IT-Shared Infrastructure

8TH Forum on Telecommunications /ICT Regulation in Africa (FTRA-2007), “INFRASTRUCTURE SHARING” Nairobi, Kenya 6 – 7 June 2007 FTRA 2007 DISCUSSION PAPER Work in progress IT-SHARED INFRASTRUCTURE COMMENTS ARE WELCOME AND SHOULD BE SENT BY 1 JULY 2007 TO FTRA07@itu.int IT – Shared infrastructure General chapter – Why share? (Pourquoi partager?) There are a number of reasons for sharing of infrastructure – both pragmatic and principled – and it is worth teasing these out before looking in greater detail at examples of how it functions in the telecoms sector. The pragmatic reasons usually focus on either reasons of cost and practicality. Whereas the principled reasons relate more clearly to policy approaches, particularly in terms of interconnection and competition. A number of industries including oil, electricity, gas and banking have all made arrangements to share various aspects of their infrastructure. In some cases, this was set up before liberalisation but in others (particularly electricity and gas) it has been a consequence of liberalisation. The single biggest reason for sharing infrastructure is cost. Two or more operators are often driven by cost pressure to share a network. The high prices paid by mobile operators in the UK for 3G licences has been a key driver for a non-binding agreement between Orange and Vodafone on sharing Radio Access Networks. As each base station costs several hundred thousand dollars to install, both have realised that those in rural areas may be uneconomic if each chose to build them alone. Both companies will continue to manage their own traffic independently and retain full responsibility for the quality of service they offer their respective customers. In short, they will share infrastructure but compete on services. A similar deal was agreed in 2004 between Optus and Vodafone Australia. But whilst these kinds of infrastructure sharing deals are motivated by cost reduction, they may also be spurred by a commercial imperative. In India Vodafone has come to an agreement to share rural telecoms infrastructure with Bharti Airtel. Vodafone was keen to increase its footprint quickly and thus increase revenues. A similar deal was put in place in Tanzania when Zantel was given permission by the regulator to roll-out its mobile service from its base in Zanzibar to the mainland of Tanzania. It agreed with Vodacom Tanzania that it could (at an agreed price) make use of the existing operator’s network. In effect, it will be an MVNO until its traffic justifies rolling out its own infrastructure. In several of the above deals there is a clear connection between saving costs and on that basis, for operators to extend the reach of their networks, particularly in rural areas. Sometimes infrastructure sharing in these circumstances is initiated by equipment vendors who agree to operate the rural extension of the network on behalf of several operators. This has happened in a number of places including both India and Tanzania. Often cost-sharing infrastructure deals are made because the infrastructure is being built for another reason and the cost of adding more capacity is marginal. This is particularly true for fibre networks used to manage things as diverse as oil pipelines, power transmission and railways. Each requires its own fibre for management purposes but it is relatively easy to add fibre pairs, either before or after construction. The resulting additional capacity can then be shared either by the operating company setting up its own wholesale fibre capacity sales operation or through it selling the right to sell the capacity to an independent organisation. In Africa, there are several examples of where this has occurred including: the Cameroon-Chad oil pipeline (known as the Doba-Kribi pipeline), Kenya Power and Light and Tanzania’s TANESCO. In the case of the oil pipeline, 12 out of the 18 fibre cables installed will be available for use by operators. Electricity companies can also offer the use of their power networks to deliver broadband access using powerline technologies: an actually functioning African example of this is GTS in South Africa. Arrangements of this kind clearly cut the cost of building network infrastructure. It has been calculated that for every US$1 million spent on building terrestrial fibre networks, only US$300,000 is actually spent on the cables themselves. The rest is spent on rights of way and infrastructure. In terms of development funding, it makes considerable sense to combine the building of things like roads, railways and water pipelines with the roll-out of fibre infrastructure. Again the digging of a ditch alongside the construction of a road or railway is a relatively modest additional cost once the construction crews are in place. The final practical reason for sharing infrastructure is to minimise the environmental impact of network equipment. In 2001 Swedish mobile operators Hi3G and Europolitan Vodafone set up an equally owned joint venture company to roll-out their 3G network in the country called 3G Infrastructure Services AB. It announced that it would open the services of the shared infrastructure company to other operators and in 2002 Orange Sverige became a third equal shareholder in the company. At the initial launch, 3G Infrastructure Services AB announced that the joint network would cover the 70% of the population outside of the three major cities: Stockholm, Goteborg and Malmo. As a result of the joint company, both of the initial founders said that it would cut the amount of masts required by 50%, a figure that obviously went down again when the third operator joined. One of the primary motivations for sharing the infrastructure was to limit the environmental impact, a sensitive issue in many European countries. Both of the founding companies wanted to speed up the application process for cell site building permits and being able to announce a shared cell site would clearly favour speedy resolution of applications. This environmental motivation is not solely a developed country preoccupation. The Nigerian Communications Commission’s guidelines on shared infrastructure specify that one of the reasons they are being put in place is to “protect the environment by reducing the proliferation of infrastructure and facilities installations.” Internet operators have a quite well developed approach to sharing key parts of infrastructure around the co-location of servers and agreed local exchange points for both data and voice traffic. In some cases, companies agree that servers will be co-located for reasons of cost and convenience or to take advantage of local loop opportunities opened up by regulation. So an example of the latter is in the UK where broadband providers co-locate in BT exchanges to provide DSL services to customers. They share BT’s capacity on agreed terms whilst at the same time competing with it at the services layer for customers. With local Internet exchange points, operators agree to exchange traffic between each other at a national level through an agreed point. This lowers the amount of traffic that has to travel outside of a country to reach its final destination. Obviously sending traffic locally through an agreed single point is cheaper than sending it through an international destination. In this instance, competing ISPs have agreed to co-operate together in the provision of the Internet Exchange