Content uploaded by Charles Kenny
Author content
All content in this area was uploaded by Charles Kenny on Dec 06, 2018
Content may be subject to copyright.
Development Policy Review, 2002, 20 (2): 141-157
Overseas Development Institute, 2002.
Published by Blackwell Publishers, Oxford OX4 1JF, UK and 350 Main Street, Malden, MA 02148, USA.
Information and Communication Technologies
for Direct Poverty Alleviation: Costs and
Benefits
Charles Kenny∗
∗∗
∗
Information and communications technologies (ICTs) are powerful tools for
empowerment and income generation in developing countries. The cost-
effectiveness of different ICTs does vary between developed and less
developed countries, however. This article reviews the potential efficacy of
radio, telephony and the Internet as tools of direct poverty alleviation in the
latter. While the requirements for their successful utilisation make radio
and telephone far more suitable technologies for the poor, traditional ICTs
can act as a sustainable intermediary for them to gain indirect access to the
power of the Internet. Governments should concentrate on opening up
private and community provision of broadcasting and widening access to
telephone services, so that they can effectively play this intermediary role.
Radio and the telephone have a long history demonstrating their utility in developing
countries. The Internet has also already proved itself useful in these countries, in
increasing both the incomes and quality of services received by citizens. Using Internet-
based systems to make phone calls has reduced the cost of international communication;
the Internet is being used to ease the export and import of goods; and countries such as
India are earning billions of dollars a year exporting IT services and software (see Grace
et al., 2001).
Because of the many uses of information and communications technologies (ICTs)
in a developing country context, many donors, governments and NGOs have long
supported roll-out programmes for these technologies – providing radios to schools,
community groups and refugees, and supporting the expansion of rural telephone
services. Some of these same groups are now experimenting with community-based
multi-purpose community telecentres (MCTs) that provide public Internet access, email
and other computer applications.1 Building on these first steps, there is a movement in
the development community pushing for the widespread roll-out of community access
points to the Internet as a tool for direct poverty relief.2
∗Infrastructure Economist, World Bank. The views presented are those of the author, and do not necessarily
represent the view of the World Bank, its executive directors, or the countries that they represent. He is
grateful to Jeni Klugman for helpful comments on earlier drafts. The usual disclaimers apply.
1. A few multi- and bi-lateral donors with active programmes related to MCT development include:
USAID’s Leland Initiative (http://www.info.usaid.gov/regions/afr/leland); IDRC’s Acacia Initiative
(http://www.idrc.ca/acacia/index.html); the International Telecommunications Union (http://www.itu.int/
ITU-D/index.html); UNESCO (http://www.unesco.org/webworld/iip/#funding); and the World Bank
(http://www.worldbank.org/html/fpd/telecoms/subtelecom/selected_projects.htm).
2. See, for example, the ITU’s Buenos Aires Action Plan (http://www.itu.int/ITU-D-UniversalAccess/
BAAP09.htm).
142 Charles Kenny
This article argues that, whilst there is a continued (perhaps growing) role for
donors to improve access to a range of ICTs in developing countries, that role probably
should not extend to the widespread provision of Internet access – at least in the poorer
regions of the least developed countries. The nature of extreme poverty in developing
countries – very low incomes, subsistence and unskilled wage labour as the dominant
income source, food as the dominant consumption good, low education and high
illiteracy, minority language group status and rural location – points to an unsustainably
high cost and relatively low benefit of direct Internet service provision through
telecentres to the very poor. This might suggest that the push for universal Internet
access as a tool for poverty relief is misplaced. Instead, the article argues that access
programmes focused on the telephone and radio might have a higher benefit-cost ratio
and lower overall cost as alternatives to and intermediaries for the Internet in poverty
alleviation programmes.
Characteristics of the poor
Taking as a cross-country definition of ‘the poor’ those living on less than $1a day,3 we
have information on the country location of a little over 1 billion such people.
Approximately 40% live in India, a further 22% in China, 8% in Nigeria and a further
14% in Pakistan, Bangladesh, Ethiopia, Indonesia, Brazil, Mexico and Russia
combined. The remaining 14% are spread throughout the rest of the world, primarily in
Africa.
