EN

Only the English text is available and authentic.

REGULATION (EC) No 139/2004 MERGER PROCEDURE

Article 6(1)(b) NON-OPPOSITION Date: 07/10/2005

In electronic form on the EUR-Lex website under document number 32005M3752

Office for Official Publications of the European Communities L-2985 Luxembourg

COMMISSION OF THE EUROPEAN COMMUNITIES

Brussels, 07/10/2005

SG-Greffe(2005) D/205408

In the published version of this decision, some information has been omitted pursuant to Article 17(2) of Council Regulation (EC) No 139/2004 concerning non-disclosure of business secrets and other confidential information. The omissions are shown thus [Ö]. Where possible the information omitted has been replaced by ranges of figures or a general description.

PUBLIC VERSION

MERGER PROCEDURE ARTICLE 6(1)(b) DECISION

To the notifying party:

Dear Sir/Madam,

Subject: Case No COMP/M.3752 - Verizon/MCI Notification of 2 September 2005 pursuant to Article 4 of Council Regulation No 139/2004

1.On 2 September 2005, the Commission received a notification of a proposed concentration pursuant to Article 4 of Council Regulation (EC) No 139/2004 by which the undertaking Verizon Communications Inc. (ìVerizonî, USA) acquires within the meaning of Article 3(1)(b) of the Council Regulation control of the whole of the undertaking MCI Inc (ìMCIî, USA) by way of purchase of shares.

I. THE PARTIES

2.Verizon provides telecommunications services to residential, small business and some large corporate and government customers in various regions of the U.S. These services include local, domestic long distance and international voice telephony services. Verizon subsidiaries and affiliates also provide internetworking, wireless, directories and telecommunications-based information services and systems.

3.MCI (formerly WorldCom) is a global provider of advanced communications connectivity to businesses and governments. It also sells communications services to consumers in the United States. MCI delivers a portfolio of local-to-global business data, Internet and voice services. MCIís portfolio includes SONET private line, frame

1OJ L 24, 29.1.2004 p. 1.

Commission europÈenne, B-1049 Bruxelles / Europese Commissie, B-1049 Brussel - Belgium. Telephone: (32-2) 299 11 11.

relay, ATM and a full range of dedicated, dial and value-added Internet services as well as audio, video, and Net conferencing services.

II. CONCENTRATION

4.On 14 February 2005, Verizon and MCI entered into an agreement, subsequently amended, whereby Verizon has agreed to acquire MCI. Verizon will acquire 100% of MCIís shares.

5.The operation constitutes an acquisition of sole control within the meaning of Article 3(1)(b) of the Merger Regulation and is therefore a concentration.

III. COMMUNITY DIMENSION

6.The undertakings concerned have a combined aggregate world-wide turnover of more than EUR 5 billion(in 2004, Verizon: EUR 57.3 billion, MCI: EUR 16.7 billion). Both Verizon and MCI have a Community-wide turnover in excess of EUR 250 million (Verizon: EUR 8.2 billion, MCI: EUR 2.4 billion), but they do not both achieve more than two-thirds of their aggregate Community-wide turnover within one and the same Member State. The proposed operation meets the thresholds of Article 1(2) of the Merger Regulation and therefore has Community dimension.

IV. COMPETITIVE ASSESSMENT

7.The main markets concerned with the operation are the markets for Internet connectivity, for Global Telecommunications Services (ìGTSî) and for International Voice Telephony Services (ìIVTSî).

A. Internet connectivity

1. Market definition

8.The Internet works as a ìnetwork of networksî. A local Internet Service Provider (ISP) who wants to offer Internet services to end-customers has to connect with other networks in order to allow his end-customers to exchange traffic with other end-customers / content providers beyond its own local network. In order to reach networks in far distance, an ISP has to connect to larger networks which can link both ISPs to each other.

2Turnover calculated in accordance with Article 5(1) of the Merger Regulation and the Commission Notice on the calculation of turnover (OJ C66, 2.3.1998, p25).

9.Such connectivity can be acquired either by peering (the mutual and free exchange of traffic between two networks) or by transit (the provision, for a fee, of access to the Internet via a network). Peering offers access only to the customers of the other network whereas a transit supplier gives access to the whole Internet. Peering usually occurs between ISPs of like size and geographical reach. With increasing asymmetry transit is used more often. Smaller networks usually can obtain connectivity to the larger ones only by paying for transit. Most ISPs need to complement the connectivity obtained through their peering relationship by purchasing some transit.

Commissionís precedents

10.In cases WorldCom/MCI and MCI WorldCom/Sprint, the Commission defined top-level (or universal) Internet connectivity as a separate market. It found that only top-level or top-tier Internet connectivity providers were capable of delivering complete Internet connectivity entirely or in the great majority through their own networks and peering agreements with other top-tier providers, thereby constituting the highest level in the Internet-hierarchy. Secondary Internet connectivity providers (or second-tier providers) may be able to deliver some of their own peering-based connectivity, but have to supplement it through bought transit. It was found that second-tier ISPs could not avoid continuing to buy transit from the top-level networks and that they could not provide a competitive constraint on the prices charged by the top level networks. A hypothetical monopolist consisting of all top-level providers would provide transit to all other ISPs and could profitably raise prices. In the more recent case KPNQwest/Ebone/GTS, the Commission referred to the provision of ìwholesale internet connectivity,î without specifying whether the party involved was a top-level or second-level provider.

11.11. With respect to the geographic market, the Commission has taken the view that the market for the provision of top-level Internet connectivity is worldwide. In case KPNQwest/Ebone/GTS, however, the Commission considered whether the evolution of Internet connectivity in Europe (e.g., the continued development of European Internet connectivity providers) altered the conditions for the supply and demand for Internet connectivity in a manner that led to a distinct European geographic market (the question was left open).

Partiesí view

The Parties argue that in the five years since the MCI WorldCom/Sprint decision, there have been changes that have substantially altered the distinction previously drawn by the Commission between top-level Internet connectivity and general Internet connectivity. These changes relate to three considerations that led the Commission to define top-level Internet connectivity as a separate market: the relationship between peering and transit, the developments in technologies and routing practices and the diminished weight of the US in the global Internet. Nonetheless, the Parties submit that whether the relevant product market should be defined as top-level Internet connectivity, or more broadly as general Internet connectivity, can be left open since the transaction will not raise any competitive issues under either approach. Similarly, the Parties submit that the geographic market definition can be left open, since the transaction will raise no competition issues on either a global or an EEA-wide basis.

The relationship between peering and transit

12.The Parties state that, in recent years, transit prices have fallen dramatically. Telegeography (a consultancy company) refers to London median prices of US$530 (Ä573)/month (for 155 Mbps ) in October 2000, falling to US$62 (Ä77)/month (for 155 Mbps) in June 2004. More recent IDC (another consultancy company) data indicates that the current monthly average price of 155 Mbps is between US$35 (Ä43) and US$50 (Ä61). In essence, transit prices have declined by approximately 90% in less than five years. MCIís listed prices and Verizonís own experience are consistent with this trend. Investments in new fibre in recent years combined with technical improvements would have resulted in significant overcapacity among Internet connectivity providers. This, in turn, has put downward pressure on the price of such capacity.

13.The Parties argue that the drop in transit prices has made purchasing transit significantly more attractive and peering relationships correspondingly less important. For instance, the number of requests for peering received by MCI has dropped [Ö] between 2002, 2003 and 2004.

Developments in technologies and routing practices

14.The technologies and routing practices at stake are mirroring, caching, content delivery networks (CDNs) and multi-homing. The Parties argue that network operators by using these content delivery management tools move content closer to end users. As a result, demand for Internet connectivity would rely less on top-level providers. They explained that for instance at least [Ö] % of MCIís customers were multi-homed with either AT&T, Sprint, Level 3 or Savvis. The Parties argued that content exchanged using peer-to-peer software is even more decentralised (and closer to the ìedgeî of the Internet) than content stored on cached or mirrored sites. The parties were however unable to identify a specific tool or study measuring the influence of these developments.

