International Mobile Roaming: Progress and Challenges in African Markets

The persistence of high prices for international mobile roaming services, in contrast to falling national mobile prices, has been a recognized item on the global regulatory agenda for half a decade. In Africa, there have been studies and discussions about regulatory options in regional economic groups and in the various networks of national regulators. As yet, there has been no transnational regulatory action. Yet the initiative of one large operator saw the introduction of transnational tariff schemes (ie without a surcharge for roaming), forcing competitors to collaborate in order to respond, if they wished to attract and to retain customers. This has both saved money for consumers and greatly reduced the need for regulatory interventions that might have proved counterproductive. In some countries this type of offer remains impossible, because international gateways are a monopoly, having yet to be opened to competition.


INTRODUCTION 1
Recent years have seen remarkable growth in cellular wireless telecommunications in Africa, rising to over 350 million reported connections, equivalent to just over one-third of the population (see Figure 1). The predominant technology has been GSM, with only some 10% being connections using CDMA (CDG, 2010), plus a very few individuals with satellite telephones, supplied by Inmarsat and Thuraya. However, these numbers are significantly and systematically overstated, due to the ownership by many individuals of multiple SIM-cards -with the resulting need to take at least 20% off official mobile teledensity figures (Sutherland, 2009  An overwhelming majority of these customers are pre-paid (see Figure 2). This reflects individuals' low levels of disposable income, their uncertainty about future availability of cash and their lack of experience in the use of credit. Moreover, the operators have access to credit histories for only a tiny minority of their prospective customers, making risk assessment entirely impractical. In contrast, top-ups for mobile phone services are widely available at locations close to customers and in small, affordable increments.

FIGURE 2: PREPAID CUSTOMERS IN SELECTED AFRICAN COUNTRIES IN 2008
Source: ITU, 2009 For Mobile Network Operators (MNOs) in Africa, International Mobile Roaming (IMR) is a very attractive service, both for inbound and outbound roamers. Inbound roaming traffic generated by tourists, business travellers, visiting journalists, government officials and the like can be highly lucrative -even in war zones -with foreigners making expensive IMR calls from airports, hotels and offices. The traffic is paid for by the home MNO in foreign currency and requires no marketing effort, with the only financial risk being fraud control for which procedures are well established (eg accelerated transfer of records) (GSMA, 2007). Outbound roaming appeals to a set of high-spending domestic customers, such as government ministers and business leaders, who wish to use their phones all over the world. To secure these influential customers and to avoid them switching to a domestic rival, operators enter into a very wide range of contracts for IMR, even in countries where the likelihood that the service will be used might seem small. For the predominantly poor customer base the option of paying very high rates for IMR simply does not exist. Instead, they find inexpensive alternatives, of which the most obvious is to purchase a local SIM-card and thus engage in "plastic roaming". This causes inconvenience for colleagues, family and friends, who are no longer able to call them on their usual number. Africa now has a number of geographically extensive operator groups, recently modified by the acquisition of Atlantique by Etisalat and by the sale of most of the Zain networks to Bharti Airtel (see Table 1). Their geographical scale allows them the possibility to internalise at least a portion of their IMR traffic, except where they cannot obtain their own international gateway. In a few countries there remains a formal monopoly, requiring all international traffic to pass through the incumbent operator. These countries are: Angola, Burkina Faso, Cameroon, Eritrea, Ethiopia, Gambia and Zimbabwe (ITU, 2010).  (EU, 2009). While politically effective, it was an economically crude instrument, one that further complicated analyses of the markets (Sutherland, 2010).
Rightly or wrongly, IMR has come to be seen as a policy and regulatory problem, rather than as a commercial challenge or opportunity. It became a feature of international policy and regulatory discussions, being raised at regional bodies for Asia-Pacific, the Americas, the Arab states and Africa. It was also taken up at the Organisation for Economic Cooperation and Development (OECD, 2009). It was discussed at the ITU-T in Study Group 3, first in 2002 and again in 2009-10 (ITU-T, 2010). The ITU-D held discussions at its 2006 Global Symposium for Regulators (GSR) and included a chapter on IMR in the 2008 edition of Trends in Telecommunications Reform (ITU, 2008). IMR was also included in the ICT Regulatory Toolkit (infoDev & ITU, 2010). The IMR problem had become part of orthodox regulatory agenda, something that ought to be considered.
However, there has been a poor level of understanding of the economics of roaming, not least because of the shortage of data, especially concerning the operations of wholesale markets. This has delayed and perhaps confounded the identification of a solution to the IMR problem in the form of an intervention in the market.
Individual governments and regulators have been caught in a form of Catch 22, being expected to act but being unable to do much that is likely to be other than counterproductive. Collective action requires a legal basis that, outside the EU, does not exist except, conceivably, in free trade agreements.
Thus while the problem of IMR has been recognised, there are only very limited analyses and no solutions. There is no robust and detailed model of roaming markets with which potential interventions can be tested, nor even satisfactory data to construct such a model. Governments and regulators are thus forced to conduct experiments in the real world with very limited certainty of outcomes, something that, understandably, gives rise to hesitancy. This is compounded by the need for collaborative action between groups of governments and regulators in which interests are often poorly aligned.

