Competition & Regulation BulletinA backbone to pick: The European Commission and MCI/WorldCom by Christian Koboldt When WorldCom bid for MCI in October 1997, topping the price that BT was prepared to pay for the US carrier, it came as a surprise to many. Still more surprising was that it should be the two companies' Internet interests that concerned the antitrust authorities. The Internet businesses of MCI and WorldCom were small by comparison to the total turnover of the two companies (less than 5 percent). And the Internet, characterised by phenomenal growth and built on open standards, might appear insusceptible to domination. Nevertheless, in March 1998, the European Commission, working closely with the US Department of Justice (DOJ), extended its investigation into the proposed acquisition into a second phase. The Commission feared that the companies' combined market share in the supply of Internet services could lead to a position of dominance. More specifically, the Commission was concerned that MCI and WorldCom together would provide more than 50% of Internet "backbone" services. Although the number of Internet Service Providers (ISPs) had grown rapidly from only a handful in the early nineties to over 4500 in 1997, it was felt that the backbone – roughly defined as a set of high-speed, high-capacity connections between the major hubs of the network - was operated by only a few companies. As a result of "network effects", the Commission argued, operators controlling a sufficiently large proportion of this backbone would be in a position to further their dominance through a variety of anti-competitive strategies. Being the first case where the Commission had to delve into the depths of the Internet, MCI/WorldCom is likely to be important for future decisions in this area. This article provides an overview of the Commission's arguments within the context of a general description of basic Internet concepts. Some basic Internet concepts The Internet is not a single network with a carefully planned architecture, but a network of interconnected networks. Each of these networks can have its own architecture and its own standards for communication. They work together by using a standard protocol for the exchange and transfer of data - the so-called Transfer Control Protocol/Internet Protocol (TCP/IP) - which is non-proprietary. Almost all ISPs, and even some final customers (e.g. larger corporate customers), operate interconnecting networks which together make up the Internet. Traffic between two points in the Internet, say between a computer containing the files that make up Microsoft's World Wide Web site and the PC of a customer who wants to download a file from this site, is not routed along a dedicated connection between these points. Rather, the data are split up into many small "packets", each of which is routed individually to the final destination where they are re-assembled into the full file. Packets are like postcards or letters - they may travel to their destination over multiple routes, depending on how the physical transport of mail is organised at any particular point in time. Unlike in "circuit-switched" telephony networks, there is no dedicated connection between two users which exists for the entirety of the time during which communication takes place; rather, "packet switching" uses so-called "dynamic routing" to connect two users, allowing for a flexible allocation of network resources. A router (which replaces the switch in traditional telephony networks) takes each incoming packet, examines its destination address, and sends it on to another router along one of the many alternative paths that eventually lead to the destination address. This is based on a routing table, which essentially contains a list of all feasible routes between the router and any destination address. If a link in the network fails, the IP routing mechanism quickly alerts other parts of the network that paths using the broken link are no longer available. Once the routers near the failure have updated their databases, which happens within seconds of the failure, traffic is routed along other paths. It is this dynamic routing, again governed by a public standard protocol (BGP 4) that makes the Internet invulnerable to the destruction of parts of its physical infrastructure. Interconnection between the individual networks results in the Internet's main feature: universal connectivity. Each user connected to the Internet can communicate with any other user, collect information or distribute it around the world. Interconnection gives rise to traffic flows in two directions and thus generates both costs and benefits for the interconnecting networks. Each network operator adds to the other's ability to offer universal connectivity, but each incurs a cost in providing the infrastructure that allows traffic to flow between customers. The balance between these benefits and costs is reflected in different forms of interconnection agreement. One important type of agreement, known as "peering", is essentially a form of barter arrangement. Under such a peering agreement, each network operator agrees to terminate traffic destined for its own customers. A peer does not have to accept traffic that is not destined for its own network or its customers’ networks. Peering agreements arise where the costs and benefits from interconnection are roughly balanced