MAKING EU NETWORK MARKETS COMPETITIVE · liberalization and privatization. When they preceded EU...

�� © 2001 OXFORD UNIVERSITY PRESS AND THE OXFORD REVIEW OF ECONOMIC POLICY LIMITED

MAKING EU NETWORK MARKETSCOMPETITIVE

OXFORD REVIEW OF ECONOMIC POLICY, VOL. 17, NO. 3

JACQUES PELKMANSWRR (The Hague), Maastricht University, and CEPS (Brussels)

Since the late 1980s the EU has made great strides in the liberalization of network markets. This articleassumes a horizontal perspective, juxtaposing different network markets while focusing solely on the rulesand policies at EU level. A six-step checklist is applied to facilitate a comparative analysis of EU regula-tory liberalization in gas, electricity, telecoms, postal services, and rail and air transport. Competitionpolicy is discussed with respect to: the relation between regulation and competition policy; the role of theEC Court; the ‘essential facility’ doctrine; defining relevant markets; and merger control in networkindustries. Finally, the question is addressed as to whether these network industries operate in an ECinternal market. The answer is no. Policy recommendations: the EU defines a well-considered overallstrategy for network market liberalization; the issue of the internal market with common regulators, at leastwhere a subsidiarity test is passed, should be squarely addressed.

I. INTRODUCTION

The tortuous move towards (more) competitivenetwork markets in western Europe is a very com-plicated process, with shifting insights and researchemphases over time. It is only recently that twomore general approaches have begun to be system-atically addressed: the horizontal perspective, jux-taposing different network markets, and the overallframework of European integration. The presentarticle combines the two in attempting to understandthe roles the European Union (EU) has graduallyassumed and is likely to play in realizing competitivenetwork markets. The reader should be aware that,

here, no attention is paid to national policies ofliberalization and privatization. When they precededEU proposals, it is likely that the latter have beenhelped by the former. The significance of EU-levelliberalization is that the laggards are to join suchefforts, and that EU competition policy and theinternal market rules can be effectively and coher-ently applied to network industries. Presumably, thisshould also promote the economic performance ofEU member states which is a ‘common concern’under the EC’s economic coordination article, 99.

The paper begins with the fundamentals of EUeconomic integration, namely the internal market, its

proper functioning (through undistorted competi-tion), and the overall aims the latter is supposed toserve. For some three decades the Communityadhered to the view that network industries, calledpublic utilities (or their analogues in other EC lan-guages), were justifiably carved out of the internalmarket. Nowadays, the Community embraces prettymuch the opposite view, although the core article onwhich these two views are based (Article 86) hasnever been changed. This is followed by a briefreminder of the nature of network liberalization inthe Union, being recent, gradual, uneven, and com-plex. Section III stylizes a step-wise approach ofhow to introduce competition in network industries,which have ‘enjoyed’ exclusive rights up to re-cently. We recognize the limitations of employing auniform scheme of steps for network markets withquite distinct properties, but use it as a tool tofacilitate a comparative analysis of the regulatoryapproach to liberalization. This framework is ap-plied, in section IV, to six EU network industries:telecoms, postal services, scheduled air transport,rail, gas, and electricity. Section V consists of a briefexcursion to EC competition policy, illustrating thedifficulties of proper application in a comparativeperspective across sectors. Section VI asks thequestions of whether and to what extent the EC hasaccomplished an internal market for these networkindustries. Section VII concludes.

II. NETWORK MARKETS ANDEUROPEAN INTEGRATION

(i) From Exclusion to a Pro-competitive Ap-proach

Until far into the 1980s it seemed as if public utilitieswere not part of economic integration in the Com-munity. Presumably reflecting widespread policyconvictions, a few early challenges of nationalexclusive rights1 were rejected by the EC Court onthe basis of Article 86 (then Article 90). Thisimpression was strengthened by the strict neutralityof the treaty with respect to private or state owner-ship of companies, including utilities. Thus, neitherprivatization nor liberalization of network industries

seemed to be influenced by basic provisions of theEC treaty. And this was upheld despite the ECtreaty’s core provisions on the cross-border freedomsto provide goods and services and the freedom ofestablishment as well as a system ensuring effectivecompetition.

An elementary appreciation of the problem can behad when focusing on two relevant EC articlesrelating to network industries (though not exclu-sively to them): Article 86, about exclusive rights,and Article 31, about distribution monopolies. Thelatter relates to goods only, as it is found in thesection on the free movement of goods. Suchdistribution monopolies are not forbidden but mustnot discriminate in procurement or sales. In the1960s and 1970s endless legal battles took place toimpose this non-discrimination rigorously upon thefew remaining such monopolies (e.g. tobacco prod-ucts in Italy), yet no action was taken againstdownstream distribution monopolies in, for instance,gas or electricity, or against prevailing import andexport prohibitions in distribution contracts in thesesectors.

Article 86 (then 90) is a compromise article balanc-ing the Community interest (integration, as ex-pressed in basic treaty provisions) and the memberstate’s interest (as expressed in the motives behindthe granting of an exclusive right). Such a compro-mise is, on the face of it, a curious balancing act,because exclusive rights preclude the hard-coreprovisions of market integration: free movement,free establishment, and effective competition. In-deed, a literal reading of Article 86/1 says that, withrespect to undertakings having been granted exclu-sive rights, ‘Member States shall neither enact normaintain in force any measure contrary to the rulesin this Treaty, in particular to those rules provided forin article 12 [discrimination as to nationality] andArticles 81 to 89 [rules of competition].’ Takenliterally and in isolation from Article 86/2, this quo-tation is a liberalization clause. The mere existenceof a legal monopoly must undermine the singlemarket and competition in it, hence infringe thetreaty (unless explicit derogation is found in thetreaty, e.g. national security in Article 296). But

1 Notably Costa vs Enel (1964) (case 6/64; 1964 ECR 585) (in which ENEL’s electricity monopoly was de facto supported aslegal) and Sacchi (case 155/73; 1974 ECR 409) (the public interest of a non-economic nature justified RAI’s broadcasting monopoly,including cable transmissions, and the exclusive right could even be extended to Sacchi’s public-display TV sets.

433

J. Pelkmans

434


read in combination with Article 86/2 the conclusionmay be the opposite one. It says that, if undertakingshave been entrusted with ‘services of general eco-nomic interest’, the subjection to treaty rules isconditional. This applies to network industries and afew other special cases.2 The derogation protectsthe public-service function, by stating that the treatyrules apply ‘insofar as the application of such rulesdoes not obstruct the performance, in law or in fact,of the particular tasks assigned to them’. For threedecades it was simply taken for granted that thetraditional public utilities needed exclusive rights forthe ‘performance . . . of the particular tasks as-signed to them’. Even if the existence of such legalmonopolies seemed beyond challenge, it was possi-ble to complain about or discipline the exercise ofexclusive rights—for instance, because of discrimi-nation or excessive prices.3 However, this occurredonly very rarely and did not cause any fundamentalchange in markets.

Nowadays, the approach based on Article 86 isquite different and the upshot is, more often than not,the opposite of the post-Sacchi period.4 A conven-ient summary for economists (though for lawyerssomewhat crude) is as follows. First, a necessitytest has been introduced since the Corbeau case(1993) about the boundaries of the Belgian postalmonopoly. The test is strict: what public serviceobligations would the utility be unable to meet,without the legal monopoly? This is familiar groundfor economists. Indeed, in Corbeau5 the crux of thecase did not turn around the violation of the postalmonopoly per se—value-added services not part ofuniversal service were viewed by the EC Court asin the competitive domain, hence allowed—butabout the possible undermining of the financialequilibrium of the Belgian post. In other words, theobligation of uniform tariffs for the items under theuniversal service, and the cream-skimming thatvalue-added services might imply, were at issue.Thus, the exclusive right was upheld in case ofcompetitive entry of value-added services only tothe extent that the latter could cause the Belgian

post to become a loss-maker. Clearly, this necessitytest opens the door for more fundamental queries,such as the necessity of the exclusive right for theuniversal service itself, with or without uniformity oftariffs. Are there not less restrictive, pro-competi-tive solutions that do not compartmentalize theinternal market and do not reduce competition to atrickle? Inspired by economic analysis and earlyliberalization examples at national level or outsidethe Community, these queries were promptly posed.Later on, liberalization in one network industry atEU level led to a closer scrutiny of the scope for pro-competitive options in another network sector. Onecould even go further and ask whether these alter-native options could not be demonstrated to servethe treaty objectives, such as the ‘raising of thestandard of living and the quality of life’, a ‘highdegree of competitiveness’, and ‘sustainable andnon-inflationary growth’, better than rigid exclusiverights, without in any way negatively affectinguniversal service (presumably reflected in treatyobjectives such as ‘harmonious, balanced and sus-tainable development of economic activities’ and‘economic and social cohesion’).6 This is where atleast the rhetoric of the Union arrived at the LisbonEuropean Council of March 2000. Second, theburden of proof for the justification of the exclu-sive right is on the member state(s). Again, this is aradical change from the past, when the existence ofa legal monopoly was seen as legal if motivated bya public service. Third, the combination of thenecessity test and the reversed burden of proofforces member states—and the Commission in itsrole as guardian of the treaty—to identify far moreprecisely than in the past what ‘particular tasks’ fallunder exclusive rights and why. If not for thewelfare benefits and for reasons of regulatoryemulation of early liberalizers, the U-turn in Com-munity law forced national policy-makers into afunctional reconsideration process about the ration-ale of utilities regulation.

All this does not mean that a few high judges inLuxembourg can and do alter interpretations of vital

2 Like the Finnish monopoly of slot machines, upheld by the EC Court on grounds of public morality, justifying the exclusiveright (Läärä, C-124/97, ruling in 1999).

3 In conjunction with Article 82, where abuse of dominance is prohibited.4 The EC Court first took the opposite view with two telecoms cases, one on goods (telecoms terminal equipment) and one on

services, in 1991 and 1992 respectively.5 Case C-320/91, in ECR (1993) I - 2533.6 The formulation of the objectives is quoted from the Amsterdam version of the EC treaty (in force since 1999).

435

J. Pelkmans

EC articles at will. In Community practice, actualobservations of the network markets, trendsetting atthe national level, and reasoned proposals from theCommission all play a role. The Court tends to allowmargins of discretion to member states which canbe considerable. As Blum (2000, p. 70) notes:

The more complex the economic issues and the morepolitical they are in the sense that they involve nationalpolicy choices, the lower the chances that the Court willbe able to rule on whether the necessity test is met. It issubmitted that it can only do so where the market pressureis such that the market itself has changed and it becomesobvious that an exclusive right is no longer necessary toprovide goods or services to customers—as the telecomscases illustrate.

The consequence may be that it is unlikely that a Courtjudgement could result in the full dismantling of a mo-nopoly over a network industry. Conversely, it might beeasier to prove, as in Corbeau, that monopolies cannot bemaintained as they are and that their scope should bereduced.

Most of the actual progress has therefore beenmade, on the front of agreed liberalization andapproximation (sometimes called harmonization).Such a regulatory approach is based on directivesthat have to be adopted by the Council and theEuropean Parliament. However, with the Commis-sion initially more often than not in the role of front-runner, it risked encountering staunch oppositionfrom captured ministers in the Council and/or ma-jorities in the European Parliament for ideologicalreasons or for fear of the protests of well-organizedpublic-service unions. Therefore the Commissionused the unique instrument of a Commission (so, nota Council and European Parliament) Directive, theonly instance of which happens to be found pre-cisely in Article 86/3. In the telecoms sector of theearly 1990s, these directives acted like crowbars.The backing by the EC Court, reversing the Sacchidoctrine much to the surprise of many lawyers, dida lot to alter the regulatory climate and to bolster theCommission’s determination to pursue the course ofnetwork liberalization where sensible. Although mostof this liberalization-cum-approximation is done on

the basis of directives adopted jointly by Council andParliament, the mere option of Commission direc-tives, and the precedent of the backing by the ECCourt, create significant leverage of power againstdelays and watering-down by Council and Parlia-ment.