THE AFRICAN ROAMING MARKET
In 2008, the global market for roaming was estimated to be worth USD24.5 billion, with some 365 million outbound roamers. Of that total, African countries represented only one percent of roamers, though forecast to grow to around three percent by 2013 (see Figure 3). A substantial majority of the African roamers were consumers rather than business travellers, though both groups were forecast to grow.

FIGURE 3: FORECAST OF TOTAL OUTBOUND ROAMERS FROM AFRICAN MNOS
Source: Stainthorpe, 2008 Unfortunately, there are no equivalent estimates for inbound roamers, making it very difficult to assess the market dynamics or to understand the net cash flows. Only when inbound and outbound roamers are both accounted for can the effects of IMR be analysed. Countries in North Africa and some of the Small Island Developing States (SIDS) are likely to have heavy volumes of inbound traffic from tourists and from return visits of migrant workers.
Unusually, there are data for roaming traffic for the Cape Verde Islands. These show the expected preponderance of inbound over outbound traffic (see Table 2). The effects of the global financial crisis can be seen in the sharp drop in traffic in 2009. Taking the 2005 data on tourists visiting the Cape Verde Islands, some 198 000 visitors (of a total population of 500 000), represented about 13 minutes per visitor over an average stay of 4 to 5 nights. In 2008, outbound roaming represented only 0.1% of total mobile voice traffic, while inbound roaming was about 2.4%, though it was likely to be a higher portion of the international calls originating on mobile networks. Given the higher charges, the revenues would not be in proportion.
The revenues earned by African MNOs from outbound roaming are shown in Figure 4. There are minimal amounts from SMS and data roaming, with the vast majority of the money coming from voice traffic. The forecast growth comes mostly from roaming between countries in Africa.  Although there are no data on inbound roamers, it is possible to consider the prices they pay, for example, the IMR tariffs charged to visitors from the US. Table 3 shows the prices charged in the summer of 2006, Table 4 and Table 5 for the summers of 2009 and 2010. The same broad pattern applies to all, with a "flat rate" charge, the same per minute rate for incoming, local and international calls. If customers pay an additional monthly fee, usually on an annual basis, lower rates are available. For the most part, the prices are quite expensive, in some cases extremely so. In particular, calls forwarded to Kenya from the US, where the wholesale cost would be a few cents, are charged at USD3.99 or 4.99 per minute. The differences between the tables suggest there is little, if any, competition between the operators in the US and give the appearance that the operators are engaged in an upward price spiral, copying the higher prices of their domestic rivals. Although some of their customers engage in plastic roaming, it would appear not to exert any pressure on the home operators.   + The higher rate is the standard roaming plan and the lower rate is the value plan.

