Prepared by: Charles J. Kenny
The World Bank
ckenny@worldbank.org

Introduction

The Internet is, as of yet, a largely insignificant medium in Africa. Radio covers about 75 percent of the region's population, television 40 percent. Access to the Internet is limited to about 0.1 percent (Pruett, 1998) --the great majority of these in cities. Having said that, the Internet is expanding rapidly. Outside of South Africa, the number of host sites on the continent has increased from 290 in five countries in 1995 to 6,510 in 32 countries last year (Network Wizards, 1998) and 21 African countries were estimated to have more than 1,000 users last year (Africa Internet Connectivity website, 1998).

What makes this expansion potentially significant is that the Internet could have a major role to play in reducing poverty --especially in rural areas. It is clear that a major cause of low rural incomes can be blamed on 'information poverty' --the lack of access to information and knowledge that could improve earnings potential. And we have long known that one reason for the poor performance of businesses in rural areas is connected with low access to information technology. For example, a study of rural business in Botswana and Zimbabwe found that a critical determinant of success or failure was access to a telephone (Kayani and Dymond, 1997). Again, in Sri Lanka, the introduction of telephone services in rural areas saw the percentage of the price for which crops were sold in the capital city actually passed on to the farmer increase from 50-60 percent up to 80-90 percent (World Bank, 1998). Recent estimates suggest that investments in rural telephony can have annual economic returns of over 40 percent (Ernberg, 1998).

The Internet, far more flexible than the telephone and, with email, a far cheaper form of global communication, offers even greater opportunities. These are already being seized in a few places in Africa. For example, in Kenya, a rural farming cooperative has established a relationship through electronic mail with EarthMarketplace, a US organization, to sell local produce direct to the American market --bypassing the distributor and increasing the revenues of local farmers. As has been reported elsewhere, the Internet is already being used for a range of other activities in African rural areas --including health, education, governance and journalism (Africa Internet Forum, 1998).

But if rural Africa is to benefit more widely from Internet access, the barriers to Internet access will have to fall far and fast. This will be no easy task in a region where rural GDP per capita is often under $100, rural energy and telephone access rates are very low and the great majority are not educated beyond the primary level. Nonetheless, we do have some idea about the bottlenecks to be overcome in improving rural access.

The computer training center in Wa, in the Upper West Province of Ghana, suggests the problems faced by Internet users far from the capital African counties. First is the cost of telephone connection to the Accra-based Internet service provider. The connection cost is 33 cents and the hourly charge $7.50. Poor line quality slows connection speeds and means that it frequently takes five attempts before successful connection. Frequent line drops disrupt downloads and necessitate restarting the whole process. On top of these problems and charges come those of the ISP (although Ghana is blessed with one of the better ISPs in Africa). The center also faces significant infrastructure problems. Electricity was rationed in 1998 due to low rains, and the seven-hour bus ride to Accra (the nearest source of repair and maintenance) is over appalling roads. Finally, the shortage of locals trained in computer skills means that they can fetch $6,000 per year --a price that the center in its present form, could never meet (at the moment it is run by volunteers) (Hirsch, 1998).

Clearly, there are a range of problems faced by potential rural Internet users (too many to be tackled in one paper). The remainder of this paper will focus on just two of them --costs of Internet provision and access. They are related, in that, unless costs fall, access will perforce remain limited. But, in a continent as poor as Africa, access issues go far beyond lower Internet connection charges to involve questions of efficient methods of increasing community access.

Costs

Where do the costs of Internet access come from? Table 1 lays out the costs (with one-off costs annualized) faced by a user in Mozambique with a computer dedicated to Internet access who uses the Internet for one hour each business day. For a user with local access to an ISP, over 50 percent of the costs come from telephone and ISP charges. This is an underestimate of true percentage costs --the computer will be used for many other things apart from Internet access and is likely to be used by more than one person. For local access, then, telephone and ISP charges are probably the largest cost consideration. This is even more the case for a user who has to make a regional call to access the Internet. Total costs of yearly access increase from $1,385 to $4,297, with the telephone bill now equal to over 72 percent of that sum.
And where do the costs of the ISP come from? Estimates for Ghana and Mozambique suggest that anywhere between 48 and 90 percent of ISP costs are accounted for by the costs of connection to the US internet backbone via an international circuit (Africa Internet Forum, 1998 and Quaynor et. al. 1998). A large proportion of ISP costs are accounted for by telecommunications costs as well, then.

A central concern is clearly to lower these telecommunications costs. For the regional connection to an ISP, there are a number of answers. Senegal's main telecommunications company Sonatel has decided to charge a fixed local rate to Internet users across the country. This generous act might be a good short term solution, but in the long term, wireless and other technologies are likely to offer a better answer. For example, Arua, a small town in Northwest Uganda, is already connected to the Internet through a wireless radio system, while two schools in rural areas outside Pretoria, South Africa use a wireless phone connection (Pruett, 1998). Direct satellite connection is another option that might become attractive in the next five years (CSIR, 1998).