Point to create savings for all, whilst competing vigorously for customers at the service level. As can be seen from several of these examples, a commitment to sharing of infrastructure does not mean a lessening of competition at the service level. Private companies understand where it makes sense to co-operate as well as compete. Obviously it does raise competition issues and these are examined at the end of this section. There are a number of underlying policy concepts that it is important to understand when looking at infrastructure sharing from first principles rather than simply examining arrangements of convenience, although of course the two often overlap. The first of these concepts is “market failure”, an acknowledgement that for whatever reason the market does not operate freely or efficiently. Where there is market failure, Government can argue that it has two roles to play. Firstly, it may seek to build those things that the market is unlikely to provide: for example, roads or water systems. Secondly, in its role as the “referee” of a market economy, to has the right to intervene or to appoint regulators to intervene. The second concept is the idea of a “natural monopoly”: in other words, the provision of certain goods and services tends favour the development of a single supplier. So for example, it is argued that it makes little sense to build two water or electricity supply systems in a country as this would not be economically efficient. Because services like this are seen as a “natural monopoly” it is important that Government has the power to regulate its operation otherwise the operator of the natural monopoly has the power to completely skew the operations of the market. In the case of telecoms, the idea of a national network being a “natural monopoly” is more open to question. In larger countries there may be three or more national network operators and even in smaller countries, it is possible to have more than one network operator, particularly for the main markets served. However, the issue of how to get network into areas that are currently economically marginal raises the issue of network sharing as can be seen above. It is argued that the provision of networks infrastructure to communicate is a “public good”, irrespective of whether it is delivered by the public or private sector. The issues are then whether the public sector delivers cheap and ubiquitous networks or the private sector provides them and how the Government through a regulator ensures that the operators do not derive a significant competitive advantage if they have a natural monopoly. Examples of public service shared infrastructure are the municipal networks in the United States that are both widely distributed at city level and also highly contested by a range of people. In Africa one of the first “muni” networks is the Knysna municipality in South Africa that has created Wi-Fi coverage across the area that it is responsible for, providing both voice and data services to its 50,000 citizens. There are also a considerable number of other municipalities in South Africa that have tendered for “muni” networks. The infrastructure created can be shared by any service provider at agreed terms. But this idea of public shared infrastructure has not only been proposed at a local level. In South Africa, the Government has proposed a new state-owned company called Infraco that will operate the national fibre network assets of two state corporations, Eskom (the power utility) and Transtel (part of the national railway company). In this particular case, this national network will compete with Telkom South Africa as a way of creating greater levels of price competition on national networks, and in so doing bring down the price of broadband for end-users. On a slightly different basis, the Ugandan Government has prepared a plan to build a fibre network designed to cover the whole country. In this case, Uganda already has two companies – MTN and utl – providing fibre networks. However, again the Government’s proposals are seen as a way of offering a wider range of coverage and bringing network costs down. (CHEC WITH TUSU DOCUMENT) Sharing of infrastructure has become a much more common experience as competitive markets develop and there is a transition to IP networks. Historically, telecom operators were vertically integrated, monopoly operations that provided both the services and the network on which they could be provided. As markets developed, they were joined by mobile operators and second national operators. These were conceived of and generally run as vertically integrated operations, mirroring the structure of the historic operator. They might in certain cases use the network infrastructure of the historic operator but would generally wish to roll-out their own infrastructure. But around this core market structure began to grow up a series of independent entities like ISPs and data carriers whose operations relied heavily on the infrastructure of others. Few ISPs built all of their own infrastructure so these were the first organisations to share the network infrastructure that others had built. As explained above, others - like utilities and railways - seeing an opportunity in fibre network development, have built out their networks to compete with the historic operator. Because they did not have their own traffic, these networks were shared between anyone that wanted to buy capacity. This led to a significant market development that supported infrastructure sharing: the separation of infrastructure and services. This might also be described as the separation of retail and wholesale functions. So for example, if an historic operator is not the vertically-integrated supplier of cyber-cafes, call-shops and telecentres, it will supply independent operators of them with either minutes or bandwidth on a wholesale basis. Indeed, it may supply bandwidth to an ISP that in turn supplies it to a cyber-café, thus creating more than one level of wholesale and retail relationship. The impact of this is already widely known and highly visible in the developed world: the growth of organizations as diverse as e-Bay, Skype and Amazon.com are a product of a lively market of providers in the service layer but rely on others to provide the infrastructure to make those services work. In Africa, wholesalers and retailers of mobile pre-paid calling cards sell to customers from street corners. In South Africa larger ISPs operate virtual ISPs (acting as wholesalers) for smaller organizations with local or specialist customer bases. Mobile Virtual Network Operators (MVNOs) like Virgin in South Africa operate on a similar basis: it buys its network access on an agreed basis from Cell-C who have built most of their own network but who in some instances, buy network access from others. These changes are messy and not clear-cut because liberalisation and increased competition are at different stages in different countries. But these underlying trends are likely to mean that there will be far more horizontal players with trading between the different network layers in the market. For example, where VoIP has been legalised a new category of African VoIP service provider has been set up. It may have little or no infrastructure but “rents” access to network infrastructure from others. Whether from market imperatives or regulatory pressures, significant numbers of historic