bureaucracy: Cultural Theory and Institutional Variety’ (1996) 15Journal of Public Policy 207.64 Hood, Rothstein and Baldwin above n 62, 9–14.65 Lessig above n 2, 235.66 Lessig above n 60, 507.67 P Drahos and J Braithwaite, Global Business Regulation (Oxford, Oxford University Press,2000) 20.

This category includes not only the social norms which exist generally orbetween particular groups, but also some elements of more formalisedregimes, as where self-regulatory standards are socially generated and writ-ten down and then combined in a hybrid form with hierarchical elements tocreate a self-regulatory control system which is a hybrid between commu-nity and hierarchical bases.

The concepts of markets and architecture as they are deployed by Lessigare each under-inclusive. Rivalry and competition provide a form of controlin environments where there is no identifiable market. Indeed recent publicsector reforms have made widespread use of what we will call competition-based controls in non-market situations.68 Additionally there is a markedelement of regulatory competition applying to the development of regula-tory standards in some domains both in the US and the EU.69 Where theconditions for such regulatory competition exist (a topic of hot debate) andstates are permitted to develop their own rules, competition for client busi-nesses is said to create a check on any tendency to ‘over-regulate’.70

The concept of architecture, referring in Lessig’s terms to the whole builtenvironment with and without intended effects,71 does not capture thewhole set of control mechanisms which are premised upon design as a basisof control. Thus there are social and administrative systems which havedesign features which create control in a way in which the regulatee cannotaffect. A key example is the deployment of ‘contrived randomness’ in theoversight of taxpayers or employees so as to reduce the scope of thesegroups to exploit a wholly predictable system of opportunities and pay-offs.72 Accordingly we relabel this fourth modality of regulation asdesign.73 The different elements of each of the four types of regulation areillustrated in Figure 6.1.


68 P Self, Government by the Market (London, Macmillan, 1993).69 D Esty and D Gerardin (eds), Regulatory Competition and Economic Integration (Oxford,Oxford University Press, 2001).70 W Bratton, J McCahery, S Picciotto and C Scott, ‘Introduction: Regulatory Competitionand Institutional Evolution’ in Bratton, McCahery, Picciotto and Scott (eds), InternationalRegulatory Competition and Coordination (Oxford, Oxford University Press, 1996).71 Lessig above n 60, 507–8.72 Hood above n 63, 211–14.73 This concept of design has a loose affinity with the deployment of the term ‘technologies’ inthe Foucauldian literature on governmentality. It is possible that the term technologies ‘linkingtogether forms of judgement, modes of perception, practices of calculation, types of authority,architectural forms, machinery and all manner of technical devices with the aspiration of pro-ducing certain outcomes in terms of the conduct of the governed.’ N Rose, ‘Government andControl’ (2000) 40 British Journal of Criminology 321, 323 infers a rather wider range ofinstrumentalities than are inferred by the concept of design in this chapter. For deployment ofthe concept of technologies in regulatory theory see J Black, ‘Decentring Regulation:Understanding the Role of Regulation and Self-Regulation in a “Post-Regulatory” World’(2001) 54 Current Legal Problems 103.

It is part of Lessig’s argument that there is scope for the use of hybrid formsof regulation which link two or more of the ‘pure’ modalities of regulationnoted above.74 In particular, he suggests there is scope to link law andarchitecture in his terms, for example by mandating software designers tobuild certain elements into software code in pursuit of public regulatoryobjectives.75 However we think he underplays the extent to which contem-porary control is already based on hybrid regulatory forms and the extentto which a wide variety of regulatory hybrids may be useful in developingregulatory control. Indeed, underlying Lessig’s argument is a claim thatthere is considerable novelty to the nature of law in Cyberspace, a viewseemingly accepted by those Cyberlibertarians who contest the normativedimension to Lessig’s work.76 Nowhere in the work of Lessig or his criticsis this claim substantiated. As Lessig himself recognises, features which wemight call design or architecture have long been fundamental to the way weare governed, whether by features of the built environment (such as theParisian boulevard system) or the Byzantine systems of an obscure public


74 Lessig above n 60, 511–14.75 Ibid 514–22.76 Lessig above n 2, 5–6; D Post, ‘What Larry Doesn’t Get: Code, Law and Liberty inCyberspace’ (2000) 52 Stanford Law Review 1439, 1443.

Element of a Hierarchical Community- Competition- Design-BasedControl Control Based Control Based Control ControlSystem

Standard Law or other Social norms Price/Quality Inbuilt designSetting formalised ratio (and, features and

rules equivalents with social andnon-market administrativedecisions) systems

Information Monitoring Social Monitoring by Interaction ofGathering (by agencies or interaction dispersed design features

third parties) buyers, clients, withetc environment

Behaviour Enforcement Social sanctions Aggregate of As forModification (eg ostracism, decisions by information

disapproval) buyers, clients, gathering etc on (self-executing)purchase,take-up, location etc

Figure 6.1 Elements of Control Systems

bureaucracy or of commercial actors such as banks and insurance companies.It is not clear that design of software is fundamentally different from designin other aspects of social and economic activity. Wherever it is deployed ithas controlling effects and a potential for those controlling effects to beturned towards different or modified effects.

If each of the four pure bases of regulation is theoretically capable ofbeing deployed on its own and with each of the other three bases (givingfour single bases, six pairings, four threesomes and one foursome) thenthere are fifteen forms of regulation in total. There is no empty set since alldomains are subject to some form of regulation (or else, by definition, theycould not be a domain since they would not hold a recognisable shape).Even regimes which apparently exhibit a pure basis of regulation may havethe dominant form tempered by another. For example, much hierarchicalregulatory enforcement is tempered by more co-operative relationshipsmore characteristic of community, and where there is a proliferation of hierarchical regulators in a particular domain (telecommunications andcompetition authorities in the communications domain for example),77

then hierarchy may be tempered by a form of institutional competition asregulators jockey for position and custom.

Among the widely observed hybrid forms are competition law and co-regulation and enforced self-regulation. Though competition law is oftenequated with competition in its control dimensions, competition law exem-plifies hierarchical control, with elements of competition possible wherethird party actions are widely deployed. Co-regulation and enforced self-regulation each link some of the strengths of community-based control(notably within self-regulatory regimes) with the use of hierarchy, for exam-ple by state approval of standards set by industry groups (co-regulation) ormandating firms to establish and sometimes enforce their own standards(enforced self-regulation). Other less prevalent forms are observable but donot have widely accepted labels. Thus mandatory design features (for exam-ple in product design) are hierarchy/design hybrids which we could refer toas ‘enforced design’. The form taken by some self-regulatory efforts toinhibit access to undesirable websites is a community/design hybrid.

One further set of remarks is necessary concerning the bases of control.Different forms of control work differently in different contexts. Markets,hierarchies, communities and design are each embedded in wider socialpractices.78 Key social networks may be a factor in explaining relations of


77 C Scott, ‘Institutional Competition and Coordination in the Process of TelecommunicationsLiberalization’ in J McCahery, B Bratton, S Picciotto and C Scott (eds), InternationalRegulatory Competition and Coordination (Oxford, Oxford University Press, 1996).78 JR Hollingsworth and R Boyer (eds), Contemporary Capitalism: The Embeddedness ofInstitutions (Cambridge, Cambridge University Press, 1997).

interdependence and thus how power is played out in particular social settings.79 Similarly the effects of controls may vary depending on how theyare perceived in the cognition of those whom they affect. Thus some indi-viduals or societies may respond with resistance to controls which are metwith compliance by others or at other times. Thus an analysis of modalitiesof regulation does not, by itself, provide a toolkit for decisions on thedesign of controls, but rather a more limited analytical understanding ofcontrols which have been observed and might be deployed in certain envi-ronments and which might be expected to be effective under appropriateconditions.80

3. PUTTING CONTROLS TO WORK

The importance of the reconfiguring and development of the modalities ofregulation argument further extends to institutional choices for seeking touse controls for public policy objectives. Whereas Lessig places greateremphasis on top-down institutional approaches, of which regulatoryagency forms represent the leading example, we contend that an emphasison hybrid forms of control will tend to lead to the deployment of hierarchi-cal controls as instruments to steer organic or bottom up developments,whether in the form of competition, community or design-based control. Insome instances successful regimes have combined three or even all four ofthe bases for regulation.

A. Hierarchy/Community

Hierarchy and community-based controls are often combined either toensure that industries effectively collaborate on controlling their sector orto give sectoral self-regulation greater authority. The hierarchy/communityhybrid bases of regulation are exemplified by the structures established toaddress scarcity in domain names. By regulating the domain name systemICANN plays a key role in the regulation of Cyberspace.81 ICANN and itspredecessors, IANA and Network Solutions Inc, have long provided regula-tory control over the domain name system not as a function of hierarchicalcontrol, but rather to assist in the development of the domain name


79 R Rhodes, Understanding Governance (Buckingham, Open University Press, 1997) esp ch 3. See also P Drahos and J Braithwaite, Global Business Regulation (Oxford, OxfordUniversity Press, 2000) (discussion of ‘regulatory webs’, ch 23); C Scott, ‘Analysing RegulatorySpace: Fragmented Resources and Institutional Design’ (2001) Public Law 329.80 We are grateful to Julia Black for this point.81 As discussed above ICANN controls the allocation of a scarce resource and therefore playsan important regulatory role.

system as required by the community and to ensure the system designremained intact.

A simple example of the deployment of hierarchical controls to assistin the development of community based controls may be seen in thepromulgation by both Network Solutions and ICANN of Domain-NameDispute-Resolution Policies.82 These procedures are used to counteractthe primary problem of misappropriation of scarce resources. The proce-dure appears to have been extremely successful in countering the problemof ‘cybersquatting’. The practice of cybersquatting was recognised at anearly stage of development of the Web. In its simplest form it is the abilityof unscrupulous individuals to register valuable domains such asDisney.com and then to offer them on at a profit to the rightful holder ofthe trade mark in question. Individuals who entered into such practiceswere quickly dubbed ‘cybersquatters’ by the Web community, a reflectionof their standing within the community as equivalent to persons whounlawfully misappropriate physical property in the real world. Communityopinion was brought to bear. These people were acting antisocially butsocial sanctions failed to affect their actions; being ostracised inCyberspace did not affect their everyday lives. Their actions were, though,more than socially unacceptable; they were also a threat to the developingarchitecture of the domain name system. By controlling domain nameswhich reflected well-known identifiers from the real world they posed athreat to the system. How could people navigate the Web if they couldnot rely on the knowledge they had developed in the physical world?83

Although courts could intervene in cases where cybersquatters had misappropriated another’s trade mark,84 regulatory arbitrage meantenforcement of orders could sometimes prove problematic.

What was required was a regulatory regime which would apply to allregistrations and could be applied whatever the jurisdiction of the parties.This led directly to the first Network Solutions Inc Domain-NameDispute-Resolution Policy, a policy which has now been adopted andrefined by ICANN. The policy has proven successful as it treats the domainname space as a separate jurisdiction, thus preventing regulatory arbitrage.Anyone who resides in the ICANN domain name space must contractuallyagree to be bound by the policy and must agree to the arbitration proce-dure contained therein. Consequently, the values of the Cybercommunitymay be upheld by ICANN through the arbitration process. Secondly, theICANN policy of using low-cost on-line arbitration at the expense of court


82 Kleinwächter n 4 above, 271–72.83 Eg if cybersquatters controlled domains such as disney.com, mcdonalds.com andmicrosoft.com, how would users navigate their way to the sites of these well-known companies? 84 See eg Panavision v Toeppen 945 F Supp 1296 (1996); British Telecommunications plc andOthers v One in a Million Ltd [1999] RPC 1.

proceedings meets the needs of the community. One of the key problemswith usable domain names was that they were unusually an inexpensivescarce commodity. Scarce commodities often carry a proportionately highprice tag, as demonstrated by the UK and German 3G mobile spectrumlicence auctions.85 This is a simple application of the economic model ofdemand, supply and equilibrium pricing. Domain names though do not fitthe economic model particularly well as the market as a whole is oversup-plied while a small percentage of that market is undersupplied or scarce. Asregistrars cannot differentiate useful (and therefore scarce) domain namesfrom the majority, it means market-based controls may be circumventedand a scarce and therefore valuable domain name may be had for as little as$25. This allows for a high degree of speculation in domain names.

The previous Network Solutions Domain-Name Dispute-ResolutionPolicy required the complainer to obtain a court order. This meant it was inmany cases cheaper to buy the disputed domain name from the defenderthan to pursue an action to recover the name, especially if the dispute hadan international element. The present ICANN Uniform Domain-NameDispute-Resolution Policy, through its use of inexpensive arbitration proce-dures, provides a regulatory process which takes account of market condi-tions. This is not to say that the policy is not without its critics. There isstrong criticism of the ICANN policy on the grounds that it now favourstrade mark holders over domain name holders who fail, for whatever rea-son, to comply with US trademark law.86 This has led to a practice knownas ‘Reverse Domain Name Hijacking’ occurring.87 This is a potential flawin the ICANN policy. As discussed, the policy was originally introduced todeal with cybersquatters who were perceived as a socially unacceptable anda potential threat to continued utility of the architecture of the domainname system. The policy now needs to develop to provide a more balancedapproach between the competing interests of the parties. Fortunately thereis evidence that the arbiters under the policy may be developing such amature and balanced approach. There were some initial claims that the pol-icy was being used to restrict free speech.88 Recently though, decisions of


85 The UK raised US$35.4 bn by auctioning 5 UMTS spectrum licences, while Germany raised$46.1 bn by auctioning twelve spectrum blocks. In both cases the number of interested biddersexceeded the number of licences available creating a scarcity of resources. This may be con-trasted with the position in the Netherlands where the auctioning of five licences was met withfive serious bidders and raised only $2.5 bn or in Italy where a similar situation to theNetherlands saw the Italian Government raise only $10 bn. 86 See eg Froomkin above n 4, 96–101; C Perry, ‘Trademarks as Commodities: The FamousRoadblock to Applying Trademark Dilution Law in Cyberspace’ (2000) 32 Connecticut LawReview 1127, 1155–57.87Examples involving American Express and QVC may be found at �http://www.ejacking.com/�(visited 4 January 2002).88 These claims are based in the so-called ‘sucks’ cases. Domains such as directlinesucks.com(D2000-0583) and freeservesucks.com (D2000-0585) were transferred to the trade marks holdersfollowing arbitration. Claims followed that decisions such as these were restricting free speech.

the arbitration panels have shown the policy has a degree of flexibilitywhich may allow them to develop the policy to meet the demands of thecommunity at large.89 Clearly the regulatory authority was implementing ahierarchical control system to support the development of community-based and design-based controls.

B. Hierarchy/Competition

The combination of hierarchical with competition based controls is wellestablished in the media and communications sectors. Thus regimes whichapply economic or content controls more intensely to some firms than toothers effectively create a continuum within which firms exerting dominanceare often located closer to the hierarchy end, while smaller and/or less pow-erful firms are located towards the market end. Within the ‘responsive regulation’ theory this approach is labelled ‘partial industry regulation’.90

The logic of the approach is that the benefits sought for regulation may besecured less intrusively by applying regulation only to a proportion of thefirms, whilst creating space for other firms to be controlled more by marketelements. Typical patterns of more intense regulation of broadcasting overprint media are said to have reduced risks of censorship and promoted pluralism.91 In the telecommunications sector ‘asymmetric regulation’ hasbeen deployed to provide tighter controls over dominant incumbents both tomaintain service levels and to promote access to the market by newentrants.92 With the new media other forms of control which mix hierarchyand competition have been developed.

With the scarcity issue related to spectrum, conventional hierarchicalcontrols have been displaced by a hierarchy/competition hybrid in somedomains. In relation to 3G mobile governments have attempted to use spec-trum allocation mechanisms to promote competitive markets, to promoteefficient allocation of resources and in some cases to secure fee-incomewindfalls for finance ministries. Attempting to set policies that were


89 Three recent decisions wallmartcanadasucks.com (D2000-1104), lockheedmartinsucks.com(D2000-1015) and michaelbloombergsucks.com (FA0097077) have all found in favour of therespondent. These cases may signal a new approach in relation to such free speech cases.90 I Ayres and J Braithwaite, Responsive Regulation (Oxford, Oxford University Press, 1992) ch 5.91 L Bollinger, ‘Freedom of the Press and Public Access: Towards a Theory of PartialRegulation of the Mass Media’ (1976) 75 Michigan Law Review 1.92 A Perucci and M Cimatoribus, ‘Competition, Convergence and Asymmetry in Telecomm-unications Regulation’ (1997) 21 Telecommunications Policy 493. Partial industry regulationin telecommunications is exemplified by US rules which apply greater restrictions to the com-mercial packaging of digital subscriber lines (DSL) provided by telecommunications compa-nies than apply to functionally equivalent cable modems provided by cable communications(formerly TV) companies: M Lemley and L Lessig, ‘The End of End-to-End: Preserving theArchitecture of the Internet in the Broadband Era’ (2001) 48 UCLA Law Review 925.

friendly to the development of advanced infrastructure, the EuropeanCommission initially recommended that Member States should allocatelicences to 3G mobile operators free of charge.93 Only Finland and Sweden,among the first movers on Universal Mobile Telecommunications System(UMTS) licensing, followed this policy course. All the other Member Statesdecided to charge for the licences. Cynical accounts claim that the decisionto charge was premised upon the greed of finance ministries. But there is amore principled explanation for the policy which is posited as a solution toone of the key problems of scarcity—that governments may fail to allocatescarce resources to those who are best able to exploit them to the generalbenefit.

The conventional instrument for the allocation of scarce spectrum is theexercise of government’s hierarchical authority to examine potential appli-cants and make a decision along the lines of a ‘beauty contest’.94 Thismethod was used in eight of the Member States.95 The weakness of thismethod is said to lie in its dependence on the knowledge and judgement ofthe applicable state bureaucracy both to guess the appropriate fee to chargesuccessful applicants and which applicants are best placed to exploit thespectrum. This ‘limited knowledge’ problem is perhaps more acute in the3G mobile sector where there is little consensus on the commercialprospects for services which are made possible in the digital environmentbut which have not yet been tested in the marketplace.

The alternative method for allocating spectrum used in the remainder ofthe member states was to auction the licences, combining hierarchy withcompetition as the basis of control. Deviating from the sealed bid methodused in previous spectrum auctions, the UK government and others decidedto use a transparent (ie no sealed bids) simultaneous multi-round ascendingauction under which bidders’ offers would be revealed at the end of eachround and whoever held the highest bid when the number of bidders wasreduced to equal the number of licences would win the particular licence. Inthis way the price mechanism is used to determine which firms should haveaccess to the scarce resource controlled by government. The outcome of theUK auction was that payments for licences totalling 22 billion pounds weremuch higher than was expected by commentators and government.96


93 European Commission, Communication from the Commission to the Council, the EuropeanParliament, the Economic and Social Committee and the Committee of the Regions: Strategyand Policy Orientations with Regard to the Further Development of Mobile and WirelessCommunications (UMTS) COM (97) 513 Final.94 The theoretical basis of the shift towards auctions, which lies in game theoretic approaches,is described in D Salant, ‘Auctions and Regulation: Reengineering of Regulatory Mechanisms’(2000) 17 Journal of Regulatory Economics 195.95 Curwen above n 36, 461.96 M Cave and T Valletti, ‘Are Spectrum Auctions Ruining Our Grandchildren’s Future?’(2000) 2 Info 347.

Details of auction rules and incentives resulted in less successful outcomesin some other Member States.97 The UK experience initially suggested thatthe auction had been successful in revealing a true value of the licences wellabove government estimates. Commentators still do not agree on whetherthe high cost of licences, particularly in the UK and Germany, will stifle themarket as operators struggle to repay the cost.98 The German regulator hasalready indicated that it may allow the operators to share infrastructurecosts and the same thing may happen in the UK.99 This divergence betweenthe actual operating conditions (and reduction in costs) over those pro-jected at the time of the auctions suggests that the injection of competitionin the licence allocation process has generally been less than successful.

