open acces ing - apcrights.apc.org/documents/open_access_en.pdf · open access lowering the costs...

A South African, Mike Jensen sent his first email more than twenty years ago. He is an independentconsultant with experience in more than 30 countries in Africa assisting in the establishment ofinformation and communications systems over the last 15 years.

APC ISSUE PAPERS

Mike Jensen

OPEN ACCESSLOWERING THE COSTSOF INTERNATIONAL BANDWIDTH IN AFRICA

A S S O C I A T I O NFOR PROGRESSIVECOMMUNICATIONS

Bandwidth is the life-blood of the world’s knowledgeeconomy, but it is scarcest where it is most needed –in the developing nations of Africa which requirelow-cost communications to accelerate their socio-economic development. Few schools, libraries, uni-versities and research centres on the continent haveany internet access. For those that can afford it, theircosts are usually thousands of times higher than fortheir counterparts in the developed world, and evenAfrica’s most well-endowed centres of excellencehave less bandwidth than a home broadband userin North America or Europe, and it must be sharedamongst hundreds or even thousands of users.

A variety of factors are responsible for this situation,but the biggest cause is the high cost of internationalconnections to the global telecommunication back-bones. This is mainly the result of the lack of inter-national optic fibre infrastructure, which is neces-sary to deliver sufficient volumes of low-cost band-width, and the consequent dependency on muchmore expensive satellite bandwidth. Less thantwenty of the 54 African countries have international

optic fibre cable connections, and these are currentlycontrolled by inefficient state-owned operatorswhich charge monopoly prices while neglecting tobuild the national backbones needed to carry localand international traffic. As a result, circuits fromAfrica to the US or Europe usually cost more thanUS$5000 a month1 , while cross-Atlantic links be-tween North America and Europe can now be ob-tained for US$2.5/Mbps/month and for US$16–30/Mpbs/month on international routes in Asia2 .

The only large-scale international fibre link in Af-rica (SAT-3/WASC/SAFE) connects eight countries onthe west coast of the continent to Europe and theFar East. Operating as a cartel of monopoly state-owned telecommunication providers, prices havebarely come down since it began operating in 2002.New fibre projects have been proposed which couldbreak this monopoly and add many more Africancountries to the global grid, but most of theseprojects are also being developed by state-ownedtelecom operators. As a result they are followingthe same high-priced SAT-3 business model. Unlessinterventions are made to reduce the cost of these

1 See for example - GISPA signs agreement with GT for new,low prices on SAT3. Balancing Act Issue 233www.balancingact-africa.com/news/back/balancing-act_233.html. and Fibre Optic Cable Systems in the ArabWorld www.aticm.org.eg/admin/Documents/ArabFiberOptics-Study2.pdf

2 See for example circuits obtained by the research networkCanarie, Canada. Note these prices are predicated on 3–5year contracts for multi-Gbps circuits. Personalcommunication with [email protected]

open acces ing 10/17/06, 7:00 PM1

2 / ISSUE PAPERS

ASSOCIATION FOR PROGRESSIVE COMMUNICATIONS

3 While demand forecasting is heavily dependent on theassumed end-user cost, the billion dollar annual paymentsfor satellite bandwidth across Africa already show thatthere is sufficient demand for international infrastructureto justify deployment for virtually every country in Africaexcept perhaps the smallest and most remote nations.

4 See for example the World Bank InfoDev Study: OpenAccess Models: Options for Improving BackboneConnectivity in Developing Countries www.infodev.org/content/highlights/detail/2568 to undercut pricing.

existing international fibre links and to ensure thatnew fibre infrastructure is quickly built, the conti-nent will be prevented from tapping its latent po-tential and will fall further behind the rest of theworld.

This problem is not unique to Africa. Other devel-oping regions suffer from the same problem, but itis at its most extreme in sub-Saharan Africa, whichhas the lowest teledensity in the world and thehighest unmet demand for telecommunication serv-ices3 . Fortunately, African governments and the in-ternational community have recently become moreaware that action is needed to improve access tocommunications and to encourage the adoption ofalternative business models that can significantlylower the cost of international links. These havecentred on what are known as open access models,which are cost-based and owned by the public sec-tor (similar to roads and rail lines), rather than be-ing operated by a club of companies aiming tomaximise profits.

Most African country telecommunication marketsare slowly moving to a more competitive environ-ment which will ultimately address pricing and na-tional imbalances in demand and supply. Howeverthe international sector in developing countries isdifferent from developed nations because the ma-jority of countries have markezts that are too smallto justify the cost of deploying many competing in-ternational fibre cables. With each cable able to carrydata at terrabit speeds, only one international con-nection to a global hub is needed, although a sec-ond physically separate link is also required for back-up (redundant connection) purposes. Howeverachieving competitive pricing between just two sup-pliers is infeasible. Thus, in order to ensure cost-basedpricing, a different model of deployment is needed,where the cable and landing points are operated ona non-profit basis, extending the models used byinternet service providers for operating national orregional Internet Exchange Points (IXs).