between two networks, and so no settlement payments need to be made. If, on the other hand, the costs and benefits of interconnection are not balanced, one network operator will need to pay the other. The network operator receiving greater net benefits becomes a "transit" customer of the other. It purchases a transit service from the vendor network, and the agreement is known as a transit agreement. As such, the customer network has the right to pass on traffic destined for any part of the Internet, not just the vendor network or its customers networks. With these concepts in mind, we can now explore the Commission's arguments in more detail. What backbone? The networks which together constitute the Internet vary with regard to their size, their geographic coverage and their capacity. Equally, ISPs vary with regard to the size and coverage of their networks, and it is not easy to identify a clear-cut distinction between classes of ISPs. In its investigation, the Commission proceeded to distinguish between "backbone" networks and other, lower level networks by defining a separate market for "top-level" networks. These are characterised by their ability to offer "top-level connectivity", in turn defined as the ability to offer universal connectivity exclusively, or almost exclusively, through peering agreements without the need to purchase substantial amounts of transit. All other ISPs outside this market needed to purchase a significant amount of transit services from the top level networks, and as such were customers of the top-level ISPs. MCI and WorldCom did not agree with this market definition. They argued that the relevant market was that for universal connectivity due to the architecture of the Internet, the lack of a clear hierarchical structure, the existence of multiple connections and the ease with which traffic could be routed along different paths. Drawing a distinction between ISPs based on their size, or even on the specific terms of the agreements through which they achieved universal connectivity, would be arbitrary, unjustified and not in line with the Commission’s own principles for defining relevant markets. In response to a hypothetical price increase, the merging parties argued, one would find that customers of MCI/WorldCom would have the opportunity and the incentive to reduce the amount of connectivity bought from the merged firm, thus rendering such a price increase unprofitable. How to measure market share? Having delineated the relevant market, the Commission had to measure the market shares of the two parties. To do this, one would ideally want to measure the volume of traffic carried on the networks of the parties as well as the overall volume of traffic. However, since the retirement of the government-funded NSFNET in 1995, there have been no systematic, reliable and comparable traffic measurements on the Internet. Although each network operator has some information about the capacity and load of its network, these figures are not necessarily comparable. Alternative metrics - such as the number of addresses allocated to network operators, the number of hosts connected, etc. - are also fraught with problems. In this situation, one would usually revert to a calculation of market shares based on revenues. However, splitting revenues along those lines - between revenues linked to the provision of "top level connectivity" and other Internet related revenues - was almost impossible. Instead, the Commission attempted to reconstruct traffic volumes and traffic shares by using the known traffic flows across interfaces between networks, and building a model based on a set of assumptions about the relationship between traffic flowing within and between networks. According to its model, the Commission found that MCI and WorldCom together accounted for over 50 per cent of traffic in the relevant market. MCI and WorldCom argued that this model was highly sensitive to particular assumptions and so unreliable. They maintained that revenues are a more appropriate measure of market share, and that using this measure the combined share of the two companies would be between 19 and 21 per cent of the total ISP market. Snowballing and the serial killer The Commission went further than stating that, due to the high market shares, the merger would create a dominant position in the Internet backbone market. It also argued that MCI/WorldCom would be able to exploit its market position by:
Such strategies would also adversely affect MCI/WorldCom’s own customers, who would experience a degradation of their connections to all hosts not directly connected to the MCI/WorldCom network. However, the Commission argued that the merged entity would benefit from "network effects". Because it would operate the largest network, MCI/WorldCom’s customers would be more likely to communicate with each other than with customers connected through the networks of other ISPs. They would therefore be less affected by degraded interconnections. In order to minimise the adverse effects on its own customers, the Commission suggested that MCI/WorldCom could pursue a strategy of attacking the other ISPs sequentially, picking off one after the other. This would lead to a "snowballing" effect, where more and more customers would switch to MCI/WorldCom, eventually leaving the merged entity with sole control over the Internet. The Commission did not accept