With the problem of regulatory arbitrage the solutions are often put interms of regulatory competition or co-ordination. In other words arbitragemay be overcome by providing co-ordinated or harmonised rules acrossjurisdictions, or arbitrage itself may be seen as a solution to the problem ofexcessive regulation. Regulatory harmonisation was for a long time thefavoured way of providing a level playing field for competition in the inter-nal market of the EU. However, this exercise of hierarchical authorityraises practical difficulties in terms of the scale of resources necessary toachieve it and is said to risk stultifying the very markets which are to beliberalised. A partial response to the practical problems of harmonisationwas the decision of the European Court of Justice in the Cassis de Dijoncase which gave judicial authority to a principle of mutual recognition.100

Regulatory competition is said to provide the flexibility for jurisdictions todevelop standards to match the local requirements (whether technical orpolitical) and the capacity to innovate in regulation while encouragingstates to adopt rules of minimum necessary burden on business or others(because of the threat that such regulatory clients might shift their businesselsewhere). A recent analysis suggests that the choice between competitionand co-ordination is a false one both in practice and normatively, and thatwhat we are likely to see is elements of competition (for example betweeninstitutions) emerging in domains that are notionally co-ordinated and vice-versa. Thus it is better to talk of ‘regulatory co-opetition’, a hierarchy/community hybrid form of control, both as description of thephenomena and as normative aspiration.101


97 Curwen, above n 36, 474–75.98 Cave and Valletti above n 96; J Bauer, ‘Spectrum Auctions, Pricing and Network Expansion

in Wireless Telecommunications’ paper presented to 29th Research Conference onCommunication, Information and Internet Policy, October 2001, Alexandria, Virginia avail-able at �http://www.arxiv.org/ftp/cs/papers/0109/0109108.pdf� (visited 19 December 2001).

99 ‘MMO2 und T-Mobile schliesen UMTS-Kooperationsvertrag’ Frankfurter AllgemeineZeitung, 22 September 2001. 100 Rewe-Zentral AG v Bundemonopolverwaltung für Branntwein [1979] ECR 649.101 D Esty and D Geradin, ‘Regulatory Co-Opetition’ in Esty and Geradin (eds), RegulatoryCompetition and Economic Integration (Oxford, Oxford University Press, 2001).

Regulatory arbitrage is a well-recognised phenomenon of Cyberspace,though commentators reach different conclusions as to its significance.102

Cyberlibertarians argue that regulatory arbitrage prevents hierarchical reg-ulation of Cyberspace. This is most clearly and famously put in DavidJohnson and David Post’s seminal article, Law and Borders—The Rise ofLaw in Cyberspace.103 For Johnson and Post the practical effect of regula-tory arbitrage is that hierarchical controls are rendered impotent. Netizensmay choose to reject hierarchical controls they find unpalatable by movingto another part of Cyberspace. As previously outlined, Netizens maychoose how they wish to be regulated much more freely than citizens ofphysical jurisdictions. The only effective regulatory system according toCyberlibertarian theory is therefore one which is acceptable to all (or thevast majority of) Netizens. Johnson and Post therefore suggest a bottom-up or organic regulatory model. They envisage a self-regulatory gover-nance system along similar lines to that developed to regulate the domainname system. Lessig disagrees with their conclusion. He agrees thatCyberspace is a separate space and can be seen as a distinct jurisdiction,but he disagrees though with the conclusion that it is a jurisdiction whichrequires the organic development of regulatory regimes. For Lessig, onceyou isolate Cyberspace as a distinct space you may use its unique architec-ture to establish a hierarchical regulatory structure. The argument of theCyberpaternalists is therefore that once a recognised regulator emerges inany given activity they may impose regulatory regimes on Netizensthrough the unique man-made architecture of the Web, its code.104

To the extent that regulatory arbitrage is a problem with new media gen-erally, and usage of the Internet in particular, it remains an open question towhat extent the balance between competition and co-ordination might bedeployed to resolve issues. For many commentators, the nature of Internettechnology makes regulatory arbitrage inevitable and difficult to forestall,


102 See eg M Froomkin, ‘The Internet as a Source of Regulatory Arbitrage’ above n 28; D Johnson and D Post, ‘Law and Borders—The Rise of Law in Cyberspace’ (1996) 48Stanford Law Review 1367; L Lessig, ‘Zones in Cyberspace’ (1996) 48 Stanford Law Review1403; P Samuelson, ‘Five Challenges for Regulating the Global Information Society’ in C Marsden (ed), Regulating the Global Information Society (London, Routledge, 2001). Aninteresting side effect of regulatory arbitrage is a regression to the least interventionist stan-dard in a given area. This can most clearly be seen in relation to freedom of speech followingthe decision of the US Supreme Court in ACLU v Reno 177 S Ct 2329 (1997) where an on-line movement towards the US free speech standard may be detected. For further discus-sion on this see D Vick, ‘Exporting the First Amendment to Cyberspace: The Internet andState Sovereignty’ in N Morris and S Waisbord (eds), Media and Globalisation: Why the StateMatters (Lanham, MD, Rowman & Littlefield, 2001).103 Ibid. On-line version available at �http://www.temple.edu/lawschool/dpost/Borders.html�(visited 4 January 2002). 104 Lessig above n 2, passim. See also L Eko, ‘Many Spiders, One Worldwide Web: Towards aTypology of Internet Regulation’ (2001) 6 Communications Law and Policy 445.

whatever may be desirable from a policy point of view. It is the high mobilityboth of providers and users within Cyberspace which makes it difficult toenvisage co-ordinative solutions. For some this is a strength militatingagainst excessive control of Cyberspace. It was argued that regulatory arbi-trage acted as a (limited) check on stringent UK legislation governing statemonitoring of electronic communications generally in the Regulation ofInvestigatory Powers Act 2000.105 Arguably any solutions here are likely tobe a product of co-operation and community-based controls involving bothgovernments and businesses rather than of co-ordination between govern-ments, as through the EU or the World Trade Organisation (WTO).

C. Hierarchy, Competition and Design

In addition to the use of hierarchical/community controls discussed earlier,ICANN is also applying a design/competition-based hybrid in an attemptto alleviate the pressure on the domain name system. As domain names area man-made rather than natural phenomenon, they do not have to berationed in the manner of natural resources such as bandwidth. Whereasgovernments cannot simply create additional bandwidth to meet the demand of mobile phone operators,106 ICANN hopes to solve thedomain name problem by creating additional resources. To this end on 16 November 2000 ICANN announced seven new top-level domains.107 Itis the hope of ICANN that by creating competition in new, more specialiseddomains, demand will be lowered in the oversubscribed .com domain and asolution will be found to the scarcity problem. There has been profounddisquiet about allegedly anticompetitive outcomes from ICANN’s alloca-tion of new top-level domain (TLD) names. Thus the allocation of some ofthese new resources (notably .pro and .info) has been made to organisa-tions already controlling other key TLDs such as .com, .net and .org. Therefusal to create other new top-level domain names (for example .xxx forpornography) has been criticised for inhibiting design-based controls overaccess or exploitation of particular sites. The solution to these problems


105 The House of Lords debated this possibility at some length at the Report Stage of the Bill(HL Deb vol 615 cols 381–88; cols 400–452, 13 July 2000). See in particular the debate onAmendment No 64 at col 408–18.106 Lessig offers an alternative solution to the problem of undersupply of bandwidth. Althoughacknowledging supply of radiocommunications bandwidth is naturally limited, he suggests weare extremely wasteful of the resource available. He rejects the use of beauty contests or auctions to propertise bandwidth as outlined above and suggests instead a design solutionallowing a more efficient use of bandwidth as a free or common resource. Lessig above n 1, ch 12.107 They are .aero, .biz, .coop, .info, .museum, .name and .pro. For further details on thesenames including who may apply for a name within these new domains see �http://www.icann.org/tlds/� (visited 4 January 2002).

posited by one key critique is to open the domain name market to greatercompetition between assignment organisations and use competition as akey form of control.108 It may, though, already be too late for any competition-based approach to work in relation to domain names. Thescarcity of resource problem in relation to domain names appears to berestricted to the .com TLD. As discussed previously, there are already alarge number of alternative top-level domains available.

Attempts previously to turn country code TLDs into generic TLDs havenot released useful domain names. The most concerted effort has been inrelation to the .ws (Western Samoa) domain, which is promoted as a ‘WorldSite’ domain. In many instances holders of current generic TLDs simplyreplicated their registration in the new domain. There is little evidence thatthe creation of manufactured additional resources deals with this particularscarce resources problem. The availability of these alternatives has notencouraged sufficient competition to effect the base of the .com domain.The relevant market appears therefore not to be the market in TLDs as awhole, or even generic TLDs, but is restricted to the .com TLD. The .comTLD is, it appears, too well established to be affected by the creation ofalternative domains. The creation of such alternatives does not appear tointroduce competition within the relevant market; instead it merely createsalternative markets in which mere replication of registration occurs.

The only possible methodology which would appear to provide a func-tioning competition-based solution to the .com problem would be toincrease the marketability of competing TLDs. The current ICANN policyis for the creation of alternative TLDs which they expect will increase inmarketability through the efforts of the registrars who deal in such names.They are relying upon a free market rhetoric which states that those withsaleable assets will work to increase the marketability of their asset throughadvertising and marketing. ICANN believes that the domain name systemis thus a free market in which demand may be created in new productsthrough advertising and marketing. Unfortunately the free market rhetoricdoes not apply to domain names in this manner. They are more than simplysaleable assets. Firstly, valuable domain names are, in many cases, a reflec-tion of currently held trade marks. As has been previously alluded to, thecreation of other TLDs fails to release alternative resources due to replica-tion of registrations by current holders of trade marks and valuable .comdomain names to prevent any risk of cybersquatting. Secondly, all domain


108 M Mueller, ‘Domains Without Frontiers’ (2001) Info 97, 99. See also M Froomkin, ‘IsICANN’s New Generation of Internet Domain Name Selection Process ThwartingCompetition?’ presentation to US House of Representatives Committee on Energy &Commerce, Subcommittee on Telecommunications, 8 February 2001. Available on-line at�http://personal.law.miami.edu/~froomkin/articles/commerce8Jan2001.htm� (visited 4 January 2002).

names are a method of indexing information and navigation. Consequently,they are streetnames not just marketable assets. As with all other communi-ties the Web has its desirable areas and its undesirable areas: in this virtualcommunity .com is the business and financial district: it is the Web’s equiv-alent of the City of London, Wall Street or Rodeo Drive. Just as businessesin the real world will pay a premium for such addresses, so the focal pointfor competition in relation to domain names will remain in the .comdomain. Due to these problems, the scarcity issue in domain names may beas ingrained as the bandwidth problem in relation to telecommunicationsand a more radical solution may be required in the future.

D. Other Forms of Control

The emphasis of current thinking on alternatives to hierarchical control islargely focused on linking hierarchy to competition or to community-basedmethods of control. This focus largely excludes two major classes of formsof control, one defined in terms of excluding hierarchy and the otherdefined in terms of including design.

(1) Design-Based Regulation

A key example which is located in both sets (employing design and exclud-ing hierarchy) is the use of regional management codes by DVD producersand equipment manufacturers. Producers and equipment manufacturershave collaborated in a regional coding system which allows for market segmentation within the DVD industry. Regional coding was developed topermit studios to control the home release of movies within different geo-graphical regions allowing the staggering of cinematic releases.109 The stu-dios required that DVD software codes included a simple code that couldbe used to prevent playback of certain discs in certain geographical regions.The equipment manufacturers assisted by producing region specific DVDplayers, each player being given a code for the region in which it is sold.The player will refuse to play discs that are not encoded for that region.


109 The coding also allows studios to sell exclusive distribution rights to a variety of foreigndistributors. See D Marks and B Turnbull, ‘Technical Protection Measures: The Intersection ofTechnology, Law and Commercial Licences’ paper presented to World Intellectual PropertyOrganisation Workshop on Implementation Issues of the WIPO Copyright Treaty (WCT) andthe WIPO Performances and Phonograms Treaty (WPPT), 6–7 December 1999, Geneva avail-able at �http://www.wipo.org/eng/meetings/1999/wct_wppt/pdf/imp99_3.pdf� (visited 7January 2002). The European Commission is currently investigating whether DVD producersare using the technology to partition the market illegally. See European Commission,Directorate-General for Competition, 31st Report on Competition Policy (Office for OfficialPublications of the European Communities, Luxembourg, 2002) para 5.2.2.

This means that discs bought in one country may not play on playersbought in another country. The addition of regional management codes areentirely optional for the maker of a disc: discs without codes will play on anyplayer in any country. These codes should not be confused with the DVDContent Scramble System, discussed below, which acts as a copy-controlmeasure. Regional management codes are not an encryption system; theyare merely one byte of information on the disc, which denotes one of eightdifferent DVD regions.110 Thus an encoded DVD bought in the US will notbe viewable on a European DVD player. There is no hierarchical element tothis. Customers are not prevented by contract or any other laws from buyingDVDs in other countries. The control is effected by features of the diverseproduct standards which make a DVD useless when paired with a playerwith a different coding.

(2) Including Design

A related example is the use of a hierarchy/design hybrid in an attemptto manage the high levels of digital piracy which occur on the Web. Copy-control devices have been employed by almost all copyright holders whotrade in digital media. These controls have met with varied degrees of suc-cess, but are supported by not only industry groups such as the MotionPicture Association of America (MPAA) and the Recording IndustryAssociation of America (RIAA), but also have been given the force of lawthrough the actions of the World Intellectual Property Organisation(WIPO)111 as enacted within the European Union by the Directive onCertain Aspects of Copyright and Related Rights in the InformationSociety,112 and in the United States through the Digital MillenniumCopyright Act 1998.113 With the legal support offered by these enactments,several copy-control systems have been developed and implemented by bodiesrepresenting copyright holders mostly against the wishes of the communityat large. One such standard developed by the MPAA for use on DVDreleases is the Content Scramble System (CSS). CSS was developed by twohardware companies, Matsushita Electric and Toshiba, for the motion


110 These are as follows: Region 1 USA and Canada; Region 2 Japan, Europe, South Africaand Middle East; Region 3 Southeast and East Asia; Region 4 Australasia, Central and SouthAmerica and Caribbean; Region 5 Eastern Europe, India and Africa; Region 6 China; Region7 Reserved and currently unused; Region 8 Special Venues (Planes, Cruise Ships etc).111 Art 11 WIPO Copyright Treaty requires contracting parties to ‘provide adequate legal pro-tection and effective legal remedies against the circumvention of effective technological meas-ures that are used by authors in connection with their exercise of rights under this Treaty orthe Berne Convention and that restrict acts, in respect of their works, which are not authorisedby the authors concerned or permitted by law’.112 Dir 2001/29/EC, Art 6.113 §§ 1201(a)(1) and 1201 (a)(2).

picture industry and was adopted as industry standard in 1996. The systeminvolves a dual key encryption system which encrypts all sound and graphicfiles contained on a DVD release. The files may be decrypted by the appro-priate decryption algorithm which is made up of a series of keys stored onboth the DVD and the DVD player. This means that only players and discscontaining the appropriate keys may decrypt the necessary files and playthe movies stored on the DVDs.114 The CSS system did not directly preventdirect copying of DVD discs; the contents of a DVD (while encrypted) couldbe copied directly from one DVD to another. CSS did however prevent theuploading of the contents of a DVD onto hard disc or a web server. Theconcern of some users was that CSS systems were only licensed for use onMacintosh and Windows based operating systems (and for dedicated DVDplayers). Users of open source operating systems such as GNU/Linux couldnot play a CSS encoded DVD on their system. This led to a campaign ofcivil disobedience leading to the development of a decryption code for CSSwhich would allow the playing of CSS encrypted DVDs on any platform.The CSS code was a quite weak 40 bit encryption system and in September1999 it was successfully hacked independently by an anonymous Germanhacker and a member of the ‘Drink or Die’ cracking community.115 Thisdevelopment meant that CSS encrypted DVDs could now be used on unli-censed DVD players and that DVD material could be placed directly ontothe Web. Such a development was an obvious threat to the continued use ofCSS by DVD producers. Action was taken immediately in Norway whereJon Johansen who had been erroneously identified as the author of DeCSSwas prosecuted and in the United States where Universal Studios success-fully obtained injunctions under the Digital Millennium Copyright Actagainst several individuals who were distributing the DeCSS code from US-based websites.116 The decision in this case has been extensively criticisedby many commentators, including Lessig who argues that ‘DeCSS didn’tincrease the likelihood of piracy. All DeCSS did was (1) reveal how bad an


114 For more detail on CSS see the opinion of Judge Kaplan in Universal Studios Inc vReimerdes et al 111 F Supp 2d 294 (2000). Affirmed Universal Studios Inc v Corley et al2001 WL 1505495 (2d Cir 2001) Available at �http://eon.law.harvard.edu/openlaw/DVD/NY/appeals/opinion.pdf� (visited 4 January 2002). 115 The media wrongly attributed the development of DeCSS to a 15 year-old Norwegian JonJohansen. Although Mr Johansen was a member of the ‘Masters of Reverse Engineering’ com-munity which released DeCSS he was not the author of the program. This is made clear in atext file which accompanied the release of the program. The text file is available at�http://www.lemuria.org/DeCSS/dvdtruth.txt� (visited 4 January 2002). On 9 January 2002the Norwegian Economic Crime Unit (ØKOKRIM) charged Jon Johansen with violation ofNorwegian Criminal Code s 145(2), which outlaws breaking into another person’s lockedproperty to gain access to data that one is not entitled to access. A campaign to drop allcharges was immediately begun by the Electronic Frontier Foundation (see �http://www.eff.org/IP/DeCSS_prosecutions/Johansen_DeCSS_case/�) and at the time of writing was still underway. 116 Universal Studios Inc. v Reimerdes et al above n 114.

existing encryption system was; and (2) enable disks presumptively legallypurchased to be played on Linux (and other) computers.’117 Lessig isextremely critical of the use of law to support these design controls arguingthat they create an ‘imbalance where traditional rights are lost in the nameof perfect control by content holders’.118

This view taken by Lessig may prove to be unduly pessimistic. There is asyet no evidence of content holders attaining the perfect control he fears inCyberspace. Indeed the victory of Universal Studios and the MPAA hasproved to date to be pyrrhic. As is often the case in Cyberspace when hierar-chical/design controls are used to regulate the community at large, the com-munity will rally in an attempt to defeat the regulatory control mechanisms.The DeCSS code may currently be obtained from any one of hundreds ofwebsites which remain out of the reach of the US authorities.119 Currentlythe producers of DVD titles and the hacking community are involved in awar of code. The motion picture industry has updated the CSS code whichmeans the DeCSS code no longer decrypts the latest DVD releases. This hassimply encouraged hackers to produce new, more powerful, second genera-tion decryption codes such as DVD-Decrypter. Both parties continue to bat-tle for the control of DVD encryption/decryption codes. The producers ofDVD titles and the community at large are both using design tools toattempt to protect their position. The producers presently have the advan-tage, due primarily to a weakness of current technology. At the moment thelack of widely available broadband technology prevents distribution ofdecrypted movie data over the Web: the producers hold the upper hand. Asdistribution technology improves, the movie industry may find that theirdesign solutions cannot effectively function without either the support of thecommunity at large or far greater reliance upon the hierarchical control ele-ments introduced by the Digital Millennium Copyright Act and the Directiveon Copyright and Related Rights in the Information Society. Producers ofDVDs will need to decide within the next few years whether they wish torely on a hierarchy/design hybrid or a community/design hybrid.120


117 Lessig above n 1, 189. See further 187–90. 118 Ibid 200.119 The website operated by Shawn Reimerdes (one of the defendants in the MPAA action)contains the following advice:

A Federal Judge removes this link by court order! We are fighting for the right to put thislink back up for you! I am not allowed to have this decryption information anymore, soI will just tell you the obvious: Go to your favorite search engine and enter ‘DeCSS.’ Youwill find one of thousands of websites that has decided to post this information.

Doing so will allow you to locate sites such as the DeCSS mirror site at �http://heavymu-sic.8m.com/� (visited 20 December 2001); Download.com �http://www.download.cnet.com/�(visited 4 January 2002) and �http://www.lemuria.org� (visited 4 January 2002)—all ofwhich currently have the DeCSS program available for downloading. 120 See further above Lessig n 1, ch 11.