(ii) EU Liberalization: Recent, Gradual,Uneven, Complex

It is good to remember how recent the liberalizationof network markets at EU level is. The famous EC-1992 White Paper of 19857 does not speak about itat all, except for a rehearsal of the 1984 GreenPaper on a common market for broadcasting andvague words about ‘the information market’.8 In airtransport, renewed stress is laid on the adoption of(what later became known as) the first liberalizationpackage in 1987, proposed in 1984. This weakproposal did not have any noticeable economicimpact.9 No word on rail, gas, electricity, or postalservices.

But that would soon change. The first proposals fora common gas and electricity market date from1988, in 1989 the broadcasting sector was liberal-ized in the TV-without-frontiers directive, in 1990telecoms liberalization cautiously began, in 1991 thefirst rail ‘liberalization’ directive was adopted, in1992 the final air transport liberalization packagewas agreed, and in 1994 the Council solemnlydeclared that the postal sector would be subjected toa ‘gradual and controlled’ liberalization process.

This outburst of liberalization activities at EU levelwas not the result of an overall plan or of a funda-mental discussion paper about the economic orother advantages of liberalization, or its adjustmentcosts, for that matter, the appropriate regulation thatshould be in place, the nature and implications ofcompetition policy when applied to these sectors, orthe optimal assignment of regulatory and competi-tion policy powers between the EU and nationalgovernments. It very much looked like an ad-hocapproach, on a sector-by-sector basis, and—in the

7 COM(85)314 of June 1985, ‘Completing the Internal Market’.8 Note that, in 1985, the Commission still held an industrial policy, rather than a liberalization, view on telecoms. The only sentence

on telecoms speaks about the necessity ‘of appropriate telecommunications networks with common standards’. See also Schneideret al. (1994).

9 Its significance is probably that competition policy, including an initially very wide exemption which could be narrowed overtime, was firmly introduced into the sector.

436


absence of overall political guidance—driven by theCompetition Directorate-General.10 This prompteda lot of anxiety, ranging from the perception ofsneaky liberalization through the backdoor (by Euro-crats rather than elected decision-makers), to anideological drive claiming the destruction of theculture and merits of public service for all byunfettered market forces with adverse redistribution-al effects. The ground-swell led to targeted lobbyingto amend Article 86 and/or to insert other articles orphrases in the treaty of Amsterdam, that would haverobbed the Commission of the power to issue its‘own’ directives, or would have come close torestore the Sacchi doctrine by inserting a clause inthe treaty.11 However, helped by the first-everCommission paper of a horizontal nature on networkindustries with public-service obligations,12 stress-ing a balance between shared values (e.g. socialcohesion, solidarity) on the one hand, and flexibilityand dynamism on the other, while adhering tosubsidiarity, this option was pre-empted. The 1997Amsterdam treaty comprises a new Article 16which reflects the essence of this Commissionpaper, without in any way adding to or changing thelegal framework of the treaty.13

That the EU approach is gradual and uneven can beread from Figure 1 which provides a highly stylizedpicture of the stages of liberalization over the last 15years for a range of network markets. Stage 1 isdefined as the first EU measure (be it a directive ora competition decision) in the sector with at leastsome liberalizing effect, e.g. a reduction of exclu-sive rights. Stage 2 combines remaining elements of(network) monopoly or essential facilities with regu-lation about access, wholesale and retail price set-ting, public-service obligations, given actual or po-tential entry. Stage 3 is that of competitive networkmarkets, combining regulation and competitionpolicy, but with an emphasis on the latter in the lightof existing rivalry in services and perhaps even ininfrastructure. That the process is gradual can beseen with one glance at the figure. In fact, it wasmore gradual still, since the actual policy debatespreceding stage 1 measures move us back into the1980s, or before, in every one of these sevensectors.

The process is also uneven. In 2001 only threesectors have reached stage 3. Yet, even this shouldnot be considered as ‘final’. In broadcasting, the

broadcasting stage 2 stage 3

telecoms stage 1 stage 2 stage 3

postal stage 1 stage 2

electricity stage 1

gas stage 1

air transport stage 1 stage 2 stage 3

rail transport stage 1

85 90 95 00

Figure 1EU Liberalization: Recent, Gradual, Uneven, Complex

10 Note that Article 86 is in the section of the treaty on competition rules, and that the autonomous Commission directives weredrafted and advocated in the Competition Directorate-General.

11 See Pelkmans and Olsen (1996, pp. 53–7 and ch. 9) for details. See also Rodriguez (1998).12 COM(96)443 of 11 September 1996, Services of General Interest in Europe. A follow-up paper with more sectoral detail was

produced on the request of the Lisbon European Council as COM(2000)580 of 20 September 2000 (same title).13 An accompanying Declaration specified the ‘full respect for the jurisprudence of the Court of Justice’.

437

J. Pelkmans

separation between content and transport serviceshas assumed a new meaning because of conver-gence in electronic communications. Thus, thetelecoms proposals of 2000 aim for technologicalneutrality of ‘transport’ (e.g. fixed copper wire,optical fibre, radio-frequencies-based (mobile orsatellites), cable) and hence apply to broadcasting,too. In air transport, the full liberalization in the EUcannot lead to the contestability one might perhapsotherwise expect, owing to the mercantilist systemof bilaterals governing intercontinental scheduledair services.14

Finally, the process is extremely complex. To beginwith, one should never focus merely on the regula-tory route of EU liberalization, without having re-gard to competition policy. With respect to the latter,it would be inappropriate to zoom in only on domi-nance and collusion, because de-facto merger con-trol has had and will continue to have a significantimpact on the effectiveness of liberalization; muta-tis mutandis, this is true for state aids supervision aswell. Beyond these two central pillars, there areseveral other domains that tend to receive relativelylittle attention from economists in the field but in theEU context have been or still are crucial. The threemain issues are competitive procurement (of equip-ment for the infrastructure or the services), networkcompatibility standards, and the vexed question ofEU agencies (for the last, see section VII). Intelecoms and rail, procurement was not only notcompetitive until the 1980s, but strictly national, withdeleterious effects upon unit cost and innovation.15

Network compatibility standards are sometimesimportant for liberalization in that incompatibilities inthe European installed base may render the actualtake-up of competitive (cross-border) services un-der liberalization unduly costly. Conversely, theemergence of competition may render the adoptionof network compatibility standards more difficultthan under monopoly. The former case can bewitnessed in conventional rail, where only recentlyand under great pressure rail standardization hasbegun to move ahead—typically, in high-speed rail,

where smooth cross-border services are vital forthe return on investment, standard setting occurredrelatively quickly. The latter case is illustrated by theelaborate Global System for Mobile Communica-tions (GSM) standard, which was largely developed(at high cost) under monopoly, yet once firmly on themarket, has greatly spurred EU competition andgrowth in mobile (Pelkmans, 2001).

III. A SIMPLE FRAMEWORK FORCOMPARING NETWORKLIBERALIZATION

It is convenient to start from the assumption that,before EU liberalization, national network marketswere monopolized by law and that the incumbentwas state-owned, often even as part of a ministry.Such a uniform picture would not do justice to someexceptions, be it the energy markets in Germany(often regional), privately owned entrants in expressmail, broadcasting, or (e.g. regional) airlines, theabsence of exclusive rights for postal services inSweden, and countries which privatized ‘early’(such as the UK). However, the assumption wouldusually be fulfilled in many member states, andhence would determine the respective steps to betaken at EU level.

The fundamental idea behind liberalization is thatpublic policy should pursue every means to improvethe incentives for network industries to performoptimally in terms of the (European) public interest.In turn, this idea emerges from the severe difficul-ties in (monopolistic) network industries of prevent-ing or minimizing regulatory failure. If networkindustries are part of the ordinary government ma-chinery, including politicization in the parliament andpolitical appointees by the government, regulationand supervision of the monopoly will be heavilyinfluenced by the ‘regulatee’. Indeed, the formula-tion of the public interest is likely to be ‘captured’ bythose having a direct interest (workers, managers,political parties) and performance may well suffer

14 Since the 1944 Chicago Convention scheduled air transport has developed in the framework of bilateral agreements on landingrights, conditional on highly restrictive provisions. Liberalization in air transport outside the EU means that the bilaterals becomeless restrictive. For EU member states, however, all air transport with countries outside the EU (and a few other European countries)is still governed by these restrictive bilaterals.

15 The (ex-ante) Cecchini and (ex-post) Monti reports provide telling examples and large orders of magnitude of welfare gainsof the introduction of competitive procurement in telecoms equipment and railway rolling stock by the EC-1992 process. SeeEuropean Economy, March 1988 and December 1996, as well as Monti (1996).

438


greatly. Typically, in such an environment, perform-ance indicators will be vague and hard to verify, letalone verify independently, and the internal organi-zation, cost revelation, and management will not beexposed to effective pressures. Incorporation (out-side ministries) may improve incentives and trans-parency, but it will be of little avail if it is not placedat arm’s length, and subjected to independent scru-tiny. Even if this is done, the fundamental principal/agent problem with asymmetric information willremain. More important still, why would themonopolist not suffer from (severe) static and dy-namic X-inefficiencies? What incentives would existto be responsive to users and consumers?

Liberalization seeks to break through this by theintroduction of competition, or, if this is not possiblebecause of (overall) natural monopoly, by the appli-cation of methods which reduce informationasymmetries and enhance incentives to perform.

This logic is no different at EU than at national level.But the method of liberalization at EU level isfounded on a special legal basis and different inter-mediate objectives than national liberalization. First,there is an existence vs exercise dichotomy, withrespect to ownership. The Union has no formalposition about privatization, whereas this is a majorissue for member states, and sometimes for theirregions or municipalities. State-owned companies,including network industries, cannot, however, be-have any differently from private ones—the exer-cise of ownership rights is the same under EC law.In addition, three caveats ought to be noted: therewere strong EU pressures for ‘incorporation’ ofutilities formerly working as departments of thepublic administration—mainly for reasons of properand verifiable cost accounting ; privatization shouldnot be conducted in ways providing hidden state aidsor an artificial initial competitive advantage;16 in therecent ‘open coordination’ approach promoting regu-latory reform by the member states (the so-calledCardiff process)17 the Commission has at timescome close to advocating privatization as a means toimprove incentives for dynamic performance.

Second, it is the establishment of the internal marketand its proper functioning (i.e. effective competitionacross intra-EU borders and, where relevant, withinnational borders) which must drive EU-level liber-alization. However, before one arrives at an internalmarket for network goods and services, a series ofsteps are required, several of which are similar tonational liberalization.

With these basic points in mind, Table 1 provides achecklist for network liberalization. One can see itas a basic framework to understand the progressand hiccups of network liberalization—it should notbe understood, however, as a blueprint issued by EUinstitutions, since this does not exist.

Step 1 asks first whether the network sector ischaracterized by natural monopoly. Nowadays, ithas become clear that few network industries arecomplete natural monopolies. Water supply andsewerage is probably the archetypical example. Forsuch sectors, the incentives cannot come fromintroducing competition because this would reducetechnical efficiency. What can be considered iscompetition for the market, via bidding for (x years)franchise (concession), as this would (a) createstrong incentives for potential entrants to show howthe sector could perform better, and (b) amount toa permanent incentive during the franchise period topre-empt later entry by showing high performance.One might also consider yardstick competition orbenchmarking as incentives to pursue good per-formance. Apart from non-discrimination, the EU isnot involved in such matters. For instance, in asector such as water, cross-border trade hardlyexists, so that EU powers hardly matter.