the contrasting view of the merging parties that what customers buy from their ISPs is universal connectivity, rather than the ability to communicate with someone connected to the same ISP. The parties argued that this gives all ISPs a strong incentive to obtain connectivity with all others, and to maintain the quality of interconnection. Otherwise, they would not be able to deliver what their customers want. The parties also failed to persuade the Commission that the effects of the alleged anti-competitive strategies would be at least as severe for MCI/WorldCom’s customers as for the customers of other ISPs. For example, in the Commission’s "serial killer" scenario, MCI/WorldCom customers would experience a degraded service quality with regard to only a small number of connections at any point in time; but, they would suffer from a deterioration of quality over a much long period of time. Therefore, the parties argued, it would be mistaken to assume that the overall damage to its own customers would be reduced if MCI/WorldCom adopted a strategy of serial degradation. And what about entry? The Commission held that these anti-competitive strategies, aimed at driving competing ISPs out of the market, would also have the effect of making it difficult for new entrants to build successful businesses and contest MCI/WorldCom’s position. Thus the threat of potential competition from new entrants was deemed likely to be ineffective, even though one competitor (GTE) had announced plans to upgrade its backbone to a capacity a hundred times that of the current Internet, and a number of other companies (such as Qwest and Level 3) were building out high capacity networks tailor-made for Internet traffic. What’s in a divestment? Having failed to convince the Commission that the proposed acquisition would not create a dominant position in the Internet market, MCI agreed to sell off its backbone business, which consisted of its network assets and the "wholesale" business (i.e. carriage of traffic for other ISPs). One might have expected this to allay the Commission's concerns since it would have avoided any increase in concentration within the market for top-level connectivity. However, the Commission apparently did not consider this divestment to be sufficient, perhaps following talks with the DOJ. In the end, the Commission was satisfied with a divestiture of MCI’s entire Internet business, including the "retail" business (which involved selling Internet services directly to corporate and residential customers). The Commission thought that this complete divestiture would be necessary for the purchaser to replicate the position in the market previously held by MCI. This remedy would appear to be at odds with the Commission’s market definition. If there is indeed a separation between the provision of top-level connectivity and the provision of Internet access to final customers, as the Commission argued, then there is no reason why a divestiture of MCI’s backbone business should not have been sufficient to allay any competition concerns. Even if one argued that a successful operator of a top-level network would need access to customers, the required divestment appears to be inconsistent with the Commission’s view of a separate market for top-level connectivity. In this case, all ISPs with a sufficiently strong customer base should have been regarded as actual or potential competitors of MCI/WorldCom, irrespective of whether they were a member of the group of top-level networks which relied almost exclusively on peering. The requirement of a complete divestiture of MCI’s Internet business, both the backbone and the retail business, would suggest that the relevant market should be the wider market for Internet access. However, in this case the combined market share of MCI and WorldCom would have been well below the threshold at which a presumption of dominance exists. Christian Koboldt is Managing Consultant in the Telecommunications & Media Team at London Economics. He acted as advisor to MCI and WorldCom during this investigation. Takeovers in water – a consistent approach? by Rob Francis and Ian Alexander When Enron announced its intention to buy Wessex Water last year, the UK Secretary of State for Trade and Industry did not refer the bid to the Monopolies and Mergers Commission (MMC). A number of high profile bids for water companies were referred during previous years, yet this latest takeover has been allowed without reference. In this article we explore the rationale for the decision, and consider whether it opens up the prospect of further acquisitions of water companies in England & Wales. Recent referral decisions Enron’s bid for Wessex was the first time that a US utility has become involved in the provision of licensed water and sewerage services in England and Wales. However, the industry has seen several other merger attempts over the past few years, as set out in Table 1. As is clear from this table, the regulator’s approach to referrals has been consistent across these attempted mergers. The water industry in England & Wales is dominated by 10 large and independent water and sewerage companies (WaSCs). There are also more than 15 smaller water-only companies (WoCs), which supply drinking water only, sewerage services for their customers being provided by the larger companies. Each of these companies has a regional or local monopoly.