E. Excluding Hierarchy

A successful example of a community using design tools to effect a regulatoryscheme is the community-based approach to protecting children inCyberspace. As discussed above, the anonymity problem raises two distinctdangers for minors in Cyberspace. One is that they gain access to materialswhich are unsuitable for minors and the other is that adults take advantageof anonymity to forge improper relationships with minors. Hierarchicalcontrols fail to remedy these problems, but a community-based solutionhas proved extremely successful, especially when linked with design-basedsolutions. Within organised Cybercommunities children may be supervisedby the community. Communities such as AOL encourage family member-ship where parents register the details of the family as a whole and eachindividual member has his/her own password. Unless the child were tocompromise an adult password, their status can therefore be made knownto the community and the community can supervise and protect the childwhile he is online. Children cannot be watched all the time and the commu-nity cannot take over all parenting responsibilities. To assist, additionaldesign-based tools may be used. In addition to the community supervision,parents may employ software solutions such as CYBERsitter and NetNanny. These products allow parents to set acceptable parameters for theirchildren when in Cyberspace.121 Combined, the role of the community andthe security provided by these products appear to provide a relatively suc-cessful solution to the access problem.

4. CONCLUSION

New and unpredictable configurations of power are among the hallmarksof the new media. It is not surprising that the problem of control hasattracted such a high degree of interest among scholars. Not only are thereinteresting problems of designing regimes to provide appropriate con-straints on undesirable activities, there are also challenges in securing themaximum benefit to the community of new technologies such as theInternet and 3G mobile (each of which is said to be subject to ‘networkeffects’ such that the more users there are, the greater the benefit to thecommunity generally). The new media phenomena present scholars with at


121 Such software programs are called alternatively Filterware or Censorware (dependingvery much upon your political viewpoint). Many programs such as the ones listed use stand-alone value judgements to categorise websites based on their content. The software providerwill review sites and put them on either an ‘allowed’ or a ‘not allowed’ list. Other programsrely upon the Platform for Internet Content Selection (PICS)—a standardised industry sys-tem which allows content to be rated in various categories including: topics such as ‘sexualcontent’, ‘race’, and ‘privacy’, under the control of the user.

least two temptations. One is to overstate the novelty of the problems presented, with a consequent tendency to reject ‘old’ forms of control.122

The second is to overstate the extent to which the media themselves ‘hardwire’or constrain the possible means of addressing the problems. Both tendenciesare prevalent in analyses of the control problem as it applies to the Internet.

The alternative, which we have argued for, is to locate problems of controlling the new media squarely within well-established analyses ofproblems of regulatory control. Such analysis encourages us to look at themechanisms of control which already subsist within the target system andto find ways to stimulate or steer those indigenous mechanisms towardsmeeting the public interest objectives of regulation. Thus a central role forhierarchy is to steer systems which involve other forms of control based incommunity, market or design (or combination thereof). This does notexclude the possibility that effective control may occur through competi-tion, design or community, together or separately, without hierarchicalinvolvement.

A key challenge presented by such novel governance mechanisms is howto deploy them in such a way that are perceived as legitimate. The legiti-macy of democratic government is linked to processes of representationand open decision making. Though other governance mechanisms may belegitimated in similar ways, in many cases it will either be alternativeprocess elements and/or outcomes which are more important in generatinglegitimacy. Judgements on the appropriate balance between democratic andother forms of legitimation are likely to differ within different political cultures. This is evidenced in markedly different responses in Europe andthe United States to the creation of ICANN. For some it represents an unac-ceptable delegation of government authority to a private body.123 For others,it is an efficient technical solution to a pressing problem, even if its decisionmaking is not wholly technical. A key challenge in deploying ideas aboutthe mixture of control forms advanced in this discussion is to balance thesetwin concerns about efficiency and legitimacy. The conditions for achievingan acceptable balance are likely to vary in different places and differenttimes.


122 A useful early consideration of the ‘newness’ issue in Cyberspace is I Trotter Hardy, ‘TheProper Legal Regime for “Cyberspace” ’ (1994) 55 University of Pittsburgh Law Review993. More recently see M Price, ‘The Newness of New Technology’ (2001) 22 CardozoLaw Review 1885. 123 Froomkin, ‘Wrong Turn in Cyberspace’ n 4 above; J Wienberg, ‘Geeks and Greeks’ (2001)3 Info 313; cf R Marlin Bennett, ‘Icann and Democracy: Contradictions and Possibilities’(2001) 3 Info 299.

7

Regulating E-Commerce in the WTO: Exploring the

Classification Issue

FIONA SMITH*

IN 1998 THE global value of goods and services traded via theInternet1 was $3 billion, with predicted growth anywhere betweenUS$100 billion2 and US$300 billion3 by 2000.4 Despite the notorious

dotcom collapses, estimates show that worldwide online trade exceededUS$2000 billion in 2002 with predicted increases in excess of US$12,800billion by 2006:5 the European Union alone is expected to experience on-line trade rising from e77 billion in 2001 to e2.2 trillion by 2006.6

Such a shift to on-line trading may challenge existing regulatory systemswhich are built on the assumption that physical evidence of a contractexists. Consequently, it may be necessary to search for alternative solutionswhen regulating commerce that are not reliant on traditional legal mecha-nisms. This shift potentially challenges existing regulatory systems because

*Dr Fiona Smith is a lecturer in law at the University of Leicester, UK, specialising in interna-tional economic law, particularly as it relates to the World Trade Organisation.<[email protected]>.1 E-commerce can encompass all transactions using telecommunications networks, but thephenomenon is more usually associated with the Internet: WTO, Electronic Commerce andthe Role of the WTO, Special Studies No 2 (Geneva, WTO Publications, 1998) 1.2 WTO, Global Electronic Commerce, WT/GC/W/86, 23 April 1998, para 2.3 WTO, Electronic Commerce and the Role of the WTO above n 1, 1. 4 Note the problems of obtaining accurate statistics: UNCTAD, Electronic Commerce andDevelopment Report 2001, UNCTAD/SDTE/ECB/1 (New York and Geneva, UnitedNations) ch 1.5 ‘E-Commerce Next Wave: Productivity and Innovation’ presentation given by Mrs JulieMeringer, Group Director, European Research and Managing Director UK Forrester Researchat WTO Seminar on E-Commerce, ‘Geneva’ 22 April 2002. 6 R Greenspan, ‘EU B2B Expected to Explode’, 28 August 2002. <http://cyberatlas.internet.com/markets/b2b/> (last visited 30 October 2002).

the participants in e-commerce need means to facilitate, legitimate andenforce their transactions. Such challenges operate on two levels: first, thosethat focus on the nature of the e-commerce transaction itself. In this case,the emphasis centres on the extent to which the transaction is legitimate inlaw, which inevitably leads to issues of its enforceability.7 For example,when is a contract entered into through electronic means concluded, whatare its terms8 and which jurisdiction governs its enforcement?9 In addition,procedural issues occur particularly in relation to the extent to which elec-tronic signatures on contractual documents are acceptable to legitimate anytransaction.10

Second, concerns arise over facilitating the transaction, as opposed toensuring its legitimacy directly. This operates in three ways: in the first case,the emphasis is on ensuring that participants in e-commerce are able toobtain relevant technology, including appropriate computer hardware andsoftware coupled with access to global telecommunication networks, whichallows them to enter into electronic transactions. In addition, participantsmust then be able to import and export products11 that are the subject ofthose transactions12 without unnecessarily restrictive border controls.Finally, once those products are within the jurisdiction, domestic regulationmust not be so restrictive as to ensure that importation is prohibitivelyexpensive. Consequently, regulation at an international level may providethe answer to this dilemma.

A common theme running through the regulation of trade via e-commerceis the international character of each element: participants can be locatedin different countries, products are exported across jurisdictions and thetelecommunications networks required to enter into the transactions arenecessarily worldwide. Some commentators13 have therefore argued that aglobal regime that can transcend jurisdictional boundaries is the most effectiveway of regulating e-commerce.

160 Fiona Smith

7 A Terrett and I Monaghan, ‘The Internet—An Introduction for Lawyers’ in L Edwards andC Waelde (eds), Law and the Internet: A Framework for Electronic Commerce (Oxford, HartPublishing, 2000) 1, 8.

8 AD Murray, ‘Entering Into Contracts Electronically: The Real WWW’ in Edwards andWaelde (eds) above n 7. 17, 18.

9 C Heaven, ‘A Proposal for Removing Road Blocks from the Information Superhighway ByUsing An Integrated International Approach to Internet Jurisdiction’ (2001) 10 MinnesotaJournal of Global Trade 373, 379.10 See UNCITRAL’s important work in this area: UNCITRAL, Model Law on ElectronicSignatures 2001, (New York, United Nations, 2002); also M Hogg, ‘Secrecy and Signatures—Turning the Legal Spotlight on Encryption and Electronic Signatures’ in Edwards and Waelde(eds) above n 7, 37.11 This is deemed to cover goods and services at this time: for the problems relating to this, seesection 2 below.12 This also covers those products imported in digital form.13 Eg Heaven above n 9, 379; C Reed, Internet Law: Text and Materials (London,Butterworths, 2000) 2.

Devising a completely new regime is complex14 and perhaps unnecessaryif Reed’s assertion that commerce based on Internet technology raises nonew regulatory issues is accepted.15 Consequently, relying on existing inter-national regulatory structures with proven dispute settlement systems couldaddress the challenges raised by e-commerce’s international character.

The World Trade Organisation (WTO) has a comprehensive and reputable dispute settlement system16 that governs its complex rules oninternational trade in goods, services and intellectual property. This makesthe WTO a relevant starting point as a potential governor of e-commerce.Indeed, the WTO itself has already recognised that its rules inevitablyimpact on e-commerce17 and instigated a work programme in 1998 toinvestigate their effects fully.18 A working party on e-commerce also wasestablished which reports progress to the WTO General Council.19 Theextent to which the WTO can regulate e-commerce has already been thesubject of academic commentaries.20 These provide a valuable overview ofthe ways in which the WTO rules on goods, services and intellectual prop-erty potentially affect e-commerce. However, it is clear from these discus-sions that the WTO is not a panacea for regulating trade via e-commerceand that significant problems exist using its rules in their current form.

This chapter builds on previous academic commentaries by focusing onthe classification of e-commerce as trade in goods and/or trade in servicesand the implications of failing to clarify where that boundary lies. It arguesthat whilst categorising e-commerce in this way is difficult, it is crucial forthe effective application of WTO rules because the rule structure adopts arigid goods/services distinction, with members’ obligations effectivelychanging dependent on whether the transaction falls within the GeneralAgreement on Tariffs and Trade (GATT) or the General Agreement on

Regulating E-Commerce in the WTO 161

14 Cf Heaven above n 13.15 See Reed above n 13, 2.16 See generally J Cameron and K Campbell (eds), Dispute Resolution in the WTO (London,Cameron May, 1998).17 WTO, Geneva Ministerial Declaration on Global Electronic Commerce,WT/MIN(98)/DEC/2, 25 May 1998.18 WTO, Work Programme on Electronic Commerce, WT/L/274, 25 September 1998.19 WTO, Work Programme on Electronic Commerce: Information Provided to GeneralCouncil, G/C/W/158, 26 July 1999; following the WTO General Council’s decision of 17 July 2000, the WTO committees on trade in goods, services, intellectual property andtrade and development now report individually on the progress on e-commerce within theirrespective competences: eg WTO, Council for Trade in Goods, Chairman’s Factual Reportto the General Council on the Work Programme on Electronic Commerce, G/L/421, 24 November 2000.20 WJ Drake and K Nicolaidïs, ‘Global Electronic Commerce and GATS: The MillenniumRound and Beyond’ in P Sauvé and RM Stern (eds), GATS 2000: New Directions in ServicesTrade Liberalisation (Harvard University and Washington DC, Center for Business andGovernment and Brookings Institution Press, 2000) 399; and MD Powell, ‘The Role of theWTO in Cyberspace’ (1998) 4 International Trade Law and Regulation 157.

Trade in Services (GATS). This problem is acute as many members have notundertaken full liberalisation commitments in all their goods and services’sectors. GATT and GATS rules do not assist the classification decisionbecause they rely on members first designating items traded as either goodsor services in their schedules consequently triggering the application of therelevant rules. It is argued that the only way to use the existing WTOscheme successfully to regulate e-commerce is by clearly defining what con-stitutes trade in goods and trade in services in the abstract. Such a designa-tion facilitates predictability in a volatile trading environment, as it isclearer which rules apply and to whom. Following the collapse of theCancun Ministerial Meeting in September 2003, it is imperative for its con-tinued legitimacy that the WTO is seen as meeting the needs of all thosewho participate in international trade.

The discussion is divided into three sections: section one gives anoverview of the WTO legal system, particularly concentrating on the rigidrule structure. Section two considers the scope of the WTO’s working defi-nition of e-commerce. Despite appearing comprehensive, it does not addresswho the participants to the e-commerce transaction are and whether tradein goods or trade in services is involved. These two issues are inextricablylinked because GATT and GATS rules impact on participants in differentways. The third section explores the classification issue in more detail, ini-tially looking at what categories products traded via e-commerce could fallinto and then assessing what effect the scheduling methodology in GATTand GATS has on the classification dilemma. Rather than resolving theproblem, it appears that the methodology only serves to blur the boundarybetween goods and services further by pushing the decision back to individ-ual members, who will make it based on domestic concerns, rather thaninternational considerations.

Consequently, the work programme’s failure to recognise the effect ofthe WTO’s regulatory structure on on-line trading and to define clearlywhat constitutes goods and services for the purposes of the rules in thiscontext means that using the WTO in its existing form will not addressall the regulatory problems presented by the changing nature of interna-tional trade.

1. BACKGROUND

Following eight years of protracted multilateral trade negotiations in theUruguay Round,21 the World Trade Organisation (WTO) finally came

162 Fiona Smith

21 On the Uruguay Round see generally TP Stewart (ed), The GATT Uruguay Round: ANegotiating History 1986–1992 (The Hague, Kluwer, 1993).

into being in 1995.22 The Marrakesh Agreement creating the WTO prescribes its scope,23 functions,24 structure,25 relationship with other multilateral agreements26 and includes details on accession27 and decisionmaking.28 The WTO itself is merely an ‘umbrella’29 organisation responsi-ble for the ‘implementation, administration and operation’30 of the multilat-eral agreements which were concluded during the course of the UruguayRound.31 These rules are primarily contained in Annex 1 to the MarrakeshAgreement and cover trade in goods,32 services 33 and intellectualproperty.34

The WTO rules on trade in goods build on the GATT originally negoti-ated in 1947 as part of the Havana Charter to establish the abortiveInternational Trade Organisation.35 GATT 1947 aimed to liberalise inter-national trade by reducing tariffs and other governmental restrictions onthe import and export of goods.36 Emphasis was placed on lowering tariffson goods listed in countries’ schedules,37 but it also contained rules restrict-ing the use of non-tariff measures including subsidies,38 quantitative restrictions39 and safeguards.40

At the core of the GATT 1947 were two non-discrimination provi-sions: the most favoured nation (MFN) clause,41 which prevented coun-tries’ offering more favourable treatment to goods from one GATT


22 Art I Marrakesh Agreement Establishing the World Trade Organisation (the MarrakeshAgreement).23 Art II Marrakesh Agreement.24 Art III ibid.25 Art IV ibid.26 Art V ibid.27 Art XII ibid.28 Art IX ibid.29 JH Jackson, The World Trading System (Cambridge, MIT Press, 1997) 47.30 Art III:1 Marrakesh Agreement.31 This is perhaps overstating the situation as some agreements were already in existencebefore the start of the Round, but they were subsequently modified during the UruguayRound and brought within the WTO scheme: eg the TBT Agreement and the WTOAgreement on Subsidies and Countervailing Measures have their roots in the earlier multilat-eral trade talks in the Tokyo Round: see JH Jackson, WJ Davey and AO Sykes (eds), LegalProblems of International Economic Relations: Cases, Materials and Text (St Paul, MN,West, 1995) 768.32 Annex 1A.33 Annex 1B.34 Annex 1C.35 See generally JH Jackson, World Trade and the Law of GATT (Indianapolis, Bobbs-Merrill,1969).36 See Jackson, Davey and Sykes above n 31, 290.37 Art II GATT.38 Art XVI GATT.39 Art XI GATT.40 Art XIX GATT.41 Art I GATT.

participant over other participants42 and the national treatment clause43

which prohibited discrimination between domestically produced andimported goods once the imported goods were within the participant’sjurisdiction.44 Following the conclusion of the Uruguay Round, GATT1947 and its subsequent amendments were incorporated into the WTOagreements as GATT 1994 and supplemented by a number of otheragreements covering trade in goods negotiated and amended during theUruguay Round.45

Prior to the creation of the WTO, regulation of international trade wasrestricted to goods and not services, so the GATS was a completely newaddition to the multilateral regime in 1994. GATS disciplines are strik-ingly similar to GATT, with emphasis placed on liberalising trade in serv-ices by reducing restrictive trade barriers46 particularly on those servicesdetailed in members’ schedules.47 Like the GATT, GATS contains MFN48

and national treatment obligations,49 but in contrast to GATT, GATS’rules are separated into general and specific commitments: the MFNclause in Article II is included in the general obligations applicable to alltrade in services, but there is limited provision for derogations from amember’s MFN commitments.50 The national treatment51 and marketaccess52 rules are included in Part III GATS on specific commitments somembers only need to comply with these obligations in relation to thoseservice sectors they have specifically agreed to liberalise in their schedulesof commitments. GATS’ general rules are also supplemented by specificagreements to liberalise certain sectors, including financial services, basictelecommunications53 and maritime transport.

164 Fiona Smith

42 There are exceptions to this rule: Art XXIV GATT on free trade areas and customs unionsand also Art XX GATT which contains a list of general exceptions.43 Art III GATT.44 Jackson, Davey and Sykes above n 31, 300.45 Some agreements were revised from other rounds, most notably the WTO Agreement onTechnical Barriers to Trade. See Annex 1A Marrakesh Agreement for the complete list.46 Para 2 Preamble GATS.47 On GATS see generally G Feketekuty, ‘Improving the Architecture of GATS’ in Sauvé andStern above n 20, 85, 90–99. 48 Art II GATS.49 Art XVII GATS.50 See generally, Jackson, Davey and Sykes above n 31, 922; also Jackson, The World TradingSystem, above n 29, 307.51 Art XVII GATS.52 Art XVI GATS.53 Annex on Telecommunications: further negotiations were carried out to liberalise the basictelecommunications sector and were concluded in 1997 and came into effect in 1998.Members’ commitments are separated into the general commitments in the annex itself andthen further specific commitments contained in the Reference Paper on Regulatory Principles(1997) 36 International Legal Materials 367, although members have to specifically accede tothe commitments in the Reference Paper: see M Naftel and LJ Spiwak¸ The Telecoms TradeWar (Oxford, Hart Publishing, 2000) at 102–17; also MG Durantez, ‘WTO/GATS

Consequently, international trade rules are now divided clearly intothree parts covering goods,54 services55 and intellectual property.56 Thismeans that the application of the rules is dependent on the categorisationof the transaction as one relating to goods, services or intellectual prop-erty. Whilst identifying an intellectual property issue might be relativelystraightforward, dividing the e-commerce transaction along thegoods/services boundary is more problematic particularly when items thatwould usually be categorised as goods are delivered in electronic formreliant on telecommunications services to transmit the final ‘product’.57

Despite some members’ assertions58 that trade in e-commerce effectivelycontinues without resolving whether it involves trade in goods, or tradein services, coupled with the Appellate Body ruling that the WTO’s rulesin GATT59 and GATS60 can be applied concurrently anyway,61 the classi-fication discussion remains at the core of the WTO work programme one-commerce.62

It is clear that resolving the classification issue is important because aWTO member’s commitments to trade liberalisation in the goods and serv-ices sectors are contained in schedules annexed to both GATT and GATS:there is no single, common scheduling arrangement where members canindicate their willingness to liberalise trade in both the goods and services sec-tors at once. Instead separate schedules are required for each agreement.63

Consequently, the scope of these commitments can differ between goodsand services dependent on which sectors the member decides to target.Potentially this means that the member’s obligations then either change, orare removed completely dependent upon whether the subject matter in dis-pute falls within GATT or GATS. For example, separating the MFN andnational treatment rules in GATS but not GATT means that the scope of


Negotiations on Basic Telecommunications—An Overview’ (1997) 3 International Trade Lawand Regulation 135. On the services covered by the negotiations see <www.wto.org/emglish/tratop_e/serv_e/telecom_e/telecom_coverage_e.htm> (last visited 11 October 2002).