Diminished importance of the US to the global Internet

15.Over the past years, the Parties state that the relative importance of the US to the global Internet service sector has decreased. The number of Internet users in the European Union (215 million) would now be greater than that of the United States (200 million). As regards content, content accessed over the Internet would have become increasingly less US-centric because of different influences, including sharp growth in the size of the non-English-speaking online language population, increasing use of national domain names, changes in end-user content preferences and the sharp growth of file sharing traffic.

6 Megabits per second.

7 A ìpushî technology that permits websites to be replicated on servers other than the origin server, to bring the content close to the end user.

8 A ìpullî technology that allows content hosted on distant servers to be sent to a cache closer to the user at the same time that it is sent to the user; subsequent queries for the same content are responded to by the cache.

9 Multi-homing allows hosts and content providers to switch some or all of their traffic among various backbone providers. Multi-homing has also reduced the impact of peering relationships, because multi-homed hosts and content providers send traffic directly to the network to which customers are directly or indirectly connected, rather than across a peering point.

10 File sharing can be defined as the activity of making files available to other users for download over the Internet.

11 Internet World Stats, Internet usage in Europe, March 31, 2005.

17.The market investigationís results generally confirm that the Commissionís previous market definition remains accurate. Global connectivity can still only be provided by the largest ISPs who still can be characterized by their ability to achieve their connectivity (almost) exclusively by peering and therefore do not depend on buying transit from a higher level ISP. The market investigation has not brought about indications that these top-tier providers could be circumvented by interconnecting the large number of regional and local ISPs directly. In order to achieve global Internet connectivity, still (direct or indirect) access to one of the top-tier ISPs is unavoidable which supports the assumption of a separate market for global Internet connectivity.

18.The market investigation has shown that 2tier ISPs are unlikely to be able to generate enough global traffic to warrant building out to a number of peering points around the world at cost effective rates. Indeed, it is still the case that 2tier providers find it difficult and even impossible to peer with tier 1 providers.

19.The market investigation has generally confirmed the global scope of Internet connectivity. Only a few market participants considered that there are global as well as regional top-level providers. A particular backbone qualified as ìtop levelî will according to this view vary by region, not paying any backbones in the US, but paying in regions where it has more limited presence. Top-tier Internet connectivity providers could be now considered on a regional basis (US tier1, EU tier1). However, as the transaction implies one of the top-level global providers, the transaction has to be scrutinized at this level.

20.The Commission has assessed the arguments submitted by the Parties which according to their view have significantly changed the market in the past years requiring a new delineation of the relevant market. While having confirmed the indicated developments in principle, the market investigation has not supported the Partiesí suggestion that these changes would lead to a new market definition.

21.As regards the relationship between peering and transit, there was general confirmation that the relative cost of purchasing transit as compared to settlement-free peering with a top-tier provider has diminished. However most respondents still considered that the economics of peering were more favourable than that of getting transit. The market investigation has shown that settlement-free peering could be cheaper than transit by a factor of 8-10 on a per Mbps basis as far as direct costs are concerned. Transit might be an option when the cost of reaching the peering point exceeds the cost of transit in that region.

22.As regards the developments in technologies and routing practices, the use of multi-homing has not changed significantly and the use of caching has decreased over the last five years. This is because bandwidth costs have declined whereas circuit sizes have increased making this technique less effective. As regards peer-to-peer traffic, respondents indicated that such traffic had grown in absolute traffic terms but that the ratio of such traffic over total traffic has actually been declining on some networks and that it was likely to generalise. In the long run, they considered that such traffic was likely to wane due to copyright enforcement actions. This trend is also foreseen by the consultancy company RHK that believes that the growth of peer-to-peer traffic has hit an inflection point. All in all, most did not believe that the reliance on top-level backbones had been or would be affected significantly by these techniques. They do not remove the 2tier Internet providersí need to buy transit services from top-tier Internet providers.

23.As to the relative importance of the US to the global Internet, the ratio of traffic either received or transmitted to the US decreased significantly over the last years. Regarding content, there has been increased supply of local content geared to local tastes (a Yankee Group report, citing Telegeography, stated that 2/3 of Europeís Internet traffic remained in Europe by the end of the year 1999, compared to less than 1/2 just one year earlier; Telegeography data indicates that in 2004, on the top 25 international European routes, 68% of capacity, and 60% of traffic was intra-European). However it is not sure this decrease in ratio has brought any change to the US-centric nature. With the global growth of customer base, the proportion of users demanding transit to the US traffic has become smaller. Despite the rapid growth of intra-Asian and intra-European traffic and network capacity, three factors continue to keep the US central to the global Internet. First, the US is home to a great deal of content which is accessed by users around the world. Second, traffic between regions often must transit via the US before travelling to its final destination (traffic between Asia and Europe is routinely sent via the US due to cheaper capacity). Third, the US is home to roughly 24% of all Internet users worldwide.

24.For all these above reasons, the changes that have occurred for the last years do not seem to have led to a significant modification of the hierarchical nature of the Internet; the previous Commissionís conclusions relating to the existence of a global top-level Internet connectivity separate market appear to be still valid. However, this question can be left open as it will have no impact on the competitive analysis of the transaction.

25.If global top-level Internet connectivity is identified as a relevant product market, there would be formally no horizontally affected market. MCI owns one of the leading Internet networks. However, Verizon is not a top-level provider as it is a regional provider of Internet access in the US (it routes a majority of its traffic through transit arrangements and, with the possible exception of AOL, none of Verizonís peers qualify as top-level Internet connectivity providers).

26.Nevertheless, an increase in MCIís market power is possible due to the potential post-merger integration of Verizonís network and traffic into MCIís global network. This might lead to a change in the relative weight of the various top level players.

27.Estimating market sizes and shares in the Internet sector raises significant methodological issues in view of the absence of a consensus on the preferred approach to, and unit of, measurement. In MCI WorldCom/Sprint, the Commission used traffic flows and revenues criteria to derive market shares. The Parties have provided estimates based on traffic flows made by RHK. RHK provides estimates of the traffic shares of the seven largest Internet backbone providers in North America. At the end of 2004 there were estimated 416 petabytes (1 petabyte = 1,024 terabytes = 2^50 bytes) of data per month being transferred over the Internet in North America. This led to the following figures:

14Company

Petabytes Per Month ofShare of TotalShare of the first 15Traffic (End of 2004)Internet Traffic seven

Company A

52.33

12.58%

21%

Company B

51.31

12.33%

20.6%

Company C

45.89

11.03%

18.4%

30.87

7.42%

12.4%

Company E

25.46

6.12%

10.2%

Company F

19.33

4.65%

7.8%

Company G

15.19

3.65%

6.1%

Others

175.62

42.22%

28

The merged entity would rank fourth behind three competitors having market shares between 12.5% and 11%. MCI would have 7.4% and Verizon [LESS THAN 3] % in a North American market. However, the figures derived from that study are probably biased to a significant extent by the fact that they add together traffic that belongs to various steps in a hierarchical order of the industry.

The parties and respondents to the market investigation have considered that at least seven backbone providers possess large networks and exchange traffic on an almost entirely settlement-free basis with their peers. These are AT&T, Level 3, Sprint, MCI, Qwest, AOL, and Savvis. According to RHK figures and adding corresponding data of Verizon, MCI would still be the fourth player with 12.4% market share.

Another means to measure the size of the respective market participants could be to look at an AS (Autonomous System) - based ranking. The AS ranking is a measure of the connectedness of an IP network to the rest of the public Internet. As of June 2004, there were about 18 000 ASs active in the world with about 42 500 connections between them. Telegeography tracks the 50 top Internet providers by AS rank. Even if it indicates the degree to which an AS is ìwell connectedî, it is not necessarily reflective of market shares (the data reflects the connections between ISPs or end-user customers with networks such as ISPs and large corporates but not the connections to customers without an assigned AS such as SMEs and consumers). But it is an indicator of the relative size of the market players. Vint Cerf, at that time a senior vice president at MCI, argued for instance that ìthe number one ranking illustrates MCIís role in delivering critical Internet services for our customers and the entire Internet communityî.