The EU framework does matter once natural mo-nopoly is incomplete or absent (competition in themarket). Step 2 then immediately poses the issue ofpotential free movement, which is blocked. Article86, EC demands that the member state(s) justifyexclusive rights, as it is a measure eliminating cross-border economic intercourse and competition. Theuniversal or public service obligation, traditionally

16 For a thorough analysis, see Harbord and Yarrow (1999).17 This is based on Articles 98 and 99. The annual conclusions of the scrutiny of national regulatory reform end up in the Broad

Economic Guidelines, with a view to improving the performance of the EU’s economic union, and so facilitating the functioningof monetary union. See, for example, ‘The 2000 Broad Economic Policy Guidelines’, in European Economy, No. 70, 2000.

439

J. Pelkmans

Table 1A Checklist for EU Network Liberalization

Regulatory issue Regulatory option(s)

1. Competition In or for the market?

2. Unbundling Independent regulatorReserved and competitive activities separated (sepa-rate accounts; Chinese walls; separate business units;separate companies; separate ownership); no cross-subsidization

3. Public-service obligation (PSO) USO (and quality aspects)Universal-service obligation (USO) PSO

4. Financing USO/PSO Via USO funds or taxes or access charges

5. EU access rules Open access regulation (capacity limited or not; non-discriminatory/objective/fair; third party; commoncarrier; etc.)

Interconnection obligations for incumbents

6. Cross-border trade (intra-EU) Removal of special, distorted cross-border prices(e.g. ‘accounting rates’; ‘terminal dues’)

EU-wide quality standardsTechnical compatibilities

taken for granted as a justification, is nowadays onlyaccepted if it can be shown convincingly that com-petition would hinder or make more costly theattainment of public-service objectives. Experiencein several network industries and economic analysishave clarified that exclusive rights tend to be sec-ond- or third-best (‘disproportionate’, in EU lan-guage) and will thus not easily pass the ‘necessity’test. Once the natural monopoly element is viewedas an ‘essential facility’ or bottleneck, entry could beallowed upstream and/or downstream and access tothat facility ought to be carefully regulated. As faras the public-service element is concerned, someformula should be agreed, in regulation, leading newentrants to assume their share of the burden, inprovision or in money. Alternatively, suppliers/pro-viders can compete in a contest or auction to ensurethe non-profitable elements of the public service fora minimum subsidy (see step 4).

Yet, in order to introduce such competitive entry,network industries must first be put on equal footingthroughout the internal market, i.e. incorporatedoutside ministries and restructured so as to focuspurely and only on the business—and not on theimplicit regulatory, pricing, and certification—tasks.

In order to level the playing field and to create adegree of confidence for new entrants, a regulatorshould be appointed, independent from the govern-ment (i.e. executing a well-specified law, equally forall, rather than letting the politics of the day prevail)and independent from the incumbent. The regulatorshould be subject to rules of accountability andtransparency. Unbundling will then split the networkindustry into the ‘reserved’ (natural monopoly) andthe competitive elements, and the exclusive rightsfor the latter are abolished. Since the incumbent willbe allowed to compete outside the ‘reserved’ activi-ties, competitors should be guaranteed that no cross-subsidization between the reserved and competitivebusiness of the incumbent takes place. The crediblesolution is to split up the incumbent into separatebusinesses. However, when Council is under pres-sure (capture?), weaker solutions are also accept-ed—for example, separate accounting for divisionswithin the incumbent, and sometimes a prohibitionof, for example, data flows between them (so-calledChinese walls: no information about competitorsusing the essential facility, or about their clients).

Step 3 defines the universal-service obligations(USO) or public-service obligations (PSO) in the

440


network industry. The liberalization experience inEurope has revealed how sloppily USOs and PSOshave been formulated in the past. Politically this wasno point of concern because parliamentary andbureaucratic interference in public utilities used tobe routine. Economically, vague and arbitrary obli-gations, as well as a never-ending flux of interven-tions, cannot possibly lead to a high level of effi-ciency; it also renders the measurement of perform-ance illusory, which was tactically useful for theincumbent vis-à-vis consumers and business cus-tomers. Under competition, obligations have to beproperly identified and subject to independent veri-fication. This provides consumers with a new op-portunity to obtain a constant set of services ofagreed quality, in addition to the hoped-for benefitsof competition. It also disciplines parliaments whichhave to face, for the first time seriously, the cost ofUSOs and PSOs. Finally, because PSOs and USOsare discussed, measured, and compared EU-wide,a healthy element of benchmarking and yardstickcompetition is introduced, which should have apositive impact on performance.

Step 4 regulates the financing of the USOs/PSOs.As is well known, there are three methods and theydo not have identical economic effects. The finan-cial problem of a USO is the net cost which a fullycommercial operator in fierce competition wouldnot be willing to take on. These net costs consist ofthe costs which could be avoided and subtracting therevenues which would be lost if uneconomic USOswere not provided. Clearly, with competition thesecosts will have to be covered via any of threemethods, or, alternatively, by tenders for ‘super-peripheral areas’ , separate enough (e.g. islands) tobe split off. The general tax route is the neutral onebut typically avoided by finance ministers. Accesscharges risk having anti-competitive effects; allow-ing them should be coupled with strict (EU) rules forcost calculation and hands-on supervision. Whendistance-based pricing matters, and is allowed, thelikelihood that USOs can be provided at (virtually)no net costs increases and the EU directive couldthen impose justification before allowing restrictivenational solutions. If uniform tariffs are part of theUSO (e.g. postal services), distance-based pricingmay be strongly resisted, rendering the net-costissue more sensitive. The Commission claims that

postal services USOs have a net cost of 5 per centof turnover in the EU.

Step 5 defines access rules. Three very differentsituations should be distinguished here. First, theinitial situation when the incumbent is the (only)network-holder, calls for pro-competitive (and non-discriminatory) access rules in such a way that allnew entrants have equal opportunities, equal to eachother and equal to the (separated, competitive) onesof the incumbent. This is neither simple nor innocu-ous. The overwhelming dominance of the incum-bent may cause entrants to plead for ‘asymmetric’regulation so as to ‘level’ effective opportunities;that is, special privileges for entrants for a period(practised in some member states in telecoms, andin EU regional air transport in the 1980s, for exam-ple). Beyond the initial stage, the fundamental choiceis between network and service competition. To theextent that natural monopoly does not exist, net-works (or parts of them) can be multiplied. Servicecompetition can take place on a single network(subject to capacity constraints) and on severalnetworks. Once there are several (parts of) net-works, regulation might recede and competitionpolicy or a regulator might only have to ‘supervise’access negotiations. However, negotiated access isa strategic game and may yield sub-optimal results(cf. Laffont et al., 1997). More generally, accessand interconnection pricing—possibly set by theregulator after deadlocked negotiations—may bedecisive regarding whether or not entrants chooseto invest in new networks or seek to go for servicecompetition. High access charges raise the incen-tives for entrants to build their own networks, but theeconomic impact on competition can only be ex-pected much later. Although the Union has neverassumed an articulated position, leaving this regula-tory choice to the member states, the Commissionhas more than once attempted to facilitate servicecompetition via a tough ‘essential facilities doctrine’(see section V) and via proposed regulation (e.g.local-loop unbundling—see Doyle, 2000).18

A third situation can be envisaged where unbundlingis pushed so far that infrastructure and services arecompletely separate businesses, possibly havingdifferent owners. This could apply to electricitytransmission (high-voltage grids). In rail it is at

18 See also Cave and Prosperetti, in the present issue of the Review.

441

J. Pelkmans

present the case in the UK for Railtrack. Althoughsuch unbundling would seem to be ideal from ahistorical and competition point of view (entrantshave full confidence that no conflict of interest withan incumbent might arise), it may lead to seriousincentive deficits. How can one design incentivesfor cost minimization, rather than using cost-plusrules? What is the optimal level of maintenance andservice and how can this be ensured? What areeffective incentives for upgrading and new invest-ment (given capacity constraints)? For actual orpotential cross-border services (e.g. electricity andgas interconnectors, international rail freight routes)these issues require an EU-framework.

The other obligation worth recalling in step 5 is thatincumbents must ‘interconnect’. Apart from stand-ards and other compatibility issues, this is mainly aquestion of price. When opting for services compe-tition (on a single network), must the incumbenthonour his obligation at any price? The Commissionaccepts that interconnection (or interlining in airtransport) should never have to be loss-making, butbeyond that the issue has to do with proper costingmodels and this has turned out to be an extremelycumbersome and as yet unresolved issue at EUlevel. Yet, the national regulators cannot wait, andhave to act. As a result, the disparities in the EU aresignificant, although ad-hoc solutions (see further)have reduced their potential impact.

Finally, step 6 is concerned with cross-border issuesin the EU. When the old public utilities were en-gaged in cross-border services, highly inefficientand distortive practices were condoned under theguise of international cooperation. Introduction offree movement, free establishment, and competitionin the internal market requires the removal of theseagreements and practices inside the EU. Telecomsoperators in the International TelecommunicationUnion agreed on very high ‘accounting rates’ forbilateral calls, postal utilities in the World PostalUnion used ‘terminal dues’ which seemed erraticand bore little or no relation to cost, and the tariffscharged for international trains (and traction overthe domestic part of the route) were much higherthan for domestic ones. Of course, these practicesmake a mockery of the internal market.

IV. THE EU APPROACH IN ASECTORAL COMPARISON

In Table 2 the EU approach, as applied thus far, isset out for six sectors, namely gas and electricity,telecoms and postal services, and air and rail. Abrief commentary is provided. It should be notedthat, in focusing on a horizontal comparison at EUlevel, two limitations have to be taken into account.First, for a full understanding of the implications ofsuch liberalization for market players and, ulti-mately, consumers, much more detail is required.Some of these specifications are sector-specific,too. In the space allotted to this article it is notpossible to enter into such details; they are thesubjects of specialized sectoral papers. Second, instudying the EU approach as laid down in direc-tives, one might lose sight of the diversity andlaboratory functions of the Union. Frequently, somemember states allow more or full liberalization; ifsuccessful, this may spread to other member statesand may eventually prompt liberal amendments oran overhaul of the approach. This wider mechanismhas played a role prior to 1998 in telecoms, and iscurrently relevant for electricity, gas, and postalservices, where the directives form a commonframework but specify no more than the minimumliberalization.

Lest it be forgotten, the regulation-based tableshould always be read in conjunction with nationaland EU competition policies (see section V).