54 Annex 1A: Multilateral Agreements on Trade in Goods (incorporating the GeneralAgreement on Tariffs and Trade (GATT) 1994).55 Annex 1B General Agreement on Trade in Services (GATS).56 Annex 1C: Agreement on Trade-Related Aspects of Intellectual Property (TRIPS).57 eg books, or CDs.58 WTO, Work Programme on Electronic Commerce: Submission from the United States,WT/GC/W/493, 16 April 2003, para 8.59 The General Agreement on Tariffs and Trade 1994.60 The General Agreement on Trade in Services.61 European Communities—Regime for the Importation, Sale and Distribution of BananasWT/DS27/AB/R, 9 September 1997, para 221.62 The latest designated discussions on e-commerce show the classification discussion at thetop of the agenda: WTO, Fourth Dedicated Discussion on Electronic Commerce Under theAuspices of the General Council, WT/GC/W/492, 8 April 2003.63 Art II:1 GATT and Art XX GATS.

the member’s liberalisation commitment differs depending on whether theproduct traded involves goods or services when a member has not under-taken full liberalisation commitments in their GATS’ schedules.

Parallel application of both agreements does not address this problemfully either, for two reasons: first, there will be uncertainty over which rulesapply until there is a finding made by the WTO dispute settlement panel.Such ambiguity ensures it is difficult for participants to operate effectivelyin the highly volatile e-commerce market if they are uncertain which rulesare applicable.64 Second, the Appellate Body has made it clear that despitesharing common core rules, the scope of members’ obligations differsdependent on the extent to which they have agreed to liberalise sectors ineither the goods or services’ sectors.65 Conflicts between the application ofthe two agreements arise therefore.

Despite the WTO’s much praised dispute settlement mechanism,66 theboundary between goods and services has not been resolved judicially,67

but has instead been left to individual members of the WTO to determinethrough the negotiating process.68 Removing the decision away from thejudicial mechanism towards the members inevitably politicises the classifi-cation decision as the boundary between goods and services is blurred.This raises broader questions: whether it is appropriate to leave such acrucial delineation to the vagaries of politics when it directly affects indi-vidual business strategies; and also, who is the most appropriate arbiter inthe regulation of e-commerce and whether this person or body is even thesubject of either GATT or GATS. The correct choice of agreement then iscrucial because restrictions on trade in e-commerce may be containedwithin domestic regulations policed by non-governmental bodies.69 GATS

166 Fiona Smith

64 A question remains whether the scope of the MFN and national treatment obligations arethe same in GATT and GATS: see A Mattoo, ‘MFN and GATS’ in T Cottier, PC Mavroidisand P Blatter, Regulatory Barriers and the Principle of Non-Discrimination in World TradeLaw (Ann Arbor, University of Michigan Press, 2000) 51.65 The scope of the MFN and national treatment obligations is outside the scope of this, butthe Appellate Body did recognise that the scope of the most favoured nation (MFN) obliga-tion found in both GATT and GATS could be different: Canada—Certain MeasuresAffecting The Automotive Industry, WT/DS139/AB/R and WT/DS142/AB/R, 31 May 2000,para 181.66 See E-U Petersmann, The GATT/WTO Dispute Settlement System: International Law,International Organisations and Dispute Settlement (London, Kluwer, 1997).67 The question did arise in the Canada—Certain Measures Affecting the Automotive Industryabove n 65, but the Appellate Body side-stepped the issue. See section 3 of this article for afuller discussion. Note that the European Court of Justice has been faced with this dilemma inrelation to European Union law and despite making several attempts to resolve the issue, diffi-culties remain: see LM Woods, Free Movement of Goods and Services within the EuropeanCommunity (Aldershot, Ashgate, 2004, forthcoming).68 WTO, Work Programme on Electronic Commerce Interim Report to General Council,S/C/8, 31 March 1999: see annex which contains Chairman’s summaries of the informal dis-cussions on classification of e-commerce issues.69 See Feketekuty above n 47, 86.

recognises that decisions can be taken at this level, whereas GATT doesnot directly make this assumption.70 It is evident from the documentationsubmitted as part of the work programme on e-commerce that there is aninextricable link between the problems caused by the classification issueand the definition of e-commerce adopted by WTO members.

2. DEFINING E-COMMERCE

At the start of the WTO’s work programme on e-commerce finding a com-mon definition of e-commerce was difficult because there was divergencebetween those members71 who suggested definitions in their submissions aspart of that programme and those who felt that ascertaining how the WTOrules applied was more important than defining e-commerce in theabstract.72

Despite these conceptual difficulties, the WTO devised a working defini-tion in which e-commerce is ‘the production, distribution, marketing, saleor delivery of goods and services by electronic means’.73 This broad viewappears to mean that all transactions conducted via telephone, fax, and allother telecommunications systems74 are included,75 even though the term‘e-commerce’ is usually more commonly associated76 only with businessconducted via the Internet.77 During subsequent discussions on issues


70 Cf the WTO Agreement on Technical Barriers to Trade (TBT Agreement) which does envis-age some action taken by other bodies in relation to technical barriers to trade in goods, butmembers’ liability for their actions appears limited: Art 3:1 TBT Agreement. 71 G/C/W/158 above n 19, para 2.1.72 WTO, Work Programme on Electronic Commerce: Submission by the United States,WT/GC/16, G/C/2, S/C/7, IP/C/16, WT/COMTD/17, 12 February 1999 at para 3: note thatthe US does recognise that certain definitional issues must be addressed before the completionof the work programme: ibid. There is still disparity amongst members on this issue, particu-larly in relation to classification of products: see WTO, Second Dedicated Discussion onElectronic Commerce Under the Auspices of the General Council on 6 May 2002,WT/GC/W/475, 20 June 2002 at para 1.73 WT/L/274 above n 18, para 1.3; see also Electronic Commerce and the Role of the WTOabove n 1, 1. This can be contrasted with the more comprehensive approach taken by theEuropean Union’s work on the information society: see European Commission, Towards aNew Framework for Electronic Communications Infrastructure and Associated Services: The1999 Communications Review, COM (99) 539.74 eg television, electronic payment mechanisms and electronic data interchange (EDI): E-Commerce and Development Report 2001 above n 4, 175 ibid.76 UNCTAD, E-Commerce, WTO and Developing Countries, Policy Issues in InternationalTrade and Commodities Study Series No 2 (Geneva, United Nations, 2000) 1.77 Drake and Nicolaïdis above n 20, 399. The Internet grew out of the development of a com-munications mechanism (ARPANET) devised by university lecturers in the US in the early1960s. Although originally conceived to assist the US military, this scheme was graduallyabandoned in favour of exploitation of commercial usage. In the early period the new com-munications system was primarily used to send e-mail, but following the standardisation of

covering a number of legal aspects including the application of GATT andGATS’ rules to e-commerce (the ‘cross-cutting issues’) conducted pursuantto the work programme, members agreed that although the definitionappears to place the emphasis on the electronic media through which thetransaction occurs, in fact, the product traded should be the starting pointfor determining how the WTO rules apply.78

This working definition seems comprehensive, but significant difficul-ties arise because it fails to take into consideration the way that WTOrules operate in practice. On one level, the definition does not acknowledgewho the participants in e-commerce are; in particular, whether privateparties or businesses are involved, or whether it is limited to governmentsand other public or private organisations.79 Ensuring that the WTO definition at least acknowledges that non-state entities take part in e-commerce is important because on the basis of statistical data available,it is evident that whilst e-commerce actors may rely on the Internet as themain tool to undertake their transactions, they are all involved to differingdegrees.

Transactions between businesses (B2B commerce) account for the great-est proportion of trade conducted via e-commerce80 and projections esti-mate growth in this sector from e77 billion in 2001 to e2.2 trillion in 2006within the European Union alone.81 Trade between business and consumers(B2C commerce) accounts for a lesser proportion of on-line trade, but isstill expected to grow at a slightly slower rate, rising to e172.4 billionwithin the European Union by 2007.82 There is also growing interest in theextent to which e-commerce occurs between business and government (B2Gcommerce), although this has traditionally been more difficult to obtainstatistics on.83

This differing participation means that such parties’ transactions maynot be covered by WTO rules at all, or they might be affected in differentways dependent on whether trade in goods or trade in services is involved.

168 Fiona Smith

information flows as a result of the introduction of the TCP/IP protocol in the 1980s, inter-connection between networks was more readily achievable. The World Wide Web was established in 1990 and this directly facilitated on-line transaction including shopping andbanking: Electronic Commerce and the Role of the WTO above n 1, 10

78 WTO, Dedicated Discussion on Electronic Commerce Under the Auspices of the GeneralCouncil 15 June 2001, WT/GC/W/436, 6 July 2001, para 1.79 Electronic Commerce and Development Report 2001 above n 4, 6.80 UNCTAD estimated that over 80% of e-commerce was conducted via B2B commerce in2001: ibid. at 7.81 R Greenspan, ‘EU B2B Expected to Explode’ Cyberatlas, 28 August 2002 <http://cyberat-las.internet.com/markets/b2b> site last visited 6 November 2002).82 Ibid.83 Electronic Commerce and Development Report 2001 above n 4, 7. Note also that e-government transactions may also be included dependent on how trade in services is defined:see the UK’s ‘e-strategy’: <http://www.e-envoy.gov.uk/EStrategy/EStrategy/fs/en>.

Whilst there is an assumption that the rules only affect states,84 this ismisleading due to the way that GATT and GATS’ rules in particular oper-ate in practice. For example, Article XIII GATS limits the application ofthe MFN,85 national treatment86 and market access87 rules to govern-ment procurement, which changes their impact on those involved in B2Gcommerce. Similarly, protecting consumers from dubious business prac-tices is more important in B2C commerce, than in B2B commerce wheremaintaining competitive conditions within the market is paramount.88

Whilst maintaining competition may protect consumers in the long term,it may be desirable to intervene directly to introduce consumer protectionlegislation to supplement this.89

Articles VIII and IX GATS address anticompetitive behaviour betweenbusinesses in relation to trade in services thereby affecting B2B commerce.However, this occurs in a very generalised way: Article VIII GATS requiresmembers to ensure that monopoly service suppliers do not abuse their dom-inant position within the domestic territory such that it leads to violation ofthat member’s general commitments under GATS,90 consequently restrict-ing commercial freedom within the relevant jurisdiction.

This potential effect on non-state entities is clear from a pending dis-pute between the United States and Mexico on the interpretation ofGATS, the Annex on Telecommunications and the TelecommunicationsReference Paper.91 The United States alleged that Mexico failed effec-tively to regulate the activities of its major telecommunications servicesupplier, Telmex, thereby facilitating anticompetitive behaviour byexcluding other potential operators from the Mexican market. AlthoughMexico is the respondent in the dispute, if the United States succeeds in


84 Only ‘states or separate customs territories possessing full autonomy’ can be members of theWTO: Art XII:1 Marrakesh Agreement Establishing the WTO 1995 (the MarrakeshAgreement).85 Art II GATS.86 Art XVII GATS.87 Art XVI GATS.88 See J Lücking, ‘B2B E-Marketplaces’, ch 4, this vol.89 A detailed discussion of the economics of competition policy is beyond the scope of thischapter, but see R Posner, Antitrust Law: An Economic Perspective (Chicago USA, Universityof Chicago Press, 1976) for a US perspective and DG Goyder, EC Competition Law 4th edn(Oxford, Oxford University Press, 2003) for an EU perspective; also note that Trebilcock andHowse make the point that the extent to which competition law based on the maintenance ofa freely competitive market may not be desirable to achieve broader public policy aims: see B Dunlop, D McQueen and M Trebilcock, Canadian Competition Policy (Toronto, CanadaLaw Books, 1987), 67.90 Art VIII:1 and 2 GATS.91 Mexico—Measures Affecting Telecommunications Services WT/DS204 (pending): seeWTO, Request for Consultations by the United States, S/L/88, 29 August 2000 for detail onthe scope of the dispute. The panel was established on the 26 August 2002, but there has been a request for more time for deliberations as of 17 March 2003 (WT/DS204/5, 17 March 2003).

its claim on this issue, then amended domestic competition legislation islikely as a result of a successful GATS claim which will inevitably affectTelmex’s business activities.92 However, the exact form of the legislationis uncertain because the WTO rules do not currently insist on effectivecompetition legislation as such. In contrast, there are no direct equivalentmeasures in GATT,93 apart from the general exhortation in Article 9WTO Agreement on Trade-Related Investment Measures (TRIMS) toconsider the possibility of incorporating competition and investment poli-cies within the WTO rules.94 This distinction means that anticompetitivebehaviour will only be regulated directly if it concerns trade in servicesnot trade in goods. There are no rules that protect consumers directlyfrom anticompetitive behaviour in either GATT or GATS.95

On a more specific level, defining e-commerce in terms of the producttraded means that the WTO rules will apply to that product, rather thanthe media through which it is traded. This categorisation does not go farenough because the delineation between WTO rules on goods in GATT andthose on services in GATS96 means that before the rules can be applied,trade in the product must be classified further as falling into one of thesecategories.97 This boundary is blurred in e-commerce particularly in rela-tion to digital products, making it unclear whether participants’ businessactivities are affected or not.98

Despite extensive discussions in the work programme,99 members havebeen unable to decide whether e-commerce concerns trade in goods or serv-ices or both, with some members even suggesting that this discussion isunnecessary.100 On one view, the e-commerce transaction is regarded as

170 Fiona Smith

92 S/L/88 ibid para 7(4).93 Trebilcock and Howse suggest that a number of the WTO agreements covering trade in

goods do contain rules which operate to restrict anticompetitive behaviour, but there is nodirect equivalent of Art VIII and IX GATS: see MJ Trebilcock and R Howse, The Regulationof International Trade 2nd edn (London and New York, Routledge, 1999) 470.

94 Working Groups on Trade and Competition Policy and Trade and Investment Measureswere set up following the first WTO ministerial conference in Singapore in 1996: seeWT/MIN(96)/DEC, 18 December 1996. The latest report of the working group is: WTO,Report (2003) of the Working Group on the Interaction between Trade and CompetitionPolicy to the General Council, WT/WGTCP/7, 17 July 2003.

95 The WTO Agreement on the Application of Sanitary and Phytosanitary Measures (SPSAgreement) allows members to impose restrictive trade measures to protect human health:Art 2:1 SPS Agreement. This is a limited exception where the consumer is considered.96 The Appellate Body has confirmed this situation: see Bananas WT/DS27/AB/R above n 61,

para 221.97 The Appellate Body does argue that the rules can be concurrently applied, but note its

practice in Canada—ertain Measures Affecting the Automotive Industry above n 65.98 On the generally distinction between goods and services see J Bhagwati, ‘Economic

Perspective on Trade in Professional Services’ (1986) 1 University of Chicago Legal Forum 45,45–5399 Fourth Dedicated Discussion on Electronic Commerce, WT/GC/W/492 above n 62.

100 See WTO, Work Programme on Electronic Commerce: Submission by the United States,WT/GC/W/493, 16 April 2003, para 8. Contrast WTO, Work Programme on Electronic

trade in goods if the point of the transaction is trade in goods purchased forelectronic commerce, such as computers, or if the transaction is conductedelectronically, but there is delivery of physical goods.101 Whilst this appearsa coherent solution to the dilemma, difficulties occur when digital productsare delivered via electronic media:102 here, items like CDs, video, computersoftware103 and books would constitute goods by analogy if they are tradedin their physical form, but if instead their delivery is dependent directly onthe relevant electronic media, they could just be seen as ‘the transmittedbytes and data streams [consisting] only of one’s and zero’s’.104 The natureof the product traded changes therefore so that it no longer resembles theoriginal book, or CD that otherwise would be delivered in its non-digitalform raising the question whether it is still ‘goods’.

Some members have tried to resolve this problem by trying to identify acentral characteristic which categorises the digital product as goods, ratherthan services:105 one interpretation sees such products as goods if they‘resemble or were close substitutes’ to the physical goods,106 or if they weredelivered personally to the consumer, rather than being generally availableon the Internet;107 and as services if they are immediately consumed by thepurchaser and are not stored.108 However, these views are not satisfactorybecause they do not try to isolate what the inherent nature of an item isthat makes it ‘goods’, but instead either try to identify the product as goodsby analogy to a nebulous ‘like product’, or classify the product by its end-use. Whilst this latter idea could be satisfactory if digital products are seenas sui generis, problems arise if all products are classified in this waybecause this conflicts with the original view that an electronic transactioninvolving physical delivery of goods is trade in goods: taking this exampleto its limits, purchase of a chocolate bar over the Internet would be goods ifit was then stored, but services if consumed immediately by the consumer.This is clearly an erroneous distinction.109


Commerce: Classification Issue: Submission by the European Communities, WT/GC/W/497,9 May 2003, paras 8 and 14.

101 G/C/W/158 above n 19, para 2.2(i)–(ii).102 Ibid para 2.6.103 See Canada’s non-paper submitted to the second designated discussion on e-commerce inMay 2002: The Classification of Software Delivered Electronically, Job No (02)38 (8 May 2002). Non-papers are not generally available, but see summary in WT/GC/W/475above n 72, para 1.5. 104 Ibid para 2.10.105 This view is not universally accepted: see WT/GC/W/436 above n 78, para 1.6.106 Ibid para 2.7.107 Ibid para 2.7.108 Electronic Commerce and the Role of the WTO above n 1, 51; and Drake and Nicolaidïsabove n 20, 408. 109 Some members seem to be advocating sui generis treatment for electronic products:G/C/W/158 above n 19, para 2.9.

This ‘subject matter only’ methodology is not universally accepted bymembers, as some favour an approach which places greater emphasis onthe electronic nature of the product traded: the European Union in particu-lar argues that the underlying need for the appropriate electronic networksto participate in e-commerce means that all products traded as a result mustbe services and not goods, irrespective of the product that is either physi-cally or electronically delivered at the conclusion of the transaction.110 Thisview seems to be supported by the WTO Secretariat despite their neutrality,as their 1998 background note does follow the European Union approachclosely.111 In addition, the view is growing in popularity amongst mem-bers,112 although there is still significant disagreement over the classifica-tion of a ‘small number of products made available on the internet … suchas books and software’.113 There is consequently still no solution to theclassification of digital products, nor to the more general issue of whatmakes a product goods or services.

3. APPLYING GATT AND GATS TO E-COMMERCE: IS E-COMMERCE TRADE IN GOODS OR SERVICES?

Achieving consensus amongst members on the correct classification of thee-commerce transaction is crucial because GATT and GATS do not definethe terms ‘goods’ or ‘services’;114 instead the rules apply to products inmembers’ schedules listed according to customs classification codes.115

Significant difficulties arise on two levels: first, defining the product asgoods or services at all, and second, finding the correct classification codewithin GATT or GATS. These problems are acute in relation to digitalproducts.

172 Fiona Smith

110 WTO, Communication from the United States—Work Programme on ElectronicCommerce—Scope and Classification Issues, S/C/W/87, 9 December 1998, para 1; this view isalso reflected in their paper on the categorisation of computer and related services: WTO,Coverage of CPC 84—Computer and Related Services, TN/S/W/6, 24 October 2002, para 8;it reiterates its view in WTO, Work Programme on Electronic Commerce: Classification Issue:Submission by the European Communities, WT/GC/W/497, 9 May 2003, para 16.111 WTO, Work Programme on Electronic Commerce: Note by the Secretariat, S/C/W/68,16 November 1998, 10; see also discussion by Drake and Nicolaidïs above n 20, 409.112 WT/GC/W/436 above n 78, para 1.3.113 Work Programme Reflects Growing Importance of Electronic Commerce 15 June 2001,available at <www.wto.org/english/tratop_e_ecom_e/ecom_briefnote_e.htm>. (last visited9 October 2002) 2.114 WTO, A Consideration Concerning the Relationship Between the WTO Provisions andthe Subjects Listed for Discussion Under Paragraph 3.1 of the Work Programme: BackgroundNote by the Secretariat, G/C/W/128, 5 November 1998 at para 1.2; note some members’opposition to the secretariat approach which was reiterated in the secretariat background note(Job No (02)37 (1 May 2002) ) to the second dedicated discussion on e-commerce in May 2002: see WT/GC/W/475 above n 72, para 1.11.115 See Art II GATT and Part III GATS.