operators are separating out their different operations into more or less free-standing subsidiaries. Commercially this is motivated by a desire to understand if providing network infrastructure is cost-effective. The actions of the mobile operators cited above show that each has clearly established a good business case for sharing infrastructure. In regulatory terms, the separation of wholesale and retail is often imposed to produce some clarity about underlying costs. Regulators usually want to ensure that a dominant operator is not cross-subsidising its service operations and therefore giving it an unfair advantage that distorts the market. A crucial question both in commercial and policy terms is: what level of infrastructure can a market support? As might be imagined, there is no simple answer that applies in all cases. Even with technology innovation, infrastructure is a large-scale investment and therefore there are unlikely to be more than one or two large-scale entities providing anything approaching national coverage in a mature market. Even with maximum competition, the costs of network infrastructure means that it will tend towards becoming a monopoly. Greater infrastructure competition depends on the value of the market and population density found in a particular country. Relatively wealthy, densely populated countries like Nigeria and South Africa have two infrastructure providers, whilst thinly populated, relatively poorer countries like Chad or Niger may only have one. Mid-scale countries with more competitive markets like Kenya have a second infrastructure operator that provides competition in and between the larger urban markets and increasingly at a national level. In a significant minority of African countries, power utilities offer alternative national infrastructure: in effect, these are carriers’ carriers. The situation is made more complex by the decision of several of the panAfrican mobile operators to go into providing data as well as voice services. Once in place, these network infrastructures will become national networks. The mobile operators are now the dominant voice operators in most markets by revenues. Therefore will these networks be accessed and shared by others in the same way as the network of the historic operator? One of the drivers for mobile operators to implement data capacity on their network is to start making the transition to IP. This has clear commercial implications for network sharing because if voice and data become the same thing – bandwidth rather minutes and bits – there is a strong argument for a single network, even if parts of the traffic are differently managed. This underlying transition to IP networks is an important driver of the sharing of infrastructure as it allows for the separation of services, transport and infrastructure in the market. IP networks are the transmission system for almost anything that can be sent in digital form: voice, text or image. They offer the ability to allow the growth of decentralised, ”intelligent” services at the edge of the network. Traditional telephony in its purest form is a highly centralising technology. The “intelligence” in the network is located centrally (in the functionalities of the switch) and usually controlled by one organisation. In its historic form, largely “dumb” devices (telephones) were attached to the network and these had only a limited set of functions. More recently, the functional attributes of devices attached to the network has increased but this functionality is still run centrally. The telephone network’s root and branch structure means that traffic flows to and from exchanges in ways that reinforce this pattern. By contrast, the IP network (the basis for the Internet) is one where no single entity controls anything but the most basic transport and relationships with other networks. The service-providing “intelligence” is deliberately designed out of the network architecture. Indeed, put simply the network is “dumb” and intelligence is at the edge of the network. For example, a computer accessing the network has a far more complex range of service functionality in its application programmes and this is not solely related to its size. This allows services to operate independently of transport and infrastructure and creates the conditions in which network infrastructure sharing at different levels is both a commercial advantage and a policy consideration. For developing countries, this paradigm shift from a closed to an open network architecture offers plentiful opportunities to carve out a competitive environment that does not need to follow a single path or necessarily copy all aspects of developed world examples, where the sharing of network infrastructure will be a central feature. The basis for these changes can be understood by looking at how those involved describe the operation of IP networks. Diagram 1 below shows a Network Layers model. Essentially there are three key layers: physical infrastructure, transmission and services (including applications and content). In practical terms each layer has a set of functional rules that allow it to interface with the other layer and for information to flow over the network. In terms of day-to-day reality, these network functions are combined with Reach (the geographic scope of providers) and Type of Customer (wholesale or retail). Layer Services Wholesale Transport Type of Customer A vertically-integrated organization like an historic operator or SNO will be selling at every layer: it will both carry traffic and offer services. In this way the historic operator in Africa finds itself selling international transmission to both external ISP customers and to its own ISP, leading inevitably to accusations of conflict of interest. An ISP may offer its own infrastructure at a local or citywide level (using Wi-Fi) to reach its customers but will buy both national and international transmission from another carrier. A relatively recent market entrant in Kenya – KDN – is providing national and international transmission to ISPs and other carriers but not selling to individual retail customers. The Layered Networks model describes a world in which there are multiple relationships between providers operating in different layers in the model. As IP networks begin to take over from telephony networks, this layered approach will become a tool for understanding how markets function. There is already significant infrastructure sharing but as the world moves towards competitive markets with this structure, the terms and conditions under which infrastructure is shared will become a vital policy consideration. In order to share infrastructure in whatever circumstances, there has to be an agreed set of principles that inform these sharing relationships. However, any veteran of interconnection discussions will know that even with agreed principles, the precise details are often hard-fought because they directly affect commercial returns. That said, a broad set of principles for infrastructure sharing may help set the terms for these discussions. There is a need for a policy and regulatory framework that offers: • a technology-neutral framework. Retail Physical Infrastructure Local National International Reach • • competition at all layers in the IP network. transparency (particularly on pricing) to ensure a fair cost of access to all those sharing the infrastructure. all