This follows a number of recent studies which haveidentified public-private partnerships and open ac-cess models as a more appropriate solution for fibre

deployment4 . These also build on precedents set bythe oil and gas industries when building pipelines,in which the basic approach is to establish a SpecialPurpose Vehicle (SPV) to operate the facilities. Themain objective of the SPV is not to make a profit,but to facilitate profits made elsewhere by the par-ticipating companies. The aim is not to exclude in-cumbent telecom operators from the process, butto allow the participation of others that might bringadditional funding or other advantages to the tablesuch as rights of way to build fibre along power orrail routes5 .

The most viable structure for this approach is likelyto be a two-part system in which national cable land-ing points are managed by national associations ofbandwidth providers, while the cable itself is ownedby a mix of operators and private or public inves-tors. Given that the most appropriate place for thecable landing point is likely to be at the facilities ofthe national operator, these would most likely beowned by the state, but operated by a managementcompany appointed by the national association ofbandwidth providers.

With the cable itself, different models can beadopted. In one scenario any entity would be freeto invest, either as an operator, in which case theinvestment would be tied to guaranteed amountsof bandwidth, or as a non-user shareholder whomight invest funds or provide a right of way (e.g. agas pipeline operator wishing to minimise the costof operating their pipeline network). Alternatively,ownership of the cable can be defined on a nationalbasis with shares held by the same special purposecompanies that operate the landing points.

In either case, sufficient investment is likely to comefrom the much broader base of operators that wouldbe able to access the bandwidth at cost, and littleadditional financing would likely be required6 . How-ever some of the smaller, more remote or less devel-oped countries might require special assistance, andgiven the general interest by the international com-munity in ensuring more universal access, along withthe positive impact on demand for national back-bones that would result from affordable interna-tional connectivity, donors could provide a demandguarantee that would meet any revenue shortfalls

5 Finding rights of way for cable is a major difficulty interrestrial projects unless they can be provided bymunicipalities and parastatals such as railway, pipeline,and electricity grid operators.

6 The potential role for private finance in these projects is alsomore limited because there is virtually no risk of failure – thedemand for international fibre bandwidth from the first twoconnections is guaranteed, and if services are offered at cost-plus, there is little opportunity to undercut pricing.


3 / ISSUE PAPERS


in the early years. This may be a risk for donors if thedemand was not met over the life of the cable. How-ever, assuming the long-term business case is sound,they might look to recoup the funding when trafficincreased at a later point. Donors could also be in-vited to meet the cost of additional add-drop unitson fibre projects to ensure small and remote com-munities along the way can be reached. The choiceof these locations would be a matter for negotia-tion between the donors and national governments.

Given the interest of governments in supporting thedevelopment of their nations such as through im-proved access to health and education, along withthe broader social improvement and enhanced publicservices which can be provided through better con-nectivity, there is a growing interest amongst a widerange of stakeholders in ensuring that open accessmodels are adopted.

The initial focus is likely to be on supporting theadoption of open access models for the upcomingEast African fibre project (EASSy, see below) whichcould then be replicated in West and Central Africa.At the same time SAT-3 and other existing interna-tional fibre cables may be declared essential facili-ties serving the public good with regulated pricing.Specific activities are likely to be:

• Increased backing for policy makers and regula-tory agencies in Africa to implement policychanges and regulations that allow open accessto international fibre

• Support to local associations of bandwidth pro-viders to establish shared international fibre gate-ways

• Increased backing for international fibre projectswhich aim to provide equal access to all band-width providers.

There is the risk that the entrenched interests of theincumbent operators and their state-owners will beable to resist efforts to change national telecompolicy, and that the EASSy project goes ahead as cur-rently planned. Nonetheless, support from a broadrange of stakeholders is expected to substantiallyimprove the chances of an alternative strategy be-ing adopted, which could have a major impact onthe way international fibre projects in developingcountries are being planned in the future.

In summary:

Most of Africa is as yet unconnected to the globalfibre backbones.

Optic fibre is the only way to supply sufficient inter-national low-cost bandwidth.

As elsewhere, the limited fibre that has been laid inAfrica is not competitively priced, and uses businessmodels developed by cartels of monopoly telecom-munication operators.

A cable planned for the East coast of Africa (EASSy)which will have a major impact on bandwidth avail-ability in the region, was being developed as a clubof mostly state monopoly operators with high pricesand low volumes in mind.

The strategy for the deployment of an open accessmodel for EASSy is in the process of being legislatedby policy makers in the region.

The adoption of a low-cost open access model forEASSy would likely have a major impact on the waynew fibre projects are planned in other regions inAfrica.


4 / ISSUE PAPERS


Communication costs in Africa are currently thou-sands of times higher than in Europe or NorthAmerica. This particularly affects those with themost limited resources: students, researchers, doc-tors, scientists, and other public servants, as wellas the general public, who are unable to take fulladvantage of the unprecedented access to knowl-edge the internet provides. Cheaper bandwidth forAfrican institutions, particularly governments,schools, universities, libraries and hospitals wouldprovide widespread access to the wealth of infor-mation available online, facilitate African contri-butions to the global economy and increase thelikelihood of successful solutions to African devel-opment problems. So in a nutshell, the constraintson development in Africa caused by the high costof communications are not being addressed dueto inappropriate business models used for deploy-ing international fibre infrastructure.