1. This table includes all mergers involving England and Wales WaSCs, other than those with small WoCs. It excludes all mergers involving WoCs only, even where these were referred to the MMC, as was the proposed merger between Mid Kent Holdings plc, General Utilities plc and SAUR Water Services plc (MMC, January 1997). 2. Now called Suez-Lyonnaise des Eaux. At the time of this bid, Lyonnaise already owned two of the larger water-only companies (WoCs) in England and Wales, North East Water and Essex and Suffolk Water, and the former was the only WoC in Northumbrian's area
In all three of the mergers involving consolidation between a WaSC and another major England and Wales water company, a referral to the MMC was made. In all other cases, where the merger was between a WaSC and a company having no England and Wales water interests, the merger was allowed without reference, albeit following undertakings. The importance of comparative competition The reason for this clear distinction between major within-industry mergers and others relates to the importance of comparative competition, also known as yardstick competition. In the water and sewerage industry, the scope for product market competition is limited, especially for the residential market, owing to the fact that the pipe networks have natural monopoly characteristics and account for the vast majority of the total costs of supply. The prices charged by water companies in England and Wales will therefore remain regulated for the foreseeable future. Comparative competition is a way of achieving some of the benefits of product market competition in a regulated environment. The regulated price set for each firm depends - at least in part - on its performance relative to that of the other firms in the industry. The Director General of Water Services (DGWS) compares the performance of the water companies in such areas as operating efficiency, capital efficiency, customer service and other quality measures. By employing these comparisons when setting regulated prices, the DGWS can simultaneously improve his ability to set prices in line with efficiently incurred costs and incentivise each company to reduce its costs and match the performance of the best. Two inter-related factors are vital for comparative competition to be effective:
Following a merger, the regulator's ability to undertake efficiency comparisons, and the firms' incentives to strive for efficiency gains, could potentially be affected by a reduction in any of the following three factors:
Water-water mergers The importance of comparative competition - and of these comparators being controlled independently - was enshrined for the water industry in the Water Act, 1989. This obliges the MMC when considering public interest arguments to have regard:
The water industry is the only regulated utility in the UK where the MMC has this duty. In the earliest of the three major cases referred to the MMC - Lyonnaise des Eaux/Northumbrian (July 1995) - the takeover was permitted despite the fact that Lyonnaise already owned North East Water and Essex and Suffolk Water. The MMC concluded that the merger would affect the regulator's ability to make comparisons and that this would act against the public interest. However, it believed that this adverse effect could be remedied by "substantial price reductions sufficient to compel the merged company to the forefront of efficiency in the industry". Further, it should be borne in mind that Northumbrian, a water and sewerage company, was merging with two water-only companies, so limiting the impact of the loss of a comparator. A year later, by contrast, when two bids were made for South West Water by existing England & Wales water companies, the MMC concluded that there was no possible remedy that would offset the loss of the comparator and therefore recommended that neither merger be allowed. One reason for this difference of result may have been the fact that the companies involved were both WaSCs. Also important, though, was the fact that the concept of comparative competition, and the likely impact of losing comparators, was developed further during the investigation. Much of this development involved fleshing out the three ways noted above in which a merger can affect the information available to the regulator. The DGWS's view now seems to be that possible remedies exist for the issues of quantity and quality of data, as discussed below, but there is no real remedy for the loss of independence. For example, Severn Trent offered to keep separate regulatory accounts and operational data for South West water. There would thus have been no reduction in the quantity of information available to the regulator, and the potential loss in quality would have been similar to that resulting from a takeover by a company outside the England and Wales water industry. Yet the MMC blocked both mergers, reflecting the importance attached to independence. Further mergers between the ten WaSCs would therefore seem unlikely. But why? In some industries, ten independent companies are far more than would be needed in order to achieve comparative competition. Indeed, at one extreme, certain comparisons can be carried out between as few as two companies, with each competing to outperform the other. This is not true, however, for the England and Wales water industry. Firstly, the companies face diverse operating conditions, both in terms of their physical geography and their demand profile. This makes direct efficiency comparisons between any two companies difficult. Instead, sophisticated statistical analysis is needed, and this in turn requires a large number of independent observations. Secondly, the key comparators at any given time will be the companies that are most efficient in each category. It is these "frontier" companies that the rest of the industry will be judged against and that are critical for ensuring effective rivalry. However, the identity of these companies is likely to be different in each category and also to change over time. Thus, each of the ten companies, with their ten independent management styles, is likely to be at the frontier in some category at some time and so have a significant effect on comparative competition. Mergers with companies outside the industry On the other hand, the DGWS has consistently allowed mergers with companies from outside the England and Wales industry, whether these be in other industries (such as electricity) or in the same industry but outside England and Wales. As noted above, there are three ways in which mergers can reduce the information available for ensuring effective comparative competition. How does a merger with a company outside of the water industry measure against these criteria? The issue of independence is clearly of less concern, since the new management is still independent of other England and Wales water companies. However, these mergers still raise two potential areas of concern relating to the quantity and quality of information:
These two factors are certainly considered by the DGWS when choosing whether to recommend referral for such mergers. In all of the non-referred cases shown in Table 1, the regulator imposed undertakings that were intended to meet these concerns. Had any of the companies refused to provide these undertakings, they might well have found themselves before the MMC, despite being inactive in the England and Wales water industry. Ring-fencing the appointed business Why is ring-fencing so important? Firstly, the DGWS will want to ensure that the appointed business would not be compromised if financial problems were to occur in other parts of the parent company. This can be achieved by undertakings prohibiting debt finance that contains cross-default covenants or requiring that the appointed business retains investment grade credit ratings. Secondly, and more importantly, some degree of ring-fencing is clearly necessary for effective comparative competition. If the operational data and regulatory accounts of the water company are to be of relevance when assessing its comparative performance relative to other water companies, it is important that this information relates to the water business only, and is not distorted by the other activities of its owner. Achieving this separation will require undertakings from the acquirer as to how transfers between different parts of the parent group are to be treated. In many mergers, synergies between the parent and the acquired firm are an important motivation for the deal and these often necessarily involve such transfers. Examples of the synergies that companies have claimed include: improved use of computer resources for billing and being able to issue several bills at one time, improved use of construction services and head-office services. Ring-fencing requires that all transfers take place at arms length and are market-tested. Ensuring that services are contracted out achieves the same result, and the boundaries of where it is possible to contract out services have been pushed back significantly over the last five years. There is, though, a limit on the effectiveness of such measures where there are costs that are truly joint between the two businesses, such that they would have to be duplicated if the businesses were run entirely separately. In this case, market-testing may not be sufficient, since in the presence of economies of scope the sum of the stand-alone market-tested costs of the two businesses will exceed the true costs. This problem will tend to be most important where the synergies between the businesses are greatest. Thus, it would tend to be a more significant problem for mergers between UK utilities than for takeovers of UK utilities by companies currently not active in the UK. Against this background, the regulator's favourable treatment of mergers between water companies and other UK utilities might seem somewhat surprising. It is worth noting, though, that the issue of ring-fencing is not limited to merger situations. It also arises when regulated companies diversify into other activities and Ofwat has spent significant time and effort in establishing a credible and effective system of ring-fencing for regulated companies. As such, the requirements for merged companies involve the refinement of an existing system, rather than the imposition of a new one Loss of financial market information Although the regulator will still be able to collect the operational data that is needed to assess the acquired company’s efficiency, a takeover by another company will usually lead to a loss of financial market information. The effect of this will be two-fold:
How serious are these factors? The first is unlikely to be of major concern. By no means all financial information would be lost; the merged company may well remain quoted and will need to go to the markets to raise finance. Moreover, to the extent that this is a problem, it could alleviated through an undertaking by the merged company to abide by Stock Exchange disclosure rules with respect to the water company. The second factor may be of greater importance. The implications of a loss of a comparator for the purpose of measuring the cost of capital are different from those related to assessing the operating efficiency of a company. In practice, the required rate of return on investment is calculated for the entire industry rather than on a company-by-company basis. Therefore, it is not so important whether a particular company is publicly quoted, provided that the regulator has sufficient data left to assess the required rate of return. At present there are five independent WaSCs (following the takeover of Wessex) and a similar number of water-only companies. As the number of independently quoted companies falls, however, the robustness of the financial estimates will also fall. There are steps that could be taken to alleviate the loss of financial market information. These include:
The regulator needs to determine at what point the information loss becomes a serious problem that cannot be remedied. The electricity regulator, Offer, last month chose not to recommend referral to the MMC for Scottish Hydro Electric's takeover of Southern Electric, the last independent REC. This suggests that the electricity regulator was not concerned about the loss of financial market information. Nevertheless, it will interesting to see how Offer overcomes the resulting loss of financial information at the forthcoming distribution price control review. Conclusions Given the importance of comparative competition for the regulation of the water industry, the DGWS is highly likely to refer to the MMC any consolidation amongst the ten England and Wales WaSCs. Moreover, any such merger is likely to be blocked by the MMC on the grounds that the resultant loss of independence cannot be remedied. The regulator has taken a more lenient view towards other mergers involving water companies, so long as it receives appropriate undertakings, and we would expect to see further bids being made without referral to the MMC. There may, though, be a limit, as the loss of financial comparators gradually erodes the DGWS's ability to regulate effectively Rob Francis is a Senior Consultant in the Domestic Utilities Team at London Economics. Ian Alexander is a Managing Consultant in the International Utilities Team. |
In this edition: MCI/WorldCom Takeovers in Water
|
||||||||||||||||||||||||||||||||||||||||||||||