providers have access to infrastructure networks. • In these circumstances, the policy-makers and regulators have to balance a range of different needs. The private sector has to make a return for its shareholders otherwise it will be unable to continue investing. The public sector represents the “public good” – the needs of all the citizens of a country. It may therefore take steps to incentivise infrastructure sharing where the private sector is unwilling or unable to take the risk of extending network in the short-term or monopoly structures are preventing the efficient operation of markets and holding prices at an artificially high level. Because of cost, the idea of a single, rational infrastructure network shared by all seems like a good idea where resources are short. As shown in the international fibre section below, there have already been moves in Africa to rationalise both international and inter-country fibre connections. At this point, the involvement of the public sector raises two questions: firstly, what is the level of provision required for sharing? And secondly, what are the governance structures that will ensure fair sharing of a single infrastructure? The first question is about what level of capacity might be shared. The criticism usually raised is that the market “over-provides” and if there were a rationally planned process of sharing then it would be possible to provide “just enough” capacity. The argument is that “over-provision” by the private sector is wasteful of resources that might be used to invest in other areas. There is some truth in this argument. Indeed, one of the central arguments for investing international donor money in the EASSy cable has been to free up private sector resources for investment in national rather than international infrastructure. However, it is arguable that some level of “over-provision” is a market mechanism that will keep prices lower and service levels higher. Furthermore Governments in general and state planning in particular does not have a good track record of predicting demand over the medium-term. The second question raised is one of governance and here the issues are a combination of the public sector’s ability to deliver efficiently and to do so in a way that does not favour particular interests. Again few Governments have a convincing track record in running “public interest” enterprises that have to operate in a commercially effective way. It is not impossible to achieve but it requires considerable skill and political subtlety and even in countries where these are available is not always successful. Nevertheless, it is perfectly possible for a Government to grant a private sector company the contract to run some part of a shared infrastructure network on its behalf. The example of the Knysna municipal network cited above is one where a municipality has contracted a private enterprise to provide the service. Another example is South Africa’s national broadband network provider Infraco mentioned above. This public sector structure may also be used to lay an international cable from South Africa to Europe, again to provide a shared infrastructure network that could provide international price competition. These may work well so long as the private contracted company in each case does not start competing with other providers at the service level. Because in these circumstances, it will almost certainly be in a position to create unfair advantages for itself. International fibre The history of cable laying goes back to the era of sending telegraphs and there were a number of private companies, some of which had to fight the historic national operators to get access to new markets. In the UK, these companies were nationalised and became the General Post Office. It laid its first submarine cable between England and France, enabling telephone conversations between specially equipped booths in London and Paris. The first transatlantic voice call was made in 1915 as an experiment by Alexander Graham Bell but it was not until 1956 that a transatlantic cable carrying voice traffic was built between the UK and the USA. The impetus for building and operating cables came in the main from Government-owned telephone companies until quite recently 1 . These were vertically-integrated organisations that relied on Government for their finance and the process of building international cables was an inter-governmental activity. The same could be said for building cables between countries across land. It was not until 1998 that the first privately owned cable was built (across the Atlantic) as a partnership between MFS and Cable and Wireless PLC. This led to the first of many cables being built by companies that were not directly connected to the historic operator. The liberalisation of the historic operators in the developed world accelerated this process. Once the market was opened, private companies like Flag Telecom began to both build and operate international fibre cables. However, it was still seen as a relatively financially risky activity. One consequence was that Flag Telecoms was acquired by Indian carrier Reliance Communications in 2004. The overbuilding of capacity on the North Atlantic route (among others) was very beneficial for users but made private investors cautious about investing in this type of infrastructure. In broad terms, two structures of financing evolved to meet the conditions of a liberalised market: construction by private companies like Flag Telecom and the Club Consortium of operators. In the case of Flag, it sought investment from operators through a combination of: a contribution to the landing stations 1 For a listing of operators see: http://www.atlantic-cable.com/Cables/CableTimeLine/index1951.htm on the route; advance purchase of capacity and an element of its own financing. Subsequently Flag operated and managed the cable. Raising finance in this way was quite time-consuming and it sought to limit its risks by selecting routes with where it felt it could see proven traffic. International cable projects were large-scale investments and it was often hard to know what the likely demand for capacity might be. This was particularly true for the North Atlantic route where arguably a bandwidth glut increased use substantially. If all of this was true for one of the busiest routes in the world, it was doubly true for anybody wishing to build international fibre cables to Africa or indeed most parts of the developing world. Liberalisation has opened the market to new competitors at both an international and national level. But competition to build new international fibre routes in the developing world was slower to arrive. It was hard to find the scale of finance required from a single investor or operator. Most finance capital or loans are spread over a five to ten year period. Like most infrastructure investment the risk on an international fibre cable is long-term and spread over the life of the cable, usually 25 years. International fibre routes are in effect a form of shared infrastructure and often shared between potential competitors. Therefore their subsequent management required a clear framework for co-operation between participants. When club consortia were formed to build international cables just before the main wave of liberalisation in many developing countries, most of the likely participants were Government-owned historic