The developed world is benefiting from the sur-plus of optical fibre cable laid during the dot-combubble which has coincided with technology ad-vances that have made speeds of over 1000

Gigabits per second routine on these fibre links.While those in the North reap the benefits of thesedevelopments, much of the South, and Africa inparticular, has not seen significant deployment ofinternational fibre. This is clearly shown in the dia-gram on page 5 which depicts improvements in thespeed of the internet over the last 10 years as meas-ured by Stanford University in the US. Africa is theregion showing the slowest improvements and isactually steadily falling behind the rest of theworld.

There is only one intercontinental fibre link to sub-Saharan Africa (SAT-3) which provides connectionsto Europe and the Far East for eight countries alongthe west coast of the continent7 (shown in the mapleft). Except for some onward links from South Af-rica to its neighbours, and from Sudan to Egyptand from Senegal to Mali, the remaining 33 Afri-can countries are unconnected to the global opti-cal backbones, and depend on the much more lim-ited and high-cost bandwidth from satellite links.Even the few countries that have access to inter-national fibre through SAT-3 are not seeing thebenefits because it is operated as a consortiumwhere connections are charged at monopolyprices8 by the state-owned operators which stillpredominate in most of Africa, and in many otherdeveloping regions.

As a result, institutions in these countries paythousands of dollars a month for internet con-nections which a home broadband user in NorthAmerica would pay US$20 a month for. Aside fromthe general dampening effect this has had onuptake, unaffordable bandwidth has actually ex-cluded African scientists from gaining access tothe services of global research networks whichnow expect their member countries to have atleast 1Gbps on international connections in or-

Source: Mike Jensen/NEPAD e-AfricaCommission. “Needed” links are definedby the NEPAD e-Africa Commissionas thosenecessary to support its strategy of establishingredundant fibre links to every capital cityin Africa, for which it is now in the processof lobbying operators and investors.

FIBRE NETWORKS & PLANS

ACTUAL

FUNDED ⁄ COSTED

NEEDED

1. THE NATURE OF THE PROBLEM

7 As detailed in Annex 2 (Stakeholders in TelecomInfrastructure Deployments in Africa). See end note.

8 The International Telecom Users Group (INTUG) observedin January 2005 that “the level of the prices seldom arisesfrom high underlying costs, but instead are imposed bymonopoly or dominant operators through the exercise oftheir market power. Often these operators are shieldedfrom potential competitors by the refusal of theirgovernments to permit any new players to enter themarket in international telecommunications”.


5 / ISSUE PAPERS


9 Telemedicine and genetic research in particular requirehigh bandwidths for the transfer of images, video andlarge data sets, other examples include high definitionvideo, super-computing, and physics, and remote sensingdata. The ATICS survey of 84 leading tertiary institutionsin Africa found 850,000 students and staff with access to atotal of only 100Mbps international bandwidth(www.atics.info). By contrast, Australia’s tertiarycommunity of 250,000 share 6Gbps of internationalbandwidth (although even this is still insufficient to meettheir needs).

der to access the advanced services and petabitdata sets they now provide9 .

In a chicken-and-egg situation, the constraints ondemand resulting from the high tariffs charged bythe monopoly operators have contributed to theslow pace of fibre deployment and the severe lackof investment in needed infrastructure. Many ofthese state-run telecom operators, often misman-aged, inefficient and suffering from much reduced

TCP THROUGHPUT MEASURED FROM NORTH AMERICA TO WORLD REGIONS

DER

IVED

TR

AN

SMIS

SIO

N C

ON

TRO

L PR

OTO

CO

L (T

CP)

THR

OU

GH

PUT

IN K

BIT

S/SE

C

profits caused by the collapse of international set-tlement rates, do not have the resources to investthe millions of dollars needed to deploy nationaland international fibre, and neither do their hostgovernments. Understandably, few private investorsor donors are interested in financing these mori-bund organisations that rest on artificially-closedmarkets. At the same time, continued state-opera-tor control over international gateways and na-tional backbones has meant there are very fewopportunities for investment in privately-operatedtelecommunication infrastructure.

Source: Stanford Linear Accelerator Center PingER projectwww.slac.stanford.edu/xorg/icfa/icfa-net-paper-jan06/report-jan06_files/image022.jpg


6 / ISSUE PAPERS


ble was limiting Sentech’s ability to provide afford-able, high-speed internet access to consumers. It saidthat SAT-3/WASC/SAFE was a strategic national assetthat was funded by taxpayers. At the time the cablewas initiated, Telkom enjoyed a statutory monopolyand was majority-owned by the state.

There are three recurring issues here: the mo-nopoly each national operator has on the land-ing stations in their country; the monopoly onthe sale of capacity; and the fact that shares inthe consortium are not tradable. Examples include:

Under pressure from Ghana’s ISP association the localincumbent with the landing station, Ghana Telecom,lowered its prices by about a third to US$8,050 amonth for an E1 leased circuit from Ghana to Portu-gal. Previously, E1s had been quoted at betweenUS$12–15,000 a month, but even with the reduction,the price is many times higher than cost. The agree-ment is again covered by commercial confidentiality.