Source: Websites of AT&T T-Mobile, Verison
The net effects of IMR in African countries are very difficult to assess in the absence of data on the inbound number of minutes and the associated revenues. Even the relatively expensive rates charged to visiting Americans cannot be fully evaluated without the wholesale rates paid by the US-based MNOs to their African IMR partners. Nonetheless, it seems likely that the bulk of the profits on IMR calls for visitors from developed countries are retained at home. Moreover, the regulators in developed countries ought to examine IMR tariffs in Africa for evidence of lack of competition on their own retail markets, and conceivably for evidence of collusion.

ONE NETWORK
Celtel was a leading African mobile operator with a substantial geographical presence, often in adjoining countries, built up at a time when the continent was considered much less inviting for MNOs. Expanding into adjacent markets. This was to achieve 3x3x3: "It is the strategy that will make Zain a global player in three stages: regional, international and global, with each stage completed in three years, with an aim of reaching a customer base of 150 million. In essence, with this expansion plan, we aim to achieve in nine years what other companies have taken more than 27 years to achieve" .
However, the expansion did not last long and having overextended its reach, Zain sold its African operations, other than in Morocco and Sudan, to Bharti of India for USD10.7 billion in March 2010 (Mookerji, 2010). These networks were rebranded "Airtel" later the same year (Telecoms, 2010).
MSI, subsequently part of Celtel and later Zain, had operations on both sides of the River Congo, in The Democratic Republic of Congo (DRC) and in The Republic of Congo (M2 Presswire, 2002). Although Kinshasa and Brazzaville, the two capitals, are separated by only seven kilometres, telephone traffic between MSI's two MNOs was routed to the respective fixed incumbent operators, which only interconnected in Europe. The costs for this were considerable and had the effect of suppressing demand for calls between the two neighbours. MSI was eventually able to obtain the necessary licences and installed a microwave link across the River Congo in 2002, allowing it to cut its charges by 80% and thus greatly increase traffic volumes. This type of problem was replicated on different scales at many borders in Africa.
In East Africa changes made by the governments gradually liberalised international telecommunications, allowing Celtel and later Zain to own and to interconnect its gateways. In 2004 Kenya joined Tanzania and Uganda in this liberalisation, allowing Celtel to launch reduced rates for calls between the three countries (Zain, 2005). Then in 2006, Zain announced a One Network offer eliminating IMR surcharges for both postpaid and pre-paid customers in East Africa (see Table 6). With all the traffic retained on its own networks, roaming had been fully internalised and with no roaming onto the networks of rivals, there could be no out-payments. A more conventional approach to pre-paid roaming, with high charges and requiring an expensive technical platform, was considered highly unlikely to stop customers switching to rivals.
The One Network offer was gradually extended westward to the Atlantic Ocean, covering an area greater than the EU and becoming available to nearly half the population of the continent. The exception was Zambia, whose government refused Zain its own international gateway, officially for reasons of national security. MNOs there were required to pay the fixed incumbent operator for their international traffic, the government forcing them to support Zamtel in preparation for its privatisation. Instead of One Network, Zain offered its Zambian customers only very limited pre-paid roaming (eg to the UK). Finally, the government relented, allowing competition in international gateways and thus One Network could be offered there (Malakata, 2010). The Zain One Network tariff introduced a special case, which rivals found difficult to copy. If, say, someone living in Uganda has a relative or friend working in Kenya, then if they both use SIM-cards from their countries of residence they must pay for international calls. However, if both use Zain Kenya SIM-cards then the person in Kenya can call the person in Uganda for the price of a domestic call, making a considerable saving (see Table 7). All that is required is a little "juggling" of SIM-cards, to ensure that calls are made on SIM-cards of the same nationality -presumably they would wish mostly to be on a local SIM-card for local family, friends and colleagues.  The One Network scheme eliminated all IMR charges for both post-paid and pre-paid customersthey simply paid the applicable national rates for outbound calls and received inbound calls free of charge, as if they were at home. Pre-paid customers were additionally able to purchase top-up cards locally to maintain their credit balance. Since customers were almost exclusively pre-paid they would never have paid traditional IMR charges, but would instead change their SIM-cards at the border. It therefore made commercial sense to abandon established IMR charging models in order to avoid customers switching to a rival operator. It also allowed customers access to all of their stored credit and ensured cross-border communications, keeping friends and families connected.
George Held, Zain's Marketing Director for One Network noted, "When we launched it in DRC, there was a surge in customers in Uganda, Kenya and Tanzania because of these cross-border activities, especially the lake area" (The New Vision, 2009).
Zain also offered conventional data roaming using GPRS, with both post-paid and pre-paid tariffs, to a range of destinations (Daily Trust, 2008). This has been presented as a premium service, intended to attract high-spending customers. An even more exotic form of IMR was available through Aeromobile, which provided a roaming service for Zain customers on flights operated by the Emirates airline (Leadership, 2008).
While it has been argued that One Network drove up traffic volumes for Zain there are only vague hints from the operator to support this view, which are undermined by the UN World Tourism Organisation reporting only a few tens of thousands of individuals crossing these borders each year (Gillwald & Mureithi, 2010). Rather, it appears that Zain obtained considerable publicity from the One Network plan, which it used to support the expensive process of rebranding, helping it to attract more domestic customers. It incurred some regulatory costs in negotiating the necessary permissions. One Network also had a significant effect on rivals, which felt it necessary to respond, even if not on the same scale. It needs to be recalled that Zain maintained its conventional roaming business for post-paid visitors, especially those from developed countries. The attractions of the One Network tariff to Zain were a complex mixture; discomfiting rivals, encouraging customers not to switch to rivals and driving up domestic market numbers.