operators. However, by the time these cable were completed, there were a number of Second National Operators and GSM companies who were unable to participate because they had not been in existence when the financing was agreed. The shareholders in these type consortia raised funds through a combination of their own resources (often supplied by Government) and donor funding to some of the smaller companies involved. In addition, a number of international companies participated. Money was raised against the advance purchase of bandwidth so each company became both a shareholder and a bandwidth purchaser. In order to protect against the high financial risk taken by some consortia, participating companies were given time-limited monopoly on the sale of that bandwidth in their own country. The participants also elected a managing agent. What is SAT3/WASC and how does it work? The building of SAT3/WASC was a considerable achievement. Its promoters notably Telkom South Africa - seized the last moments before the end of the telecom investment boom to raise finance for it. It combined those wanting to transport African traffic with those needing traffic to transit around the world from other destinations. It is in effect two cables: SAT/WASC which travels down the west coast of Africa to Cape Town and SAFE which goes from Cape Town to Asia. In terms of Africa, the latter connects Mauritius and Reunion (a French "department") to the rest of the world. The consortium that built these two connected stretches of cable has 36 shareholders. The SAT3/WASC section has 11 African shareholders: Angola Telecom, Camtel, Cote d'Ivoire Telecom, Ghana Telecom, Maroc Telecom, Nitel, OPT Benin, OPT Gabon, Sonatel, Telecom Namibia and Telkom South Africa. Two do not have landing points (Maroc Telecom and Telecom Namibia) and among the full list of shareholders is the Indian carrier VSNL that is owned by Tata that has bought into the soon to-be-set-up South African SNO. What does the "club consortium" mean in practice? It is very difficult to say precisely as its basis is a shareholder agreement that is "commercially confidential". However the broad outlines of how it works are fairly clear. The consortia's shareholders appoint a managing agent who runs it on their behalf, taking care of day-to-day performance and maintenance issues. In the case of SAT3/WASC this is Telkom South Africa which also has the largest amount of traffic among consortium members. Consortium members have a monopoly on selling fibre in their own country. If the consortium builds more capacity – as it has done recently - then its members can take up (on which they have first right of refusal), then the "pool" capacity will be sold off as IRUs (Indefeasible Rights of Use). Whilst the club consortia structure got cables built when it was difficult to raise finance, times have changed. This form of international infrastructure sharing may well have been appropriate for when it was built but there is now a widely held consensus that it should not be repeated in the future. There are a number of issues that sharing international infrastructure using a club consortium model raises, both in terms of raising finance and operationally. These include: Investment and access issues: In one country where there was a single international fibre cable built by a club consortia, a second national operator was set up it needed access to international fibre capacity. Without this access, the only supplier of international bandwidth would be the historic operator, its main competitor. To overcome this problem, it sought to buy its own shareholding in the club consortium. It was told that it would not be able to buy a shareholding in the club consortium. In order to overcome this problem, it has had to announce that it will build its own international fibre cable. Whilst this move is laudable in terms of developing a more competitive international infrastructure, it has solved the second national operators’ access problem but not the underlying issue of access and investment in a liberalised market. The questions still remain: what guarantees do other operators (who have not built their own cable) have of obtaining fair prices for international bandwidth, if both of the main international cables are controlled by their rivals? What opportunities do they have to invest jointly with others in capacity? Faced with this dilemma, some parts of the Government talked about hiving off the historic operator’s international fibre capacity as a separate operation but this did not become publicly accepted policy before it was privatised. There are three recurring issues: the impact of the monopoly on landing stations; the monopoly on the sale of capacity; and the fact that shares in the consortium are not tradeable. Therefore the following issues need to be tackled: Monopoly supply and pricing issues: Where the historic operator is the sole supplier of international fibre bandwidth in a given country, the access given to this capacity and the prices set for it are entirely in the hands of a single organisation. In the early years of such a cable, prices for bandwidth are often set an extremely high by any comparable international benchmark. There may have been some justification for setting prices in order to make a speedy return on the investment. But in cases of this kind it is arguable that the initial prices and subsequent reductions do not reflect the speed with which the operators recouped their investment. However on one of these club consortia systems, fibre prices were pegged at a level that either mirrored or were more expensive than competitive satellite prices in the same country. In properly competitive markets, international fibre prices are more usually cheaper on volume traffic than satellite. So for many countries in the beginning, the cost of buying bandwidth means that muchneeded foreign exchange is still going outside of the continent to satellite operators. This situation still remains the case now for a much smaller number of countries. Not using the international fibre capacity makes it a wasting asset. Although its life is 25 years, each day part of its capacity is not used means that some of the value of that asset is lost. If the prices set were lower and the volumes higher, then a greater part of that capacity would have been used in the earlier years. No landing station: Where international fibre routes have been built by Club consortia, those operators in seaboard countries are often able to exercise undue influence the price paid to those without landing stations. In one case, the cost of transiting through a country of this type and for the stretch from that country to Europestill make it cheaper for the operator without a landing station to send a significant part of its traffic by satellite. From a purely investment point of view, the country involved should have bought its own landing point. But there is a wider point of principle here in terms of infrastructure sharing: why does each country need its own landing station if there are clear “ground rules” for accessing other countries’ landing stations? The costs of each country having its own landing station significantly increases the cost of large international fibre projects so why not share a landing station with another country? Landlocked countries: This same problem recurs when landlocked countries are put under the spotlight. The