AS connections Share of totalShare of top-tierTrend 2004/2002

3034

7.1

29.7

-5.5

AT&T

1966

4.6

19.3

38.1

Sprint

1842

4.3

18

14.9

Level3

1167

2.7

11.4

15.6

Qwest

1074

2.5

10.5

10.3

Savvis

664

1.6

6.5

145.9

AOL

452

1

4.4

118.3

* top-tier providers as considered by MCI: MCI, AT&T, Sprint, Level3, Qwest, Savvis, AOL = 10 199 AS connections.

It can be seen that the MCIís relative competitive position in the Internet sector has declined since 2002 but that according to the AS ranking criteria it still retains a leading position. Verizon does not appear on the list at all. At worst it would have a share of total AS below 0.3%. MCIís market share of top-tier AS would increase by only 0.8 % if Verizonís connections were added to those of MCI.

In revenues, from IDC data, the Parties would represent in 2003 on a total backbone North American market (gathering dedicated Internet access and wholesale upstream transit into total backbone revenues) a share of 14.3% (9.1% for MCI and 5.2% for Verizon). Their main competitor would be ATT/SBC with 19.8% (14.7% for ATT and 5.1% SBC).

18.Case still under procedure in the US (on parallel tracks with Verizon/MCI). It did not have a community dimension and has been cleared in several Member States in the EU.

19[Ö]

8

33.The Commission also sought to compute market shares on the basis of traffic volumes exchanged between various peers. Such computations needed to make some assumptions for the repartition of traffic exchanged by smaller peers for which data was not available. Independently of the assumptions made, three conclusions may be reached. First, there are always three players with higher market shares than MCI. Second, MCIís market share remains below a 20% level. Thirdly, none of the other market players exceeds a market share of 25%.

Commissionís conclusions from the market investigation

34.On the question whether this accretion of market share could have an adverse effect on competition, the Parties state that the growth in the Internet sector combined with the decreases in the prices of the various inputs necessary to provide Internet services encourage new entry and expansion. According to the Parties, this can be seen in the volatility in the rankings of the Internet connectivity providers that currently occupy the top ten positions in the Telegeography AS Rankings (two such providers were not on that list in 2000, and two on the list in 2000 are no longer on the list in 2004; in addition, Savvis replaced Cable & Wireless USA after the former acquired the latterís assets).

35.The majority of the respondents see no material effect of the merger on MCIís position in the global Internet connectivity market. Many respondents assume that its position will be only slightly strengthened. They also tend to confirm that MCIís position has been weakened since 2000 notably because of the bankruptcy procedures it went through.

36.Anticompetitive concerns have been raised by a few respondents. The first issue relates to the fact that Verizon has control over local and special access to business customers requiring high speed connections as part of GTS packages in its regions. It is alleged that Verizon would have the incentive and ability to favour its downstream affiliates (MCI) to the detriment of the other Internet backbone providers. This question is assessed in detailed below in the GTS part of the decision.

37.A second concern deals with the ability and incentive for Verizon/MCI, having allegedly reached a greater size than all the rest of Internet connectivity providers, to replace peering arrangements with other Internet backbone providers by transit arrangements or to decrease the quality of connectivity for the peering partners. The merged entity would then remain the only major global Internet connectivity provider independent of transit with an ability to raise rivals costs or decrease their qualities.

It has been also argued that the merged company would be ìeyeball-heavyî and that would confer to it an increased market power. ìEyeballsî are meant to characterise primarily residential customers with high incoming and low outgoing traffic flows, in contrast to content providers who generate high volumes of outgoing and relatively low volumes of incoming traffic. As a consequence, this would create a traffic imbalance likely to create incentives for the ìeyeball-heavyî network to de-peer content-heavy backbone service providers. However, end users will not stay on a network that does not provide them access to the content they seek. As a result, a backbone that is ìeyeball heavyî has incentives to ensure that its peering and transit decisions enable the delivery to end-users of the content that they want.

39.The small increment in market shares by the addition of Verizonís traffic does not factually support this concern. The Commission has checked nevertheless whether the combined entity would enjoy, post-merger, such traffic with some peers that its incentives to continue peering with that peer would be affected. It must be noted that even if such incentive would arise, it is unlikely that de-peering would have a significant effect on competition given the market share of the peers in question.

40.MCI originally required a 1.5:1 traffic exchange ratio in considering peering applications. In August 2004, MCI adopted a ratio of 1.8:1. MCI notes that, since the revision to the traffic ratio assessment criteria, it has not de-peered any peer on the basis that it does not meet the traffic ratio. Neither has it refused to upgrade or threatened to terminate any peering agreement. This ratio is comparable to the competitorsí ratios as published in their peering policies: the maximum traffic imbalance runs from 1.5:1 (Qwest) to 2:1 (AOL, Level3, Savvis, Teleglobe) or 2.5:1 (Broadwing). So there is no reason to fear at first sight any change in the peering policy of MCI/Verizon.

41.The parties argue that assuming all Verizonís transit traffic would eventually be migrated to MCIís network, it would not negatively impact MCIís current peers. The integration of Verizonís traffic into MCIís network would lead to a slight increase for the majority of MCIís peers. However, for those peers with current ratios below 1.8:1, the new ratios would remain below that level. For three of the four peers whose ratios currently exceed 1.8:1, the ratio would decrease. The ratio of the fourth peer would increase (however, its traffic ratio with MCI already exceeds 1.8:1). However it is highly unlikely that if that peer was de-peered there would be any material impact on competition.

42.A third concern mentioned in the market investigation relates to an alleged risk that the merger could lead to a duopoly of Verizon/MCI and SBC/AT&T. The parallel mergers would give both companies the ability and incentive to favour each other in peering arrangements and potentially engage in free peering only with one another and not with smaller, in light of the greatly increased size and scope of both leading ISPs due to the mergers. The widening market share gap would then give the merged firms an increased incentive to degrade connections with and/or to stop peering with other top-level providers to impose discriminatory and above-cost transit fees and to engage in other practices that would eventually lead to shared dominance of the tier 1 ISP business by Verizon/MCI and SBC/AT&T.

43.The traffic market shares as seen above do not support such a concern because of their low levels. It is only when considering market shares on the basis of AS ranking that the two new merged entities would together reach a level of 50% of the market.

44.In any event, irrespective of the merits of the other transaction, the Verizon/MCI merger does not lead in itself to a modification of the structure of the market and of the incentives of market players. As seen above, Verizon does not bring to MCI any weight that can change its actual incentive as regards peering. In relation to Europe, the combination of MCI and Verizon will not increase the traffic carried by the merged entity because in Europe, Verizon does not carry traffic at all. Additionally, there remains other players with sizes not dissimilar to that of the two alleged duopolists.

In light of the above, it can therefore be concluded that the transaction will have no material impact on competition in the market for global Internet connectivity and will not lead to a significant impediment of effective competition in the common market and the EEA.

B.Global Telecommunication Services (GTS)

46.The notified transaction may also have effects on the market for the provision of global telecommunication services where MCI is active. Verizon is not a provider of GTS. However, because it owns the local loop in a number of areas in the US, it is a provider of an input used by GTS providers. GTS customers generally require high-capacity local connectivity ñ that is, they require the physical connection to be capable of carrying high volumes of voice and data traffic. These connections are normally provided over dedicated facilities that run between the customerís various premises, as well as from those premises to the local, regional and international networks. They are generally referred to as ìdedicated accessî or ìspecial accessî. Verizon provides special access in a number of US-states; MCI is also active on the special access market but to a much smaller extent (see below). If GTS providers want to connect their international networks to customer locations in Verizonís area in the US, a major provider of special access as necessary input is Verizon. The potential effect of the merger in the GTS market is, therefore, a vertical one.