Landlocked African operators who have tried topurchase international fibre capacity directly fromone of the consortium’s international members havefound themselves being charged as much to reachthe SAT-3 landing point as they were charged toget from the landing station to Portugal. Sadly, thehigh costs have made it cheaper to send the trafficdirectly by satellite, even for SAT-3 shareholderssuch as Telecom Namibia, which has no landingpoint of its own.

Nigeria’s SNO, Globacom, realising that to be com-petitive it would need access to its own internationalfibre capacity, tried to buy a shareholding in the con-sortium. It was told that it could not do so, presum-ably because Nitel would be threatened by its accessto capacity. In the meantime, Globacom has an-nounced that Alcatel will build it a fibre between Ni-geria and England. The Nigerian government hastalked about separating Nitel’s SAT3 capacity as anindependent operation but this has not yet becomepublic policy while staff at Nitel have objected to theproposed separation saying that this would jeopard-ise the proposed privatisation of Nitel.

A rough budget for SAT-3 shows that the investmenthas already been recouped and running costs shoulddrop to US$30 million a year. Taking into accountthe new upgrade costing US$30–50 million, whichwill result in a doubling of capacity to 40 gigabitsper second, charges could come down to somethingcloser to those found on the North Atlantic but areunlikely to occur in the absence of competition fromother new cables.

The first large-scale international fibre project in sub-Saharan Africa, SAT-3/WASC’s first segment connectsPortugal to the Cape in South Africa reaching eightcoastal countries along the way: Senegal, Ivory Coast,Ghana, Benin, Nigeria, Cameroon, Gabon and An-gola. A second segment, in the Indian Ocean, con-nects South Africa to Malaysia while passing throughMauritius and India (SAFE). Jointly funded by 36members11 and spearheaded by South AfricanTelkom which invested US$85 million for a 13 percent stake, the project cost about US$650 milliondollars. The cable was expected to lead to much re-duced international bandwidth costs, but so far thishas not occurred due to the business models used todevelop the project.

The ownership of the cable was established as a clubconsortium, which is a confidential shareholder agree-ment about which little is known12 . The shareholdersappointed Telkom South Africa as the managing agentwho runs it on their behalf, taking care of day-to-dayperformance and maintenance issues. Telkom SouthAfrica also has the largest amount of traffic amongconsortium members. Consortium members have amonopoly on selling access to the fibre in their owncountry until April 2007. If the consortium builds morecapacity than its members can take up (on which theyhave first right of refusal), then the “pool” capacitywill be sold off as IRUs (Indefeasible Rights of Use).

Being the only international fibre cable available hasput the consortium’s owners in a relatively unassail-able position. Sentech, the South African state-ownedbroadcasting and telecom provider, argued in parlia-ment last year that Telkom’s monopoly over the ca-

11 The SAT-3 shareholders are: Angola Telecom, AT&T Corp(USA), Belgacom SA, Communications Global NetworkServices Ltd (BT), Cable & Wireless Global Network ,Camtel, China Telecom, Chunghwa Telecom Ltd Co, Côted’Ivoire Telecom, Cyprus Telecommunications Authority,Deutsche Telekom AG, France Telecom, GhanaTelecommunications Co, Global One Communications,Maroc Telecom, Korea Telecom, KPN Royal Dutch Telecom,Marconi (Portugal), Mauritius Telecom, MCI WorldcomInternational, Nigerian Telecommunications, OPT Benin,OPT Gabon, Reach, Singapore Telecommunications,Societe Nationale des Telecommunications du Senegal(SONATEL), Sprint Communications Co, Swisscom Ltd,Telecom Italia SpA, Telecom Namibia, Telefonica deEspana, Teleglobe (USA), Telekom Malaysia Berhad,Telkom South Africa, The Communications Authority ofThailand, and Videsh Sanchar Nigam Ltd (India).

12 Despite the continued efforts of the South Africangovernment to obtain details of the nature of theagreement, the consortium has not released details.

2. THE EXPERIENCE FROM SAT-3/WASC/SAFE


7 / ISSUE PAPERS


Approaches to opening up access to the cable usingcompetition legislation have been discussed, but cur-rently it appears that only South Africa has the ap-propriate legislation. There is also lack of clarity re-garding lapsing of the monopoly on the landing sta-tions in 2007 and this will still likely require legisla-tive change to allow other shareholders and whole-salers direct access to these international gateways.In the meantime the South African government isdiscussing declaring the SAT-3 landing point an es-sential facility ahead of the 2007 date, but is debat-ing whether to do this through legislation or throughregulation. ICASA could draft the necessary regula-tions under sections 44 and 51 of the Telecommuni-cations Act. As an essential service, ICASA would beable to set the prices for access to the cables andregulate these according to Telkom’s costs.