landlocked countries are in the same position as the seaboard countries without landing stations. There is no framework for agreeing the terms of access or the cost of transit to the landing station. So in one case the cost of transit from the landlocked country to the landing station is greater than the cost from the landing station to Europe. In another case the transit charges to the landing station mean that the overall cost of access until recently was higher than for using equivalent capacity over satellite. Again there is no framework for sharing privately-owned terrestrial infrastructure resources and no clear mechanism for addressing the issue. Transparency: The shareholders’ agreements that govern club consortia controlled international fibre systems are often described by their members as “commercially confidential”. As a result it is difficult to know the exact terms and conditions under which shareholders have agreed to operate. Efforts to obtain the shareholders’ agreement by national governments have not always been successful. In one case, a shareholders’ agreement explicitly defines confidential in a way that means members should be able to show it to their national governments. Although it describes the agreement as “confidential”, it does make an exception under the following conditions: “Where a signatory party is a legal entity in a national jurisdiction it should abide by the laws and regulations of that jurisdiction and render up the required information to all governmental agencies requiring it and to all competent judicial tribunals”. However, despite this definition, there have been members who have continued to insist that it remains confidential. Options for addressing these issues There are a number of ways of addressing the issues raised above, some that are relatively simple and others that are more complicated. These measures are designed to ensure a greater level of sharing international infrastructure at competitive prices that will allow all African operators to compete more successfully at the services layer in their own countries. The more competitive choices operators have in a single country, the better the range of available prices. So for example, Onatel in Burkina Faso has access to three countries (Benin, Cote d’Ivoire and Mali) where there is an operator with an international landing station. The price of international fibre bandwidth in each country varies. By contrast, Mali only has access through one country to an international landing station. As there is a greater roll-out of African infrastructure links – not only fibre but also microwave – this pattern of competitive links should increase. But there needs to be an acknowledgement that it is not only the historic national operator that should be allowed to operate these links but any operator prepared to share its network. Alternative inter-country network suppliers would include: carriers’ carriers, utility companies and mobile operators with networks in several adjoining countries. Lobbying pressure from national ISP associations in at least two West African countries (Ghana, Nigeria) has resulted in the lowering of the prices offered by the national historic operator. This has usually resulted from a combination of political and commercial pressure. The commercial argument advanced by the ISP associations has been that if they “bulk-buy” for all their members they should be eligible for better discounts. On the regulatory side, after the national monopolies end, there is no reason why national regulators should not be able to insist on the co-location of access points from other carriers under agreed cost-based terms. Pricing has already been challenged on the SAFE section leading to Asia, the Mauritius regulator, the Information and Communications Technology Authority (ICTA) conducted a price investigation into the international fibre prices offered by the historic operator. The latter is privately owned but continues to have a substantial Government shareholding. On the basis of information supplied by the historic operator, an uncontested decision was reached to lower prices substantially. Indeed at the point at which the price investigation was announced, the historic operator made a first reduction in its prices. Other approaches to addressing this problem are more draconian and timeconsuming. As the club consortia shareholders are monopoly suppliers, the issue could be addressed using competition law. In Europe, European Union competition policy talks about fair access for all users. Unfortunately in Africa there is no continental competition institution and at present there is only competition legislation in a limited number of countries. However, competition challenges in any circumstances are complicated and drawn-out affairs. The same issues occurring elsewhere These issues affecting sharing international fibre infrastructure are not ones that solely affect Africa. They occur anywhere there is monopoly supply or access to international cables. Examples from two very different parts of the world (the Caribbean and Asia) may help illustrate the point and suggest other ways those involved can address the issues to create fair access to shared infrastructure. Until relatively the Caribbean island of Jamaica had only one international fibre cable. It was both owned and operated by the historic operator Cable and Wireless Jamaica. According to the Office of Utilities Regulation (OUR), the pricing of local and international data transmission circuits and Internet services has not been as competitive as it could be. According to OUR, this was contributing to a lower than desired access to the Internet for the Jamaican population. In view of the need to address Jamaica’s shortage of submarine fiber-optic facilities for international telecommunications connectivity, the Minister of Commerce, Science and Technology directed OUR to develop a long-term policy and a regulatory framework for international cable landings in Jamaica based upon considerations of the public interest, environmental soundness and national development imperatives in a competitive market. In response to this directive, the OUR in 2004 issued invitations requesting the submission of applications for a license to build, own and operate an additional Fiber Optic Cable Carrier Facility (FOCCF) landing on the shores of Jamaica. The construction, ownership and operation of this additional facility required specific approval and permission from the Minister. At first, the aim was to attract and promote the development of one additional service provider but this was later changed to allow more than one. Among the preliminary requirements for the additional providers was the provision of detailed network drawings and a roll-out timetable to include important milestones such as a start-up date, completion of construction and the target launch of full service. The provision of fiber-optic cable connectivity had a number of technical requirements. The planned route of the cable was subject to scrutiny with regard to matters such as its originating point, points of interconnection with other international undersea cable facilities and integration into regional networks. By the end of 2004, three entities had submitted applications for providing international submarine fiber-optic connectivity. They were: 1. 2. 