A. GATT

GATT generally uses the Harmonised Commodity Description and CodingSystem (HS) nomenclature,116 which aims to list all goods that could pos-sibly be subject to tariffs. Devised by the World Customs Organisation,117

the HS nomenclature has 97 separate chapters describing goods in termsof physical characteristics; these are sub-divided into headings and sub-headings, with the individual goods identified by a six-digit code. Thiscode is then used in the member’s schedule to identify the goods subject toany tariff reduction commitments.118

This methodology apparently resolves the classification issue because ifthe product is listed in the HS nomenclature, it must be goods; but despiteadding explanatory notes to each category, it is difficult to place e-commerce accurately within a specific code. Several problems arise: the HScode does not have a separate classification for the content of any electronictransmission, but instead focuses on the carrier medium used to transmitthe content.119 The WTO Secretariat notes that this problem is acute forcomputer software purchased on the Internet, because the HS code doesnot have an individual listing for software as such,120 but instead classifiesit in terms of either the laser disc, magnetic disc or tape on which it is trans-ported.121 This means that software could be classified under three HSheadings.122 Similar difficulties arise in terms of books transferred electron-ically: here a book physically transported would be covered by the HS code.If the book is purchased on the Internet instead and the content transmittedby laser or magnetic disc, is the book covered by the original code forbooks, or is it now like the software, classified in terms of the physicalmedia on which it is transmitted to the purchaser?123 Rather than resolving


116 International Convention on the Harmonised Commodity Description and Coding System,14 June 1983 amended by Customs Co-operation Council Recommendation, 25 June 1999which entered into force 1 January 2002. See <www.wcoomd.org/ie/En/Topics_Issues/ top-ics_issues.html> (last visited 10 October 2003).117 See Jackson, Davey and Sykes above n 31, 394 for a brief history of the development of theHS system.118 G/C/W/158 above n 19, para 2.1. eg HS Heading 33.01 covers essential oils, and HS Code2905.44 is the customs classification code for sorbitol: see WTO Agreement on AgricultureAnnex 1 defines product coverage in the agreement on terms of the HS Code. See HarmonizedSystem Convention, 12 February 2003, <http://www.wcoomd.org/ie/En/ Topics_Issues/top-ics_issues.html> above n 116. The HS Code was amended on 8 January 2002, but there wasno change to the essential oils and sorbitol codes. See <http://www.wcoomd.org/ie/En/Topics_Issues/topics_issues.html> ibid. 119 WTO, Considerations Concerning the Relationship Between WTO Provisions and theSubjects Listed for Discussion Under Paragraph 3.1 of the Work Programme, G/C/W/128,5 November 1998, para 2; also G/C/W/158 above n 19, para 2.2.120 On the classification of software see Job No (02)38 above n 114.121 This situation has not been resolved in the amendment.122 Above n 120 at para 2.2.123 Electronic Commerce and the Role of the WTO above n 1, 51.

the goods/services dilemma, the uncertainty in the HS coding means thatthe categorisation decision rests with the member, who will therefore makea pragmatic decision on the status of the product dependent on the impor-tance he or she places on the electronic element of the transaction, conse-quently bringing the debate back to whether the electronic nature of thetransaction changes the inherent characteristic of the product traded fromgoods to services.

Some progress has been made on the liberalisation of trade in certaininformation technology products under the auspices of the WTOMinisterial Declaration on Trade in Information Technology Products(ITA).124 This agreement lists products essential for the e-commerce trans-action including hardware, some software and certain telecommunicationsproducts. However, the ITA fudges the classification issue by just listing theproducts covered, thereby designating them as goods, rather than definingthe essential nature of information products. The ITA uses the HS nomen-clature and categorises the products based on agreement on code coveragebetween signatories to the agreement.125 Some members have expressed theview that products already classified under the ITA should remain as‘goods’ for the purposes of the GATT rules, although this does not actuallyresolve the problem for non-ITA products, as there is no classificationmethodology in the ITA as such to use as an analogy in respect of thoseproducts not already covered by the agreement.126

Relying on the GATT rules to resolve the classification decision eitherat the goods/service level, or within the individual codes does not resolvethe problem because there is no absolute requirement that members adoptthe HS nomenclature at all. This complicates matters further as GATTdoes not even require a common starting point. In Spain—TariffTreatment on Unroasted Coffee,127 the panel made it clear that there was‘no obligation … to follow any particular system for classifying goods,and a contracting party had the right to introduce in its customs tariffnew positions or subpositions as appropriate’.128 This position was fol-lowed in Canada/Japan: Tariff on Imports of Spruce, Pine Fir (SPF)Dimension Lumber129 where the panel went further and stated thatalthough both parties to the dispute had adopted the HS nomenclature, this‘had brought about a large measure of harmonisation in the field of cus-toms classification of goods, but this system did not entail any obligation as

174 Fiona Smith

124 WT/MIN(96)/16, 13 December 1996. The ITA aims to eliminate tariffs on specified ITproducts between 1997 and 2000 (2005 for Costa Rica): see NE Scott, ‘WTO Declaration-Trade in IT Products’ (1997) 3 International Trade Law and Regulation 45.125 WT/MIN(96)/16 ibid. para 5.126 WT/GC/W/475 above n 72, para 2.2.127 BISD 28S/102, 11 June 1981, 111.128 Ibid.129 BISD36S/167, 19 July 1989.

to the ultimate detail in the respective customs classification’.130 GATTtherefore adopts a laissez-faire approach to classification where the cru-cial aspect is not how the product is classified per se, but whether the tar-iff treatment complies with the member’s MFN and national treatmentobligations, so that ‘like products’ are treated equally, irrespective of theiractual classification.131

B. GATS

Classifying the e-commerce transaction as trade in services under GATS ismore complex than GATT, as it raises problems on three levels: first,whether the e-commerce transaction is trade in services at all; second, if itis, how is that service supplied; and third, how should the e-commercetransaction be categorised for the purposes of members’ schedules? UnlikeGATT, which does not define ‘goods’ at all, GATS includes a definition ofservices. However, it does not define ‘services’ in terms of the inherent char-acteristics of the products themselves,132 but instead focuses on the mannerby which those products are supplied, referred to in GATS as the mode ofsupply.133 Under Article I GATS, the rules apply to the supply of a serviceeither traded across borders,134 where the consumer is present in anothermember’s territory to receive the service,135 if the service is suppliedthrough the presence of a commercial entity within the territory of anothermember,136 or if the service is supplied by the presence of a ‘natural person’in the other member’s territory.137

The Appellate Body considered GATS’ coverage, particularly the scopeof Article I, in Canada—Certain Measures Affecting the AutomotiveIndustry.138 At issue was Canada’s duty free importation scheme for cer-tain manufacturers of cars, buses and other commercial vehicles. Underthe scheme, these vehicles enjoyed duty free tariff treatment if they con-formed to the terms of the Motor Vehicles Tariff Order 1998 and certain


130 Ibid paras 5.7–5.10.131 GATT, Panel On Vitamins BISD 28S/102 at para 4.4. See also European Communities—Customs Classification of Certain Computer Equipment, WT/DS67/AB/R andWT/DS68/AB/R, 5 June 1998, para 90.132 This is despite the fact that GATS says it covers the product traded: WTO, WTOAgreements and Electronic Commerce, General Council, WT/GC/W/90, 14 July 1998, para 3.133 Art I GATS.134 Mode 1: Art 1:2(a) ibid.135 Ie consumption abroad: Art I:2(b) ibid.136 Art I:2(c) ibid.137 Mode 4: Art I:2 (d) ibid.: on the general problems raised by the various modes of supplysee G Karsenty, ‘Assessing Trade in Services by Mode of Supply’ in Sauvé and Stern above n 20, 33.138 WT/DS139/AB/R and WT/DS142/AB/R, above n 65.

Special Remission Orders:139 the manufacturer had to have built vehiclesin Canada of the type it wanted to import duty free during a designated‘base year’; ensure that the ratio between vehicles it produced in Canadaand those it imported in any given period was ‘equal to or higher than’the ratio from the designated ‘base year’, but that this should not fallbelow a 75:100 split in any event; and finally, that the value of theCanadian input into the domestic production of the vehicle concernedwas ‘equal to or greater than’ the equivalent value put into the vehicle’smanufacture in the base year.140

In addition to the general requirements in the Motor Vehicles TariffOrder, Canada also granted duty free treatment to other manufacturers ifthey complied with specified requirements in the Special Remission Orders(SROs). These SROs provided different ratios to those in the MotorVehicles Tariff Order: some operating on a like-for-like basis, that is100:100, whilst others offered more favourable treatment, with some man-ufactures only required to meet 60:100 or even a 40:100 ratio instead ofthe 75:100 in the original order.141

Although allegedly origin-neutral, Japan and the European Union arguedthe de facto application of the measure meant only certain manufacturersbenefited from the duty free treatment. In particular, Japan pointed to thesuccess of Swedish and Belgian manufacturers, Saab and Volvo in obtainingduty free access in contrast to the Japanese manufacturer, Lexus.142 In addi-tion to violations of Article I:1 GATT 1994 and Article 3:1 WTOAgreement on Subsidies and Countervailing Measures, the panel found thatthe measure breached Article II GATS, as the measure affected trade in serv-ices under Article I GATS.

On appeal, Canada argued that its duty free tariff scheme failed toimpact on ‘wholesale distribution service suppliers in their capacity as serv-ice suppliers’.143 Consequently, it claimed the scheme did not come withinGATS because the measure did not concern trade in services at all underArticle I, but only affected trade in the vehicles as goods, so the legality ofthe scheme should only be considered under GATT.144 Clearly, Canada wasarguing for a definitive categorisation of the products involved based on itsown classification criteria, and was not content to allow both the GATTand GATS to apply concurrently to the measure. Both the panel and theAppellate Body rejected this rigid approach, albeit for different reasons.

176 Fiona Smith

139 Ibid para 7.140 Ibid paras 8–10.141 Ibid para 14.142 Canada—Certain Measures Affecting the Automotive Industry—Report of the Panel,WT/DS139/R and WT/DS142/R, 11 February 2000, paras 5.271–6.143 WT/DS139/AB/R and WT/DS142/AB/R above n 65, para 20.144 Ibid para 147.

Superficially, the panel145 adopted a two stage analysis: first, it considered whether the measure affected trade in services at all to satisfythe threshold test in Article I GATS. Second, it evaluated the breach ofCanada’s substantive obligations, particularly under Article II.146

Although it reiterated Canada’s arguments, the panel did not go on andconsider their veracity, but instead pointed to the Appellate Body’s asser-tion in Bananas147 that there are certain measures which can be assessedsimultaneously under GATT and GATS, thereby negating the need toevaluate whether the measures in fact affected trade in services orgoods.148 The fact that the measure could affect trade in services wasenough and a determination of whether it did affect trade in services wastherefore unnecessary.149 By moving straight to the violation of Article IIGATS, the panel fused the threshold test with the assessment of the sub-stantive breach, arguing that as the measure breached Article II it must bewithin GATS for the purposes of Article I.150 This interpretation wasrejected by the Appellate Body.

In contrast to the panel, the Appellate Body reiterated the need to consider GATS’ threshold test in Article I:1 and the substantive obligationsseparately.151 Whether the transaction involves trade in services must beestablished first, followed by a determination that the measure at issueaffects services trade.152 In the Appellate Body’s view, trade in services willbe at issue where a service is provided through one of the designated modesof supply listed in Article I:2 GATS. It went on to state that as the issue indispute was ‘wholesale trade services in motor vehicles’, which was a listedcustoms classification category and not disputed as such by Canada, theissue involved trade in services, so no further discussion of the scope ofArticle I:2 was needed. There was no substantive discussion on the natureof trade in services in the abstract, or whether the customs classificationcategory covered the Motor Vehicle Tariff Order at all.153 It appears thatrecognition by Canada that it used this customs classification in some waywas enough for the application of GATS, irrespective of whether the meas-ure in dispute actually came within it.

By adopting this view the Appellate Body therefore side-stepped the clas-sification issue, instead focusing on the mode of supply to determinewhether the issue comes within GATS. Although this approach follows the


145 WT/DS139/R and WT/DS142/R, above n 142, paras 10.223–45.146 Ibid para 10.228. 147 WT/DS27/AB/R above n 61, para 221.148 WT/DS139/R and WTDS142/R above n 142, paras 10.233–34.149 Ibid para 10.235.150 Ibid para 10.235.151 WT/DS/AB139/AB/R and WT/DS142/AB/R above n 65, paras 150–51.152 Ibid para 155.153 Ibid para 157.

wording of GATS closely, this does not resolve the classification issue forthe purposes of digital products.154 This is because placing the emphasis onthe mode of supply, rather than any inherent characteristic of the productmeans that there still must be an initial decision that the product is servicesbefore the mode by which that product is traded is ascertained. This isbecause a product traded across borders could equally be regarded as goodstherefore applying GATT, or as services by virtue of mode 1 GATS, so thatthe mere border transfer in itself cannot be determinative of the product’sstatus as services.

GATS’ technological neutrality155 is arguably irrelevant to this dilemmabecause it only impacts once the product is within the scope of GATS, sothe fact that the product is supplied electronically will not prevent it frombeing services; but equally, it does not automatically lead to the productbeing designated as services either.156 Without further guidance from theWTO on the designation of e-commerce products as goods or services per se, members will make the decision based on their own definition,thereby raising the same concerns seen in the GATT context.157

Little assistance is given in the scheduling guidelines adopted by theCouncil for Trade in Services in March 2001 to assist in the GATS renegoti-ation discussions.158 Paragraph 28 states that ‘the supply of a servicethrough telecommunications or mail and services embodied in exportedgoods (ie services supplied in or by a physical medium, such as a computerdiskette … ) are all examples of cross-border supply’ under mode 1GATS.159 Whilst this seems to be arguing that e-commerce transactionscould be categorised as services because they are conducted via telecommu-nications networks, this presumes that the product transmitted via that net-work is services in the first place because it places the emphasis on how thatservice is supplied, rather than whether it is a service at all. Adopting thisapproach without further clarification is difficult therefore because somemembers do not even accept that transactions conducted on-line in this wayare trade in services in the first instance.160

Even if e-commerce is seen as trade in services in the first instance, fit-ting it within a particular mode of supply to determine what a member’s

178 Fiona Smith

154 The Appellate Body noted that the scope of GATS’ coverage was problematic. Although itrejected the panel’s finding on the scope of Art II:1 GATS, it did not go further and completethe analysis, arguing that this should be left to later dispute proceedings where GATS formed acentral element of the proceedings: ibid. para 184.155 WT/GC/W/90 above n 132, para 3.156 Members recognise the problems of technological neutrality in relation to the applicationof GATT: WT/GC/W/436 above n 78, para 1.8.157 See section 2 above.158 WTO, Guidelines for the Scheduling of Specific Commitments under the GeneralAgreement on Trade in Services (GATS), S/L/92, 28 March 2001.159 Ibid (emphasis added).160 WT/GC/W/475 above n 72, para 1.11.

obligations are in GATS is difficult because the boundary between the fourmodes becomes blurred. This problem is acute in the boundary betweenmodes 1 and 2.161 Here the mode of supply is determined by the origin ofboth the consumer of the service and its supplier and the extent to whichboth can be said to be territorially present within a member’s territory.162

Physical presence of either the supplier or consumer is not a requirement,therefore, as Drake and Nincolaidïs observe, problems arise in relation toprovision of services via the Internet because ‘millions of customers can“virtually visit”a foreign country and import services’ without ever leavingtheir own jurisdiction physically.163 This makes it hard to decide if theconsumer has in fact crossed the border to enjoy the service, or whetherthe service is imported to the consumer instead for the purposes of decid-ing whether it is mode 1 or 2.164

The Committee on Trade in Financial Services discussed the boundarybetween modes 1 and 2 in 1997,165 particularly raising the classification ofthe electronic supply of financial services where the physical presence of theservice provider was not required for the delivery of the service.166 Ratherthan making a definitive suggestion on the classification, the note suggestsfive interpretations of the transaction which would then designate the trans-action within either mode 1 or 2: the service will be mode 1 if either theconsumer is resident within the member’s territory, if the transaction takesplace under the laws of the member, or if the supply of the service alsoinvolves ‘solicitation’. The final solution fudges the issue by suggesting themerger of modes 1 and 2, which could involve the renegotiation of existingreduction commitments if members have only made commitments in rela-tion to either mode 1 or 2 and not both.

The financial service paper was later built upon by the Secretariat167

who suggested that rather than trying to distinguish between the modes ofsupply in the abstract, the distinction should be based on the commitmentsmade in members’ GATS schedules, because the mode is only relevant todetermine whether a member has made a commitment in a specific areawhere its measures are in dispute.168 This does not resolve the problem inrelation to e-commerce because there are disparities between members’


161 Ibid para 1.12.162 GATT, Scheduling of Initial Commitments in Trade in Services: Explanatory Note,MTN.GNS/W/164, 3 September 1993 (see also Addendum, MTN.GNS/W/164/Add.1, 30 November 1993).163 Drake and Nicolaidïs above n 20, 413.164 S/C/8 above n 68, Annex para 2.165 WTO, Informal Note by the Secretariat for the Committee on Trade in Financial Services:The Distinction Between Modes 1 and 2, incorporated into S/FIN/W/14, 24 June 1997.166 Ibid para 3.167 S/C/W/68, above n 111.168 Ibid para 8.

views on its correct classification within their schedules, so the proposedsolution only moves the original classification decision to a later stage.

Classification within members’ schedules is more complex under GATSthan GATT: members are required to enter those sectors in which theywish to make commitments in their schedules. They are assumed to makea full commitment to liberalising these sectors, unless they add any reser-vations in regard to market access or national treatment. If the member’sschedule is silent on a particular sector, then they are not deemed to makeany specific commitments in relation to that sector.169

This methodology was more complicated prior to 2001 because therewas no universally agreed method of scheduling amongst members assuch, which meant that the way that individual commitments and reser-vations were expressed was left to the discretion of the member.170 Thescheduling guidelines adopted by the Council on Trade in Services inMarch 2001171 do ‘explain, in a concise manner, how specific commit-ments should be set out in schedules in order to achieve precision andclarity’,172 although the guidelines state that ‘(t)he answers should not beconsidered a legal interpretation of GATS’.173 Even with the adoption ofthe guidelines, GATS scheduling for products traded as a result of e-commerce is problematic. This is because there is an inherent conflictbetween the assertion that the guidelines promote ‘precision and clarity’,174 and the prospect that they perpetuate the ambiguity whichexisted prior to 2001.

Before the guidelines, members could either adopt their own system ofclassification for scheduling purposes, use the Services SectoralClassification List (W/120) prepared by the WTO Secretariat during thecourse of the Uruguay Round,175 or adopt the United Nations CentralProduct Classification System (CPC)176 on which W/120 is based.177 Thisfreedom meant that there were diverse scheduling practices amongstmembers making it difficult to ascertain in some cases exactly what

180 Fiona Smith

169 Feketekuty above n 47, 98.170 See Electronic Commerce and the Role of the WTO above n 1, 51.171 S/L/92 above n 158.172 Ibid para 1.173 Ibid para 1.174 Ibid para 1.175 WTO, Services Sectoral Classification List, Note by the Secretariat, MTN.GNS/W/120, 10 July 1991.176 UN, Provisional Central Product Classification, Statistical Papers Series M No 77Department of International Economic Affairs (New York, United Nations, 1991) currently inversion 1.1 ST/ESA/STAT/SER.M/77/Ver.1.1, March 2002. This chapter uses the numberingthat members adopt in their submissions to the work programme on e-commerce. Note how-ever, that this does not always correspond with the numbering in Version 1.1 CPC, which onlyobfuscates the issue further. 177 Electronic Commerce and the WTO above n 1, 51.

commitment a member was making in which sector.178 To some extentthe guidelines address the scheduling problem: they make it clear thatgenerally members should adopt W/120 as the basis of their commit-ments, which lists each sector according to the CPC’s classification. In theevent of an omission in the CPC classification, the guidelines state thatmembers then should rely on another internationally recognised sys-tem.179 The guidelines also recognise members’ right to rely on their ownmethodology for ‘sub-sectoral classification or definitions’ provided thatthese either comply with the CPC where possible, or they give a ‘suffi-ciently detailed definition to avoid any ambiguity as to the scope of thecommitment’.180 Whilst retaining flexibility within the classification sys-tem is important, especially in relation to e-commerce where technologychanges rapidly,181 ascertaining the level when a member will be deemedto have given a ‘sufficiently detailed definition to avoid any ambiguity asto the scope of their commitment’182 is uncertain. Although more detail isgiven on the general methodology concerning scheduling commitments inthe guidelines,183 this does not overcome the basic problem of classifyingproducts traded by e-commerce using W/120 and the CPC.