The following paragraph examines the effects of the vertical integration of Verizonís special access activities with GTSí activities of MCI.

Market definition

GTS

Relevant product market

48.The Commission has in the past examined the GTS sector in several decisions. In its MCI WorldCom/Sprint , the Commission defined GTS as: "telecommunications services linking a number of different customer locations, generally in at least two different continents and across a larger number of different countries. They are generally purchased by MNCs [Multinational corporations] with presence in many countries and a number of continents. The services provided are enhanced services - going beyond the provision of simple services such as basic voice and fax - to provide customers with package solutions including virtual private networks for both voice and data services and advanced functionalities". In the most recent Commission decision dealing with this sector ñ M.3641-BT/Infonet- this definition was confirmed.

49.According to the Parties, GTS are usually provided as packages of different single services, which may include for example retail end-user access; local, national, and international voice telephony and data communications; virtual private networks; intranets/extranets; enhanced voice and data; audio, video, and net conferencing; data security and content management; call and data centres; and advanced facilities and service management. Every bundle of GTS as well as its geographic reach is tailored by the GTS providers to meet the individual customersí requirements. GTS providers are usually able to offer all main types of GTS on their own. In addition, they regularly purchase third-party inputs to complete their offers where they cannot provide a specific service by themselves. Due to these characteristics, the definition of one GTS market comprising all different single services is ñ according to the Parties - appropriate.

50.The market investigation has broadly confirmed this view even though customers indicated to buy GTS services not only in bundles, but also where needed as individual services. However, all major competitors stated to be able to offer all significant different GTS including newly evolving technologies such as MPLS (multi-protocol label switching) which accommodates the simultaneous use of different platforms and therefore provides converged solutions of different services.

51.GTS are moreover different from telecommunications services provided on a merely national basis. It appears that in particular the international ìfootprintî is not easy to obtain by a provider. This footprint covers ñ apart from the cross-border network infrastructure ñ in particular the know-how about business conditions in other countries as well as the corresponding contacts. The internationality of the services provided therefore constitutes an essential element which distinguishes national services and providers from the global ones.

In the light of this, a single market for all GTS is assumed which has to be distinguished from merely national telecommunication services.

23Case No. COMP/M.1741 - MCI WorldCom/Sprint, 28.06.2000.

24Case No. COMP/M.3641 - BT/Infonet, 25.01.2005.

25MPLS is an IP (Internet protocol) -based private platform. However, it is not a part of the public Internet. MPLS allows customers to consolidate multiple local access circuits that have previously been dedicated to particular types of traffic (e.g., data, voice, Internet and video) to a single port connection (without compromising security or performance/quality of service). The use of a single port reduces access requirements, simplifies network design and increases utilisation of individual ports.

Relevant geographic market

53.The Parties explain that the relevant geographic market for GTS is global. This premise is based on the fact that customers consider offers for GTS solutions from suppliers irrespective of the geographic region where the potential supplier originates.

54.In past decisions, the Commission has mostly considered a global market, although, there has never been the need to define the geographical scope. The market investigation has shown that each supplier apparently has a certain focus on one region (e.g., BT and Equant are comparatively stronger in Europe, while NTT and SingTel have a focus on Asia). Moreover, some GTS providers stated that a regional focus and therefore the corresponding specific know-how is of advantage for the provision of GTS which are required mainly in a regional scope (e.g. Europe). However, the market investigation also confirmed that most suppliers have an international portfolio of customers and appear to be in the position of provisioning GTS on a worldwide basis which corresponds to the global nature of the service provided.

The exact definition can, however, be left open since the assessment of this case does not change regardless of the geographic market assumed.

Special access

Background

56.The main traditional providers of local access in the US are the Regional Bell Operation Companies (ìRBOCî) as Incumbent Local Exchange Carriers (ìILECsî). Until the mid 1980s, local access including special access was provided in the US by the former AT&T ñ also known as the Bell-System - which until then had functioned as a regulated telephone monopoly. In 1984, as a result of antitrust proceedings, the monopolistic local access business was split from AT&Tís long-distance business. AT&T was divided into seven RBOC providing local access in regionally defined areas and one long-distance company holding the name AT&T. A number of mergers between the original RBOCs resulted in four large RBOCs, Verizon, SBC, Bell South and QWest, who provide today local access services mainly in four different regions in the US.

Since the break-up of the Bell-System, Competitive Local Access Providers (ìCLECsî) have entered in particular the special access markets by building own networks mainly in urban areas to provide special access service to business customers. Apart from constructing own fibre, CLECs also re-sell special access which they purchase from ILECs often in combination with other services, including various forms of local and long distance voice and data services. ILECs are legally required to provide co-location. This permits CLECs to co-locate in an ILEC central office or ìwire centreî.

58.Despite the described entry of some competitors, local access including special access provided by ILECs is in the US regulated by the US Federal Communication Commission (ìFCCî). ILECs are subject to a general non-discrimination and tariffing obligation. Moreover, the Section 272(e) of the Telecommunications Act requires non-discrimination between affiliates and non-affiliates in providing access services. In addition to these general obligations, specific regulation applies. Switched and special access is regulated to differing extents: While switched access - comprising termination and origination services - is subject to direct price controls and is regulated under a comprehensive price-cap regime, relief from price cap regulation is available to varying extents on request by the respective ILECs for special access.

59.The special access prices charged by the ILECs are regulated in one of three ways, depending on the level of competition that each faces in a particular geographic area. The region in which a RBOC operates is according to the FCC regulatory framework geographically divided into numerous so-called ìMetropolitan Statistical Areasî (ìMSAsî). Competition in each of these MSAs is gauged under a two-phase inquiry that measures the extent to which competitors have obtained fibre-based co-location. In order to achieve relief in regulation, different requirements have to be fulfilled by the ILECs for channel termination, i.e. the link between an ILECís central office and a customerís premises, and for other dedicated transport and special access services (i.e. connections between wire centres and to other networks).

Depending on the extent of co-location by CLECs in the respective MSA, Phase I or Phase II relief from full price-cap regulation can be granted by the FCC. Phase I relief allows to offer contract tariffs and volume and term discounts. The conditions negotiated with a specific customer in the contract tariff are subsequently available to all customers that qualify for the tariff. An ILEC may not offer a contract tariff to an affiliate unless it certifies that a non-affiliate purchases under the same contract tariff. Phase II relief introduces full price flexibility by allowing to offer special access services free from price-cap rules. With Phase II relief, ILECs in essence only remain under the general non-discrimination and tariffing obligation. The FCC retains in addition also in Phase II the right to intervene directly or control prices where the market situation changes and price control should become necessary to ensure non-discrimination.

Relevant product market

61.As opposed to switched access, special access circuits generally have a high capacity and predominantly work with fiber facilities. Special access mainly refers to the US line types DS 1 (ìDigital Signalî circuit) for a capacity of 1.544 Mbps or above (up to 44.736 Gbps). In the European Union other line types for special access are used (ìE1î with a capacity of 2.048 Mbps, ìE3î with a capacity of 34.369 Mbps). Below DS1, other circuits are used to provide voice and data services, ranging from a POTS line (ìplain old telephone serviceî) at 28.8 kbps, an Integrated Services Digital Networks (ìISDNî) line at 64 to 128 kbps or a cable modem connection (at 1 Mbps or higher).

62.From a demand-side perspective, special access can therefore not be regarded as exchangeable with switched access. It is apparent that for the high amount of data traffic which is in tendency even increasing over time, no low-capacity lines can be reasonably used. The differing degrees of regulation applicable to special and to switched access moreover reflect and at the same time create very differing competitive environments for the two main forms of access.

In light of this, a separate market for special access will be considered in the following.