But any move by the government to have the cablesdeclared an essential service, or to amend legislation,is likely to be met with strong opposition fromTelkom. Telkom says it is a public company and that

the cable is its asset. “It would be an unfortunateprecedent to nationalise this cable landing, as itwould discourage Telkom from any further invest-ments in projects of this nature,” it said in a state-ment. Some observers have echoed this concern andwarned against hasty decisions by government, sug-gesting thhat declaring the cables an essential serv-ice could raise concerns among foreign investors thatgovernment is interfering unnecessarily in the mar-ket. Telkom says access to the capacity in the cablecould be classified as an essential facility only if itmeets certain characteristics. But, it says, it doesn’tmeet these characteristics for various reasons, onebeing that satellite links are available as substitutes.

Thus most observers have concluded that improvingaccess to low-cost international bandwidth is morelikely to be achieved through new projects which willalso put competitive pressure on the old models,rather than to focus purely on the problematic areaof legislating access to existing facilities which wereestablished under different regimes.

3. THE CASE OF THE EAST AFRICAN SUBMARINE SYSTEM (EASSY)

Recent efforts to establish a fibre project serving thecountries on the east coast of Africa is one suchproject, which provides an ideal case study in newmodels for telecommunication infrastructure provi-sion, while underlining the problems describedabove. Known as the East African Submarine Sys-tem (EASSy), the project is being developed by about25 telecom operators, of which twenty are majorityowned by African governments13 in the region, fourare private operators which have recently receivedinternational gateway licences (in South Africa,Kenya, Somalia and Tanzania), along with recentlyexpressed interest from international operators in-cluding British Telecom, Teleglobe and Etisalat.

With their sole franchise on international links, thestate-owned operators have adopted a closed con-sortium ownership model, similar to that of SAT-3,which raises the spectre of continuing the strategyof selling small quantities of bandwidth at high mar-gins. Currently prices on SAT-3 are up to US$15,000 /Mbps/month, while it is estimated to cost the con-sortium only about US$300/Mbps/month14 . With new

technologies and a shorter cable, EASSy will be ca-pable of up to 640Gbps and bandwidth should costthe operator less than US$15/Mbps/month to pro-vide15 , while current indications are that pricing willinitially be in excess of US$1,000/Mbps/month16 .

13 In early June 2006 Mauritius was included in NEPAD’sproposed alternative to the EASSy consortium.

14 The cable is estimated to cost about US$1500 million over itslife: (US$600m to lay + US$35m/year x 25years for financing,maintenance, and upgrades) i.e: Assuming only about halfthe total bandwidth (75Gbps) of the cable is sold over itslifetime, the annual cost of the cable is: US$1500m / 25 years= US$250m/year. Monthly cost /Mbps: US$250m/75 000Mbps/12months = 280 US$/month per Mbps

15 The cable is estimated to cost US$1030 million over its life:(US$280m to lay + US$30m/year x 25years financing,maintenance, and upgrades) i.e: The annual cost of the cableis: US$1030m / 25 years = US$41m/year. Assuming only half thetotal bandwidth of the cable is sold over its lifetime, themonthly cost /Mbps: US$41m/300 000Mbps/12months = 11US$/month per Mb/s. This does not include the cost ofupstream bandwidth to gain access to the global backbonesfrom the EASSy termination points, currently planned forSouth Africa, Djibouti and Sudan

16 The pricing target proposed in June 2006 by a group of DFIsworking with the EASSy consortium is for bandwidth toinitially be priced at about 30% of current satellitebandwidth prices and then as demand increases, to reducethe tariffs to around US$750/Mbbps/Month by 2012.


8 / ISSUE PAPERS


20 Such as the World Bank’s ATICS and Open Access Modelstudies, NEPAD e-Africa Commission studies and IDRC’sPAREN programme.

21 See Annex 3 for the Declaration of the Policy Makers

mit on the Information Society (WSIS) in Tunis in200520 , national policy makers in Africa are nowaware that traditional strategies to telecommunica-tions backbone deployment have not worked andthat a new approach needs to be adopted. At thesame time the DFIs made it known that they wouldnot finance the EASSy consortium if it continued asa closed club.

While the EASSy consortium continued to pushahead with its strategy, the seeds of an alternativeinitiative were laid at the November 2005 meetingof Southern African ICT policy makers in Botswanawhich requested its representative body, the NEPADe-Africa Commission, to develop an open access non-discriminatory model to build the necessary fibre in-frastructure in the region21 . The meeting proposedthat SPVs, consisting of public-private partnerships,be used to operate the infrastructure. Followingthrough on these recommendations, studies on regu-latory and business models for SPVs were conductedon behalf of NEPAD by the DFIs, the CommonwealthTelecommunications Organisation (CTO) and theSouth African parastatal investment agency, the In-dustrial Development Corporation (IDC).