Table 1 looks at the average characteristics of a country weighted by the percentage
of the global poor that reside there. In other words, it is a measure of the conditions
prevalent in the average country occupied by poor people. This is compared with the
global population-weighted average – the conditions prevalent in a country occupied by
the average person, poor or not.
The table suggests certain features that poor people share, beyond their extreme
poverty:
•Poor people are concentrated in countries and regions with low average income
and higher than average poverty rates. The average poor person lives in a
country with a little over one-third of the income of the average global citizen,
and approximately double the population living under $1a day. There is also
evidence that the poor are concentrated in particular regions within those
countries.4
•About two-thirds of the population in countries where the poor reside are rural.
The percentage of the poor that are rural dwellers is even higher,5 and the rural
poor frequently live in population-sparse environments.6
3. The $1 a day standard is an arguable definition, based only on absolute income, taking no account of the
non-income dimensions of poverty, nor the depth of poverty below $1. Nonetheless, it is a widely accepted
proxy measure of absolute deprivation.
4. In India, incidence of poverty by its 61 agro-climatic regions is immense, from 8.4% in UP Himalayan to
77% in Orissa Southern (World Bank, 1997a). For evidence from Brazil see World Bank (1995a).
5. 93% of the very poor in Bangladesh are rural (World Bank, 1999), as are 80% of the poor in Ghana, 74%
in Zambia and 89% in Zimbabwe (World Bank, 2000 and author’s calculations from data in WDI).
6. Two-thirds of households in the poorest quintile in Peru are in the mountain region, compared with less
than a tenth in the densely populated coastal region (World Bank, 2000). In Ghana, 59% of the poor live in
rural forest and savannah (World Bank, 1995b).
Information and Communication Technologies for Direct Poverty Alleviation 143
Table 1: Characteristics of countries of the poor
Poverty-weighted
global average
Population-weighted
global average
Income
GNP per capita (PPP) 2293 6200
Poor people % population 36.4 17.6
PCs/(1,000) per capita 7.4 58.4
Fixed telephone lines/(1,000) per
capita
36.5 144
Mobile phones/(1,000) per capita 7.9 40
Radios/(1,000) per capita 196 380
Education and language
Female adult illiteracy (%) 46 33
Male adult illiteracy (%) 27 18
Population not speaking the most
widely used language (%)a
48 26
Population not speaking the official
language (%)
53 36
Rural
Rural population % total 67 54
Rural population density (per km2) 483 515
Fixed lines/(1,000) in largest city 171 231
Source: World Bank (2000) except for language.
Note: a) Source for language variables is Easterly and Levine (1997). The language variable
poverty figure is calculated from data on India, China, Nigeria, Bangladesh, Ethiopia,
Indonesia, Brazil and Mexico alone, the global figure is an unweighted average.
•A third feature of the global poor (not immediately apparent from the table) is
that most are self-employed (and) subsistence farmers or unskilled agricultural
labourers.7 Linked to this, in percentage terms the poor buy fewer services and
more physical goods (especially food) than the average consumer. Much of this
‘commerce’ is carried out on an in-kind basis.8
•Related to their unskilled labourer status, the average poor person is illiterate.
Average literacy rates in the countries occupied by the poor are 54 and 73% for
women and men respectively. Within countries, the poorest are much more
likely to be illiterate (and vice versa).9 Unsurprisingly, the same is also true of
7. In Tanzania, producing for subsistence is one of the best correlates with low income, and 83.5% of the
poorest are crop producers – much of that for subsistence (World Bank, 1996). For evidence from India,
Peru and Ghana, see World Bank (1997a), Harrell et al. (1989) and World Bank (1995b) respectively.
8. In Cote d’Ivoire and Peru in 1985, about 70% of the expenditure of poor households went towards food
(World Bank, 1990), and 56% of food expenditures for the poorest in Tanzania (rural and urban) are in
kind rather than cash (World Bank, 1996).
9. Compared with a national average of 35% headcount poverty in India as a whole in 1994, poverty rates
were 45% for households where all were illiterate (World Bank, 1997a).
144 Charles Kenny
education rates, both past education amongst adults and current education of
children.10 Even those in school see very low per-student expenditure.11
•Poverty is also frequently correlated with minority language status. The
average poor person lives in a country where half the population (and doubtless
a greater percentage of the poor) do not speak the official or most popular
language. In the great majority of cases, the languages the poor speak are
minority, not global, languages.