Relevant geographic market

32Channel Termination: The Phase I triggers require an ILEC to demonstrate that competing carriers have obtained fibre-based collocation (i) in 50 percent of the wire centres in an MSA or (ii) in wire centres accounting for 65 percent of the ILECís revenues from these services. The Phase II triggers require an ILEC to demonstrate that competing carriers have obtained fibre-based collocation (i) in 65 percent of the wire centers in an MSA or (ii) in wire centres accounting for 85 percent of the ILECís revenues from these services.

Transport: The Phase I triggers require an ILEC to demonstrate that competing carriers have obtained fibre-based collocation (i) in 15 percent of the wire centres in an MSA or (ii) in wire centres accounting for 30 percent of the ILECís revenues from these services. The Phase II triggers require an ILEC to demonstrate that competing carriers have obtained fibre-based collocation (i) in 50 percent of the wire centres in an MSA or (ii) in wire centres accounting for 65 percent of the ILECís revenues from these services.

Megabits per second.

Gigabits per second.

Kilobits per second.

69.In order to estimate market shares of the providers on the GTS-market, the Parties as well as some market participants refer to two studies concerning the GTS market prepared by two independent consultancy firms: the report ìMNC providers in Europe ñ 2004î by Ovum (ìOvum reportî), which was used as a basis for market share estimations in the decision BT/Infonet; and the report ìForrester Wave: Global WAN Services, Q2 2005î (ìForrester reportî). Both series of estimates suffer from drawbacks. The Ovum report does not cover all GTS competitors. It moreover includes non-GTS revenues in the analysis of two companies (AT&T and Cable & Wireless) and thereby over-estimates their turnover to an unknown extent. The Forrester report includes ìsystem integratorsî into the analysis and the market share estimations.

36The parties believe that also MCIís market share is overestimated [Ö].

70.The Parties consider that system integrators, such as IBM, Atos Origin and Siemens, have to be regarded as competitors to the facilities-based providers, such as AT&T and MCI, who have own extensive networks, on the GTS market. They only provide the management and specific services and in lease the large majority of lines needed from the facilities-based providers in order to offer their telecommunications services. In addition, they usually do not provide the full range of GTS but only specific services. In the past Commission decisions dealing with this sector, these companies have not been considered as market participants in the GTS market.

71.The market investigation has shown that even though some system integrators appear to compete partially against the facilities-based providers to acquire special access services, the large majority of system integrators do not consider themselves as being competitors to MCI. They rather sell individual IT / telecommunication services which are often provided in combination with GTS purchased from the facilities-based GTS providers. They therefore, at present, mainly have to be regarded as customers and / or resellers of GTS providers rather than competitors. (An important exception appears to be the company Vanco, which has only very limited own facilities but nevertheless is generally perceived as a full-fledged GTS provider. Vanco describes itself as ìthe first virtual network operatorî being able to provide the full range of main GTS.)

72.In light of this, system integrators will in the following not be included into the analysis of market shares. The market analysis does, however, not change regardless of their consideration as competitors:

Estimated market shares on the GTS markets in %

Worldwide Market

EU-wide Market

Forrester report 2005 (system integrators excl.)

Supplier

Forrester reportOvum report 2003 2005

Ovum Report 2003

AT&T

33

12

19

9

19

7

11

19

T-Systems

17

2

3

20

BT

15

11

17

14

Cable & Wireless

6

2

3

5

Equant (France Telecom)

4

12

19

14

Global Crossing

4

5

Colt

3

12

Vanco

1

1

System integrators

37

Others

17

27

Even though the market share estimates provided by the consultancy firms are not consistent, some conclusions can be drawn taking also into account estimations supplied by the competitors and the Commissionís calculations on the basis of revenue figures provided by the market participants: MCIís market share is never estimated as being higher than 20% and it competes on a global as well as on a European market at least against three companies whose positions can be regarded as either comparable or stronger. AT&Tís estimated position will - according to these studies - range roughly between 20% and 30% on a global market. BTís position is estimated at 15% to 17%.

Special access

75.No publicly available market share estimations exist on the market for the provision of special access in Verizonís area. It is however very likely, that Verizon as an incumbent local access provider generally holds a very strong position in its region. Being asked to estimate Verizonís market shares in its area, most market participants only provided estimates for local access in general. According to these estimates, Verizon has a market share for local access between 70% and 90% in its territory.

76.It can be assumed that competition is of a higher level for special access (which addresses a market with better economics) than for other forms of local access in a same area. Consequently, Verizonís market shares in each MSA are most probably lower than that for local access. CLECs have in the past to a larger extent entered the special access markets than the switched access business. The market investigation has confirmed that CLECs preferably enter the special access market in dense urban areas where network infrastructure can be used by a larger group of (potential) clients and where entry is more profitable than in remote areas.

77.Customers for special access tend to be highly concentrated geographically. In the case of Verizon, nearly 80 percent of the demand for high-capacity special access services (as measured by revenues) is concentrated in 8 percent of the wire centres where Verizon bills high-capacity special access. At the same time, the areas of high demand for special access show a significant number of CLEC entries. In the 20 MSAs in Verizonís region with the highest demand for special access there is an average of 10 competing providers. MCI ñ active as a CLEC for special access in 30 Verizon MSAs ñ has also concentrated its networks mostly in dense urban areas. According to the Parties, in 87% of these 30 MSAs (26 MSAs) there are five or more additional providers of competitive fibre apart from the Parties.

78.The differing competition conditions in the special access markets are reflected by the achieved Phase I and Phase II relief of regulation in Verizonís 175 MSAs. It has obtained Phase II pricing flexibility for end-user channel terminations and other services in 26 MSAs, all of which are MSAs in which MCI is also active as a CLEC. For other components of special access circuits, Phase II relief has been achieved in additional 36 MSAs. The areas within which Verizon has obtained (partially or completely) Phase II relief represent [50-60]% of Verizonís special access revenues. If Phase I relief is added (obtained in 12 MSAs), then Verizon has obtained some form of relief from regulation in a total of 74 MSAs which represent approximately [75-85]% of Verizonís total special access revenues.

It is highly likely that in those areas where special access is most relevant for the provision of GTS, some form of regulatory relief has normally been achieved. However, it cannot be derived from this that in MSAs having reached Phase II according to US regulation rules, Verizon does not anymore hold a very strong or even dominant position (in the meaning of EC competition law) for the provision of special access. While it might be the case that its strong position has been challenged by CLECs in some MSAs to a significant extent, the fact that Verizon remains subject to regulatory rules requiring general non-discrimination, indicates its still outstanding market position. This question, however, does not need to be resolved in the context of this case, since the merger does not lead to a significant impediment of effective competition in the common market and the EEA.

Analysis

Concerns raised

80.It has to be noted that no GTS customer has raised any competition concern linked to the vertical integration of Verizon and MCI. In the course of the market investigation competition concerns were nevertheless raised against the proposed merger by some competitors of MCI who buy special access in Verizonís area. It was submitted that Verizon has a very strong or even dominant position in the markets for special access which could be further strengthened through the elimination of the CLEC MCI in the 30 MSAs where the activities of the Parties overlap and through the elimination of MCI as a potential competitor in the other MSAs. According to the concerns raised, GTS providers are very dependent on special access in the Verizon territory since it covers important business areas in the North-East of the US. The fear was expressed that the merged entity could discriminate against other GTS providers by granting special access to competing GTS providers only to less favourable terms than to the own GTS branch MCI. This could significantly impede competition on the GTS market. It was also argued that the merged entity could not only discriminate on price but also on quality.

82.With respect to both above mentioned concerns, the FCC regulation is not considered as being a sufficient means to prevent such discrimination. It was in particular brought forward by complainants that in the US the possibility of Phase I / Phase II relief from regulation has been opened up to the ILECs too early in the development of the special access markets towards more competition. Moreover, it was doubted whether the criteria used by the FCC to assess and decide Phase I / Phase II relief adequately reflect the real degree of competition.