W/120 was prepared during the Uruguay Round and categorises servicesaccording to a generic list with sub-divisions then listed according to theCPC classification. Consequently, once a member makes a commitmentusing W/120 in a sector that carries a CPC code, the interpretation of theclassification lies in the scope of the CPC listing.184 The CPC classificationmethodology operates on a coding system that allocates a code number toeach main product category. This first category is referred to as the sectionheading and carries a one-digit code. The section heading is then sub-divided into four further levels. Level 2 is the division heading and carries atwo-digit code; level 3 is the group heading with a three-digit code; level 4refers to classes and has a four-digit code and finally level 5 is referred to asthe sub-classes and carries a five-digit code.185 Each category includes defi-nitions of the scope of each level of the code. Like the HS nomenclatureused in GATT,186 the CPC was designed to be exhaustive, but there areinstances where a product is not listed at all and several instances where theCPC has an illusive category listed as ‘other’: for example, computer and


178 See P Low and A Mattoo, ‘Is There a Better Way? Alternative Approaches to LiberalizationUnder GATS’ in Sauvé and Stern above n 20, 449, 468.179 S/L/92 above n 158, para 23.180 Ibid para 24.181 WT/GC/W/475 above n 72, para 1.2.182 S/L/92 above n 158, para 24.183 Ibid paras 23–25.184 Electronic Commerce and the WTO above n 1, 51.185 See <http://unstats.un.org/unsd/cr/family2.asp?Cl�16>. 186 See section 2 above.

related services includes CPC 845 and 849 specified vaguely as ‘other computer and related services’.187

Several problems arise when this nomenclature is applied to e-commerce.First, it is possible for e-commerce to be listed under several CPC headings,including communication services, telecommunication services,188 com-puter and other related services, as well as professional services. Inevitably,this means that there is the likelihood of considerable disparity between thetreatment of e-commerce in members’ schedules. The European Union pro-posed clarification in relation to the coverage of computer and other relatedservices,189 so that CPC 84 would cover ‘basic functions used to provide allcomputer and related services: computer programs … data processing andstorage, and related services, such as consultancy and training services’.190

This still raises the possibility of conflict between CPC 84 and CPC 86 ifsome members regard the provision of technical assistance to facilitate bet-ter use of the computer networks as technical testing and analysis serv-ices191 or under the generic ‘other business services’ category,192 rather thanas computer services per se, because a narrow reading of CPC 841 restrictsits coverage to consultancy services regarding the installation of the com-puter hardware only.193

The second problem is illustrated by the conflicting classification prob-lems raised in the European Union’s proposal on computer and relatedservices: if a digital product is traded via the Internet and classified as serv-ices in the first instance, should those services be individually classifiedand if so, how? Should it be by provision of telecommunication services,the provision of access by the Internet Service Provider, the availability ofencryption services, data processing services, or even on-line informationand data retrieval systems?194 The scheduling guidelines suggest focusingon the type of service provided,195 rather than the service provider, butthis does not help identify exactly what type of service is provided whendigital products are traded.

Finally, despite the classification system in both GATT and GATS, thefundamental problem is that there must still be a decision that a producttraded via e-commerce is de facto goods or services before the member can

182 Fiona Smith

187 W/120 above n 175, 2.188 The Telecommunications Agreement only addresses classification of basic telecommunica-tions: see Naftel and Spiwak above n 53, 102–15.189 TN/S/W/6 above n 110, and WT/GC/W/497 above n 100.190 Ibid para 7.191 CPC 8676.192 CPC 8790.193 CPC 841.194 See Drake and Nicolaidïs above n 20, 412. Also CPC 7523 (on-line information and/ordata processing) and 843 (data processing).195 S/L/92 above n 158, para 23.

then go on and schedule it under the classification system. This is becauseboth the CPC and the HS nomenclature operate on a descriptive product-specific basis, rather than on the inherent characteristics involved in theitems traded. GATT and GATS rules’ silence on this basic issue means thatthe ultimate classification decision rests with the member, rather than withany independent judicial body and, as such, the decision is removed fromthe scrutiny of the WTO dispute settlement system, thereby circumventingone of the primary reasons for using the WTO scheme in the first place.

Consequently, at this level, the decision shifts from one solely aboutWTO law, to one incorporating members’ domestic political and policyconcerns, including the commercial needs of those participating in e-commerce within the members’ territory. In this way the purpose ofusing the WTO system to create an international harmonised frameworkfor e-commerce is undermined because regional disparity remains. Thisbrings the debate back to the lack of agreement on the classification ofproducts traded via e-commerce, because without further guidance, mem-bers are likely to classify on the basis of their stance on whether they seee-commerce as goods or services.196 Arguing that inevitably membersmust retain competence over some decisions unless the WTO adopts afull ‘market integration’ strategy does not take the argument furtherbecause the operation of the GATT and GATS means that liberalisationin e-commerce could be patchy without further elaboration on the classi-fication issue.

4. CONCLUSION

Successfully regulating e-commerce is complicated by the need both toenforce the transaction so that contracts concluded are valid and to facili-tate trade conducted through electronic means by the removal of unneces-sary border controls and internal restrictions. Using the WTO regulatorystructure appears to be the answer to this regulatory dilemma as it has acomprehensive system of rules that govern most aspects of internationaltrade and a proven dispute settlement mechanism. However, it is clear thatthe inherent rigidity of the WTO rule structure along the goods/servicesboundary means any product traded must be categorised as involving tradein goods or services before the rules to apply at all.

Whilst this structure poses few problems for traditional products, signif-icant disagreement remains amongst WTO members into which category


196 The issue has been fudged to some extent in relation to the imposition of customs duties, asmembers have agreed not to impose customs duties on ‘electronic transmissions’ until 5thMinisterial Conference in September 2003: WTO Doha Ministerial Declaration WT/MIN/(01)/DEC/1, 20 November 2001, para 34.

products traded on-line fall. This dilemma is not resolved by the GATT andGATS’ rules, as they only impact on products listed in members’ schedulesannexed to the relevant agreements, with the likelihood of significant diver-gence between members over whether a product is goods or services and inaddition, in services trade, whether they are prepared fully to liberalise therelevant sector in any event. Although GATT and GATS adopt detail sched-uling nomenclature to assist the classification of a product as either trade ingoods or services, doubts exist where products do not fall firmly within onecategory: this is acute for digital products. In such marginal cases, the clas-sification decision lies with the member who arguably will make a decisionbased on the needs of its domestic economy, rather than on the basis ofglobal considerations. This leads to the politicisation of the decision andquestions must be asked whether it is appropriate to allow domestic politi-cal considerations to impede products traded via e-commerce.

Despite the parallel application of GATT and GATS, the classificationdilemma remains at the heart of the discussions in the work programme one-commerce, as members appear reluctant to leave the crucial classificationdecision to the WTO dispute settlement mechanism to resolve.

Following the collapse of the 5th Ministerial Meeting at Cancun inSeptember 2003 the WTO’s relevance to the post-September 11th tradingenvironment is being questioned.197 The application of the WTO’s rules toe-commerce is a significant element of this debate. It is only by recognisingthe limitations of the existing rule structure that the WTO can be an effec-tive arbitration mechanism for the needs of the on-line trader.

184 Fiona Smith

197 WTO, ‘Conference Ends Without Consensus’ press release, 14 September 2003: <http:// www.wto.org/english/thewto_e/minist_e/min03_e/min03_14sept_e.htm> (last visited9 October 2003).

8

Public Services in the New Economy

ERIKA SZYSZCZAK*

1. THE PROVISION OF PUBLIC SERVICES

PUBLIC SERVICES OCCUPY an uneasy space in the new economy,caught between the processes of liberalisation and the Member States’willingness to recognise the role of public services but not finance

their existence in a world of budget constraints and privatisation. Duringthe 1990s public services underwent a revolutionary transformation moti-vated by the pressures for greater liberalization of markets but also by thegrowing disinclination of states to continue to finance public services. Thisled to a fundamental reconsideration of the role of the state in the provisionof public services and an active policy of ‘re-inventing government’,1 result-ing in privatisation, fragmentation, hybridisation and restructuring of publicfunctions.2

Electorates have been persuaded of the capacity of markets to providepublic services and this in turn has led consumers to have differentexpectations3 of the nature and quality of such services. In short, to expectpublic services to be more like ‘private’ services, coupled with a belief that

* Erika Szyszczak is the Jean Monnet Professor of European Law ad personam and Professorof European Competition and Labour Law at the University of Leicester, UK. She is alsoDirector of the Centre for European Law and Integration.1 See D Osborne and T Gaebler (contributor), Reinventing Government: How the EntrepreneurialSpirit is Transforming the Public Sector (Boulder, Perseus Books, 1992); C Hood, C Scott, O James and T Travers, Regulation Inside Government (Oxford, Oxford University Press, 2000).2 New forms of public management include dividing policy formation from policy implementation, the creation of new independent regulatory agencies and experimenting withnew initiatives to fund public services.3 Technological change is one of the greatest stimulants for these demands, but also newrequirements (for example the provision of financial services to a wider sector of the popula-tion) as well the ability of the market to provide new services without the intervention (orfinancing) from public authority. Dissatisfaction with the bureaucratic handling of public services has also played an important role in persuading electorates of the ‘better’ service theprivate sector can supply. There is an amusing example of this culture in David Landes,

private enterprise is more efficient than state provision and may be moreresponsive to consumer demands.

This belief in the market to deliver public services complements goalsbeing pursued by the European Union, particularly the exercise of stricterfiscal discipline on public spending. Such goals nestle alongside a tendencyto see the liberalisation of the public sector as an unavoidable consequenceof the internal market.4 The exposure of public services to the competitivemarket through litigation using outdated, inadequate and inappropriatelegal tools to handle the complex questions has forced the EU to take agreater proactive and central role in organising an agenda for the moderni-sation of public services in the new economy.5

This agenda was stimulated by the internal market programme andfuelled by the increasing use of Article 86(1) EC by litigants at the nationallevel challenging the hegemony of many monopolists and incumbents in thenational markets. Tensions arise with national interests wishing to ‘defend’public services from erosion in the new economy. One response has beenthe inclusion of Article 16 EC and the inclusion of Article 36 on access toservices of general interest in the Charter of Fundamental Rights.6 This hascreated the need to establish how national concepts of public services canbe transferred into modern ideas in terms of definition, delivery and assess-ment of the efficiency of such services. The Commission has used soft lawprocesses to iron out differences of opinion on the role of public services inthe new economy. In December 2000, the European Council of Nice,approved a declaration developing the principles set out in Article 16 EC,inviting further study of the issues at the EU level. This was followed up inDecember 2001 by the European Council of Laeken, and Barcelona inMarch 2002. The Commission was invited to study the feasibility of a

186 Erica Szyszczak

The Wealth and Poverty of Nations (London, Abacus, 1998) 306: ‘The meanness of the FrenchPost Office was notorious. Until the 1990s, airmail letters … paid a surcharge above a weightof 5 grams, stamps included. That meant using specially thin and pricey paper—a boon to thestationery industry. Even so the PO would not have a single stamp for the postage requiredand would combine 2 or 3 [stamps] to make the amount, and these would tip the scale. Onehad to experience these exercises in petty tyranny to understand the retardive effects of bureau-cratic constipation. Fortunately for the French, the European Community has imposed newstandards.’

4 W Devroe, ‘Privatisations and Community Law: Neutrality versus Policy’ (1997) 34Common Market Law Review 268. 5 Two Communications (1996 and 2000) set out the position of services of general economicinterest. A Report, (2001) was followed by a Communication of June 2002 on the methodol-ogy of evaluation of services of general interest. These can be accessed at: �http://europa.eu.int/comm/competition/liberalization/legislation/� (visited 4 June 2002).6 Art 36 CFR reads: ‘Access to services of general interest. The Union recognises and respectsaccess to services of general economic interest as provided for by national laws and practices,in accordance with the Treaty establishing the European Community, in order to promote thesocial and territorial cohesion of the Union.’

Framework Directive, providing a clearer legislative definition of servicesof general interest by December 2002. The political process was jeopardisedby increasing litigation at the national level challenging the behaviour offormer public service monopolists and by uncertainty in the EuropeanCourt of Justice (the Court) case law on the role of public financing forpublic services.

The Commission prepared a Green Paper in March 20037 which wentbeyond the initial remit of looking at the feasibility of a FrameworkDirective to exploring the way public services are to be provided and regu-lated in the new economy.8 This is double-edged. On the one hand theCommission is attempting to reassure the Member States and nationalinterests that the new economy recognises the role of public services, but onthe other, many of the policies pursued by the EU towards the new econ-omy intrude upon the Member States’ autonomy to set the national agendafor the delivery of public services.

2. THE POLITICAL ECONOMY OF STATE INTERVENTION

Article 295 EC states that ‘The Treaty shall in no way prejudice the rulesgoverning the system of property ownership.’ Article 295 EC is interpretedas being neutral or agnostic about the role of state intervention in the econ-omy, leaving the choice to the Member States over whether to use publicresources or private markets to underpin economic growth. But in responseto the fears by the Member States of the ever-creeping competence ofCommunity activity the Court has not been sympathetic. Until the recentgolden shares cases, the scope of Article 295 EC had not received a defini-tive interpretation by the European Courts. Although Advocate-GeneralRuiz-Jarabo Colomer saw Article 295 EC as a limitation on Communitycompetence, the Court, in contrast, has not allowed Article 295 EC to beused by Member States to remove areas of economic organisation andactivity from the reach of Community law.9

A limitation on the scope of Article 295 EC is hidden in the State Aidrules. The wording of Article 87(1) EC is wide, prohibiting virtually allstate activity which affects trade between the Member States. There are

Public Services in the New Economy 187

7 COM (2003) 270 final.8 The Green Paper proposes 30 questions with a reply deadline of 15 September 2003. Therange of issues covered includes competence issues as well as external aspects involving theWTO, the place of services of general interest in the policies of co-operation with the South,the relationship between sector based rules and general rules.9 Case C–367/98 Commission v Portugal, Case C–483/99 Commission v France, CaseC–503/99 Commission v Belgium [2002] ECR I–4731; Cases C–463/00 and C–98/01Commission v Spain and Commission v United Kingdom judgment of 13 May 2003. The casesare discussed later.

exceptions to the State Aid rules10 and the use of the private-investor principle provides a buffer to Article 87(1) EC11 but there is no specific ref-erence to shield public services.

In addition, the idea that there is neutrality over whether to use public orprivate systems of ownership to deliver goods and services may be illusorywhen we consider the implications of the goals of the March 2000 LisbonSummit which committed the EU to becoming the most competitive,dynamic, knowledge based society in the world by 2010.12 Article 4 ECinstructs the Member States and the Community to conduct their economicaffairs ‘in accordance with the principle of an open market economy withfree competition’.

This goal is reinforced by Article 98 EC.13 In the hierarchy of treatynorms it is assumed that principles established in the early Articles of theTreaty will take precedence over later provisions.14 The important positionof Article 4 EC allows free market principles to dominate all policies of theEU. But since the Treaty of Amsterdam and the evolution of the LisbonProcess the Court is aware of the political need to realign the balancebetween economic and social goals. It has signalled that the political natureof the substance of the open market economy is an area where the Courtwould prefer the lead to be taken by the institutional decision-making struc-ture, not the Courts. The Court stated that Articles 4 and 98 EC establish a‘general principle whose application calls for complex economic assessmentswhich are a matter for the legislature or the national administration’.15

This interpretation of the hierarchy of norms in the Community suggest-ing a continuing imbalance between the logic of competition and the objec-tives of public services may be resolved in the new EU Constitution whereservices of general interest were discussed in the Convention on the Futureof Europe.16 In the draft constitution, Article II-52 reproduces Article 36

188 Erica Szyszczak

10 These are to be interpreted narrowly: C–730/79 Philip Morris v Commission [1980] ECR–2263.Note also Reg 69/2001 establishes a de minimis principle, set at e100,000 granted over athree-year period. The Commission has used soft law processes to regulate this highly politicalarea of state regulation.11 See the critique by M Parish, ‘On the Private Investor Principle’ (2003) 28 European LawReview 70.12 Known as the ‘Lisbon Process’.13 Art 98 EC reads: ‘Member States shall conduct their economic policies with a view to con-tributing to the achievement of the objectives of the Community, as defined in Article 2, and inthe context of the broad guidelines referred to in Article 99(2). The Member States and theCommunity shall act in accordance with the principle of an open market economy with freecompetition, favouring an efficient allocation of resources, and in compliance with the princi-ples set out in Article 4.’14 Cf the different view of Advocate-General Colomer in the golden shares cases above n 9. He argues that Art 295 EC should be attributed a higher constitutional status because of itsposition in the final and general provisions of the treaty which affect all treaty rules.15Case C–9/99 Echirolles Distribution SA du Dauphiné and Others [2000] ECR I–8207, para 25.16 The discussion of services of general interest was focused on in the Working Group XI onSocial Europe where there was disagreement as to how to regulate such services at the EUlevel.

CHR and III-55 addresses public undertakings and undertakings with special or exclusive rights. The current Article 16 EC has been changed toread:

Without prejudice to Articles III-55, III-56 and III-136, and given the placeoccupied by services of general economic interest as services to which all inthe Union attribute value as well as their role in promoting social and territo-rial cohesion, the Union and the member states, each within their respectivepowers and within the scope of the application of the Constitution, shall takecare that such services operate on the basis of principles and conditions, inparticular economic and financial, which enable them to fulfil their missions.European laws shall define these principles and conditions.

This provision (Article III-6) is placed under the heading ‘Clauses ofGeneral Application’ and is a significant change of wording allowing forgreater EU-level involvement in the delivery of public services.

3. CHALLENGES TO THE PROVISION OF PUBLIC SERVICES

The use of state intervention to provide public services, especially throughthe use of public monopolies or companies granted special or exclusiverights was viewed in the post-war period as a necessary response to the dev-astation reaped upon the European economy. It was seen as a necessarystrategy if Europe was to match the United States’ economic dominance.The pervasiveness of state intervention in the economy through public com-panies and monopolies is seen well into the 1990s until Article 86 EC cameinto prominence,17 followed by increased scrutiny of public finances underthe State Aid rules.

Article 86 EC permits the use of public monopolies but subjects themonopolies to the law of the market unless the Member State can showthat the application of the free market rules would hinder the provision of aservice of general economic interest. This justification is found in Article 86(2)EC and is derogation to the fundamental free market principles. This nar-row derogation from the free market rules is the legal vehicle whereMember States can justify anticompetitive behaviour: it is the lynchpinupon which public services can survive in competitive markets.18


17 Eg the role of state intervention was questioned where the national monopolies were ineffi-cient or not receptive to the market: Case C–41/90 Höfner [1991] ECR I–2017; CommissionDecision (EEC) 90/16 Courier Postal Services—The Netherlands [1990] OJ L/10/50;Commission Decision (EEC) 90/456 Courier Postal Services—Spain [1990] OJ L/233/22;where there were allegations of failure to move with technological change Case C–179/90 Portof Genoa [1991] ECR I–5928.18 See A Winterstein, ‘Nailing the Jellyfish: Social Security and Competition Law’ (1999)European Competition Law Review 324.