Due to the experimental nature of liberalization inthe beginning as well as to the—sometimes fierce—resistance by the sector (gas and electricity, at first,airlines back in the 1980s) or the labour unions inparticular (e.g. postal services and rail), liberalizationhas proceeded, and still does, in stages. The differ-ences between sectors in Table 2 are due, in part, tothe different stages the various sectors find them-selves in today. The first stage usually looks sym-bolic but invariably appears to have salutary effectson management (including better costing of goods,services, and intermediate supplies) and the famil-iarization of decision-makers with the nature andfurther requirements of the process. Crucially, too,time is bought during which a steady adjustment can

442

OXFORD REVIEW OF ECONOMIC POLICY, VOL. 17, NO. 3T

able

2T

he E

U L

iber

aliz

atio

n M

etho

d fo

r N

etw

ork

Indu

stri

es

Gas

Elec

trici

tyTe

leco

ms

Post

alA

ir tra

nspo

rtRa

il

Com

petit

ion

Unb

undl

ing

Fina

ncin

g PSO

/USO

‘In’

the m

arke

t for

‘non

-re

serv

ed’

serv

ices

(al

lcr

oss-

bord

er);

licen

sing

;ra

ilway

inf

rast

ruct

ures

seen

as n

atur

al m

onop

o-lie

s; in

fras

truct

ure

man

-ag

ers,

plus

inde

pend

ent

char

ging

bod

y

‘In’

the m

arke

t; ne

twor

kco

mpe

titio

n‘I

n’ th

e m

arke

t (bu

t re-

stri

cted

; su

bjec

t to

PSO

s);

infr

astr

uctu

reco

mpe

titio

n pos

sible

; for

natu

ral

mon

opol

ies

(tran

smis

sion

, di

strib

u-tio

n, s

tora

ge),

no tr

ans-

mis

sion

syst

em o

pera

tor

(TSO

)

‘In’

the

mar

ket

(as

for

gas)

; inf

rast

ruct

ure c

om-

petit

ion

poss

ible

; TS

Ofo

r hi

gh-v

olta

ge g

rid

(nat

ural

mon

opol

y) a

nddi

strib

utio

n sy

stem

op-

erat

or (D

SO) f

or di

strib

u-tio

n ne

twor

ks (

natu

ral

mon

opol

y)

‘In’ th

e mar

ket; i

nfra

struc

-tu

re c

ompe

titio

n po

ssi-

ble;

lic

ensi

ng (

gene

ral

and i

ndiv

idua

l)

‘In’

the m

arke

t, for

‘non

-re

serv

ed’ s

ervi

ces (

up to

now

a tin

y sha

re);

licen

s-in

g

Inde

pend

ent r

egul

ator

orm

embe

r st

ate;

dis

pute

settl

emen

t by

inde

pend

-en

t au

thor

ity;

sepa

rate

acco

unts

and

Chi

nese

wal

ls (

no m

anag

emen

tun

bund

ling)

Inde

pend

ent r

egul

ator

orm

embe

r st

ate;

dis

pute

settl

emen

t (s

ee g

as);

sepa

rate

acc

ount

s

Inde

pend

ent

regu

lato

r;se

para

te co

mpa

nies

(un-

til 19

98)

Inde

pend

ent

regu

lato

r;se

para

te a

ccou

nts

Inde

pend

ent r

egul

ator

orm

embe

r sta

te;

Inde

pend

ent r

egul

ator

orm

embe

r st

ate;

sep

arat

eac

coun

ts (s

ervi

ces v

s in-

fras

truc

ture

); d

etai

led

prov

isio

ns a

bout

rai

lse

rvic

es’

inde

pend

ence

from

the

stat

e

PSO

s, U

SOs

PSO

onl

y ge

nera

lly d

e-fin

ed at

EU

leve

l; sy

stem

oblig

atio

n for

the n

atur

alm

onop

olie

s (in

clud

ing

lique

fied

nat

ural

gas

com

pani

es);

unifo

rm ta

r-iff

USO

s per

mitt

ed

PSO

(se

e ga

s);

syst

emob

ligat

ions

for T

SO; u

ni-

form

tarif

f USO

s per

mit-

ted

USO

s de

fined

in

som

ede

tail

Bas

ic U

SO de

fined

; uni

-fo

rm ta

riffs

allo

wed

(yet

,‘p

rices

mus

t be g

eare

d to

cost

s’)

Nat

iona

l PS

Os

only

in

som

e cas

es, a

nd o

pen

li-ce

nsin

g for

such

city

-pai

rro

utes

PSO

reco

gniz

ed, w

ithou

tan

y sp

ecifi

catio

n; s

pe-

cial

rig

hts

perm

itte

d(n

ote:

for

cro

ss-b

orde

rfr

eigh

t or

int

erna

tiona

lpa

ssen

ger

serv

ices

, of

little

rele

vanc

e)

Reg

ulat

ed d

istri

butio

nta

riffs

allo

wed

; no

deta

il(n

ote:

at E

U le

vel,

dist

ri-bu

tion

to c

onsu

mer

s not

yet li

bera

lized

)

Reg

ulat

ed d

istri

butio

nta

riffs

(see

gas

)U

SO f

und

or a

cces

sch

arge

s (af

ter 1

998)

USO

fun

d al

low

ed;

li-ce

nsed

ope

rato

rs f

orno

n-re

serv

ed s

ervi

ces,

yet w

ithin

the

USO

, can

be f

orce

d to

pay

to th

efu

nd

Subs

idy

poss

ible

(st

rict

rule

s)N

o det

ail

443

J. Pelkmans

Tab

le 2

con

tinue

d

Gas

Elec

trici

tyTe

leco

ms

Post

alA

ir tra

nspo

rtRa

il

EU a

cces

s rul

es

Cro

ss-b

orde

r tra

de(in

tra-E

U)

Sour

ces:

For g

as: D

irect

ive 9

8/30

EC

(in O

J EC

of 27

July

1998

, L 20

4); f

or el

ectri

city

, dire

ctiv

e 96/

92 of

19 D

ecem

ber 1

996 (

OJ E

C of

30 Ja

nuar

y 199

7, L

27);

for t

he 19

98 te

leco

ms p

acka

gean

d al

l ref

eren

ces,

see P

elkm

ans a

nd Y

oung

(199

8); f

or p

osta

l ser

vice

s, di

rect

ive 9

7/67

in O

J EC

of 2

1 Ja

nuar

y 19

98—

the R

eim

s II a

gree

men

t bet

wee

n (m

ost)

post

al in

cum

bent

s, se

eno

tific

atio

n to

and

appr

oval

by

the C

omm

issi

on in

OJ E

C C

371

of 1

Dec

embe

r 199

8; o

n ai

r tra

nspo

rt, E

EC re

gula

tions

240

8/92

on

mar

ket a

cces

s, 24

09/9

2 on

fare

s and

rate

s and

240

7/92

on lic

ensin

g, as

wel

l as r

egul

atio

n 95/

93 on

slot

allo

catio

n; on

rail,

dire

ctiv

e 200

1/12

of E

P & C

ounc

il (am

endi

ng di

rect

ive 9

1/44

0, on

, for

exam

ple,

unbu

ndlin

g ), d

irect

ive 2

001/

13 (a

men

ding

dire

ctiv

e 95/

18 on

lice

nsin

g) an

d dire

ctiv

e 200

1/14

( on t

he al

loca

tion o

f rai

lway

infr

astru

ctur

e cap

acity

, cha

rges

, and

safe

ty ce

rtific

ates

), al

l of 2

6 Feb

ruar

y 200

1 and

publ

ishe

d in O

J EC

L 75

of 1

5 M

arch

200

1. S

ee al

so th

e int

er-o

pera

bilit

y di

rect

ive 9

6/48

and

the d

raft

for c

onve

ntio

nal r

ail,

on th

e tab

le ea

rly in

200

1.

Third

-par

ty a

cces

s, ne

-go

tiate

d or

reg

ulat

ed;

only

for l

arge

indu

stria

lbu

yers

and

gas

-fir

edpo

wer

plan

ts; t

hree

step

sup

to 3

3 pe

r cen

t in

2009

Third

-par

ty a

cces

s, ne

-go

tiate

d or r

egul

ated

, the

‘sin

gle

buye

r’ m

odel

;on

ly f

or in

dust

rial b

uy-

ers;

thre

e st

eps

up to

33

per c

ent in

2003

; rec

ipro

c-it

y be

twee

n m

embe

rst

ates

Neg

otia

ted

inte

rcon

nec-

tion

with

ope

rato

rs w

ith‘s

igni

fica

nt

mar

ket

pow

er’, w

hich

mus

t mak

ea

publ

ic re

fere

nce

offe

r;co

st-o

rient

ed i

nter

con-

nect

tarif

fs; s

uper

visi

on/

arbi

trage

No d

etai

l (ye

t); tr

ansp

ar-

ent a

nd n

on-d

iscr

imin

a-to

ry a

cces

s; f

ollo

w-u

ple

gisl

atio

n if

need

is e

s-ta

blis

hed

Inte

rlini

ng fr

eely

nego

ti-ab

le (s

ubje

ct to

com

peti-

tion

polic

y);

cons

erva

-tiv

e ru

les o

n sl

ot a

lloca

-tio

n (no

auct

ions

/ no t

rad-

ing)

App

roxi

mat

ion

of r

ules

for i

nfra

struc

ture

man

ag-

ers

(and

/or

char

ging

bodi

es)

e.g.

net

wor

kst

atem

ents

, ch

argi

ngfr

amew

ork

(inc

ludi

ngpe

ak-lo

ad p

ricin

g),

ca-

paci

ty

allo

cati

onsc

hem

es, e

tc.; f

ew de

tails

on m

etho

dolo

gy;

reci

-pr

ocity

for

obl

igat

ions

beyo

nd

EU

; in

ter-

oper

abili

ty re

quire

men

tsfo

r hig

h-sp

eed

and

con-

vent

iona

l rai

l, in

clud

ing

stan

dard

s

Tran

sit

(with

neg

otia

-tio

ns);

excl

usiv

e im

port/

expo

rt rig

hts a

bolis

hed

See

gas

Rem

oval

of fr

ontie

r pric

es(‘

acco

untin

g ra

tes’

)R

emov

e di

stor

tions

in

‘term

inal

due

s’ (c

ost r

e-la

ted)

; 19

98 R

eim

s II

agre

emen

t

Free

(in

clud

ing

cabo

-ta

ge)

No

spec

ifics

444


be initiated, but without social strife. These earlyexperiments were not infrequently accompanied byCommission infringement cases, or Court cases,heightening the sector’s awareness that matterswere changing.

In combining the checklist of Table 1 with Figure 1,the liberalization dynamics at EU level can be betterunderstood. In airlines three packages were adoptedbetween 1987 and 1992, the latter not allowing fullcabotage until 1997. In telecoms and postal serv-ices, services were at first divided into ‘reserved’and competitive ones. In telecoms, the period be-tween 1992 and 1998 was used gradually to reducethe size of the reserved segment, with free move-ment and full competition introduced in 1998. Inpostal services, the same idea is pursued, but progressis very slow indeed. Currently, the competitivesegment consists merely of special express mail—not counting courier services of parcels—and let-ters above 350 grams, some 2 per cent of turnover.New proposals19 to widen the competitive segmentto 16 per cent have run into staunch opposition inCouncil and Parliament. However, at the memberstates level, progress is a good deal faster, thoughdisparate. It is tempting to suggest that the crucialdriving force in telecoms is technology, which islacking in postal services. It should not be forgotten,however, that competition in postal services hasemerged due to technology: fax, e-mail, broadcast-ing, and the Internet (the last competes with directmail). In electricity the reluctance to go for fullliberalization is waning in Europe (half of the mem-ber states already (plan to) do it soon); in gas thismay well happen, too. In rail, finally, the first step (in1991) consisted in a symbolic reduction of exclusiverights by allowing access for certain railway serv-ices from other member states. It forced much-needed restructuring and some weak unbundling, aswell as a new regulatory package for access andinter-operability, adopted in early 2001 and 1996,respectively.

The enormous progress in this field can be observedfrom the fact that, in all six sectors, competition isnow allowed in the market, something not only

resisted but described as impossible by many in the1980s.