Finally, it should be noted that the majority of the poor share the majority of these
features.12 The following sections will discuss what these correlated characteristics
imply for the costs, benefits and sustainability of various ICT programmes.
The cost of providing ICTs to the poor
Poor people have very low potential expenditure on ICTs. The poorest quintile in Chile
(living on considerably more than $1 a day) are willing to spend approximately 2-3% of
their income on communications (de Melo, 1999). This appears to be the high end of
expenditure worldwide. Assuming the same holds true for people on $1 a day, it
suggests maximum yearly communications expenditures of approximately $10. What
does this mean for the affordability and sustainability of various ICTs?
The radio is by far the cheapest electronic communications technology. Receivers
cost perhaps $10 plus the cost of batteries (or a wind-up model, which does not need
batteries, can be purchased for $70 to $100). They do not require an electrical
connection and (unlike fixed line telephony or the Internet, for example), they are stand-
alone appliances. This helps to explain why 40% of rural households in the low-income
countries of sub-Saharan Africa and Asia already own a radio. On the transmission side,
programming and broadcasting are also relatively inexpensive; a low-power
transmission system can cost as little as $1,000 (www.nlgcdc.org/articles/cdcnews-
695.html). Digital sound recordings can be made on equipment that costs $800 or less.
10. There are at least 12 countries in the world where the average number of years of schooling in the 15-19
age group is zero for the poorest 40% of households (World Bank, 2000). For evidence from Bangladesh,
Vietnam and Sri Lanka, see World Bank (1999), World Bank (1995c) and Datt and Gunewardena (1997).
11. In Ghana in 1992, educational spending per capita in the poorest population quintile was approximately
US$26.50, and in the richest quintile, approximately $41.50 (calculated from World Bank, 1995b).
12. In Panama, for example, indigenous populations, primarily rural, make up 41% of the extreme poor, 29%
of the poor and 3% of the non-poor. A significant proportion (approximately one-sixth) are monolingual in
their indigenous language, and virtually all of them are poor. Close to one-third are illiterate, with four
years less schooling than the non-indigenous, and credit access is approximately half (Vakis and Lindert,
2000). For evidence from Vietnam, see van de Walle and Gunewardena (1999). It should be noted that the
majority of poor in most countries are also women (although this is not always the case at the country level
– see Datt and Gunewardena, 1997, World Bank, 1996, World Bank, 1999). The gap in school enrolment
rates and health data would suggest that women constitute a majority of the poor (World Bank, 2000). This
is not a problem per se, except to note that, because women also face lower literacy and education levels
as well as a range of other features of poverty, without specific efforts to overcome these barriers the
Internet will be a poor tool to promote gender equality. Another important minority group over-
represented amongst the poor are the disabled. Disabled people are estimated to make up 15-20% of the
poor in developing countries – compared to perhaps 10% of the population at large (Elwan, 1999). To the
extent that the disabled poor have conditions that affect their motor skills or vision, their ability to access
the Internet will be impaired.
Information and Communication Technologies for Direct Poverty Alleviation 145
In central Mali, a station supported by Oxfam is broadcasting information to 92,500
people a year at a cost of just US 40 cents per person (www.oneworld.com). This
translates into affordable access for the poor.
Because community radio fulfils a role as a ‘community telephone’ it has a major
source of income (www.commons.somewhere.com/rre/2000/RRE.Radio.and.the.Intern.
html), and allows radio stations even in poorer areas to become sustainable. Even in a
country as poor as Liberia, local stations have achieved profitability through
advertising; people are willing to pay $1 to have an obituary read on the radio, for
example.
Providing telephone access is considerably more expensive, especially in rural
areas. The average cost of building out a telephone line used to be estimated at about
$1,000. Recent advances in mobile telephony and wireless local loop have cut these
average costs dramatically. James (2000) reports that wireless local loop technologies
might cut costs of roll-out by a factor of between two and six over conventional
technologies in (population-dense) India. Nonetheless, costs in rural areas where the
majority of the poor live remain higher, because providing networked services to low
population-density rural areas – and especially to mountain and forest regions – is
significantly more complex and expensive than providing those services in urban areas.