Two examples reveal early thinking by the Court in its use of Article 86 ECto control state intervention in the market. Höfner and Elser v Macrotron19

is a decisive turning point in the attitude towards state intervention in themarket. The Court suggests that in certain situations by merely creating amonopoly in certain kinds of markets the state is infringing competitionlaw.20

Another example of the Court’s willingness to monitor the necessity fora state monopoly and to scrutinise the behaviour of the monopoly is seen inRTT.21 This is an example where the Belgian monopolist RTT held amonopoly in the telecommunications network as well as the monopoly inthe ancillary market in the equipment used for interconnection to the net-work (telephones). This monopoly was attacked by GB-Inno, a large super-market which engaged in wide-ranging litigation strategy, not only againststate monopolies, but also against state laws and regulations which italleged were anticompetitive and restrictive of trade. GB-Inno had importedand sold telephones in its supermarkets without RTT’s approval. This wasa conflict of interest case since RTT also sold telephones and indeed had themonopoly in deciding which telephony equipment could be connected to itsnetwork. GB-Inno raised the argument that RTT’s behaviour was contraryto the free movement of goods and competition rules. The Court chose todeal with the case under Article 86 EC. The Court found an abuse of adominant position. As a result of the Belgian legislation RTT had been ableto extend the dominant position it had on the network for providingtelecommunications into the ancillary market of providing and approvingequipment which could be used. There was no justification as to why it wasnecessary to extend the monopoly in this way.

This sort of litigation provided the impetus for a European-wide agendafor the liberalisation of a number of sectors which were dominated bynational champions: transport, telecommunications, broadcasting, postalservices and the utility sectors of electricity and gas. But the Member Stateswere not willing to relinquish total control of national monopolies to themarket; most Member States opted for a staged and/or partial liberalisationof their sensitive monopolies. The case law on Article 86 EC is inconsistent,but a more fundamental criticism can be raised at the political institutionalstructure of the EU which relies upon individual litigation creating a casuis-tic approach to policy formation in such a fundamental area of traditionalstate activity.

190 Erica Szyszczak

19 Case C–41/90 [1991] ECR I–2017, para 2. See also Case C–260/89 ERT [1991] ECRI–2962, para 37.20 See L Gyselen, ‘Commentary on the Port of Genoa and RTT Cases’ (1992) 29 CommonMarket Law Review 1239 who argues that Höfner and ERT are confined to the particular sit-uation and that the mere existence of exclusive rights does not necessarily lead to a breach ofthe competition rules.21 Case C–18/88 [1991] ECR I–5981.

4. THE CHALLENGE TO STATE AUTONOMY

The Court and the Commission continue to assert that Member Statesretain the freedom to define public services. Of crucial importance for thecontinuance of state intervention in the market is the boundary betweennational competence, where Member States can protect public services fromcompetition, and the reach of the market through Community law compe-tence. Recent case law on the privatisation of state assets shows that evenin the initial choice of how to provide public services, the autonomy of theMember States is restrained.22 Subjecting state monopolies to the competi-tive market took place alongside the most significant political and economicchange in the restructuring of the state and the market in the latter part ofthe 20th century: the sale of state assets to the private sector. How muchfreedom does a Member State of the EU have in choosing how to privatisepublic companies? In particular, may the state retain a share in a privatisedcompany: the use of golden shares?

Until it was recognised that proper regulation is a better way of main-taining competition, state shares in privatised companies were a usefulinsurance against anticompetitive activity by the privatised company.Golden shares may also be justified as a temporary measure to allow man-agement to adjust to private sector activity or to protect a sensitive sectorfrom take-overs by foreign companies, especially where such companies arein a dominant position as national champions, for example, in the electric-ity and postal services sector.

As a consequence of the liberalisation process, some Member Statesencouraged the national champion, not only to continue to dominate thenational sector, but also, to increase dominant positions and expand intoother markets. Golden shares are used frequently to justify the state’s tradi-tional duty to provide public services and, more generally, to intervene inthe economy where the public interest so dictates. Thus, golden shares havebecome a key tactic in the liberalisation and privatisation programmes of anumber of the Member States and continue to be used in the new accessionstates.

These tactics raise a basic constitutional issue for EU law: where to drawthe boundaries between the sovereign rights of the Member States to choosehow to allocate the ownership of property in their territory and the scopeof the EU. The Court23 brought the use of golden shares within Community


22 Above n 9.23 The Commission questioned the use of golden shares throughout the 1990s. See theCommunication of the Commission on Certain Legal Aspects Concerning Intra-EUInvestment [1997] OJ C/220/15. The view of the Commission was that the free movement ofestablishment rules were infringed by the use of golden shares. In the cases the Court restrictsits analysis to the free movement of capital.

law scrutiny, not wanting to leave such an important area of state activityoutside of the control of Community law. Arguing, without much evidenceto show how the use of golden shares is a restriction upon the free move-ment of capital, the Court declared that the use of golden shares is liable toact as a deterrent to investors from other Member States from making suchinvestments.24

Italy conceded the case that special powers reserved for the state andpublic bodies in a number of public service sectors, including defence, trans-port, telecommunications and energy resources, was contrary to the ECTreaty provisions on capital, movement and establishment.25 In later casesthe Court reinforced its view that restrictions on the four fundamental eco-nomic freedoms could not be justified for economic purposes, but concededthat golden shares could be justified provided certain procedural guaran-tees were implemented.

What motivates the Court therefore, is not so much restricting the use ofgolden shares in privatisations but ensuring that other economic actors aregiven the opportunity to participate in privatizations of strategic undertak-ings. The Court’s approach is to employ concepts of good governance tostate intervention in the market.

Member States have used the idea of defending core public services fromthe full thrust of competition by accepting that there are certain universalservices which should be protected. The Europeanisation of ideas of univer-sal service obligations has generated a new dynamic to the concept of pub-lic services and opened up an analysis of the need to scrutinise the deliveryof universal service obligations. The concept has also allowed for the devel-opment of universal service obligations. Such an obligation can be carriedout by private actors, but some Member States have chosen to maintain anumber of core State monopolies on the market as incumbents. Theseincumbents have market power and pose problems for the new economy.Where incumbents have used universal service obligations to protect theirmarket position, there are allegations of anticompetitive behaviour becausethe incumbents can use their dominant positions, in terms of market andfinancial power, as well as illegal State Aid, to keep competitors out of theliberalised sectors within their own national territory. They may also use

192 Erica Szyszczak

24 Cf the comments of Advocate-General Colomer in Case C–98/01 Commission v Spain andCase C–463/00 Commission v United Kingdom (BAA), opinion of 6 February 2003:‘Contrary to the Court of Justice’s finding, … the resulting restriction of the free movement ofcapital is incidental, rather than inevitable. If that is the case as regards measures affecting thecomposition of the membership, it is even more true as regards measures restricting the adop-tion of company resolutions (change of company object, disposal of assets). In the latter cases,the link with the free movement of capital is hypothetical or very tenuous’ para 36.25 Case C–58/99 [2000] ECR I–3811. See the criticism of the Court’s judgment by AG Colomerin the later cases, above n 9, where he argues that it was inappropriate for the Court to decidesignificant novel issues of Community law on the basis of a concession by a Member State.

that market strength to invade other national territories where the sectorhas been opened up to competition. The notable examples are the aggres-sive policies of the German Post Office, Deutsche Post, and the Frenchpostal service and electricity monopoly, EDF.

The Courts have been generous when asked to appraise the financing ofpublic services in the light of the State Aid rules. In FFSA26 the CFIdeclared that State Aid to a public service provider could be justifiedunder Article 86(2) EC. However, this ruling did not settle the matter ofwhether such aid should be notified in the usual way under the State Aidprocedures. The Court in CELF II27 confirmed that they should. Not onlywould this subject the aid to Commission investigation, but the standstillarrangements28 would also come into play until the aid had been cleared.In Ferring SA29 the Court took a different line, accepting that where aid isgiven to offset the additional costs incurred by providing public services,this was compensation for providing the service and not an economicadvantage within the meaning of Article 87(1) EC. This approach chal-lenged the political commitment by the Member States to monitor closelythe use of State Aid generally within the Community.30 It challenged theCommission’s growing hegemony in the area of controlling state interven-tion in the market, generating a number of soft law communications, aswell as new forms of monitoring such as the State Aid Score Board.31

The Court revisited the issue in Altmark.32 While the Court confirmedthe compensation approach of Ferring, it attached a number of conditionsto the receipt of the compensation in order to secure compatibility withCommunity law. First, the recipient undertaking must actually have publicservice obligations to discharge and those obligations must be clearlydefined. Second, the basis on which the compensation is calculated must beestablished in advance in an objective and transparent manner. Third, the compensation cannot exceed what is necessary to cover all or part ofthe costs incurred in the discharge of public service obligations, taking intoaccount the relevant receipts and a reasonable profit. Finally, where theundertaking is not chosen in a public procurement procedure, the level of


26 Case T–106/95 [1997] ECR II–229.27 Case 332/98 Commission v France [2000] ECR I–4833.28 Art 88(3) EC.29 Case C–53/00 [2001] ECR I–9067.30 Ninth Survey on State Aid in the EU COM (2001) 403; Stockholm European Council,March 2001 SN 100/01 point 20 and 21; Report to the Seville European Council on the statusof work on the guidelines for state aid and services of general economic interest; Report fromthe Commission on the State of Play in the work on the Guidelines for State Aid and Servicesof General Economic Interest (SGEI) progress report concerning the reduction and reorienta-tion of state aid. Available at �http//:europa.eu.int/competition/state_aid/others/�.31 Available at �http://europa.int/comm/competition/state_aid/scoreboard/� (visited 4 June 2002).32 Case C–280/00 Altmark Trans GmbH and Regierungspräsidium Magdeburg v Nahverkehr-sgesellschaft Altmark GmbH, judgment of 24 July 2003.

compensation must be determined by a comparison between an analysisof the costs which a typical undertaking in this sector would incur, withthe amounts received and a reasonable profit from discharging the obli-gations. All four conditions must be met for an undertaking to show thatit has not enjoyed a real financial advantage which could be considered aState Aid.

5. HYBRIDISATION

Partial liberalisation and privatisation of state activity has resulted in anexperimental approach, using a mixture of public/private legal and finan-cial structures and initiatives to create a set of new hybrid institutions.Member States argue that this new form of economic activity is not, andshould not be subject to the law of the market. The Court has not provideda helpful solution to clearly delineating between state activity beyond thereach of the law of the market and state economic activity which should besubject to the market rules. A number of legal techniques have been used toclose off the sphere of state activity immune from the market33 but the caselaw and methodological approach used is not consistent.

In Höfner34 the Court took a functional approach, arguing that the legalstatus and financing of a particular activity are not conclusive to the statusof the activity in Community law. The decisive criterion was that the activ-ity was capable of being provided by private market actors. The limitationof this test is seen if we observe the experimentation with privatisation inEurope. This has altered the frontiers between state/market activity; elec-torates have accepted that a number of public services can be provided bythe market in the moves towards private provision of health, education,maintenance of roads, social security.35 Even the coercive powers of thestate, for example, the control over immigrants and prison management,

194 Erica Szyszczak

33 These ideas are discussed in E Szyszczak, ‘State Intervention in the Market’ in D O’Keeffeand T Tridimas (eds), EU Law for the 21st Century: Rethinking the New Legal Order(Oxford, Hart Publishing, 2004).34 Case C–41/90 [1991] ECR I–1979.35 To take one example, the classification of health care as an economic service. The issue hasbeen discussed in the context of the free movement provisions where overriding public interestjustifications provide the means to allow the Member States to experiment with new forms ofmedical services provision but also allow for the scrutiny of the compatibility of such schemeswith the objectives of the internal market: see Case C–158/96 Raymond Kohll v Union descaisses de maladie [1998] ECR I–1931; Case C–120/95 Nicolas Decker v Caisse de maladiedes employés privés [1998] ECR I–1831; Case C–157/99 BSM Geraets-Smits and HTMPeerbooms v Stichting Ziekenfonds VGZ and Stichting CZ Groep Zorgverzekeringen [2001]ECR I–5473; Case C–368/98 Abdon Vanbraekel and others v Alliance nationale des mutualitéschretiennes (ANNMC) [2001] ECR I–5363; Case C–385/99 VG Muller Fauré v OnderlingeWaarborgmaatschappij OZ Zorgverzekeringen UA and EEM van Riet v OnderlingeWaarborgmaatschappij OZ Zorgverzekeringen, judgment of 15 May 2003.

have been handed over to private actors. The use of the market and newactors, private bodies, charities, hybrid bodies to provide goods and serv-ices previously delivered by the state has brought a new dimension to thelitigation. Courts are now faced with the issue of how far public service justifications can be extended to private actors who are involved in supplyingand regulating public services.

Wouters36 is an example of a global tendency to attack restrictive practicesof regulatory bodies in the service sector. This was a reference challengingthe Netherlands’ Bar prohibition on multidisciplinary partnerships (MDPs)between lawyers and accountants. The Court ruled that The Netherlands’Bar was an association of undertakings and subject to Article 81(1) EC.The Court found that the regulation of the legal profession was a decisionof ‘an association of undertakings’ which restricted competition and wastechnically caught by Article 81 EC since the rules relating to the prohibi-tion of MDPs regulated an economic activity. The prohibition on MDPswas liable to limit production and technical development and affect compe-tition. The Bar was not acting as a public authority because it was acting ina purely regulatory capacity for its own profession; it was not exercisingtypical public authority powers. The Bar’s governing bodies were electedwithout government intervention and it was not obliged to act in the publicinterest. The Court went on to find that the restrictive effects of the regula-tion did not go beyond what was necessary in order to ensure the properpractice of the legal profession.

The case reveals the lack of attention paid in competition law to theexemptions and justifications which may be raised by the state and dele-gated bodies to provide public services. Advocate-General Léger pointedout that any justification for the infringement of the competition rulesunder Article 86(2) EC would come up against the problem that an absoluteprohibition on MDPs would not satisfy the proportionality requirement.The Advocate-General was willing to conclude that Article 81(1) EC didnot apply by virtue of Article 86(2) EC: lawyers performed services essen-tial to a Member State governed by the rule of law. MDPs threatenedlawyers’ independence and thus the prohibition of these forms of partner-ships could be justified under the competition law rules. However, theCourt ruled that The Netherlands’ Bar was not an undertaking within themeaning of Article 82 EC, as it did not carry out an economic activity. Norcould the Bar be considered to be a group of undertakings for the purposesof Article 82 as its members were not linked to constitute a group. Themembers were not linked to constitute a group as there were insufficientties to enable the members of the Bar to adopt the same conduct on themarket and eliminate competition between themselves. The Court ruledthat whilst the prohibition on MDPs was in principle contrary to the


36 Case C–309/99 [2002] ECR I–1577.

freedom of establishment (Article 42 EC) and the freedom to provide services(Article 49 EC), the restriction on trade, as with the restriction on compe-tition, was justified. But the Court found that the restriction on trade, aswith the restriction on competition, was justified as being a necessary condition for the practice of the legal profession. Consequently, it readsacross the public interest justification which may be used in relation to non-discriminatory rules found in the free movement provisions of Article 49 EC37

into the competition rules.The public financing of health care, social protection schemes and pen-

sions has come under pressure in the new economy. The political process ofmodernisation has used the open method of co-ordination to force theMember States to rethink the ways in which social protection (and pen-sions especially) are provided.38 This political process has been pre-emptedby litigation. Here the Court has used a new tool, a concept of ‘solidarity’to organize its thinking on where to draw the line between state autonomyand market law, as well as utilising the flexibility within the concept of ‘sol-idarity’ to allow for justifications and exemptions from the application ofthe market rules. An example of the latter is Albany.39 Pension schemesinvolve a balance of public and private sector provision and the EU isincreasingly looking to involve private parties in pension provision. TheNetherlands used the social partners (employer and employee representa-tives) to establish sectoral pension schemes. A number of employers in thetextile sector objected to the compulsory nature of these sectoral schemes,arguing that if pension provision was being subject to the market then theyshould be free to opt out of the sectoral scheme and choose their own pen-sion provider, largely because the sectoral pension funds were too costly.

An earlier case, Poucet and Pistre,40 saw the successful use of the con-cept to defend national social insurance schemes against attacks from EClaw. The concept of solidarity is used where there is no apparent inter-stateelement. The rationale seems purely to protect national social insuranceschemes from the application of the rules of the market. The Court inPoucet and Pistre argued that there was no economic activity involved totrigger Community law:

13. It follows that the social security schemes … are based on a system of com-pulsory contribution, which is indispensable for the application of the principleof solidarity and the financial equilibrium of those schemes.

196 Erica Szyszczak

37 Case C–3/95 Reisburo Broede [1996] ECR I–6511.38 Objectives and Working Methods in the Area of Pensions: Applying the Open Method ofCo-ordination (Joint Report of the Social Protection Committee and the Economic PolicyCommittee, November 2001) available at: �http://europa.eu.int/comm/employment_social/soc-prot/social/index_en.htm� (visited 4 June 2002).39 Case C–67/96 Albany International Stichting Bedrijfspensioenfonds [1999] ECR I–5751.40 Joined Cases C–159/91 and C–160/91 Christian Poucet and Daniel Pistre v AssurancesGénérales de France and Caisse Mutuelle Régionale du Languedoc-Roussillon [1993] ECR I–637.

18. … [O]rganisations involved in the management of the public socialsecurity system fulfil an exclusively social function. That activity is basedon the principle of national solidarity and is entirely non-profit-making.The benefits paid are statutory benefits bearing no relation to the amountof the contribution.19. Accordingly, that activity is not an economic activity.41

‘Solidarity’ is therefore used to delineate clearly between state autonomyand EC law.

In Albany ‘solidarity’ is used in a different way. The Court was willing tofind that the agreements between the social partners establishing the pen-sion schemes were immune from the competition rules but the pensionfunds themselves were economic in character. A monopoly with exclusiverights was created and so Article 86 EC could apply. The Court found thatbecause the funds were obliged to offer pensions on the basis of the solidar-ity principle, meaning here that individuals were not fully risk-rated, itmade them less competitive than the comparable services offered by insur-ance companies. This justified a finding of a public service to be protected.A factor which seems to have influenced the Court is that the sectoral pen-sion schemes displayed many features of Dutch social democracy and thisraises a question of how far private providers of public services must mirrorthe State provision of public services which they are either supplementingor supplanting in order to fall within the definition of a public service.

Pavlov42 is an example of another route to avoid the full thrust of thefree market rules to a set of economic actors providing a social function.This time medical practioners objected to being made compulsory membersof the supplementary medical pension scheme established in The Netherlandsunder Dutch law. The Court found the activity was an economic activityand therefore the bodies established to administer the pension scheme andthe decisions taken were subject to the competition rules of the EC Treaty.But the Court took the view that the effect of these agreements on the mar-ket was minimal and applied a de minimis rule, finding also that there wasno abuse of a dominant position under Article 82 EC.


41 See also Case T–106/95 FFSA and Others v Commission [1997] ECR II–229; CaseC–218/00 Cisal di Battistello Venanzio & C Sas v Istituto Nazionale per L’assicurazione con-tro gliIinfortuni sul Lavoro (INAIL) [2002] 691. For a free movement case see Case C–70/95Sodemare SA and others v Regione Lombardia [1997] ECR I–3395. At para 29 the Courtstates: ‘It is clear from the documents before the Court that that system of social welfare,whose implementation is in principle entrusted to the public authorities, is based on the princi-ple of solidarity, as reflected by the fact that it is designed as a matter of priority to assist thosewho are in a state of need owing to insufficient family income, total or partial lack of inde-pendence or the risk of being marginalized, and only then, within the limits imposed by thecapacity of the establishments and resources available, to assist other persons who are, how-ever, required to bear the costs thereof, to an extent commensurate with their financial means,in accordance with scales determined by reference to family income.’42 Case C–180–184/98 Pavlov v Stichting Pensioenfonds Medische Specialisten [2000] ECRI–6451.

Recently Advocate-General Jacobs argued that the competition rulesapplied to the German statutory health insurance scheme which inter aliafixes the price of certain medicines.43 Under German law the majority ofemployees must belong to a statutory health insurance system, unless theirincome exceeds a certain level. The system is funded by compulsory contri-butions from the insured persons and the employers. The insurance fundsare required to purchase medical services and supplies and supply these tothe insured persons. However, certain products have maximum fixed pricesand where the product exceeds this maximum price the insured person mustbear the remainder of the cost.

The fixed amounts are decided by a two-stage process. The first stageinvolves a committee deciding which types of products are subject to thefixed amount. This committee is composed of representatives of the leadingsickness fund associations and associations of doctors. The choices madeare then approved by the Ministry of Health. At the second stage, the asso-ciations of sickness funds determine the fixed amounts following criteria setout in law. Once set, the fixed amounts are subject to annual review andmust be adapted to reflect changes in the market. The fixed amounts mustbe published and are open to challenge before the courts.