Although principles such as non-discrimination andtransparency have become well accepted, and ‘regu-lators’ should not suffer from conflicts of interest,20

regulators are independent from the sector and thegovernment only in telecoms. In other sectors, an‘independent’ regulator should have no conflict ofinterest in providing services or in capacity alloca-tion (but note that, in rail, capacity was at firstallocated by the incumbent), but the independencefrom government is less clear. In these instances,actual independence is better achieved by their needto be credible and via judicial review.

Public (or universal) service issues are dominatingin postal services, and to a lesser extent, telecoms,but hardly play a controversial role (as yet) inelectricity (since consumers are still captive), in rail(as domestic passenger services are not liberalized),or in air (because they exist only in special cases,e.g. permanent service to the Greek islands). As acorollary, the co-funding of the net costs of the PSO,potentially so problematic in postal services (and,until 2000, in telecoms), plays virtually no role as yetin EC rules for other sectors.

In electricity and, somewhat later, in gas, this isbound to change soon. Although at European Coun-cil (i.e. prime-ministers’) level, the political consen-sus for the Commission to prepare full liberalizationproposals for electricity has not yet been accom-plished, more and more member states acceleratetowards full liberalization. In such a setting, let alonewith a fully liberalized internal market, the universalor public service aspects will have to be carefullyregulated so as to minimize adverse effects on freemovement and competition while allowing memberstates a reasonable discretion for these objectives.The EU is moving towards a common frameworknow.21

Access rules and pricing is the crux of the successof network industry liberalization, although this maynot be so crucial for postal services and airlines

19 See COM (2000) 319 of 30 May 2000.20 Moreover, they are always subject to judicial review based on EC law.21 Communications by the Commission on public service in electricity and in gas are expected in the course of the autumn 2001.

A considerable amount of detail about PSOs in electricity, currently applied by the member states, can be found in a CommissionWorking Paper on the common energy market, SEC (2001)438 of 12 March 2001.

445

J. Pelkmans

(where networks are not physical but logistical).The heavy-handedness of regulation and/or super-vision is a function of natural monopoly—the ques-tion to ask remains: is infrastructure competitioneconomically sensible at all? If not, like in gas orelectricity transmission and distribution as well as inrail, except for dedicated pipelines or lines, accessregulation and supervision are a prerequisite forcompetition to emerge in the first place.

As Table 2 indicates, the 1996 Electricity Directivegave member states the choice between negotiatedor price-regulated (and published) third-party ac-cess (TPA), or the ‘single buyer’. For reasons ofnon-discrimination between market players, as wellas predictability (about tariffs) when buying elec-tricity without the threat of renegotiations, regulatedTPA is often preferred—indeed, 14 out of 15member states have opted for it.22 In the 1998 GasDirective the single-buyer option does not exist. Nomember state has opted for a purely negotiatedTPA. Instead, eight chose regulated TPA and theother seven various mixtures of regulated and nego-tiated, but always with a form of regulatory super-vision.

In electricity, not in gas, the EC directive contains areciprocity clause. The political background of thisprovision is the monopoly of Electricité de France(EdF), a fully integrated company, also fully state-owned. The sensitivity in electricity is higher than ingas (with Gaz de France, also integrated and nearlyas dominant at home) because the calendar forliberalization in electricity is much shorter and thehigh share of nuclear energy of EdF poses a majorproblem of proper cost price calculation. The sheersize of EdF (the biggest electricity company inEurope) creates anxiety, too. The reciprocity clausefollows from the disparate progress in electricityliberalization among the member states: with arange of countries going faster than the EU calen-dar, hence going beyond the 33 per cent before2003, the fear was that some countries, but inparticular France, would stick to the minimum obli-gations, and otherwise exploit the many loopholes23

in the 1996 directives. When the French parliamentwas too late with implementation into national law in1999, it induced howls of protest from a chorus ofnational energy ministers. However, EdF wentaround the trade restrictions implied by reciprocityvia a systematic acquisition strategy in neighbouringGermany, Spain, and Italy. Since it dominated theFrench market and had stakes in all cross-bordertransmission, it would result in asymmetric liberali-zation and distorted competition. When France, bothin the Lisbon (2000) and the Gothenburg EuropeanCouncil (2001) summits, refused to commit to fur-ther opening up, the incentives of the reciprocityclause were proved to be illusory. Italy and Spainswiftly adopted ad-hoc legislation of a dubiousnature (under EC law) and EdF reduced its stakesto small minority shares when faced with loudpolitical and business protests. The Commissionwas in a difficult position because, with the evasionof the directive, all it could rely on was competitionpolicy (see further).

Even when infrastructure competition is possible,and does take place, as in telecoms (both fixed-wireand mobile), a firm regulatory framework and promptintervention by national regulators in case of delay-ing tactics or excessive interconnect tariffs is shown,by experience, to be indispensable. By August 2000some 1,362 fixed/fixed and fixed/mobile intercon-nection agreements had entered into force in theEU,24 clearly some measure of success of ‘telecoms-98’. Yet, new entrants in many member statescontinued to experience difficulties, such as hightariffs, price squeeze (the combination of high inter-connect tariffs, particularly at local level, with lowend-user tariffs by the incumbent, ‘squeezing’ theentrant’s margin, while the incumbent earns fromthe interconnect tariff), delaying tactics, reluctanceof national regulators to intervene (for example,because the cost accounting systems had not beenimplemented, pre-empting a careful judgement of‘cost orientation’), a lack of freedom about the level(in the network hierarchy) of interconnection, (ex-cessive) requirements to have many interconnec-tion points (e.g. Spain, Germany), etc.

22 Germany has negotiated TPA, with published (but non-binding) prices. There is no supervisory approval. The ‘single buyer’(insisted on by France) has not been chosen by any member state, not even by France itself.

23 See, for example, Hancher (2000a,b) and the Commission proposals of May 2001 refining the 1996 directive (see COM, 2001,125).

24 Sixth Telecoms Implementation report, COM(2000)814 of 7 December 2000 (p. 16).

446


In air transport, access/interconnection has differ-ent connotations. A kind of dichotomy is emergingbetween network-based and point-to-point airlines.Network-based airlines include (many) flights whichde facto are largely point-to-point, but on the wholeintra-network transfers and interlining with othernetwork airlines are of considerable importance.Economies of scope via hub-and-spoke, loyaltyschemes, as well as high frequency in a networkwith many connections, together cause networkeffects not to be exhausted before a very large sizeis reached. Point-to-point airlines, on the other hand,do not ‘interline’ (as a rule) and do not develop theirown networks. Inside the EU (and a few neigh-bouring countries) this is not due to the bilateralsbecause, among these countries the bilaterals nolonger apply. It is true, however, that the mercantil-istic bilaterals between member states and non-European third countries severely restrain the op-tions for any serious network development by newentrants. In section V we briefly discuss the strate-gic alliances (of network-based alliances) whichpose difficult choices for competition authorities.

In rail it is too early to assess the new access regime.It is useful to realize, however, that the ‘iron’assumption of the absence of infrastructure compe-tition is rusting away, once one considers long-hauldedicated lines. It now appears that investment inlong-haul dedicated freight lines, next to existingtrack, is feasible. If dedicated, double track freightcorridors could be exploited through the EU plusSwitzerland, much like the TGV, very high capacityutilization rates could be achieved, without manycoordination and logistical problems (this wouldassume efficient terminals, also intermodal). Ongo-ing feasibility studies suggest that private ownershipof such corridors (if located immediately besideexisting track) and pay-back periods of less than 15years are entirely possible! The European freight

‘freeways’ system initiated in 1998 is not yet suc-cessful because access and its pricing are not wellregulated, there is no single regulator, and theaccess tariffs are said to be very high, yet cost (letalone efficiency) verification does not have a solidaccounting basis. Moreover, apart from complexlogistical coordination with passenger rail traffic,which raises costs compared to dedicated lines,today’s rail monopolists run around 50 per cent oftheir freight wagons empty, because of lack ofcabotage25 or simply a lack of commercial interest.This combination of X-inefficiency and regulatoryconstraints make it exceedingly difficult to assesswhat ‘competitive’ access tariffs should be for thefreeways, not to speak of the cost reductions envis-aged when exploiting dedicated long-haul corridors.

Finally, the intermediate objective of the EU inliberalization should be the establishment and properfunctioning of the internal market. The problem withthe various measures in Table 2 is that they arenecessary but not often sufficient. In section VI webriefly address the internal market question.

V. EC COMPETITION POLICY

EC competition policy is exceedingly hard to isolateas a stand-alone policy when applied to networkindustries. If and in so far as one attempts to do so,inevitably regulatory or quasi-regulatory aspectswill have to be taken into account, or competitionpolicy itself deviates from traditions in assumingquasi-regulatory functions.

In a survey paper as broad as this one, all we canhope to do is to guide the reader through a fewimportant corridors of what is rapidly becoming alarge labyrinth. The following aspects26 are brieflytouched upon:

25 However, on the (now four) trans-EU freight freeways, cabotage rights exist.26 There are many other questions one could address. Three examples provide no more than an illustration. First, has EC

competition policy had to engage in assessing the comparative merits of costing models (LRIC-plus; fully distributed costs; historicalvs current; or, as in today’s electricity debate, transaction/distance-based costing vs the ‘lake’ model)? Since these costing modelsare up to the discretion of the national regulators, does it mean that distortions of competition in the internal market or, for thatmatter, a ‘too high’ access tariff as an abuse of dominance, cannot be tackled? Second, how should competition policy be applied,if at all, to often highly restrictive long-term gas contracts before competition becomes effective, as they could frustrate access-to-grids for decades? (Note, in recent cases the Commission accepted 15 years duration, see Slot (2000).) Third, as Hancher (2000a)notes, the decentralization of EC policy with respect to Article 81 (widely resisted anyway) would seem to be particularlyunwelcome for gas and electricity, given the many access and other issues coming up (see also Albers, 2000).

447

J. Pelkmans

• the general relation between EC competitionpolicy and EC regulation in network industries;

• the thrust of the rulings of the European Courtof Justice on network industries;

• the ‘essential facilities’ doctrine (EFD) as analternative to regulation;

• the problem of defining ‘relevant markets’;

• is EC merger and alliances control tantamountto network market regulation?

(i) Competition Policy and Regulation

The traditional distinction between competition policyas ex post and regulation as ex ante is blurred in EUnetwork markets. On the one hand, regulation cantake the place of competition policy if rigid ex-antedefinitions of who is dominant and in what (relevant)market, and hence who is burdened with specialobligations to pre-empt anti-competitive behaviouragainst new entrants, are inserted in EC directives.In telecoms this was done under the heading ofcompanies with ‘significant market power’. To dothis under competition policy, via ‘abuse of domi-nance’, is bound to be very slow and costly, andperhaps not easily predictable. This would frustratethe effectiveness of the liberalization as entrants willnot be prepared to invest or seriously challenge theincumbent. Even after the first few years, regulationmight well be preferable to competition policy, insome cases. Take call termination from fixed tomobile, where very high interconnect tariffs werediscovered in the EU in 1996–8. Today, under ECdirectives, national regulators simply declare thetwo or three leading companies to enjoy significantmarket power (SMP),27 which obliges these marketleaders to charge cost-oriented interconnect tariffs.As a result, tariffs for call termination on mobilenetworks in the EU fell dramatically in 1999.

On the other hand, EC competition policy in networkmarkets has often had a quasi-regulatory flavour.