RESPONSES TO ONE NETWORK
The other large MNO groups in Africa, notably MTN, Orange and Vodafone, have all felt themselves to be under sufficient pressure from Zain to respond to One Network. Clearly, Zain anticipated that this would require difficult and protracted negotiations between firms that normally saw each other as competitors.
MTN, a rival pan-African operator, launched a special "low" roaming tariff for its customers based in South Africa, who were charged ZAR5 per minute for both making and receiving calls across the rest of Africa (MyBroadband, 2008). Sending an SMS cost ZAR1.50, while receiving one was free 2 . At the end of 2010 the prices on its partner networks were ZAR5 for a local call, ZAR7 to call back to South Africa, ZAR4 to receive a call and ZAR2 to send an SMS (MTN, 2010).
In 2007, MTN Rwanda launched a seamless roaming service with partners in East Africa (see Table 9) (Highway News Agency, 2007). This allowed customers free roaming between the networks, receiving calls without charge, making calls at home rates and being able to use airtime vouchers purchased from local operators. However, the scheme was modified in 2009, so that customers paid the local rates rather than the home rates (MTN, 2009 (The New Vision, 2008. In addition to the South African offer described above, MTN One World offers local country rate roaming tariffs in West Africa, but with very different prices depending on both the operator and the country (see Table 10) (MTN, 2010).  (Okojie, 2008). In May 2009, Glo introduced Two Nations, One Call Rate, with no roaming charges between the two countries, aimed primarily at local travellers with a view to locking them in (Ukodie, 2009 Orange created a zone of West African countries in 2007, comprising Guinea, Guinea Bissau, Ivory Coast, Mali and Senegal, with reduced prices for roaming (see Table 11). The operators offer limited pre-paid roaming, but extensive post-paid roaming (Orange, 2010). In comparison with Zain, the responses appear piecemeal but pragmatic, focusing on what can be delivered and what is significant. Ovum has noted that the volumes of traffic being generated and the revenues won and lost by such deals did not seem very significant (Obiodu, nd). Nonetheless, when faced with a non-roaming offer a significant group of customers who are nomadic or migrant or whose family and friends are nomadic or migrant see the benefits and move to or remain with operators offering beneficial tariffs.