A number of pharmaceutical companies challenged the decisions of theleading association of sickness funds in Germany fixing the price of thecompanies’ products. The Advocate-General argued that this was an eco-nomic activity. There was a degree of competition between the sicknessfunds inter se, and between the sickness funds and private insurers demonstrating that the economic activity could be carried out for profit.The fixing of the price for medicines also fell within the sphere of economicactivity, as a sickness fund’s decision regarding the parameters of the serv-ices to be offered is indissociable from the core activity of the provision ofhealth insurance.

The Advocate-General considered that the second stage of price fixingactivity, when the association of sickness funds together determine the fixedamounts, could be considered to be an act of the associations: they actindependently of the ministry, there is no requirement to obtain priorapproval, the decision making body is composed of the appellants’ repre-sentatives and the applicable criteria are insufficiently distinct from theappellant’s interest in setting the fixed amounts at a low level. Advocate-General Jacobs found this behaviour to be contrary to competition law,having the object and effect of restricting competition, behaviour expresslyidentified in the EC Treaty as an anticompetitive practice.

The issue in this case is whether the undertakings were acting independ-ently on their own initiative. If this is the case then Article 81 EC applies.

198 Erica Szyszczak

43 Joined Cases C–264/01, C–306/01, C–354/01 and C–355/01 AOK Bundesverband and others v Ichthyol-Gesellschaft Cordes and Others, opinion of 22 May 2004.

But if German law required the conduct, then the competition rules wouldnot apply. This is a question of fact for national courts to decide. TheAdvocate-General suggested that the undertakings were unable to avoid fix-ing an amount and were not entirely free to choose the fixed amount becauseof certain rules such as the requirement to determine prices on the basis ofthe lowest price of comparator groups. In order to make competition lawbite, the national courts should look to see if the undertakings had used anyremaining discretion to create an appreciably greater restriction on competi-tion than would have resulted from another permissible decision. If thisquestion was answered in the affirmative, the Advocate-General argued thatthe defence found in Article 86(2) EC was a possible justification.

The case law reveals that the transition towards the liberalisation ofmarkets for public services and the consequent reliance upon private mar-kets to provide an increasing number of public goods and services has notbeen smooth and is far from complete. The existing competition law toolsof the European Union were drafted in a different age and seem ill-definedto manage the complex questions raised by the new economy, both wherethe state acts in a competitive manner and where non-state actors are calledupon to supply publics services. Public interest defences are available underthe core four economic freedom rules and cases such as Wouters suggestthat such public interest defences need to be considered in relation to com-petition law. Article 86(2) EC has been stretched to cover a number of publicinterest situations under Article 86 EC and the State Aid rules and it couldprovide a model for a general public interest defence in the development ofthe new economy.44 But Article 86(2) EC is only available in limited situa-tions: where there is a public undertaking, or an undertaking granted spe-cial or exclusive rights performing a service of general economic interest.The Member State must be able to show why the application of the freemarket rules will obstruct the performance of the tasks assigned to theundertaking and trade must not be affected to such an extent as would becontrary to the interests of the Community. The concept of proportionalityplays an important role in applying these criteria.45 With liberalisation andthe privatisation of many state functions notions of public services arechanging and a wider idea of services (and goods) which should be pro-tected from the full thrust of the market is necessary.

6. THE PROVISION OF PUBLIC SERVICES BY NON-STATE ACTORS

The provision of public services by non-state actors in the new economyraises many questions. How far, and on what terms can we ask private


44 A Gagliardi, ‘United States and European Anti-trust Versus State Regulation of theEconomy: Is There a Better Test?’ (2000) 25 European Law Review 353.45 See J Jans, ‘Proportionality Re-visited’ (2000) 27 Legal Issues of European Integration 239.

companies to provide public services? And, if we do so, should we call them‘public services’? In addition, what are the principles of competition law tobe applied where public services are provided in competitive markets?

Governments are increasingly using competitive markets to deliver newkinds of public services, for example, the use of Internet access to connectremote geographical areas or as a means of implementing a regional, socialand economic cohesion policy, or as part of an education policy. Thesedevelopments are complicated by the tendency to link the delivery of serv-ices through competitive markets to notions of citizenship and as amedium for delivering social justice. Consumers of public services nowhave different and higher expectations of services which can be suppliedby the market.

The response to this dilemma from the European Union has been touse a mixture of regulation and competition tools to protect the provisionof sensitive public services in the liberalised sectors. Regulation was ini-tially portrayed as a conduit to facilitate competition. It envisaged thatwhen a competitive market was established, regulation should give wayto competition and that the work of the special regulatory bodies couldbe transferred to general competition principles and authorities.Consequently, in the liberalisation process we see a complementary mix-ture where competition and regulation tools are used often within thesame liberalising directive.

One way in which public services have been ring fenced in the new econ-omy is the use of the idea of universal service obligations. Member Statesare given some flexibility as to how these are to be delivered and how suchservices are financed. In some sectors, the notion of universal service obli-gations creates a new dynamic to the traditional delivery of core services.46

Rather than being a temporary idea, universal service obligations haveestablished a permanent place and are an example of the synergy betweencompetition principles and regulation principles in the new economy. Whatis emerging in the telecommunications sector, the most advanced area ofliberalisation in the EU, is that this combination of regulation and competi-tion principles may be needed for a fairly lengthy transition period toaccommodate the new market dynamics of the liberalised sectors. Althoughcompetition principles are being used—in regulatory form and in theEuropean Courts’ case law—a number of new ideas and concepts areemerging in the principles applicable to the new economy.47

200 Erica Szyszczak

46 For example, see VV Comandini, ‘The Provision and Funding of Universal ServiceObligations in a Liberalised Environment’ and S Rodrigues, ‘The French Postal Sector: OldMissions, New Challenges’ in D Geradin (ed), The Liberalisation of Postal Services in theEuropean Union (The Hague, Kluwer, 2002).47 See P Larouche, Competition Law and Regulation in European Telecommunications(Oxford, Hart Publishing, 2000); S Baker and J Dodgson, ‘Market Definition in Postal Services’and J Derenne and C Stockford, ‘Abuse of Market Power in Postal Services: Lessons from theCommission’s Decisional Practice and Court of Justice Case Law’ in Geradin above n 46.

An example of new competition principles emerging is seen in theCommission Decision in Deutsche Post.48 Litigation against postal serviceincumbents has taken place at the national level in the absence of theCommission’s reluctance to respond to the complaints from competitorsalleging the use of illegal State Aid through cross-subsidisation betweenuniversal service obligation finances and liberalised services.49 DeutschePost has been the focus of long running complaints by UPS alleging thecross-subsidisation of the parcel service by illegal State Aid.50 A landmarkin this litigation is a decision adopted by the Commission in March 2001against the German incumbent Deutsche Post. UPS, like a number ofAmerican companies engaging in competition litigation in Europe, tried toimport an American law remedy of divestiture to break up the DeutschePost monopoly in order to split the market activities of Deutsche Post fromthe universal service obligation. Such a remedy does not exist as such in EUcompetition law. The Commission has power to adopt Directives underArticle 86 (3) EC and these could be used in a normative way against aMember State. Under the general competition rules, the Commission canmerely issue cease and desist orders and fine companies breaking the competition law rules, but cannot order divestiture.

UPS’s complaint was that while it held a substantial share in ‘Business-Business’ parcels market, it alleged that it was unable to carry over this suc-cess to the ‘mail order parcel services’ market because Deutsche Poste wasusing cross-subsidies from profitable letter mail where it held the monopolyto supply the universal service. UPS argued successfully that Deutsche Postewas using the cross-subsidy to finance a strategy of below cost selling inbusiness parcel services which had been opened up to competition: a formof predatory pricing.

The decision is significant since it breaks new ground. The Commissionestablished that the use of a cross-subsidy which results in a pricing strategythat does not cover the additional or incremental costs of branching outinto the commercial sector of parcels delivery is a form of predatory pricingwhich it calls the ‘economic costs concept’. Without the cross-subsidy


48 Commission Decision 2001/354/EC of 20 March 2001, [2001] OJ L/125/27. See alsoCommission Decision 2002/180/EC Hays/La Poste [2002] OJ L/61/32. Here the Commissionfound that the French incumbent in breach of Art 82 EC where the reserved sector (uso) usedpersuasive methods to force business-to-business customers for Hays’ DX service to change toLa Poste’s DX service by threatening to withdraw preferential postal rates offered to such cus-tomers in the separate (reserved) business-to-consumer postal market.49 See generally Geradin above n 46.50 UPS has also brought a complaint against the Commission for failure to act: Case T–175/99UPS Europe v Commission [2002] ECR II–1915; Case T–253/01 pending. The cross-subsidisation of the parcels’ sector through uso resources was tackled as a State Aid by theCommission in June 2002. The original complaint by UPS was made in 1994. See Press ReleaseIP/02/890 ‘Deutsche Post must repay Euro 572 million used to subsidise price undercutting incommercial parcel services’, 19 June 2002.

Deutsche Post would not have been able to stay in business for long bycharging such low prices. Deutsche Post was not fined for this abuse, how-ever, as the Commission admitted that the new test of ‘economic cost con-cepts’ was not sufficiently developed at the time the abuse occurred. But inorder to comply with the requirement of transparency in financial arrange-ments and also to give legal certainty to competitors in the parcel deliverymarket Deutsche Post agreed to a restructuring plan which created a newseparate company to supply business parcels services. The new companycould use the Deutsche Post infrastructure—sorting, delivery, transportservices—but must pay for services at market prices.51

The ripple effect of this decision spread across the Atlantic since it alsoallowed UPS to ask for the revocation of a foreign air freight forwarderlicence which had been granted to DHL World Wide Express, which isowned and controlled by Deutsche Post, and which would have allowedDHL to establish an inter-state package delivery service in the US in compe-tition with UPS.52

A. Economic Regulation

The analytical framework which evolved is one described in a recent speechby Commissioner Monti as one of ‘economic regulation’.53 Economic regu-lation is based on the perspective that intervention in the market is neces-sary and beneficial only when it addresses certain kinds of market powerand, in particular, market failures which derive from formerly monopolisticmarket structures. But Commissioner Monti also reveals that theCommission is aiming towards the elimination of sector-specific regulation

the same set of tools, the same competition-based philosophy and the sameconcerns may soon govern regulatory intervention in all sectors where someform of economic regulation can still be useful.

202 Erica Szyszczak

51 Note also the application of Case 85/76 Hoffman La Roche v Commission [1979] ECR 461to fidelity rebates used by Deutsche Post: the rebates were not quantity rebates which areacceptable as part of a normal commercial strategy—but fidelity rebates. These foreclose mar-kets are a disincentive for the customer to ‘shop around’ for better service.52 Press Release, 20 March 2001: ‘EC Ruling Results in Deutsche Post Restructuring and Fine;UPS calls on DoT to Revoke D.P/DHL Licence’ available at: �http://www.pressroom.ups.com/pressreleases/archives/archive/0,1363,3862,00.html� (visited 4 June 2002). The language ofthe press release issued by UPS makes interesting analysis. Nowhere is UPS’ commercial inter-est mentioned. Appeals are made as to why the Commission’s Decision is for the good for thecompetitive market—the neutrality of the market—but what is also interesting is that UPSthanks EC Commissioner Monti, describing Deutsche Post purely in terms of a commercialcompany, while UPS takes it upon itself to explain why the Commission’s action is for thegood of America.53 ‘Competition and Regulation in the New Framework’ Public Workshop on The ‘ElectronicCommunications Consultation Mechanism’ provided for by Art 7 of the Framework Directive2002/21/EC, Brussels, 15 July 2003. Available at: �http://europa.eu.int/comm/competition/index_en.html� (visited 4 June 2002).

B. Market Power and Special Responsibility

An example of the synergy between competition law and regulation is seenin the way the Court handled demands for an ‘essential facilities’ doctrineto be imposed under general competition policy. The Court has beenresponsive to regulating companies with dominant positions which havepower to influence the market. In Michelin the Court signalled that privatefirms who acquire market power are regarded as having a ‘special responsi-bility’ which forbids the firm from abusing a dominant position.54 This ledAmato to comment:

Market power, just because it is conceptually accepted, is thus loaded withthe burdens and limits which, according to the general principles more ofpublic than of private law, bear upon whoever holds power.55

If we examine how the Court has held back from applying an essential facil-ities doctrine in mainstream competition law we see the Court does not nec-essarily view competitiomn law as the vehicle for steering the market inways which would impose new obligations upon dominant firms or inter-fere with vested property rights of dominant firms. An essential facilitiesdoctrine would be useful in monopoly situations where utilities are deliv-ered through networks or grids. The questioning of natural monopolies hasled to the understanding that even where a grid or network is an efficientway of delivering essential services, such services do not necessarily have tobe provided by a monopolist. There can be a competitive market in deliver-ing different goods and services and even competition in different parts ofthe network.

The Commission imported the idea of an essential facilities doctrine ina decision of 1993.56 The Commission deemed Holyhead port in NorthWales to be an essential facility. Holyhead was owned by Sealink whichwas also the main ferry operator from Holyhead to Dublin. The decisionobliged Sealink to make the port available to Sea Containers, a rival ferryoperator who had argued that there were no effective substitutes to createa competitive market for ferry services to Dublin. The Commissionaccepted that an alternative service leaving from Liverpool was not aneffective substitute since it would be a much longer sea journey. Here theCommission was looking for a flexible idea of essential services since the


54 Case 322/81 Nederlandsche Banden-Industrie Michelin v Commission [1985] ECR 3461.55 G Amato, Antitrust and the Bounds of Power (Oxford, Hart Publishing, 1997) 66.56 Commission Decision 94/19/EC of 21 December 1993 relating to a proceeding pursuant toArt 86 of the EC Treaty (Sea Containers/Stena Link—interim measures) [1994] OJ L/15/8. TheCommission also used the essential facilities doctrine in an Art 86(3) EC Decision, Port ofRødby [1994] OJ L/55/52.

implication of the decision is not that it was physically impossible for competitors to replicate the essential facility, but rather it was commer-cially unattractive to do so.

The Court has not been receptive to the essential services doctrine. InBronner57 a publisher of a daily newspaper in Germany argued thatMediaprint, another publisher of daily newspapers which had a marketshare of 45 per cent, was in a dominant position and abused that dominantposition by refusing Bronner access to its national home delivery service.Bronner’s sales were too small to justify investing in its own national deliv-ery service and other alternatives (for example, the post or newspaper retailoutlets) were inferior. Advocate-General Jacobs and the Court rejected thisattempt to muscle in on a competitor’s success at creating an efficient, verti-cally integrated distribution system for its products. AG Jacobs pointed outthat while granting access to the delivery system might increase competitionin the short term there would not be long-term gains: there would be noincentive for a competitor to create alternative facilities and no incentivefor successful dominant undertakings to invest in efficient facilities if theyhad to share them with competitors. The Advocate-General therefore sawessential facilities as a last resort: to be used only where all hope of normalcompetition had been abandoned. Otherwise, this would lead the Courtsdown the road of detailed regulation involving the fixing of prices, and con-ditions of supply to large sectors of the economy. He argued that this wasunworkable, anticompetitive and scarcely compatible with a free marketeconomy.58

The outcome of Bronner is not surprising. Courts do not like to be con-scripted to serve as de facto regulatory bodies. But second, there is anunderlying premise of competition law that liability rules and remediesshould share a common logic. Determining a ‘just’ price for access to anessential facility is essentially an economic decision: the setting of too higha price would not be of benefit for would be competitors, the setting of toolow a price for access would tip European competition laws into punitiverather than civil sanctions.

We can compare Bronner with the different approach taken where thereis regulation of a liberalised sector. The ruling in Telefónica de España SA v Administración General del Estado59 is an example of the synergy

204 Erica Szyszczak

57Case C–7/97 Oscar Bronner GmbH & Co KG v Mediaprint Zeitungs-und ZeitschriftenverlagGmbH & Co KG and others [1998] ECR I–7791. See also the CFI ruling in Joined CasesT–374/94, 375/94, 384/94 and 388/94 European Night Services v Commission [1988] ECRII–3146. Cf Case T–139/98 AAMS v Commission [2001] ECR II–3413 Case C–258/98 ReCarra [2000] ECR I–4217; Commission Decision 2002/344/EC La Poste/SNELPD [2002] OJL/120/19.58 The Court will be given the opportunity to revisit the essential services doctrine in the pend-ing IMS Health Inc case: COMP D3/38.044 NDC Health/IMS Health; Case C–481/01P NDCHealth Corp & NDC Health v IMS Health Inc & Commission, order of 11 April 2002.59 Case C–79/00 [2001] ECR I–10075.

between competition and regulation which provides the key to understandinghow public services are to be provided by private actors in the market. Herea company authorised to provide telecommunications services challengedthe way the Spanish regulator had implemented the EU directive on inter-connection, one of the universal service obligations in telecommunicationsliberalisation.60 Spanish law required a telecommunications operator dom-inant on the public telecommunications network to provide access to thesubscriber loop and offer interconnection to local and higher level switch-ing centres. The Spanish government argued that it had the discretion torequire interconnection where it considered the market would not be totallycompetitive and that users’ interests would not be guaranteed where anoperator having significant market power could refuse interconnection tocertain levels of the network.

The contested provisions of the directive are that dominant operatorsare required to satisfy only a ‘reasonable request for interconnection’.61

Telfonica argued that this transposition of the directive was ultra viressince the directive did not permit the national regulatory authorities toimpose, ex ante, on an operator having significant market power, obliga-tions concerning access and interconnection points on the network. Thedirective merely allowed national regulatory authorities to advocate inclu-sion of this question in the agreements negotiated between the operators.In other words, the market would determine the conditions governinginterconnection

The Court’s judgment shows the synergy between competition and regu-lation in the liberalised markets. In order to reach the directive’s objectivesof the provision of a universal service in an environment of open and com-petitive markets the directive did rely primarily on commercial negotiationsbetween operators providing telecommunications services. The directiveallowed the Member States to limit the freedom of those operators to decidewhether to enter into interconnection agreements in order to ensure thatthe directive’s objectives are reached.

7. CONCLUSION

Public services are changing dramatically in their function, form and deliv-ery in the new economy. Part of this dynamic change can be attributed tothe fact that public services are being delivered in competitive markets. Thisessay argues that a major task in deciding how public services are to survive


60 Dir 97/33/EC of the European Parliament and of the Council of 30 June 1997 on intercon-nection in telecommunications with regard to ensuring universal service and interoperabilitythrough application of the principles of Open Network Provision (ONP) [1997] OJ L/199/32.61 Art 4(2) of Dir 97/33.

in competitive markets begins with a need to understand and clarify wherethe boundaries end for the state to claim total immunity from market andwhere the market rules begin. The essay argues that under the existing legaltools inserted in the original 1957 Treaty of Rome, the Court has beenunable to offer either clarity or consistency in addressing this question. Theuse of ad hoc litigation is not a coherent method of addressing the com-plexity of the issue. The process of liberalization is leading to greater dis-cussion, transparency and understanding of how public services shouldwork in competitive markets and some clarity in definitions is offered bythe use of soft law communications. There is less emphasis upon how thetransition from national concepts of public services to European market-oriented processes should be facilitated. Since the insertion of Article 16 ECthe EU has signalled a greater respect for the role of public services but hasnot provided the legal base from which a positive set of EU values in rela-tion to public services can be established in legislative form.62 By putting somany questions out to consultation, the Commission’s Green Paper revealsan attempt to focus the modernisation of the regulation of public servicesupon a holistic approach. This may bind the hitherto fragmentary discus-sion of the role of public services, which is the legacy of the litigationapproach of the 1990s.

206 Erica Szyszczak

62 Cf Art 36 of the Charter of Fundamental Rights of the European Union. This is anotherroute for securing the role of public services in the EU. This route is limited. Currently thecharter has not been integrated into the treaties and has not been used by the European Court,despite a number of references to the charter in Opinions of Advocate-Generals. Art 36 is limitedin scope since it does not grant a right to services of general interest and does not guaranteeaccess to such services to everyone in the EU.

Competition, Regulation and the New Economy

Documents

Transcript of Competition, Regulation and the New Economy