In order to make these options available, there is a clear need for a flexible regulatory regime. And indeed, liberalizing the telecommunications sector is likely to be central to attempts to get the costs of access down. Widespread evidence already suggests that introducing well-regulated competition widens access and lowers the price of telecommunications (World Bank, 1998, Wellenius, 1997a). In Ghana, for example, the introduction of a second mobile operator increased the rate of rollout in the capital, Accra. The second operator had plans to expand services to Kumasi, the second main city, only when the investment could be financed from retained earnings. But when a new competitor announced plans to operate services in both Accra and Kumasi, the second operator rushed to provide services --and halved its connection charges (Smith, 1995).

India's proposed new policies on the Internet suggest methods for liberalization to cut the direct costs of ISPs, as well. The Indian government is to allow ISPs to set up their own international gateways to the US backbone (as Ghana has begun to do). This has already lowered the costs charged for international bandwidth by the country's main telecommunications operator. India also plans to allow power companies and the railway system to offer Internet backbone capacity (Gairola, 1998). Although India has some way to go to fully liberalize its telecommunications sector, this latest set of reforms suggests the extent to which policies can change toward a flexible and competitive provision of services to Internet users and ISPs.

Given the benefits of competition to telecommunications services --and the centrality of telecommunications costs to Internet provision-- it is unsurprising that there is a link between the price and extent of Internet access and sector liberalization. This result has been found for the OECD elsewhere (OECD, 1996). In Table 2, similar results are reported for an African sample. Average results are reported for two sets of African countries split by a liberalization index. A country is considered liberal if it has four or more of the following in 1997: liberal equipment trade, separate posts and telecommunications companies, independent regulator, new sector law, private cellular, telecommunications company privatized and competition in basic services (source: World Bank, 1997). Despite this rather low hurdle, it is clear from the table below that liberalized telecommunications regimes in Africa foster more Internet host sites, lower costs of Internet access (when downloading either 10 and 50 pages a day), more ISPs, more international bandwidth and more users. It also appears that African countries with a more active competition in Internet service provision also see lower prices and higher access (Africa Internet Forum, 1998). Clearly, liberalization and active competition is already important --and will only become more so-- in introducing low cost rural access options, then.

Access

We have methods to reduce the cost of Internet provision, then --especially in rural areas. But that will be far from sufficient to widen access. Table 3 lays out the annualized Internet access costs for a number of African countries (calculated in the same way as the Mozambique local access calculation above). It also estimates the approximate average amount that a citizen would be willing to spend on communications --approximately 2 percent of GDP per capita. Even South Africa, with very cheap access rates and comparatively wealthy citizens, sees annualized Internet access costs at more than tenfold willingness to pay. Clearly, if the Internet is to be anything more than the preserve of the very rich in Africa, efficient methods of providing public access will be vital.

We do have models of public access regimes for both telecommunications and the Internet, which suggest lessons for future efforts. Three countries that offer elements of a model are Senegal, South Africa and Chile.

A 1997 survey of telecenters in South Africa found that 67 percent of these centers had a telephone, 31 percent a computer and 8 percent Internet access (World Bank, 1998). Since then, the country's Universal Service Agency has set a target of establishing 1,500 telecenters, largely in underserved areas, by 2002/3. The agency provides multipurpose community information centers with equipment (computers, fax machines, a photocopier, modem, printer and scanner, all of which remain the property of the USA), two years of startup costs and field workers to provide technical support (CSIR, 1998). Clearly, public support will be vital in expanding rollout, and The Universal Service Agency is keen to work with local entrepreneurs and NGOs to ensure sustainable and efficient access provision. However, South Africa is considerably richer than the rest of the continent, as we have seen. Elsewhere it will be even more important to reduce costs of provision to a minimum.

In the case of telephones, Senegal has shown that allowing local entrepreneurs to offer public access is an important first step in reducing costs of public access provision. In 1995, the country had more than 2,000 privately owned 'telecentres,' each with a telephone and many a fax machine --four times the number just two years before (World Bank, 1998). By 1998, that number had increased to 6,000 (Ernberg, 1998). Sonatel franchises phone service to the telecentre owner, who is allowed to charge a tariff up to 140 percent above the Sonatel price per call unit. On average, telecentres paid $3,960 to Sonatel and kept $1,584 each year (CSIR, 1998). These public access points, beyond rarely offering Internet access, are also largely based in urban areas --but they still provide a hint as to the direction that efficient access techniques should take. A similar scheme is also operating successfully in the state of Punjab in India (Ernberg, 1998). Grameen Telecom of Bangladesh operates a variant of the telecentre concept in rural areas, in which a Grameen Bank member borrows money from the microfinance institution to purchase a cellphone and sells air time to villagers to repay the loan and earn income. As of October last year, 140 villages had access to telecommunications through the Grameen mobile phone network (Uddim, 1999).