Indeed, the market feasibility of networked service provision is even more
dependent on geography and population density than it is on income. Because density
varies across regions within a country more than income per capita, it tends to have a
larger effect on income per unit area – perhaps the most important variable in
determining the feasibility of public network access.
Chile’s universal access scheme provides evidence to this effect. Chile supported
the provision of telephone access to unserved areas by providing subsidies through a
reverse auction to private companies; the lowest bidder received the bid subsidy in
return for building out the service. The amount of subsidy bid for was dependent on the
companies’ calculation of costs and revenues, looking at potential usage (a function of
demand density) and cost of service provision (a function of geography and, again,
demand density). Subsidies demanded were highest in rural and remote (rather than
particularly poor) neighbourhoods. The five districts of the country requiring the highest
subsidies per locality were those with the lowest population densities, ranging from one
to seven people per square kilometre.13 These were not the poorest regions; indeed,
incomes per capita were above the median national average.
Again, the importance of population density, feeding through to demand density, on
the sustainability and cost of service provision suggests that poor people, many of
whom live in population-sparse environments, will face particular challenges in
obtaining access. Table 2 makes it clear that, whilst in aggregate South Asia and China
are both population- (and so income-) dense, countries in sub-Saharan Africa, as well as
the population-sparse regions of Brazil, China and India, will frequently find it difficult
13. The average subsidy per population was approximately $9.7 per capita ($3,550 per locality), although this
rose to as high as $100 per person in the more inaccessible communities. Sources: localities, subsidies,
estimated from Intven (2000), income per capita World Bank (1997), population, area
http://www.ine.cl/chile_cifras/chile_cifras.htm (Chile en Cifras: Sintesis Geografica de Regiones). A
similar programme in rural Peru provided access at an average cost of $19 per person, or $5,674 per
locality (estimated from Intven, 2000).
146 Charles Kenny
to sustain telecommunications services in the absence of significant subsidies (Chile’s
GDP/km2, for comparison, averages about $155,529).14
Table 2: 1996 GDP density: Selected regions and income levels
Region Average GDP/km2
Sub-Saharan Africa 39,046
South Asia 498,086
Poverty-weighted (top ten) 486,503
Low-income 175,437
Low-income, excl. China and India 64,337
Middle-income 154,675
High-income 658,367
World 280,005
Source: World Bank (2000).
The difficulty of providing telephone access in the remote areas of poor countries
explains why teledensity in the rural areas of poor countries is so strikingly lower than
in urban areas. The data in Table 1 suggest that in the countries of the poor the total
telephone gap between the whole country and the largest city is 44 phones per thousand
compared with 202 phones per thousand – or approximately a fivefold difference. The
average global difference in access is 171 phones in the country as a whole compared
with just 231 in the largest city, or a 26% difference. We have seen that 40%
(approximately) of rural households in low-income Africa and Asia have a radio. Pigato
(2001) notes that only around 0.18% of those same households have a telephone,
compared with 5.22% of urban households.
In turn, this helps to explain the huge difference in telephone access by income
group within countries. Poor people, predominantly rural, face the linked access barriers
of low income and low density of demand. In South Africa, 75% of households in the
richest income quintile have a telephone, compared with 0.6% in the poorest quintile
(the same numbers are 11% and zero in Nepal and 73.8% and 1.7% in Panama) (World
Bank, 2000). We have seen that providing rural telephony is becoming more
straightforward with technological advance. Nonetheless, cost factors alone will still
make access prohibitive for the poorest in rural areas.
What are the additional average costs of Internet access, above the requirement for
access to telephony? A recent estimate for the cost of Internet access in Mozambique
(based on Africa Internet Forum, 1999) suggests that the annualised fixed costs of
access to one Internet-enabled computer (excluding telephone installation and rental but
including equipment costs plus yearly Internet fee for unlimited use) were $1,172.15
14. The Grameen telecom model has been impressive in rolling out services to peri-urban Bangladesh, but
even in one of the most population-dense and flat countries in the world, with a strong micro-credit
institution providing loans to rural villagers to buy phones and supporting the collection of phone bills,
service has yet to expand far beyond the major cities (see http://www.telecommons.com/villagephone/
section1.html for a map of present and planned roll-out).