83.The above mentioned concerns could potentially, indeed, only materialize if Verizon had significant market power and despite the existing regulation some degree of discretion which would leave sufficient room for discriminatory behaviour. Such discretion might be conceivable where Phase I or Phase II relief from regulation has been achieved and thereby price-flexibility with respect to discounts or even to the complete pricing is granted. Where full price-cap regulation still applies, however, the ILECs ñ although apparently not facing significant competition ñ appear to have leeway for price discrimination only to a much smaller extent. This affects not only the MSAs without any form of relief from regulation for special access, but also the complete provision of switched access which is subject to comprehensive regulation and is not eligible for Phase I or Phase II relief.

84.Assuming that the above mentioned discretion exists in the instances described, the merger could theoretically lead to a significant impediment of competition. This could theoretically be the case if post-merger Verizon ñ already on a stand-alone basis or through the combination with MCIís special access business - had market power to discriminate and raised prices for special access for its GTS competitors. A significant impediment of competition could then occur if this price-increase created significant disadvantages for MCIís GTS competitors leaving them with lower incentives and / or possibilities to (successfully) bid against the merged entity for GTS customers and in last consequence exclude them from the GTS market.

Without any need to assess the effectiveness of the FCC regulatory measures, the market investigation has, however, shown that the effects of the vertical integration caused by the merger will not lead to significant effects on competition even when the above mentioned assumptions were considered to be realistic.

Discrimination by Verizon/MCI

86.For each customer, GTS providers usually have to acquire special access from third parties in many countries. The locations which GTS customers require to be connected to a GTS network are always widely dispersed across different countries and continents. The analysis of GTS customersí replies showed, that virtually all of them include numerous different locations into their GTS network - some of them indicated to cover up to more than 100 different countries. GTS customers regularly have GTS networks which connect multiple locations at least in the US and Europe. The large majority of GTS customers has stated to cover all continents.

Cost analysis

87.The differing and dispersed geographic distribution of GTS customer locations to be connected shows that GTS providers usually have to purchase special access in many different countries with Verizonís area being only one among numerous relevant areas. The analysis of the competitorsí cost structures reflects this situation. Special access costs GTS providers spend with Verizon represent according to almost all submitted figures provided by the Parties and their competitors less than 10% of the providersí respective total GTS revenues and also of their total GTS costs. Most percentages even range below 7%. This already confirms the GTS competitorsí limited degree of dependency on Verizon and exposure to price discrimination.

88.However, Verizon sells special access to a significant extent on a wholesale market where other telecommunication companies buy it in order to re-sell it later to GTS providers and other customers of special access. It is not relevant for the assessment of this case whether GTS competitors and customers depend on Verizon directly or on its resellers. If Verizon had a dominant position and decided to increase prices, this would affect GTS competitors and customers regardless whether they buy special access from Verizon or a reseller of Verizon special access. Moreover, there are a number of CLECs providing special access, among them also MCI, whose special access business activities need to be counted to the ones of Verizon. The more appropriate figure for the assessment of the GTS competitorsí dependency on Verizon post-merger is therefore not the share of costs that is charged by Verizon for special access in the US, but the share of costs that is charged by Verizon and its resellers as well as MCI for special access in the Verizon region where Verizon acts as an ILEC. As approximation, the cost share of the whole Verizon region was taken. It has to be noted, however, that this figure overstates the potential dependency on Verizon and MCI in the provision of special access since it also includes special access supplied by other CLECs over their own fibre.

89.The market investigation has confirmed that also in this respect no critical proportion of costs level is reached. Even though in tendency higher than the other cost items, the special access to the Verizon region also accounts for a proportion of total GTS revenues and total GTS costs in most cases of below 10%. This shows, that special access to only one area ñ here Verizonís territory - is only of limited importance for the provision of global services which require special access in a large number of countries, even if a dense commercial area as the North-Eastern region of the US is covered. The effect of a potential increase in prices charged by Verizon (to the extent that regulation allows such an increase in price) to the GTS competitors would therefore also be very limited.

90.It should be also noted that Verizon would in fact only be able to increase prices to varying degrees in the different MSAs and the different services (channel termination, other services). As described above, full price flexibility is only given in MSAs with Phase II relief for channel termination and other services. Where only Phase I relief has been reached or even full regulation still applies, the leeway for price increases or price discrimination in other form is restricted to differing degrees accordingly. As noted above, [75-85] % of Verizonís special access sales are made in MSAís where price flexibility has partially or completely been achieved. Verizon has extensive price flexibility (under the general non-discrimination obligation) in only 26 of the 175 Verizon MSAs where Phase II relief has been fully achieved. If partial Phase II relief (only for other services than channel termination) is added then 62 MSAs are included ([50-60] % of Verizonís special access revenues). In 12 other MSAs, flexibility with respect to discounts is given in Phase I either for all services (1 MSA) or for services other than channel termination (11 MSAs). In total, [75-85] % of Verizonís special access revenues are made in MSAs with some form of regulatory relief. This shows that the increase in price does not affect the full cost positions of the GTS competitors relating to Verizon access. The competitorsí real dependency on Verizon has to be considered as lower than the cost shares indicate.

Quality considerations

91.An additional concern mentioned by one competitor is that GTS competitors could post-merger receive special access at lower quality standards from Verizon. Verizon could according to this concern provide less prompt installation of new circuits or the less effective maintenance and repair of existing services. It seems, however, that the scope for a discriminating treatment in terms of quality is rather limited. Every ILEC is obliged to provide special access in a non-discriminatory manner. This general obligation also refers to the quality of the service provided.

92.Moreover, the services to be provided are usually contractually fixed. It appears that Verizon could not impose less favourable service terms to the GTS competitors. First of all, Verizonís so-called ìService Response Creditsî (ìSRCsî) under which a customer receives a credit for a deficiency in service performance (e.g., a credit for the monthly recurring charge if there is an outage of an hour) are generally part of the wholesale tariff, and Verizon could not unilaterally change these provisions. Instead, it must file any change to a tariff with the FCC. Customers have the opportunity to comment and object to the changes, and the FCC has the authority to suspend the implementation of the new tariff and open an investigation. Discrimination on the basis of quality is therefore most unlikely.

93.In addition, it has to be noted, that no GTS provider can offer a complete GTS bundle alone. Every GTS provider is also dependent on other telecommunications companies ñ among them also other GTS providers - in order to complete its offers. This becomes most apparent in the case of other vertically integrated operators, such as BT, Equant and T-Systems. If Verizon degraded the quality of its services for them, it would have to fear retaliation when asking for local access in the UK, France and Germany.

94. In any event, the small proportion of costs that special access from Verizon / Verizonís area represents of the total GTS revenues and costs indicates that also the impact of deterioration in quality would have only marginal effect on the overall quality of the GTS offers.

Multi-sourcing

95.The market investigation has, moreover, shown that GTS customers have some possibilities to counteract a potential price increase through multi-sourcing. Several GTS customers have confirmed that they engage in multi-sourcing. They do not purchase all of their required GTS from one supplier but divide their need between two or more providers either according to geographic criteria and/or individual services. For a geographical split of GTS, the customer has to determine how the regional providers are to interconnect and interact to ensure that both networks and services are seamlessly provided. There are several methods to ensure this which do not appear to constitute major obstacles for splitting. With this strategy, the customers can use the providersí differing strengths and gain better negotiation positions vis-‡-vis their providers since a switch to another GTS company is easier if it already works for a specific customer. In addition, GTS customers often require detailed cost break-downs by the GTS providers. They are therefore able to identify the reasons for any price increase that might occur in the GTS competitorsí bids due to a price increase by Verizon.