After a series of subsequent meetings earlier thisyear, involving the East and Southern African policymakers, regulators, DFIs, NGOs, and a few of theEASSy consortium members, an intergovernmentalworking committee (IWC) was established at theApril 2006 meeting of policy makers and regulatorsconvened by NEPAD. The IWC was chaired by RITA(the national ICT promotion agency of Rwanda), andcomprised regulators and policy makers represent-ing Botswana (as deputy, also representing NEPAD),Kenya, Lesotho, South Africa and the East AfricanCommunity (EAC). The IWC was charged to:

• Make recommendations on the viability and func-tions of a proposed Inter Governmental Assem-bly (IGA) which would have oversight over inter-national fibre infrastructure in the region througha ‘golden share’ investment in the SPV

• Make recommendations on the functions andscope of the SPV

• Conclude the dialogue with the DFIs and theEASSy consortium membership

• Take actions needed to move the project forward,including defining the process for registering theSPV

While most of the EASSy project members have amonopoly on international links in their own coun-tries, even where there is more than one EASSymember in the same country, it is possible that thestate operators could leverage their position inboth the wholesale and retail markets at the ex-pense of the other bandwidth retailers without in-ternational gateways, thus increasing the costs tothe end-user. In addition, their sole rights to thesale of international bandwidth allows the opera-tors to integrate the wholesale and retail chains,giving them an unassailable position in the mar-ket, making it more difficult for new private play-ers to gain market entry.

The French consulting group, Axiom, which carriedout the Detailed Feasibility Study17 for the EASSyproject, admit in their report that: “It is usually diffi-cult to find a Consortium’s majority ready to agreeon attractive or even fair capacity pricing policy forsale to non-Consortium members.” Axiom did dis-cuss an alternative Special Purpose Vehicle (SPV)model with a more diversified shareholding, includ-ing national operators, but they concluded that theirmodel for an SPV had problems because interna-tional gateway operators would be in competitionwith the other investors, and because of the unclearregulatory environment for an SPV18 .

Currently the EASSy consortium has raised pledgesfor the bulk of the funds from within the group ofoperators, but it has been delayed in finalising theproject because an additional US$60m-$140m in fi-nancing is still needed. Many of the smaller EASSyparticipants were expecting to obtain soft financefor their stake in the project from multilateral andbilateral development finance institutions (DFIs) suchas the World Bank and the European Union.

Globally, telecom infrastructure provisioning strate-gies are still under debate19 , and recently, the mod-els for financing the EASSy cable have come underscrutiny by governments in the region as well as theDFIs, who are also keen to see lower charges on net-work infrastructure. Culminating in the highlight-ing of African bandwidth issues at the World Sum-

17 Along with the other reports it is available on the E-AfricaCommission site http://www.eafricacommission.org

18 To be fair, at the time of the EASSy project conception theregulatory environment encouraged a closed club model asnational telecom policies precluded anyone but the licencedinternational operators to provide international infrastructure.While this is still the case, following the June 2006 meeting thecommunication ministers of the region have adopted inprinciple a modified regulatory environment which wouldallow the provision of wholesale services by an SPV.

19 See Annex 7 – Teletopia: A New Regulatory Agenda forAmerica


9 / ISSUE PAPERS


to work out the modalities for taking over the workthat has already been done by the consortium mem-bers. This meeting was scheduled for the first weekof July.

The key features of the open access model and asso-ciated strategies that were adopted in principle bypolicy makers at the June meeting is that:

1) An SPV would be established in Rwanda to im-plement, own and manage the cable with a cost-based tariff regime, a regulated rate of return oninvestment and ultimate control which rests withthe governments of the region. The establishmentof the SPV to operate the cable in this way isaimed at what is seen as the key deficiency withthe consortium model, which is that the domi-nant bandwidth retailer in each country and thewholesaler (the consortium) are not separatedand thus the consortium member retailers are ableexploit vertical integration in their markets to thedetriment of other service providers which do nothave this advantage. Thus, ensuring that thewholesaler (the SPV) is fully separated from theretailers and being licenced in each country ef-fectively spells the end of international gatewaymonopolies and allows the retailers to concen-trate on their core business while profiting fromthe lowest possible international bandwidth costs.The model also accommodates the variety in na-tional policy environments, so that local retailersare free to charge what they like for internationalbandwidth, with the expectation that competi-tion will drive down prices and in countries wherethere is still a restricted number of retailers in themarket, there will be increased incentives to openthe market to benefit from low prices.

2) The SPV will have equal shareholding from eachcountry in the region to underline the commonownership of the infrastructure by all the nationsof the region and to ensure that small nationsare not at a disadvantage from larger nations.Currently it has not been decided if the same SPVwill operate the terrestrial backhaul infrastruc-ture, or if a separate SPV will be established tocarry out this function.

3) Each country can elect how it will allocate itsshares to the local operators and can allow anynew operator to invest in the SPV at any time.This principle was adopted to eliminate one ofthe main objections to the consortium model,namely that the region is undergoing a majorprocess of liberalisation and that the many op-erators that are expected to be licensed in thenear future would not have the opportunity toinvest in or directly benefit from the cable. Not-withstanding the benefits of having a say in howthe SPV is managed, this deficiency is also ad-dressed by the additional provision that there will

• Raise the interim funding to continue the workof the IWC

• Produce a report for the planned Ministers’ meet-ing.

At the same time, a group of regulators lead byKenya was tasked with responding to the CTO re-port on the proposed regulatory model.

The recommendations emanating from these twogroups were subsequently incorporated in a draftregional protocol22 and associated declarations thatwere hammered out in an intensive three-day meet-ing convened by the NEPAD e-Africa Commission inearly June and attended by most of the PermanentSecretaries and Director Generals from the relevantICT ministries in the Southern and Eastern Africancountries.