REGULATORY INITIATIVES
Following the investigation launched by the European Commission in 1999, the persistently high charges for IMR came to be perceived as a "regulatory" issue, one that was more likely to be solved by an authority than by the market. A number of regional economic groupings joined the EC in seeking out a solution. Some terminological confusion may have been caused by the EU Roaming Regulation, which is a transnational statute, adopted under the European Community Treaty, its connection with regulators being instructions to them to collect data.
In 2005, the African Telecommunication Union (ATU) and the African Development Bank (AfDB) began a project for a single African SIM-card (East African Standard, 2005). Subsequently, the ATU indicated it would "Develop a regulatory framework for the implementation of cross-border networks and pan African services such as regional roaming" in the period 2008-09 (Africa, 2006). While still considered a work item, no progress has been made for some years.
The Economic Community for the West African States (ECOWAS) has taken various steps to harmonise policies and regulations in order to facilitate regional integration of ICT markets. It adopted a road map for regulatory harmonisation and regional mobile roaming (ITU, 2006). As of 31 July 2005 there were 268 roaming agreements made by 23 of the 42 West African Operators (Sanou, 2005). At that time, in three countries not a single operator had a roaming partner in the ECOWAS area. For post-paid customers, heavy security deposits were required, between USD340 and USD1 500, while even pre-paid customers faced one-time charges of between USD19 and USD47. There was one innovative marketing offer, with Telecel, present in six countries, called @SIM, in which the customer was given one SIM-card for the home network and others for the networks to be visited.
WATRA (the association of regulators) organised a feasibility study jointly with ECOWAS, on roaming and interconnection in the region (Ndukwe, 2003). This concluded that pre-paid roaming was a "honeypot" for operators, if they could provide an IMR service (Aihe, 2007). Conferences on roaming were held in 2007 and 2008. However, this work subsequently ground to a halt, with its focus shifted to cutting costs for international calls within West Africa. Its primary task was to investigate possible mechanisms to reduce the high charges for IMR within the region, with a view to a final decision to be taken by SADC Ministers. CRASA hired consultants to undertake an impact assessment of its roaming initiative, whose report was discussed at its meeting in early 2010 and published later that year (Analysys Mason, 2010). It noted the usual strange variations, and sometimes a lack of transparency, in prices, largely attributed to high wholesale prices, though complicated by problems with international gateways and exchange rate fluctuations 3 . Additionally, there were technical problems, including poor quality of service.
The regulatory approaches to IMR have been less than productive. Possible penalties or interventions have been seen by MNOs as highly unlikely to be imposed and thus failed to convince them to act. Moreover, the capacity for any one country to regulate is limited and many already have a transnational commercial offer from Airtel/Zain or from a rival. Indeed, it has been commercial actions and reactions that have driven down IMR prices and consequently it has become progressively more difficult to justify an intervention, requiring complex work on regulatory impact assessments. The real challenge would be to develop a regulatory intervention that would build on the One Network approach, for example by widening its scope, adding more MNOs or by encouraging MNOs to create a roaming exchange or spot market.

OPEN CONNECTIVITY
Established in 2005 by the GSM Association, the Open Connectivity (OC) programme was intended to facilitate easier and faster outbound roaming agreements for MNOs, helping new and smaller operators increase the scope of the IMR service they offered their clients. With more than 700 operators, the traditional bilateral approach was claimed to have reached its limits. Within the OC framework a number of hubs would provide access to multiple IMR partners via a single commercial agreement; ultimately these are expected to interconnect through peering arrangements (see Table 12).  (Rwandatel, 2009). Rwandatel customers were to be issued with SIM-cards with both Rwandatel and Proximus International Mobile Subscriber Identities (IMSIs). Where Rwandatel has no bilateral roaming agreement, then the SIM-card would automatically present the roaming customer as being from Proximus, becoming a virtual Belgian, to use its wider set of IMR agreements. The deal is not bilateral, so that Rwandatel does not immediately benefit from incoming roaming customers or lower prices -these must still be negotiated bilaterally.
While the hubbing arrangement appears to open the way to easier access to outbound roaming, there is no evidence that it reduces prices. Indeed, even where a hub has access to regulated wholesale roaming prices in the EU, there appears to be neither a legal obligation nor any obvious commercial incentive to pass on the lower price to non-EU operators. Competition between the hubs appears to focus on increased coverage for a few high-spending outbound roamers, rather than on reducing prices.