96.96. Due to these practices, a price increase by Verizon to its competitors might lead to the reaction of the customers to split their GTS needs in order to profit from MCIís comparatively lower access prices, but from other competitorsí advantages for those parts of the network where special access to Verizon MSAís is not needed. The merged entity might therefore gain additional business from its competitors by raising their special access costs only to a limited extent. GTS customers are big multinational corporations who are able to compare the offers even with respect to the differing parts of the networks ñ this ability at least clearly exists when GTS consultants are used who are often hired in order to assist the GTS customers in the bidding process. Multi-sourcing would therefore limit the gains that the merged entity could derive from discriminating and it would at the same time further limit the harm to GTS competitors.

97.97. These considerations are confirmed by the fact that most competitors as well as customers do regard ownership of local access facilities in general as an advantage but clearly not as a critical factor of success in the GTS business. Price is a main factor (apart from quality and reliability) in competition for customers, but it appears to become a decisive factor only after a pre-selection of comparable bids has been made in the tendering process. In order to be pre-selected, the GTS providers have to prove other advantages than a low price, such as global reach, financial stability, technical competence and responsiveness towards the customerís specific wishes. This is supported by the fact that so far non-vertically integrated companies (MCI, AT&T, Global Crossing, Colt, Cable&Wireless) have equally competed against vertically integrated companies, such as Equant (France Telecom), T-Systems (Deutsche Telekom) and BT.

Discrimination by tacit collusion between Verizon/MCI and SBC/AT&T

98.The second main concern raised during the market investigation relates to a combined effect of the two mergers Verizon/MCI and SBC/AT&T. It was claimed that both merged entities could tacitly agree to give each other more favourable prices for special access than to other GTS competitors, i.e. give discounts to the other merged entity and increase special access prices for the GTS competitors, and thereby jointly discriminate against the others with the result of a significant impediment of effective competition. In consequence, the two companies would ñ according to this theory - be able to offer significantly lower prices than the other GTS competitors in the GTS bidding processes.

99.In order for tacit collusion to be plausible and stable, several criteria have to be fulfilled. In essence, tacit collusion requires a mechanism by which the companies allocate and control the jointly achieved additional profits. The structure and functioning of the affected markets in this case, however, do not support such an assumption.

100.MCI and AT&T have according to the Ovum report a combined market share of 52% on the global GTS market (MCI: 19%; AT&T: 33%). As discussed above, the Ovum report, however, probably overestimates the indicated GTS providersí market shares by not covering all market participants. Moreover, Ovum indicates to overestimate AT&Tís market share in this calculation due to the consideration of non-GTS revenues. Therefore, it has to be assumed that the combined market share of the MCI and AT&T is lower than 50%. The Forrester report indicates a combined market share of 19% (MCI: 7%; AT&T: 12%) which would turn into 30% if system integrators are taken out of the analysis (MCI: 11%; AT&T: 19%). On a European market, a share of 28% for both merged entities together would emerge (MCI: 19%; AT&T: 9%).

101.In the light of these figures, it can be doubted whether the market exhibits at present a high enough degree of concentration from where it would result a duopoly structure prone to tacit collusion. However, the concerns raised do not relate to a joint price increase or decrease in quantity by the two ìduopolistsî on the GTS market as usual oligopoly theory suggests. The concerns are directed towards the reverse, namely a strategy according to which the merged entities would in an action of tacit collusion raise input prices for local access for the other GTS competitors and thereby offer lower prices to the GTS customers in the bidding procedures. This could subsequently lead to a gain in market shares for the duopolists and in the long-run exclude GTS competitors from the market.

102.It is, however, first of all not evident that a combined price increase by SBC and Verizon could effectively impede or even foreclose GTS competitors. The combined cost positions of the GTS competitors for special access in Verizonís area plus in SBCís area would clearly be higher and a price increase would surely have a larger effect on their competitiveness than in the case of Verizonís individual cost position alone. However, taking into account the limited scope for price increases due to the FCC regulation, it has to be doubted that the GTS competitors, and their widespread customers, are dependent on both companies to an extent which is large enough to allow for ìsuccessfulî price discrimination.

103.It was shown in the above calculations that the majority of GTS providersí special access cost positions relating to Verizon cover shares below 10% and mostly even below 7% of their total GTS revenues. If Phase II price flexibility is only given for roughly half of the sales that Verizon makes with selling special access to competitors and this Phase II price flexibility in most of the relevant MSAs only applies to a part of the services, the real dependency and potential for discrimination has to be regarded as weaker accordingly. In the concerns raised, SBCís and Verizonís positions were described as being largely symmetric. If therefore similar cost shares for SBC special access were assumed, even the combined cost positions would remain with some probability in an uncritical area.

104.It is, moreover, not obvious that a strategy of tacit collusion on price discrimination against the GTS competitors would clearly lead to a balanced increase in profits for each of the colluding parties. Under the assumption of tacit collusion, the other GTS competitors might have lower chances to succeed due to the jointly increased access prices in the bidding process for new GTS customers. However, both merged entities would have cost advantages in competition for the respective GTS customer. The

105.extent of mutually granted advantage would depend on the dispersion of locations of this GTS customer. Verizon could for example greatly profit from a newly gained customer with strong focus on SBCís region, since Verizon would according to this theory of harm enjoy a low price for access in SBCís region. In this way, SBC would ñ to its own disadvantage - support Verizonís success with the new customer. It is therefore not certain whether the two merged entities would have an incentive to engage in tacit collusion.

106.Therefore, for this theory of harm to be realistic and tacit collusion to be plausible and stable it seems necessary that an easy method for the division of the GTS market exists which provides for a balanced sharing of potential additional joint profits on GTS market and special access market. However, such a mechanism cannot be expected to develop since GTS customers vary not only with respect to their dispersion of GTS locations but also in other aspects due to the customized nature of GTS that are usually provided as packages of different single services ranging for example from local telephony to virtual private networks and service management. It is therefore most unlikely that a stable allocation mechanism of new GTS contracts which would be a necessary condition for this theory of harm could not be achieved in a tacit manner.

107.In light of the foregoing, there will be no adverse affect on competition on the GTS market as a result of this transaction which could lead to a significant impediment of effective competition in the common market and the EEA.

International Voice Telephony Services (IVTS)

108.International voice telephony services (ìIVTSî) are supplied by telecommunication companies as both retail and wholesale services, i.e. to end-customers as well as to other telecommunication companies who then resell these IVTS to end-customers. Retail IVTS are offered exclusively on an ìend-to-endî basis (from call set-up to termination), i.e. retail customers only buy international calls that the provider undertakes to ensure are completed, by connection to the called party.

109.On the wholesale level, IVTS are provided to other telecommunications customers either on an ìend-to-endî basis or broken down into the three separate segments: call origination, call termination, and carrier services. Depending on the extent and location of their own local networks, wholesale IVTS providers purchase call origination and call termination services from local telecommunications operators in order to provide end-to-end service to their end customers.

110.MCI provides both end-to-end wholesale and end-to-end retail IVTS to end customers in the EEA. Verizon is not active in retail IVTS in Europe. As a local access provider in the US, it offers call termination services for international calls from Europe. Since origination services and international carrier services on a stand alone basis are sold by neither party in the EEA, only the provision of end-to-end wholesale and end-to-end retail IVTS in the EEA as well as wholesale termination services in the US will be assessed in the following.

Market definition

Relevant product market

110.International voice telephony services have traditionally been provided by means of public switched networks in both the originating and terminating countries of a call. Interconnection between the domestic networks of any pair of countries is provided by means of international carrier services between the countries concerned.

111.The Parties submit that there might be two different markets for end-to-end retail and end-to-end wholesale IVTS. Carriers act as resellers when they buy end-to-end IVTS on a wholesale market and later provide them as retail IVTS to end-customers. Providers of end-to-end retail IVTS are therefore customers of end-to-end wholesale IVTS providers; this supports the assumption of two separate markets.