South Africa has shown that for-profit public Internet access can work outside the Internet cafes present in many African capitals. Zokode Distributors, a multipurpose information center owned by a local entrepreneur in Daveyton township, Gauteng province, serves between 16,500 and 18,000 people per month. The majority use the telephone alone --but increasing numbers are turning to the two Internet-connected computers (CSIR, 1998). Preliminary studies in Uganda and India also suggest that local entrepreneurs in some rural areas might well be able to provide Internet access while remaining financially viable.

Nonetheless, it is likely that the costs of setting up rural access to the Internet will remain prohibitive in large parts of Africa in the absence of at least a startup subsidy. The important element is to use the subsidy to harness local entrepreneurial talent and resources. A high level of direct government involvement has been found detrimental in the telecommunications industry more broadly, and recent studies in developed countries suggest that the record is mixed with providing Internet access as well (Ernberg, 1998). Chile has provided a model for providing government support while still exploiting entrepreneurial talent. The country has introduced a system of auctioning subsidies to pay for rural telecommunications rollout. In 1994, the country set up a limited-life fund to support the provision of the first payphone to remote and rural areas. Companies were asked to bid for the lowest subsidy that they would accept to provide service. Within two years, the fund had achieved 90 percent of its rollout objectives using only about half of its $4.3 million budget --largely because it received bids to provide service with no subsidy to about half of the chosen locations. Just over $2 million in public funds had leveraged $40 million in private investment to install telephones in 1,000 localities at about ten percent of the costs of direct public provision (Wellenius, 1997b).

Government support, local entrepreneurial talent and efficient targeting can be combined, then. Using the Chilean experience as a model, countries could auction subsidies toward the initial provision of Internet services in a rural area. Clearly, conditions would have to be placed on levels of access and service, perhaps pricing. But by and large, conditions should be as limited as possible, to allow local entrepreneurs to take advantage of the most efficient technology available in a liberalized telecommunications market. Rather than forcing an inappropriate method or level of service, a flexible auction subsidy would allow for the most efficient level of service at the lowest possible price --extending the benefits of a liberalized sector even to the most disadvantaged rural poor.

Liberalization -combined with a pro-competition regulatory system-can be a key to lowering Internet costs and increasing access, then. Combined with an efficient subsidy system, this could greatly increase rural access to an important new tool for development.

Bibliography

Africa Internet Connectivity Website (1998) user statistics online at http://demiurge.wn.apc.org:80/Africa
Africa Internet Forum (1998) Economic Toolkit for African Policy Makers, available on line at http://www.worldbank.org/infodev/projects/finafcon.htm
CSIR (1998) Knowledge in Development: Multi-Media, Multi-Purpose Community Information Centers as Catalysts for Building Innovative Knowledge-Based Societies, Background paper for the 1998 World Development Report, the World Bank, Washington DC.
Ernberg, J. (1998) Universal Access for Rural Development: From Action to Strategies paper presented at the ITU Seminar on Multipurpose Community Telecentres, Budapest, 7-9 December, 1998.
Gairola, M. (1998) India Spreads the Net, Communications International 25, 9 (September) 25-26.
Kayani R. and A. Dymond (1997) Options for Rural Telecommunications Development, World Bank Technical Paper 359.
Network Wizards (1998) host site statistics available online at http://www.nw.com/zone/WWW/report.html
OECD (1996) Information Infrastructure Convergence and Pricing: The Internet, available online at http://www.oecd.org/dsti/gd_docs/96_xxe.html.
Pruett, Duncan (1998) The Internet and Poverty, Panos Briefing no. 28, available online at http://www.oneworld.org/panos/briefing.
Quaynor, N., Tevie, W. and Buley, A. (1998) The Expansion of the Internet Backbone in Ghana, mimeo, Network Computer Systems, Ghana.
Smith (1995) End of the Line for the Local Loop Monopoly? World Bank Viewpoint Note No. 63.
Uddin, S. (1999), The 'Village Phone': Bringing Universal Service to Rural Areas in Bangladesh, paper presented at the NTCA First International Conference on Rural telecommunications, Washington, DC 30 November to 1 December. Available online at http://www.ntca.org/intlconf/report_main.html.
Wellenius, Bjorn (1997a) Telecommunications reform: how to Succeed, World Bank Viewpoint Note No. 130.
Wellenius, Bjorn (1997b), Extending Telecommunications Service to Rural Areas --the Chilean Experience, World Bank Viewpoint Note No. 105.
World Bank (1997) Measures of Telecommunications Liberalization, mimeo, Industry, Energy, Telecommunications and Informatics Division.
World Bank (1998) World Development Report 1998/9 New York: Oxford.