Telecoms remains the leading example because inthis field the Commission employed Commissiondirectives (on the basis of Article 86/3) as ‘crow-bars’ in 1989 and 1990, and the European Court ofJustice sided with the Commission in 1991 and 1992.But the interested reader could also verify the FullCompetition directive (96/16 of 13 March 1996), thenumerous introductory recitals of which cover eve-rything then under regulatory preparation. In fact,the directive refers to a host of procedures, withdeadlines (at the latest, 31 December 1997), whichare all meant to lead to regulatory directives. In thelast few years several Notices and Guidelines havebeen published by the Commission (e.g. on criteriafor costing schemes, access, etc.) which comeclose to ex-ante rules rather than ex-post policy.Again, under the new telecoms package,28 there isan explicit attempt to give EC competition policy a‘regulatory’ role. The proposals turn SMP intowhat, under competition policy, would be ‘domi-nance’. However, this is to be employed in an ex-ante fashion, namely, to be able to designate acompany as having SMP, with a legal basis forregulatory obligations. In a detailed Working Docu-ment29 the Commission has surveyed Commissionpractice and the Court’s case law, and in so doingoffered guidance to national regulators for the newapplication of SMP. This is an obvious hybrid,employing dominance without ‘abuse’, and marketanalysis which is not a-priori case-by-case (as incompetition law).

In electricity, however, the crowbar strategy of theCommission has never been used. Given the lowdegree of technological upheaval in this sector,compared to telecoms, and the remaining naturalmonopolies in transmission and distribution, a cum-bersome and slow regulatory approach in stages hasbeen chosen. Competition policy has been littleinvolved thus far, except via mergers (see further).In airlines, where neither USOs nor natural mo-nopoly were an issue, the crowbar tactics wereobviated by a stronger reliance on the Single Act’sdefinition of the internal market30 in combinationwith a block exemption based on Article 81/3 (inter-

27 Such a company should have 25 per cent market share or higher.28 A common regulatory framework for electronic communications networks and services, draft directive, COM(2000)393 of

12 July 2000; and draft directives on access, user rights, licensing, and data protection (all issued on 12 July, COM numbersrespectively, 384, 392, 386, and 385).

29 Draft guidelines on market analysis and the calculation of significant market power (COM (2001)175 of 28 March 2001).30 The internal market in the Single Act (now Article 14) is defined as an ‘area without frontiers’ (i.e. also for services) in which

‘free movement . . . of services . . . is ensured’. This is far stronger language than in the Rome treaty.

448


firm cooperation) which was sharpened over time,along with the next steps of liberalization.

The relation between competition policy and regu-lation is often discussed in terms of the latter givingway to the former over time. This assumes thateffective competition has developed and that re-maining market failures can be addressed by com-petition policy. For many networks markets thesetwo assumptions are not easily fulfilled.

The best example of ‘waning’ regulation, combinedwith increasing prominence of competition policy, isscheduled air transport. The only piece of regulationwhich still restricts intra-EU competition (not count-ing the external bilaterals discussed before) is the1993 slot allocation one, which grandfathers existingslots, if they are used, and forbids trading.31 For therest, air transport is a matter of competition policy.The exemption to Article 81/3 has once again aquasi-regulatory character because of the nature ofcooperation in the airline business (e.g. interliningand its tariffs; ground-handling; etc.) Also guide-lines, an ex-ante approach with a threat of ‘interimmeasures’ (a very swift Commission intervention),on predatory price and non-price moves were pub-lished as early as 1991. The latter is remarkable,because very few Article 82 predatory action caseshave been pursued in ‘ordinary’ markets. A justifi-cation of such ‘regulatory competition policy’ is thegreat vulnerability of new entrants in the airlinebusiness. Non-price predation in the form of(re)scheduling departure times to a few minutesbefore those of the entrant, while selectively offer-ing deep-discount seats may kill the competitionbefore it has a chance to take off.

Other competition policy aspects in airlines relate tostate aids and alliances.

(ii) Court Rulings on Network Industries

Despite the potentially significant powers underArticle 86/3 (Commission instead of Council/Euro-

pean Parliament directives) and despite some po-tentially far-reaching rulings of the European Courtof Justice (ECJ) (and the Court of First Instance), inactual practice, the Court’s judgements have had arelatively limited impact (see Blum (2000) for de-tailed analysis). One crucial reason for this outcomeis that it has turned out that the ‘necessity test’ (seesection II(i)) is notoriously difficult to apply. Typi-cally, in cases which reach the ECJ, strong (anduncontested) empirical evidence against the main-tenance of an exclusive right might not be available,and if this is so, the Court would tend to allow themember states a considerable margin of discretion(as noted before). Moreover, in the landmark gasand electricity cases32 the Court widened the neces-sity test. This makes it more difficult for the Com-mission to use its special powers for, for example,electricity, as it did so successfully in telecoms.

In sectoral terms the ECJ has probably had thegreatest impact on telecoms and broadcasting. Inpostal services Corbeau has limited the scope ofmonopolies on express mail services. In gas andelectricity, openings exist to challenge the validity ofnetwork monopolies, but the burden of proof is withthe Commission and empirical. In rail, where it mightperhaps be less easy to challenge certain monopo-lies, the Court has considerably reduced the Com-mission’s discretion to impose an ‘essential facility’doctrine and in so doing force third-party access.

(iii) Essential Facilities, Competition Policy,or Regulation?

The transposition of the notion of ‘essential facility’from the USA to Europe (both at EU level, and inthe competition law of some member states, e.g.Germany) has caused a lot of confusion.33 First,even in the US the Supreme Court has neverembraced the ‘essential facility doctrine’(EFD).Second, the EC has a well-developed ‘classical’case law on ‘refusals to deal’ based on Article 82 (incontrast to the greater reluctance in the USA), sothat an EFD should be narrowly constructed. In the

31 A proposal by the Commission, in June 2001, to liberalize slot allocation somewhat and, conditionally, allow slot trading,is said (by the former flag carriers) to lead to greater entry by American and Asian airlines, not the no-frill competitors. This maybe correct if slots become costly, an outcome which (for the relevant periods of the day) is not improbable.

32 About export/import monopolies of Spain, France, the Netherlands, and Italy, cases C-157(to 161)94; ruling in 1997, see ECRI-5699 and further.

33 The literature on the EFD is large. See Larouche (2000a, pp. 165–217) for a rich and illuminating survey. This section drawsfrom his work. See also Temple Lang (2000)

449

J. Pelkmans

1990s, however, this is not what the Commissiondid. Quite the contrary, it made a number of at-tempts to use the EFD to impose obligations oncompanies owning the ‘essential facility’, usually ina vertical relationship, and, more often than not,those were in network industries. In terms of classi-cal competition law, the EFD is frowned uponbecause the routine requirements of proper marketdefinition and the assessment of dominance aredropped and ‘replaced’ by a test of ‘essentiality’ ofa facility. In doing so, it turned out that one (theCommission) risks making serious mistakes, as inEuropean Night Services (passenger rail, cross-border) and Bronner (a national newspaper insistingon ‘access’ to the home distribution system of alocal newspaper, as this would be an essentialfacility).34 EC case law now requires a double test:the lack of third-party access to the facility shouldengender negative effects on competition in thedownstream, end-user market, so that a structuralmeasure based on EFD would generate benefits toend-users; in addition, there should be no (economi-cally) viable alternative for an ‘objective’ competi-tor (in other words, not necessarily the applicant,who could be a free rider). There is general agree-ment that, since Bronner, the EFD is subject to suchstrict conditions that its application will be rare andproblems might be better tackled with sector-spe-cific regulation based on careful cost–benefit analy-sis.35

(iv) The Problem of Defining Relevant Markets

The difficulties in defining relevant markets wouldseem to be greater in network markets than in‘ordinary’ markets. Of course, one could argue thatthe classical opposition between the Commissionfavouring ‘narrow’ and (most of) industry favouring‘wide’ definitions is simply replicated. Although thisis undoubtedly so, the underlying economics of

demarcations seem to be less than clear. Twoexamples are airlines and telecoms (the followingdraws from Larouche, 2000b).

The basic idea of a relevant market definition in(EU) competition law is, as the Commission puts itaptly, ‘to identify in a systematic way the competi-tive constraints that the undertakings involved face’.36

For the product market the key criterion is demandsubstitutability, with a limited secondary role forsupply substitutability. Potential competition is, ofcourse, taken into account—not at the relevantmarket definition stage, but at the (later) competitiveassessment stage in that relevant market. In fact,the same applies to the relevant geographical mar-ket, although—certainly in network markets—regu-latory constraints may cause fragmentation. Here,we are not concerned with the standard economicsof relevant markets,37 but rather with the specialfeatures caused by network markets.

In airlines the Commission has gradually movedfrom an individual-route approach to a more differ-entiated approach. What is different from ‘ordi-nary’ goods markets is that some geographicalelement is (inevitably) part of the product definitionand that network effects are taken into account.What is similar, on the face of it, is that demandpreferences (e.g. time-sensitive travellers vs non-time-sensitive but price-sensitive ones) may lead todistinct relevant markets, even though they pertainto consumers finding themselves in the same planeafter they have bought their tickets.

In other words, the relevant market will be made upof all routes that are considered equivalent by theconsumers. It is the network effects that the Commis-sion has only recently begun to recognize—com-petition can develop between networks or hubs ofairlines. The hesitation on the part of the Commission38

34 CFI Judgement of 15 September 1998, Cases T-374, T 375, T-384, and T-388/94, European Night Services vs Commission;ECJ judgement of 26 November 1998 Case C-7/97 Bronner vs Mediaprint.

35 Nevertheless, there are important (EFD) cases, all prior to the 1998 Bronner ruling in rail (combined transport, including rail,Deutsche Bahn fined for discriminatory tariffs; combined rail transport with containers, a French/UK joint venture ACI wasinstructed to provide rail services to others (even though no others requested or signalled this at any time) on a non-discriminatorybasis) and telecoms (e.g. access to the SWIFT banking telecoms network for LaPoste; Atlas and GlobalOne, two telecoms alliancesfrom the mid-1990s—but not, as Larouche (2000a, p. 188), points out, on a similar alliance, namely Unisource, where the ‘classical’Article 82 approach is employed).

36 Relevant Market Notice, in OJ EC C 372 of 9 December 1997.37 See, for instance, Church and Ware (2000, ch. 19).38 See, for instance, KLM/Alitalia, Decision of 11 August 1999, in OJ EC C 96 of 2000.

450


has everything to do with its indecision about airlinealliances and the ‘efficiency test’ applicable to them(see further under mergers).

This notion of further differentiation among relevantmarkets in the light of (sufficient) differentiation ofcustomer preferences, while allowing an inevitablegeographical element in the product market defini-tion, is harder to apply in telecoms. First, with rapidlyincreasing customer orientation of providers/opera-tors and the highly specific preferences of multina-tionals, it becomes increasingly difficult to define therelevant product market(s) in the telecoms sector.Second, the network weighs far more heavily intelecoms than in airlines. For packet-switched serv-ices (such as the Internet) tariffs are independentfrom distance and destination, so route-based choicesare irrelevant. The latter have some relevance forcircuit-switched service (e.g. voice telephony) butthe competitive overall tariff package (which is abundle of services combining distance and location)will be designed so as to keep the large bulk ofcustomers ‘in’ the network. Depending on howdemanding coverage and quality needs are (seeLarouche (2000b) for elaboration), the geographi-cal aspects of relevant markets may differ, too. The‘death of distance’, moreover, may render theactual routing of the electronic services irrelevant,as long as price and quality requirements are met.

All of this means that the classical textbook ap-proach of identifying a geographical market as anarea where conditions of competition are homoge-neous no longer applies (except for regulatory bar-riers); the product and geographical analysis over-lap in (these) networks markets. There is goodreason to be concerned about the market definitionproblem in the light of the new telecoms package, tobe adopted late 2001 or early 2002. In the draftframework directive for telecoms (cf.COM(2000)393 of 12 July 2000), a kind of ex-antemarket analysis is proposed, but, by definition, notbased on the specific offering of the firms in aspecific case, and the demand substitutability atstake in such a case.