CONCLUSION
Visitors to Africa who elect to use the roaming service from their home mobile network operators can maintain their usual telephone number and remain connected. They have to pay heavily to do so and any local African wishing to call them has to pay the international rate to call their home network. If visitors give up their home number and roam instead with a SIMcard acquired locally, they can save considerably on the charges, but then have to advise colleagues, family and friends of a new and temporary number, plus they should periodically check their home voicemail. Likewise, Africans with post-paid subscriptions who leave the continent will pay very high roaming charges in order to remain in seamless contact, perhaps being asked to pay a substantial deposit for the privilege. Alternatively, they too can switch to local prepaid SIM-cards with the associated lower charges, but at increased inconvenience.
Prepaid roamers have a more limited choice -largely because of the costs they are not offered a service in more exotic locations. Each operator has typically set up one or two dozen bilateral deals with operators in major travel destinations, usually with neighbours, significant trading partners and, especially, the former colonial powers. The importance of these arrangements is difficult to assess, since there are very few data on the levels of use of pre-paid roaming.
The Open Connectivity initiative by the operators simplifies some wholesale arrangements for roaming. This means that customers should have access to a wider range of networks and to advanced roaming services, though it has done nothing to reduce prices.
While the European Commission was able, if ill-advised, to block the introduction of transnational tariffs without a roaming surcharge, Zain did this in Africa without hindrance (EC, 2002).
Where rivals saw the need to respond in order to retain customers, they have done so, though generally on a smaller scale and on specific and commercially important routes.
The vast majority of individual Africans cannot afford expensive roaming rates, at several dollars per minute, whether post-paid or pre-paid, so that it is reasonable for mobile operators to abandon traditional IMR charges. A prerequisite is that each operator be allowed its own international gateway, with which it can make deals within a corporate group or with commercial partners. Moreover, it does not preclude charging high wholesale roaming rates to operators in developed countries, which easily pass these on to their customers, admittedly with a large and seemingly increasing mark-up by the foreign partner.
The failure to form a wholesale roaming market remains something of a mystery. With international voice telephony and Internet traffic there are intermediaries and aggregators to facilitate smaller and niche market players. There appear to be significant structural obstacles and, possibly, anti-competitive practices that impede the formation of such a market for roaming services.
There is nothing to stop operators such as Orange and Vodafone giving African customers a secondary IMSI, for example from one of their European networks, to allow them access to the low regulated roaming rates. While this might increase their competitiveness on the retail market it seems to be considered unlikely to generate a sufficient number of new customers to overcome lost revenues.
While regional economic groupings and associations of regulators have taken an interest in high IMR charges, they have yet to achieve any significant results. They have not even coordinated the introduction of national measures that are known to work: • Requiring the sending of an SMS with IMR prices on arrival abroad • Capping retail prices for call forwarding • Capping spending to avoid "bill shock".
There are concerns that interventions might further distort the poorly understood market dynamics. Further detailed study of the economics of IMR markets is required to ensure a level of understanding that is sufficient to evaluate policy options. Minimally, this requires the collection of considerable data from the operators.
Work at the African Telecommunication Union (ATU) and in West Africa ground to a halt because of the complexity of the problem, the poor alignment of interests and the opposition of the operators. The recent study for CRASA illustrated the complexity of the problem, including the lack of data.
The scope for further research is considerable. On one level the retail prices charged by non-African operators require further study to understand why they appear to be rising. On a very practical level, survey work with Africans crossing borders would help to explain attitudes towards costs and behaviour, in particular ownership of SIM-cards from foreign operators. Surveys of tourists and business travellers from developed countries would also provide insights into their communication needs and their willingness to pay very high prices. It would be helpful to map flows of visitors within Africa onto the various special tariffs and to identify any remaining obstacles to offers of roaming without surcharges. With the growing adoption of mobile broadband, analyses of the prices for, and the use of, data roaming are becoming increasingly important. The activities of the economic groupings and of their regulatory groups require further analysis, in order to better understand the roles they can usefully play in the governance of transnational telecommunications markets, their requirements for capacity building and any improvements that can be made to institutions and procedures.