3. Trans Caribbean Cable Company (TCCC); Digicel Jamaica and FibraLink. On January 5, 2005, the Jamaican Government announced that TCCC and FibraLink had been successful and the two additional cables were subsequently built. TCCC is a management organization comprised of more than 35 telecom companies for planning, building, operating, and maintaining the TransCaribbean Cable System (TCCC). The project was conceived to service the ever-growing Internet, data and voice traffic demands of the Caribbean and will connect in the USA to a point in Miami. Furthermore TCCC has plans to become the region’s common, carrier-neutral platform that will provide the capacity needed to stimulate the growth of the Internet and other telecom services in other countries in the region. Members of TCCC include: AT&T Wireless (now Cingular), Gotel Communications, Oceanic Digital, Digicel and Verizon, which already have a presence in Jamaica. Funding was agreed for the first phase in August 2006. FibraLink is a joint venture between Jamaican-based Merit Communications (33%) and Bahamas-based Caribbean Crossings. FibraLink Jamaica owns and operates the sub-sea fibre optic network that links Jamaica with the continental United States and fourteen countries in the Caribbean basin. So Jamaica has been able to encourage greater fibre competition in the region by calling for competitive tenders and this way allowed the full range of operators in its market to share international infrastructure on a fairer and more competitive basis. In India the regulator TRAI decided to take action because monopoly control of its international submarine landing cable was causing high prices. Its report on the situation concluded (among other things) that: • • • • New operators (need to) have access to the information about available capacity in the same way as consortium members. Activating IRU (Indefeasible Rights of Use) capacity is not unduly delayed by consortium members. Tariff conditions must be transparent and non-discriminatory to consortia members or non-members. Restoration and maintenance services need to be ensured/provided through a Service Level Agreement. Furthermore it concluded that there needs to be:”Close monitoring and scrutinizing (of) the situation of possible anti-competitive behaviour in order to ascertain whether the incumbent operator continues to control most of the submarine landing facilities in this country.” Within this context, Flag Telecom, a subsidiary of Reliance Infcomm challenged cable operator VSNL in court in May 2006 to give it certain access rights. These included the right to upgrade and sell international bandwidth capacity on undersea cable Flag-Europe Asia (FEA) and the right to access VSNL’s landing station in Mumbai. With this, Flag can upgrade its international capacity and sell this new capacity to customers in the Indian market. Additional capacity is likely to lead to further falls in international bandwidth prices. Prices have come down 70% since 2002. The Indian regulator TRAI had set a ceiling on international fibre prices and most fibre is now bought at or below the celing price. The International Court of Arbitration of the International Chamber of Commerce (ICC) gave the arbitral award in the dispute between Flag and Tatas-owned VSNL. Flag had accused VSNL of not allowing it to upgrade and sell capacity in the undersea cable. The dispute was referred to ICC in December 2004, holding among others that Flag has the right to access any landing station including that of VNSL in Mumbai. In its judgement, the Court said:“International Court of Arbitration of ICC has rejected VSNL’s counter claim in its entirety.” By setting both a clear regulatory framework and setting prices, TRAI has allowed prices to fall and created the environment in which monopoly access can be challenged. Again this means that the full range of operators in India can share international infrastructure on a fairer and more equitable basis. EASSy – Finding another way to share international infrastructure When those proposing to build EASSy initially came together, it was initially thought that the shareholding and governance for the new cable would simply mirror the traditional club consortium structure. However as the project went forward it became clear that this older way of doing things was perhaps not “fit-for-purpose” in the twenty-first century. For example, there were a range of investors who wanted to get involved who were not historic operators. A wide-ranging debate took place that was in part spurred by the involvement of the World Bank in offering to finance part of the project. In essence, it was arguing that for it to put donor money into the project, it could not be structured in a way that among other things created monopoly advantages and kept prices high. In effect, donor money is a combination of monies from Governments and its use should therefore reflect “public good” intentions. The term it used to describe the principles of its approach was Open Access. What is EASSy? The EASSy submarine cable will connect eight coastal countries in Eastern and Southern Africa (that are currently unconnected) to other global submarine cable systems: SAFE in the South Africa and SEA-ME-WE 4 (and potentially others) in the North. The eight coastal countries are: Sudan, Djibouti, Somalia, Kenya, Tanzania, Mozambique, South Africa and Madagascar. Eleven land-locked countries will also be connected to the system: Ethiopia; Uganda; Rwanda; Malawi; Burundi; Zimbabwe; Zambia; and Botswana. The system will be capable of carrying 320 giga bits per second end-to-end, and the cable length is approximately 9,900km. The project cost is estimated to be US$300-million. The EASSy consortium believes that the system will be ready for service by the end of 2007. The 23 members of the consortium signed the supply contract in March 2008. By interconnecting with Sea-MeWE 3, Sea-Me-We 4 and SAT3/SAFE, the EASSy undersea cable will also serve as a supporting infrastructure for these networks. Initially the structure proposed for EASSy was a single Special Purpose Vehicle (SPV) that would be financed through a mixture of loan and equity. But various of the larger investors felt that they would prefer a structure that reflected their particular commercial goals and protected their investment against risk. As a result, a compromise structure was created, the details of which are now available but which may still change as things go forward. In its investment announcement, the IFC, the investment finance arm of the World Bank described the EASSy structure as follows: “The Hybrid Model for the financing and ownership of EASSy is composed of two parts: • a Special Purpose Vehicle (the SPV), which will receive funding from the private-sector arms of the involved DFIs, including IFC, to create a commercially viable structure which meets open access policy objectives as well as supporting some operators (the SPV Operators) to fund their investment in the Project; and a Consortium, which will be composed of those operators (the Consortium Operators) that elect not to join the SPV and, instead, make direct investments in the EASSy cable. • The SPV has recently been incorporated in Mauritius with the name of West Indian Ocean Cable Company Ltd. It will be majority owned by the SPV Operators. The SPV will have its own by-laws, Board of Directors and professional management team. It will function as a commercial entity which