Beyond the examples of telecoms and airlines,matters are still hazy or untested. One could arguethat ‘electricity’ is no longer a homogeneous good inso far as customer preferences are concerned.Indeed, if the relevant product market is made up ofall goods which are equivalent in the eyes of thecustomer, then different relevant product marketsfor electricity could be distinguished when peak-load contracts, switch-off contracts, and back-upcontracts are widely used. For gas, back-up con-tracts have also emerged; furthermore, gas con-tracts are more often separating gas supply (orpurchase) and storage obligations. Unlike telecomsand airlines, however, the blurring between geo-graphical and product aspects of relevant marketsdoes not seem to occur. The geographical market inelectricity and gas is likely to remain national for awhile because of a series of obstacles to cross-border trade (see also section VI, below).39 Indeed,the recent Commission approval of Fiat’s acquisi-tion of Montedison, with a tiny stake of EdF, solelyfocused on the Italian energy market.40

In the postal sector, new services are emerging.Moreover, as alluded to before, (non-parcel) postalservices are subjected to sharp competition fromfax, e-mail, and the Internet, for example. Thisprompts the question of what the relevant productmarket is, in terms of substitutability. Particularly forlarge business mailing (including direct mail), thedegree of substitutability could be such that therelevant market would be defined much more widely.However, the Postal Notice41 does not seem todiscuss this issue. Given the fact that, nowadays, thereserved postal services are still defined ratherbroadly, the inclination will be first to verify whethernew services fall outside the scope of monopoly oruniversal service. Indeed, in the Corbeau case, theservices at stake were clearly of the value-addedtype (e.g. collection at the premises of the client) butotherwise traditional. In hybrid mail, however, postalitems are generated electronically. In a recent caseagainst Italy,42 a decree had reserved these activi-ties entirely for the incumbent operator. The crux isthat this hybrid mail carries the guarantee that the

39 See Slot (2000) for similar considerations about the relevant markets in energy. Note that a recent survey by a Commissionofficial—in his personal capacity—does not even mention the problem of defining relevant markets in energy (Albers, 2000).

40 See Press Review IP/01/1229 of 28 August 2001, European Commission.41 Commission Notice on the application of the competition rules to the postal sector, OJ EC C39 of 6 February 1998.42 See 30th Report on Competition Policy 2000, EC (2001) 694 of 7 May 2001, under item 2.4.5 (pp. 46/7).

451

J. Pelkmans

delivery will be completed at a pre-determined dateor time. This feature means, according to the Com-mission, that it is a separate market, very differentfrom traditional delivery services.

Finally, as to rail, it is too early to discern any ‘policy’at EU level. In high-speed rail services and (long-haul) freight services, competition ‘on track’ isdesirable. In high-speed rail, given the distinctions inbusiness vs economy (or ‘tourist’) and peak/off-peak, pretty much as in airlines, slot trading couldeventually lead to distinct relevant markets, with ablend of product and geographical characteristics.There is both complementarity with all airlinesservices (as feeders to airports) and some degree ofsubstitutability with point-to-point short-haul air serv-ices. In freight, competition on track would probablyrequire ‘on-line’ trading of slots and capacity alloca-tion on alternative routes, and currently these con-ditions may be too demanding.43

(v) Regulation via Merger and AlliancesControl?

Merger control has rapidly become the most impor-tant tool of EC competition policy. Cases of abuseof dominance are rare, anyway, at EU level. AndArticle 81 cases may total perhaps around 500 or soever since the EEC began, whereas for the firstdecade of EC merger control (since 1990) the orderof magnitude is around 1,000, with annual notifica-tions nowadays running at more than 250 a year. Innetwork markets the introduction of competition atEC level, even if very partial, often gives rise torepositioning via mergers, as well as to new com-mercial formulae often via alliances. It can beshown that merger control is used by the Commis-sion to promote competition and access in networkmarkets.

First, although mergers and alliances should beanalysed on their (anti-)competitive aspects as theEC Merger Regulation prescribes, and subject togiven regulatory constraints, in several networksectors there would seem to have been cases where(national) liberalization has been spurred to facilitate

merger approval. An early and clear example is thedomestic acquisitions of Air France (of Air Inter andUTA) in 1991, accepted by the Commission only oncondition of faster deregulation and sufficient ac-cess to Paris airports. In 1999, the Commissionprohibited a joint venture between EdF and Dreyfusto do business in France (the business here iselectricity trading), because French eligible custom-ers suffered from a lack of choice. Only aftersufficient French liberalization will this prohibitionbe lifted. Conversely, the highly limited (or strictlynational) progress in liberalization may hinder thepositioning of forerunners, anticipating the newcompetition (e.g. early takeovers by KLM whichhas hardly any home market to ‘give access’ to).Second, the customer-based relevant market defini-tion, gradually emerging from airlines and telecomscases, might begin to be applied to postal servicesand rail, later in the liberalization process.

Third, mergers and alliances have been ‘used’ bythe Commission to impose certain degrees of‘unbundling’ and/or otherwise ‘create’ access orentry for third parties. In pay-TV and media/telecomscases the approach has been dictated by a searchfor separation of content and networks, for exampleby legal/accounting separation requirements. Boththe networks and the content should be open(ed) inspecified ways. This overall approach is in line withthe new telecoms package, based on the samedistinction.44 In airlines, the issue is about alliances,in particular the recent ‘deeper’ ones. Such alli-ances are likely to generate benefits such as highernetwork advantages for long-haul and sometimesshort-haul passengers, in addition to better service(e.g. frequency, wider reach with more viable con-nections) and cost reduction. However, they raisebarriers to entry in congested airports and boostmarket power via the greater advantages of fre-quent-flyer programmes. Airport slots become aneven greater problem because member airlines in analliance will ‘internalize’ slot swapping in majorairports—allocation is improved among the mem-bers only. As to ‘thin’ routes (e.g. to the Balticcountries), the so-called competition between a fewairlines flying there (from western Europe) reduces

43 I am grateful to Loris di Pietrantonio for insightful discussion.44 In the absence of sector-specific regulation and any comparable precedent, especially the merger of Vivendi/Canal+/Seagram

proved to be extremely demanding for the Commission. Many relevant product markets were identified, and the problematic onesin terms of competitive concerns all had to do with the strength of the (joined) parties in content, which could be leveraged at thenetwork level. See Larouche (2001) for a critical assessment.

452


to a trickle once alliance membership spreads toseveral of those airlines. Therefore, alliances havebeen subjected to a range of commitments, includingthe freeing of slots and, in a few cases, access tofrequent-flyer programmes for passengers fromcompetitive feeders or new entrants. A problem hashowever emerged for transatlantic flights, a keyissue for the activities of new alliances. All sched-uled airlines except Virgin are network-based op-erators—and Virgin survives probably in part be-cause of the persistence of very high fares on theLondon–USA routes (owing to congestion inHeathrow and the absence of an open-sky agree-ment). Once network benefits of alliances arerecognized in full, the efficiency test might well leadto rather favourable terms of approval (typicallysupported by the Directorate-General for Energyand Transport in the Commission and the flagcarriers, as well as, broadly, by the USA); if therelevant market for time-insensitive travellers, andperhaps even for a part of time-sensitive ones, doesnot include one-stop (hub-network-based) flightsvia (e.g.) Amsterdam or Paris, merger control willbecome very restrictive (typically supported by theDirectorate-General for Competition and the UKcompetition authorities).45 For the proposed alli-ance (OneWorld) between American Airlines andBritish Airways, the Commission would impose asignificant release of slots, something the Commis-sion could neither do in the case of a fully fledgedmerger (for those slots would be assets, sold for aprice) nor on the basis of EC regulation.

In electricity, the Commission views cooperationagreements between suppliers as pro-competitive ifthey allow the respective companies to enter newelectricity markets for trading at exchanges ofnetwork services, or if they enter new geographicmarkets (see Albers (2000, p. 279) for many exam-ples). Multi-energy mergers may well be anti-com-petitive, for instance, as in the cases of Tractebel/Distrigaz and Nesti/IVO, when a dominant electric-ity supplier gains control over a (dominant) gassupplier or wholesaler; these cases were approvedonly after divestiture of the bulk gas sales business.VEBA/VIAG was the first merger between incum-

bents which had to be assessed by the Commission.Following two simultaneous mergers (VEBA/VIAGand RWE/VEW by the German Kartelamt) a duopolycontrolling well over 80 per cent of the Germanmarket for electricity delivered from the intercon-nected grid would emerge. Given a host of factors,46

the Commission found a case of collective domi-nance, conducive to coordinated effects. Approvalfollowed upon a range of divestments as well as animprovement of the basic rules governing transmis-sion through the network operated by the twoleading interconnected entities.

The quasi-regulatory function of merger controlbecame even more prominent after a series ofannounced takeovers by EdF in Germany, Italy, andSpain. EdF being still de facto a vertically integratedmonopolist, fully state-owned, and operating in theleast-opened-up national market, posed a greatchallenge to EC merger control. Despite the enor-mous political pressures exercised on the Commis-sion, EC merger control alone cannot and should notsubstitute for the failure to achieve legislative progresson further opening up at EU level. Hence, thepolitical tension in the Gothenburg summit of June2001 when France did not (yet) budge under thepressure of the Commission and most other prime-ministers. The ‘solution’ in the Italian case illus-trates the difficulties. Italy hastily adopted DecreeNo. 192 on 25 May 2001, severely limiting the votingright of incumbents from other EU countries inItalian (gas and) electricity companies. Thus, EdF’sparticipation (via a Fiat-led consortium, calledItalenergia) was reduced to a mere 2 per cent whentaking over Montedison, the second-largest elec-tricity generation company in Italy (though far be-hind ENEL). Given this new configuration, theCommission was of the view that EdF does not haveany controlling position, and hence the (consider-able) exports of EdF to Italy do not have to be takeninto account. This constrained ‘solution’ was prob-ably inevitable. However, it is unsatisfactory for atleast three reasons. First, if anything, reciprocitynow works the wrong way: rather than serving as anincentive for further liberalization it has now givenrise to major restrictions. Second, the Italian Decree

45 In Brueckner (2000), an empirical analysis shows that intra-network fares compare highly favourably with the fares of identicaltrips based on interlining. The case at hand was the Star-Alliance. See also Ryan (2000) for a critical assessment of the Commission’sremedies for airlines alliances.

46 See Press Review IP/00/613 of 13 June 2000 or 30th Report (op. cit.), item 26.

453

J. Pelkmans

is likely to be incompatible with EC law under thefree movement of capital and the right of establish-ment. Third, if the decree is withdrawn or foundincompatible, EdF’s voting rights may increase to acontrolling position. In that event the Commissionrequires re-notification under the merger regulation.

VI. INTERNAL MARKET

For all the progress and benefits EU-level networkliberalization has accomplished since the late 1980s,none of the network industries is operating in agenuine internal market, as yet. This failure isessentially due to two factors. First, as noted at theoutset, network industries were not part of the EC-1992 programme as first conceived in 1985; theywere an afterthought. As a result, an ad-hoc ap-proach emerged, initially driven by the Competition

Directorate-General of the Commission, not theInternal Market one. Some broad, target-settingstrategies have attempted to incorporate the liberali-zation of network markets, but in a far too rhetoricaland general fashion. This was true for Delors’sWhite Paper of December 1993 on growth, employ-ment, and competitiveness and, again, with theambitions of the 2000 Lisbon summit. The conse-quence is that, time and again, sectoral battles haveto be fought which enhance the influence of sectorlobbies, including labour unions. Second, nationalgovernments and their civil servants in sectoralcommittees in the EU circuits often take ambivalentviews on the liberalization process. While there isnowadays a much wider and more forceful supportof liberalization, for the sake of economic perform-ance, the implication that, under fundamental ECprinciples, this must be pursued in an internal marketframework is (purposefully?) ignored.