The protocol covered the establishment of the SPValong open access principles, an IntergovernmentalAuthority to govern it and the changes needed tothe national regulatory environment to accommo-date the SPV. The protocol was then presented tothe June meeting of Communication Ministers in Jo-hannesburg23 which agreed to house the SPV inRwanda24 , include Mauritius in the project25 , and toadopt the draft protocol as a working documentwhich would be taken home for further review andthen formal signing and ratification at a meeting tobe convened in August 2006 under the auspices ofthe African Union. The e-Africa Commission was alsorequested by the Ministers to convene a meeting inthe interim with the EASSy consortium members toensure they are on board with the new strategy and

22 Officially known as the Draft Final Protocol on Policy andRegulatory Framework for NEPAD ICT BroadbandInfrastructure Network for East and Southern Africa. Thecountries party to the protocol are: Angola, Botswana,Burundi, DRC, Djibouti, Eritrea, Ethiopia, Kenya, Lesotho,Madagascar, Malawi, Mauritius, Mozambique, Namibia,Rwanda, Somalia, South Africa, Sudan, Swaziland,Tanzania, Uganda, Zambia and Zimbabwe.

23 Officially known as the Draft Resolutions of the Ministersresponsible for ICTs and / or Telecommunications inEastern and Southern Africa, 2–3 June 2006.

24 The housing of the institution to manage the submarinecable in the landlocked country of Rwanda was adoptedas a sign of commitment that the cable project was aimedat benefitting the region as a whole and not just thecoastal countries, as had been the case with SAT-3.

25 Although Mauritius had been involved in the earlydiscussions of EASSy, it had not participated in theconsortium meetings and thus was left out of the Axiomfeasibility study. Given that Mauritius already has access toSAT-3, its interest in EASSy is indicative of the problemscountries are experiencing in obtaining affordablebandwith from the SAT-3 consortium.


10 / ISSUE PAPERS


be one price for bandwidth, regardless of whetheror not the purchaser is an investor in the SPV. Thisaddresses the other key limitation of the consor-tium model, being that only the licenced interna-tional operators investing in the cable at the timethe of finalising the financing of the project wouldbenefit from lower cost bandwidth.

4) There will be one price for bandwidth on the ca-ble regardless of distance. This may seem to runcounter to basic business principles, but the issuewas debated extensively at the policy meetingsand it was ultimately agreed that this would bethe best strategy to further the interests of re-gional integration, so as to not penalise the moreremote and isolated nations. However the impactof this policy on the backhaul terrestrial links tothe submarine cable has yet to be examined andthere may be some need to reconsider the policyfor these links. The reason for this is that in con-trast to the submarine cable, which has only onelanding point in each country and can thereforeonly be used for international traffic, the terres-trial links could have add-drop points all alongthe route and could be used by bandwidth retail-ers to carry domestic traffic as well as internationaltraffic. Clearly if this occurs, then tariffs shouldnot be uniform across the network, otherwisedomestic charges will effectively subsidise inter-national traffic.

In finalising the details of the SPV and determiningthe strategy going forward there are still a numberof outstanding issues that will be need to be ad-dressed before the cable is built. The chief issues are:

• The need to re-examine the traffic forecasts onwhich the business model is based. Currently themodel adopted projects the existing slow growthin high-priced international bandwidth and onlyexpects to provide for the low capacities that themonopoly operators expect to carry. i.e only20Gbps by 2016. Regionally, the academic insti-tutions alone currently need about 20Gbps (1Gbpsper country) to participate in the global researchand education networks, and the internationalcommunity is ready to fund it if it can be pur-chased at a reasonable price26 . But the currentmodel assumes that in total only 20Gbps will becarried on the cable by 2016. Clearly if so littlebandwidth is sold, the high prices envisaged bythe consortium would be justified to cover thecost, but this does not take into account the tre-

mendous latent demand which which expandsubstantially if affordable pricing is available. Inthe case of Uganda for example, the Axiom re-port estimates that by 2010 only about 1Gbps willbe needed by the whole country for internationalbandwidth, whereas the research and academicnetwork alone is likely to need this amount, andthe current rollout of infrastructure by the pri-vate mobile, fixed wireless operators and ISPs inthe country will shortly result in much greaterpublic demand if bandwidth is affordably priced.In many respects this is a true chicken-and-eggsituation – continue the current high pricing andbandwidth demand will most likely increase veryslowly. However it will take a leap of faith to pricethe bandwidth ten times more cheaply and as-sume that the uptake will be ten times greater. Ithas been pointed out that alternative competingcables may emerge in the future to reduce de-mand for bandwidth on EASSy, however if thebandwidth on the EASSy cable is priced at cost-based tariffs, there will be no business case for acompeting cable.

• As the legislation envisaged will effectively levelthe playing field and dispense with internationalgateway franchises, such legislation is likely to beresisted by the incumbent state operators and thefew others with international licenses. In manycases these operators have been given or pur-chased licenses contingent on periods of exclu-sivity which would need to be ended. Negotia-tions between government and the internationalgateway licensees to bring forward the end oftheir periods of exclusivity will likely be needed.This could entail a negotiated settlement for theaffected operators based on estimates of their lossof revenues, with funds provided by governmentand/or backed by the international community.This process would probably need to precede theestablishment of the SPV.