will manage, market and sell or lease its capacity to SPV Operators and any other authorized third parties. IFC has been asked to provide financing to the SPV, along with five other DFIs including EIB, AfDB, DBSA, AFD/Proparco, and KfW. The funds will be used to finance that share of the Project cost resulting from the projected capacity requirements of the SPV Operators (the SPV Project Cost), as well as certain preparation and start-up expenses. The SPV Project Cost will ultimately depend on the composition of the SPV, which has yet to be finalized; however, in the current scenario which assumes 14 SPV Operators, the SPV project cost is approximately $120 million”. Put simply, there are two elements to the funding of EASSy: firstly, those that have the financial strength to invest directly in the EASSy consortium with their own money; and secondly, those not able to make the full contribution who are members of the SPV funded by the IFC and other donors. The SPV will also own all the capacity not pre-sold to consortium members. In terms of pricing, those who have invested most will benefit from the equivalent of a volume discount. The SPV members pay a minimum equity component for shares in the SPV rather than the Consortium. Their bandwidth purchases are on a sliding scale based on volume. The mechanism that will help keep prices low is that the EASSy SPV will sell surplus capacity to authorised purchasers using a cost plus-based pricing formula. In this way, it will help keep markets competitive and ensure low wholesale prices. The latter should in time lead to much lower retail prices for end-users. This cost-plus based business model is not perfect and has a number of features that are worth noting: • High levels of bandwidth take-up will result in decreased prices being charged. But over the life of the cable it will probably ensure higher levels of return. Low levels of bandwidth take-up will lead to higher prices as the nonConsortium bandwidth would not probably be called upon. High prices will keep demand down and result in a downward cycle where low demand will not produce the necessary cash required to cover the loans. However it is understood that there is a degree of contingency financing and the sale of SPV surplus capacity can be used to help adjust market prices downwards. Since the surplus bandwidth is to be sold on a cost-plus basis, this is the main source of return for the SPV. The cost-plus formula can be adjusted during the life of the cable. • • The EASSy story is not a simple one. The arguments over the structure of the vehicle to build the cable have excited much interest and some disagreement. Alongside the commercial structure described above, African Governments in the countries to be linked by the cable have decided (through NEPAD) to provide a policy and regulatory framework for EASSy by putting in place a political protocol. Not all countries have signed this political protocol, most notably Kenya. It has fundamental disagreements with some parts of the text and has taken a position that it will both encourage its own project and stay involved with the EASSy consortium to ensure that a cable is built to the country in the quickest time possible. Its own initiative is called The East African Marine System (TEAMS) and it will connect Mombasa and Fujaira in the United Arab Emirates. It has signed an MOU with Dubai-based Etisalat to build it. In terms of ownership, Etisalat will take 40%, the historic operator Telkom Kenya 20% and the remaining 40% will be floated on the Nairobi Stock Exchange. The Kenyan Government says that capacity on the cable will be available on an Open Access basis. India’s Reliance Communications is planning to spend US$1.5 billion on four more new undersea cables, rolling out the FLAG Telecom network to an additional 23 countries and upgrading its entire system to next-generation IP infrastructure. An African route will connect Kenya, Mozambique, South Africa, Tanzania, Madagascar and Mauritius. Flag is committed to a “low price, high volume” strategy. When making the announcement, its Chairman said:”We live in a world where there is too much bandwidth for some and none for many – there is unequal access to bandwidth in and across continents, countries and communities. The Flag NGN will democratise digital access and give nearly everyone in the world the opportunity to be part of a massive lifestyle change….What’s more, it will make this bandwidth available…at strikingly affordable prices.” Exact prices have not yet been made available. Finally there is Sithe’s SEACOM that aims to connect up a broadly similar set of countries to EASSy to Italy or India. It will act as a wholesaler of bandwidth – in other words be a “carriers’ carrier” – and it has already lowered its anticipated pricing to unlock what it believes will be pent-up demand. It is offering a range of carriers IRUs and going into partnership with others where it requires licences to operate. Although the EASSy process was not designed to stimulate interest in the market from other investors, it has had precisely this impact. Whether all four schemes will be built is doubtful but it is likely that at least two cables will be built, giving at least some level of competition at an international level in these newly connected countries. Clearly if all four of these cables were to be built, there would be a high risk of capacity “over-building” and everyone investing would stand to lose the money they had invested. With this in mind, one organisation 2 has pointed out that it is perfectly possible for competing cable operations to share the same cable system. This is both technically possible and has happened elsewhere. There are at least two methods technically of achieving this: multiple fibre pairs can be assigned to separate parties and operated independently; or multiple wavelengths of light on the same fiber can be assigned to separate parties and operated independently, especially via the Optical Add-Drop Multiplex (OADM). In either case, the different systems may be given separate names and may serve different markets, or may compete charging different prices for capacity on the same system. Conclusions Policy-makers and regulators have a variety of levers that they can use to ensure that there is fair sharing of international infrastructure. These might include: 2 International networks for Sub-Saharan Africa – A new era? – A presentation given at a United States Trade Development Agency event in San Francisco in March 2007 by David Ross of The David Ross Group Inc. - Insisting that carriers can co-locate at landing stations with cost-based charges. This will allow both international and African operators to be able to offer international capacity at competitive prices in a range of countries. Allowing alternative operators – like power utilities and railways – to offer connecting fibre between different countries. For example, in southern Africa there is a well-developed, cross-border power grid that would enable this to occur. Encouraging mobile operators to connect their networks between countries and under agreed terms, allowing other operators to share the use of that infrastructure. As with the Jamaican example quoted above, encouraging investors to create inter-country links that compete with existing links, enabling all countries to have two or more shared international infrastructure links. - - -