112.According to the parties, end-to-end IVTS providers on the wholesale market buy international carrier services, origination and termination services separately and bundle them as an end-to-end offer for retail IVTS customers. The Commission has in past decisions dealt with differing aspects of voice telephony which has undergone significant changes in the process of liberalization in the past years. In the market for upstream wholesale services, the Commission has noted, in decision JV.15 BT/AT&T, in order for operators to carry international calls from end-to-end, a need for unbundled elements has developed. The Commission has consequently distinguished between the market for retail end-to-end IVTS and a separate underlying wholesale market for different international carrier services.

113.In the decision M.2803 Telia/Sonera, the Commission looked at the wholesale call termination input required to provide retail IVTS. The Commission concluded that there are separate markets for call termination on each partyís fixed network, because the operator seeking to terminate a call can only do so on the network to which the called party is connected.

114.In light of the above, this decision will consider the markets for: call termination; end-to-end retail IVTS and end-to-end wholesale IVTS.

Relevant geographic market

115.Based on previous Commissionís decisions (Telia/Sonera and TeliaSonera AB/Orange A/S), the Parties submit that the geographic market for termination of IVTS calls in the United States could either be national or the Verizon region. Considering that the national definition of termination markets in past decisions did not predominantly relate to the boundaries of a country but to the boundaries of the incumbent telephone operatorís network it is appropriate to consider a geographic market for call termination in Verizonís area which relates to the scope of Verizonís network.

116.In BT/MCI (II), the Commission noted that, from the consumersí point of view, the relevant geographic market for end-to-end IVTS in general should be defined by reference to call traffic routes between any country pair, since different international routes cannot be considered as viable demand substitutes. A further distinction was made between traffic originating from each country since calls terminated in different countries are not substitutable with each other.

117.With respect to end-to-end retail IVTS, the Parties submit that in the light of demand for international calls to all destinations, a relevant retail end-to-end IVTS market for global IVTS calls from each Member State should be considered. According to the Parties, MCIís retail IVTS customers generally acquire calls to multiple destinations rather than on specific country pairs. This would speak in favour of a market definition not distinguishing between country pairs but considering wider markets for IVTS originating in one country and terminating worldwide.

118.On the wholesale level, the Parties believe that it is important to consider European-US IVTS traffic as a whole, i.e. not split into country-pairs, since traffic is to a significant extent routed through hubs and not directly to or from the respective countries. According to the Parties, in particular the UK plays a role as a hub for wholesale traffic originating all over Europe, since a large number of trans-Atlantic cables land in the UK. The Parties, however, and most of the competitors, also indicate that wholesale customers usually buy wholesale end-to-end IVTS originating in one country but terminating worldwide rather than terminating in one specific country. This would speak in favour of a market definition similar to the one proposed by the Parties for the retail end-to-end IVTS.

119.For both product markets ñ retail and wholesale end-to-end IVTS ñ this question may be left open since even under the narrowest market definitions according to country pairs no competition concerns arise.

Competition assessment

120.While MCI offers retail and wholesale end-to-end IVTS in the EEA, Verizon is not active in these fields in the EEA. As the operator of a local access network in the US, Verizon offers termination services on its network for all calls, including international calls originating from Europe and other parts of the world. MCI offers origination and termination services only to a limited extent since it controls only a small number of direct customer access connections. Therefore, the merger mainly leads to a vertical integration of a provider of termination services into the provision of retail and wholesale end-to-end IVTS. The relevant IVTS to be assessed are therefore retail and wholesale end-to-end IVTS originated in the countries of the EEA and terminated in the US, or more precisely in Verizonís area.

121.MCI does not have a strong position in retail and wholesale IVTS in Europe. Its main focus is clearly in the US with respect to these services. In 2004, MCI generated [Ö]

million Euro in revenue from the supply of retail IVTS voice calls originating in Europe and terminating worldwide, only [Ö] million Euro of which reflect revenues generated from retail IVTS calls originating in Europe and terminating in the US. Its wholesale revenues were approximately [Ö] million Euro through the supply of wholesale IVTS from Europe (terminating worldwide). Only [Ö] million Euro out of these were generated through the supply of wholesale IVTS between Europe and the US.

122.MCI reaches the following market shares in the Member States (ìMSî):

MCI 2004 shareRetail IVTSRetail IVTSWholesaleWholesale IVTS by country market marketIVTS marketmarket MS - WorldMS - USMS - WorldMS - US

Austria

[0-10]%

[0-10]%

[5-15]%

[10-20]%

Belgium

[0-10]%

[5-15]%

[0-10]%

[0-10]%

Denmark

[0-10]%

[0-10]%

[0-10]%

[0-10]%

France

[0-10]%

[0-10]%

[5-15]%

[10-20]%

Germany

[0-10]%

[0-10]%

[10-20]%

[10-20]%

Ireland

[0-10]%

[20-30]%

[10-20]%

[40-50]%

Italy

[0-10]%

[0-10]%

[5-15]%

[0-10]%

Netherlands

[0-10]%

[15-25]%

[0-10]%

[0-10]%

Norway

[0-10]%

[0-10]%

[0-10]%

[0-10]%

Spain

[0-10]%

[0-10]%

[10-20]%

[15-25]%

Sweden

[0-10]%

[15-25]%

[5-15]%

[5-15]%

United Kingdom

[0-10]%

[0-10]%

[5-15]%

[5-15]%

123.The merger could potentially give rise to competition concerns with respect to the termination market in Verizonís area. After the merger, Verizon could have an incentive to discriminate against MCIís IVTS competitors. This incentive could be higher, the larger the MCIís market share in the respective Member State is.

124.However, it cannot be assumed that the merger will give rise to competition concerns. Even if Verizon increased its price for termination in its area, it is not likely that an IVTS competitor on the retail or on the wholesale market could suffer significant losses which would exclude competitors from the market. First of all, all IVTS providers usually offer IVTS originating in one or more countries and terminating worldwide. The specific IVTS affected by this merger (calls terminating in Verizonís area) regularly only represent a small share of the suppliersí total IVTS volumes. This can be seen when comparing the volumes of the IVTS markets originating in a Member State and terminating worldwide with those with termination in the US. On the wholesale level the market volume of IVTS for the connection MS-US represents in the EU only 10% of the market volume of IVTS for the connection MS-World. This percentage is presumably significantly lower if only Verizonís area is taken into account, which is only one of broadly four incumbent territories in the US.

125.In addition, it has to be noted that it would be difficult for Verizon to price discriminate IVTS providers for termination services. Most IVTS providers are telecommunication companies providing a large range of services which require termination services. Verizon does not know from the outset, for which application a telecom provider buys the termination service.

126.Moreover, it has to be noted that Verizon is subject to full regulation by the competent authorities in the US. The regulatory framework in the US prohibits terminating access carriers like Verizon from discriminating or charging supra-competitive rates in the provision of terminating access services. As opposed to special access, there is no regulatory relief for terminating access services in the form of Phase I or Phase II regulation. Termination is subject to direct price controls and is regulated under a comprehensive price-cap regime. Although price discrimination cannot be excluded, regulation nevertheless needs to be regarded as a restraint on the ILECís behaviour which limits the degree of possible price discrimination or other anticompetitive behaviour. In combination with the small share that the affected business (MS ñ Verizon area), it can be concluded that a foreclosure by the merged entity is most unlikely.

127.A discrimination appears moreover unlikely against the background that MCIís business in IVTS originating in Europe and terminating in the US has a volume for both retail and wholesale IVTS together of [Ö] million Euro. Verizon reaches sales from switched access, which comprises termination and origination, of around [Ö] billion Euro. It is at least questionable, that Verizon would discriminate against its main customers in favour of the comparatively small IVTS business of MCI.

128.In light of the above, it can be concluded that the transaction will have no material impact on competition in the market for global IVTS and will not lead to a significant impediment of effective competition in the common market and the EEA.

V. CONCLUSION

129.For the above reasons, the Commission has decided not to oppose the notified operation and to declare it compatible with the common market and with the EEA Agreement. This decision is adopted in application of Article 6(1)(b) of Council Regulation (EC) No 139/2004.

For the Commission (signed) Neelie KROES Member of the Commission

29