• The need to ensure that the governance and eq-uity structures for the SPV maximises African own-ership, but minimises the cost of finance byleveraging the best mix of equity and debt fromthe different players. The financing models pro-vided by the DFIs and the IDC will need to be con-sidered for their ability to balance the nationalinterests with the funds available in such a widerange of country sizes, regulatory environmentsand levels of economic development, which re-sults in diverse levels of demand, sources of sup-ply and costs of finance.

• Apart from the decision to look at the modalitiesfor adding Mauritius to the network, there is amore general need to re-examine the landingpoints for the EASSy cable and the plans forbackhaul network. Insufficient attention has beenpaid to the costs of getting from EASSy to the

26 Developed by a group of national research networksknown as the Ubuntunet Alliance.


11 / ISSUE PAPERS


27 Which has now joined its fibre network to Sudan andDjibouti.

global backbones, partly because the cash-strapped state operators were focusing on reduc-ing their investment needs by limiting the lengthof the cable. As currently envisaged, the cablewould terminate in South Africa on SAT-3 in thesouth and in Djibouti and Sudan on SEA-ME-WE-3 and Falcon in the north. The additional transitcosts that must be paid to the operators of thesecables for onward links to the global fibre hubsin Europe, Asia and North America could reducemuch of the cost-saving potential of the EASSycable. Also, potentially cheaper alternative ter-restrial routes for transcontinental traffic, such asvia Sudan, Ethiopia27 and Egypt, have not beenconsidered, and the strategy for linking the landlocked countries is currently undefined. Similarlyto SAT-3, which only lands in eight of the twentycountries along the west coast, the EASSy consor-tium model has not aimed for completeness in itscoverage of countries within the region, andmembership simply reflects the individual opera-tor’s interest in participating in the project. Oncethe cable is designed, the addition of new land-ing points at a later stages is usually not an op-tion with marine fibre as these projects need tobe designed as an integrated system.

• Considerable work will need to be done to de-velop a strategy for the terrestrial backhaul net-work. In addition to the terrestrial transmissionnetworks of incumbent fixed-line operators, anumber of other providers also own and operatetelecom transmission networks for their own pur-poses. These include cellular operators, electric-ity operators, pipeline operators, road and rail-way operators. In one or two cases these organi-sations have obtained licences to provide whole-sale services to other licenced operators, and in afew other cases they are leasing long-haul trans-

mission capacity to telecom operators. In severalkey places where gaps exist in the regional net-work, this alternative infrastructure offers sub-stantial capacity which is currently unused. Inmany countries in the region mobile operatorshave built the most extensive national fibre back-bones. Thus cellular operators may lease capacityon the backbone of incumbent operators whereit is cost-effective but in many cases have had toconstruct their own transmission networks in or-der to carry traffic between base stations andswitches. In total, cellular operators such as Celtel,MTN and Vodacom have invested in the regionof over US$500m on the construction of thesenetworks within the last five years. Furthermore,some of the cellular operators such as Celtel areactive in a number of countries with contiguousborders and are in a good position to provide in-ternational backbones across these countries. Asa result representatives from the national retail-ers would likely need to be brought together torethink the strategy for the terrestrial network.

• The EASSy consortium members have recentlysaid that the urgent need to implement theproject means that the region cannot wait forthe restructuring that an SPV would require.Whether an SPV would take much longer to con-stitute is open to question, but the Axiom De-tailed Feasibility Study has pointed out that dueto weather conditions, the most appropriate timeto build the cable is between December and May.So there is actually still adequate time to makethe necessary preparations for an SPV to initiatelaying the cable in December 2006. In any eventthe end-user would certainly be prepared to waita few months longer if it meant much betterlong-term prices. ■

This paper was commissioned by the APC as part of the Catalysing Access to ICTs in Africa (CATIA) initiative(supported by DFID) and to contribute to APC’s efforts to promote open access to ICT infrastructure in Africa(supported by the IDRC, InfoDev, OSI, OSIWA and UNDP).

The paper, which was finalised in late June 2006,contains annexes available online:

http://rights.apc.org/documents/fibre_bandwidth_annexes_EN.pdf


APC is an international network of civil society organisations foundedin 1990 dedicated to empowering and supporting people workingfor peace, human rights, development and protection of the envi-ronment, through the strategic use of information and communica-tion technology (ICTs).

We work to build a world in which all people have easy, equal andaffordable access to the creative potential of ICTs to improve theirlives and create more democratic and egalitarian societies.


Internet and ICTs for Social Justice and Sustainable Development

www.apc.org [email protected]

OPEN ACCESS: LOWERING THE COSTS OF INTERNATIONALBANDWIDTH IN AFRICA

APC “Issue Papers” Series 2006October 2006

APC-200610-CIPP-I-EN-P-0027ISBN 92-95049-27-6

Creative Commons Licence: Attribution-NonCommercial-ShareAlike 2.5

Gra

ph

ic d

esig

n: M

ON

OC

RO

MO