Table 3Why No Internal Market for Network Industries?

Broadcasting Cable/commercial stations remain national/regionalPay-TV nationalPublic broadcasting national/regional; via discs or(optional) cable cross-border reception possible

Telecoms Licensing is still too discretionaryNo European licencesNo mutual recognition(Hope) (mobile?) companies may create virtual Euro-pean networks

Postal services Large part still reserved, including all cross-borderReims II, quality up, prices converge, yet no removalof frontiers

Air Air traffic control, absurd splinteringCongested airports; slot ‘rights’ grandfatheredDistortions in the internal market owing to bilaterals

Rail Only beginning in high-speed passengerFreight limited to freeways (with distortions)Inter-operability problems hugeSevere capacity constraints

Electricity Capacity constraints for interconnectors (built forsecurity of supply in EU, not for free trade)Access and transmission fee issuesInvestment incentives are lacking

Gas As for electricityInter-operability and gas quality problems

454


Table 3 provides a bird’s-eye view on restrictions orbarriers in seven network industries, which preventthe internal market from coming into being.

Some of the listed obstacles form major restraintsfor the liberalization process and its potential wel-fare benefits. One example consists in the distor-tions of the internal market owing to the bilaterals inair transport with third countries. Suppose a Euro-pean no-frills airline finds that competitive entry inselected inter-continental routes from (say) Italy orGermany is profitable. This challenge is almostcertainly prohibited by the respective bilaterals withthe relevant third countries. In fact, this also appliesto the incumbents. Though there is free movementof services and free establishment, even if (say)KLM would create an Italian subsidiary, it wouldalmost certainly not be allowed to compete on suchroutes from Rome or Milan unless special permis-sion (from the third countries) is obtained. Memberstates have thus far not been keen to ‘Europeanize’the bilaterals from the EU end (which, admittedly,would require renegotiation), although the ChicagoConvention formally allows it. Without a break-through agreement with the USA (the proposedTransatlantic Common Aviation Area), there is nohope for the other bilaterals. In electricity, oneamong a range of problems consists of the invest-ment incentives for cross-border grid inter-connectors—when the incumbents own the presentones, would they invest aggressively, only to seecompetition become more fierce as a result? Ofcourse, these incentives also depend on the (future)decisions about cross-border transmission fees.

The negative feedback of the failure to establish aninternal market for network industries is consider-able. More competitive challenge, more infra-structural investment, and greater variety and choicewould be generated by a vigorous pursuit of a singlemarket, exactly the kind of benefits the EU leaderspromised the Union would deliver in the period up to2010 according to the seductive rhetoric of theLisbon summit.

VII. CONCLUSIONS

Since the late 1980s the European Union has madegreat strides in the liberalization of network indus-tries. In part this was prompted by early national

initiatives and regulatory emulation by a few EUmember states. But also when concentrating solelyon EU-level regulatory liberalization as well compe-tition policy in these sectors, the progress is appre-ciable. The contrast with three decades of maintain-ing a taboo on addressing national exclusive rightsfor utilities as if they were carved out of the internalmarket is striking.

The present article assumes a horizontal perspec-tive, juxtaposing different network markets whilefocusing on the rules and policies at EU level only.All major network industries except water andsewerage (because there is no cross-border trade)are now subject to processes of liberalization andminimum regulation. Also EC competition policy isactively applied.

Despite the panoramic approach of this article,which precludes detailed analysis on the manytechnical questions involved, a few more generalconclusions can be drawn.

First, the EU should come down from the high-handed rhetoric of the Lisbon European Council(March 2000) and formulate a well-consideredstrategy for network industries. The strategy oughtto make the economic case for liberalization, takingaccount of adjustment costs (but not as an argumentof—temporary—losers to block progress) in themedium run. While it is undeniable that sectoralspecifics render a simplistic horizontal liberalizationapproach impossible, this does not mean that acommon overall framework with clear goals andcalendars should not be developed at EU level. Firmcommitments backed up by a sense of urgencycomparable to that of the EC-1992 programme or,nowadays, of the further liberalization of financialservices and capital markets (a tight calendar up to2005) should greatly help to remove the inhibitions torestructure, to improve business performance andconsumer satisfaction, and to invest in cross-borderand other bottleneck infrastructure. At the moment,apart from the vague generalizations of the Lisbonsummit, the only horizontal ‘forum’ is the Cardiffprocess, based on a so-called ‘open coordination’method. However, the great weakness of this proc-ess is that it lacks a coherent analytical and strategicframework against which the multitude of issuescan be assessed. For network industries such aframework should be developed. The socio-political

455

J. Pelkmans

REFERENCES

Albers, M. (2000), ‘Competition Law Issues Arising from the Liberalisation Process’, Journal of Network Industries,1(3), 269–85.

Blum, F. (2000), ‘The Recent Case-law of the European Court of Justice on State Monopolies and its Implications forNetwork Industries’, Journal of Network Industries, 1(1), 55–87.

Brueckner, J. (2000), ‘The Benefits of Code-sharing and Anti-trust Immunity for International Passengers, with anApplication to the Star Alliance’, Working Paper, University of Illinois (Urbana-Champaign).

Church, J., and Ware, R. (2000), Industrial Organisation, A Strategic Approach, Boston, MA, McGraw-Hill.Doyle, C. (2000), ‘Local Loop Unbundling and Regulatory Risk’, Journal of Network Industries, 1(1), 33–54.Hancher, L. (2000a), ‘A Competitive European Energy Market: Creating a Culture of Competition’, Journal of Network

Industries, 1(3), 259–67.

underpinning has been agreed since the Amsterdamtreaty (cf. Article 16, EC) but clear aims, means,and credible calendars are lacking.

Second, a renewed commitment to establish aninternal market for goods and services (if not capitaland technology) of network industries is essentialfor such a strategy. Today’s inconsistencies, omis-sions, and distortions hinder the fully fledged devel-opment of competition and, in causing uncertainty,inhibit especially the infrastructural investment re-quired. Besides the reasons discussed above, theestablishment and proper functioning of the internalmarket in network industries is also made verydifficult by the vexed question of ‘independentagencies’ under the treaty. The 1958 Meroni doc-trine prohibits the establishment of independentagencies at EU level, except via ratified treatyamendment. The underlying, purely legal logic isthat the treaty does not allow the EU institutions todelegate more than ‘executive’ power—the EUinstitutions ought to be able to keep control. Todayin network markets this can be counterproductivebecause it always precludes the option of an ECregulator. In particular, member states want to haveit both ways: they refuse (thus far) to include asimple provision in the treaty that independent regu-lators (agencies) can be established, say, underunanimity but without treaty amendment; at thesame time, they refuse to create efficient andeffective coordination mechanisms at the national/EC interface which would be reasonable alterna-tives for a Community regulator, given that theCommission remains the guardian of the treaty and

can use competition policy, too. While memberstates have regulators, the internal market hascomplex, incomplete, and slow mechanisms; whilethe USA has federal regulators for its internalmarket, the EU does not, although the nature andfrequency of the problems are often the same.47

Beyond these strategic policy issues, what thispaper hopes to show is that a horizontal perspectiveon EU network markets may help to clarify how theweaknesses and omissions in the slow-moving sec-tors might be overcome and how good solutions inone industry may, mutatis mutandis, be used inanother industry. A further refinement of Table 2would be helpful in this respect. In principle, this isalso true for the application of EC competition policyto network industries. However, the problems hereare considerable, not least because it is difficult toconclude what the optimal ‘quasi-regulatory’ pos-ture of competition policy should be. For one thing,this also depends on the stage of the liberalizationprocess. For another, in the EU context, withoutcentral regulators and with recalcitrant memberstates, it is tempting to employ the crowbar ofspecial powers under Article 86/3 or a wide inter-pretation of essential facilities, or to stretch thediscretion under merger control. In the final analy-sis, these matters can best be resolved in an unam-biguous internal market context, with clear regula-tion about remaining market failures and appropri-ate mechanisms to supervise actual or potentialcross-border activities. It underscores once againthe paramount importance of a long-run strategy forEurope’s network industry liberalization.

47 An independent regulator should only be established when a careful subsidiarity test is met. For the application of this testto telecoms, see Pelkmans (1998).

456


Hancher, L. (2000b), ‘Cross-border Transmission of Electricity in the European Energy Market: Regulatory Issues’,Journal of Network Industries, 1(2), 127–55.

Harbord, D., and Yarrow, G. (1999), ‘State Aids, Restructuring and Privatisation’, European Economy, Reports andStudies, No. 3, 89–121.

Laffont, J.-J., Rey, P., and Tirole, J. (1997), ‘Competition Between Telecommunications Operators’, EuropeanEconomic Review, 41, 701–11.

Larouche, P. (2000a), Competition Law and Regulation in European Telecommunications, Oxford, Hart. — (2000b), ‘Relevant Market Definition in Network Industries: Air Transport and Telecommunications’, Journal

of Network Industries, 1(4), 407–45. — (2001), ‘Communications Convergence and Public Service Broadcasting’, unpublished paper.Monti, M. (1996), The Single Market and Tomorrow’s Europe, A Progress Report, Luxembourg, OOPEC, and London,

Kogan Page.Pelkmans, J. (1998), ‘A European Telecoms Regulator?’, in P. Vass (ed.), Network Industries in Europe: Preparing

for Competition, London, Centre for Regulated Industries. — (2001), ‘The GSM Standard: Explaining a Success Story’, Journal of European Public Policy, 8(3), special

issue, 432–53. — Olsen, O. J. (1996), ‘Towards a Single Market in Utilities’, CEPS Working Party Report No. 14, Brussels, CEPS. — Young, D. (1998), TELECOMS-98, CEPS, Brussels, 248.Prosperetti, L. (2000), ‘Some Structural Issues in Regulatory Reform and Market Functioning in the European Electricity

Supply Industry’, in G. Galli and J. Pelkmans (eds), Regulatory Reform and Competitiveness in Europe, Vol.II, Vertical Issues, Cheltenham, UK, Northampton, MA, Edward Elgar, 431–42.

Rodriguez, S. (1998), ‘Les Services Publics et le Traité d’Amsterdam’, Revue du Marché Commun et de l’UnionEuropéenne, No. 414, January.

Ryan, A. (2000), ‘As the Airline Industry Evolved to a Network Industry, How should Competition Policy Adapt?’,Journal of Network Industries, 1(2), 157–83.

Schneider, V., Dang-Nguyen, G., and Werle, R. (1994), ‘Corporate Actor Networks in European Policy-making:Harmonising Telecommunications Policy’, Journal of Common Market Studies, 32(4).

Slot, P. J. (2000), ‘The Impact of Liberalisation on Long-term Energy Contracts’, Journal of Network Industries, 1(3),287–303.

Temple Lang, J. (2000), ‘The Principle of Essential Facilities in EC Competition Law, the Position since Bronner’, Journalof Network Industries, 1(4), 375–405.

MAKING EU NETWORK MARKETS COMPETITIVE · liberalization and privatization. When they preceded EU...

Documents

Transcript of MAKING EU NETWORK MARKETS COMPETITIVE · liberalization and privatization. When they preceded EU...