15. Based on a 20% cost of capital, four-year depreciation for a computer costing $1,300 and a modem costing
$175, $500 cost of connection to a telephone line and a $10 one-time and $600 annual fee for Internet
Information and Communication Technologies for Direct Poverty Alleviation 147
Services of a part-time technician to support users are likely to add at least another
$2,000. On top of these costs are those for electricity, call charges and housing, which
can be very significant in rural areas. Shakeel et al. (2001) estimate off-grid power costs
per Internet-enabled computer at about $4,000 fixed and $200 recurrent costs, for
example. Providing Internet access is made more complex if technicians have to travel
long distances to maintain remote facilities or there are no roads on which to travel;
again, this is frequently the case in poor rural areas.
A simple model may help to illustrate the economics of telephony and Internet
provision in sparsely populated areas. The model calculates the fixed costs of provision
per capita for a population receiving public telephony or Internet access, depending on
the cost of installation and the population density.
The economic fixed cost of providing telephone service is allowed to vary. The
model assumes that the fixed cost of Internet access and technical support (above
telephony costs) is $3,172 per year (which excludes electricity costs). In order to come
up with a yearly fixed cost, these figures are discounted at 20%. It is assumed that each
telephone or Internet-enabled computer has a potential catchment area of 80km2, or
approximately a 5km. radius.16 The fixed costs of telephone and the Internet are
assumed to fall equally on this population within that radius. This is unrealistically
favourable, given that, even with access, much of the rural population will not use the
Internet because of the barriers of need, interest, age, language, literacy, skills and
disability. The fixed costs of service provision per capita are shown in Tables 3 and 4.
At a line installation cost of $10,000 and a population density of 10/km2, the annual
per capita fixed costs of telephony are approximately $2.50 – or 25% of the estimated
$10 annual expenditure on communications of a person living on $1 a day. The costs of
the Internet are significantly higher – about 65% of annual expenditure. Even at a
relatively low cost of rural telephone installation, about $2,500 for the first line, fixed
Internet costs at a population density of 10/km2 are still 46% of total expenditure
(compared with 6.3% for the telephone).
connection. We will see later that costs of access might fall below this level with policy and technological
advance. These costs might already be considered too high. The Sustainable Access in Rural India Project
(SARI: http://www.tenet.res.in/rural/sari.html) hopes to provide access to a simple Internet-enabled
computer and a public telephone over wireless local loop in the Madurai district of India for around $1,000
in fixed costs and $360 annual recurrent costs. The computer for the system uses so little electricity that it
might be feasible to create a wind-up model. The programme is still in development, and cannot offer the
type of technical support that would allow for extensive support of poor people’s Internet use, but might
suggest the potential for Internet-enabled technologies to support a considerably larger percentage of the
world’s poorest. Proenza (2002) reports costs in urban (connected) telecentres in Peru, Hungary and
Jamaica which carry annualised costs (including technical support) of between $1,496 and $4,440 per
computer – larger centres carrying significantly lower per computer costs. Looking at an operational rural
system, however, suggests the above estimates might be too low. LINCOS’ solar powered telecentre in
Costa Rica cost approximately $20,000 per computer in an area without grid electricity and $18,000 in an
area with grid electricity for a six-computer centre. Ongoing costs excluding manpower were
approximately $5,000 per computer (Shakeel et al., 2001).
16. Few areas of the world have rural population densities as low as one person per sq. km., but country
average rural population per sq. km. of arable land is frequently below 50, and we have seen that in some
areas of Chile, for example, population densities do fall this low. For example, average rural population
density is below 50 in Argentina, Belarus, Kazakhstan, the Russian Federation, Ukraine and Uruguay.
148 Charles Kenny
Table 3: Fixed costs of telephony provision per capita,
varying installation cost and population density
Cost of line installation ($) Population density/km2
1 5 19 50
fixed costs of provision/capital/year
2,500 6.25 1.25 0.63 0.13
5,000 12.50 2.50 1.25 0.25
10,000 25.00 5.00 2.50 0.50
25,000 62.50 12.50 6.25 1.25
50,000 125.00 25.00 12.50 2.50
Served population 80 400 800 4000
Table 4: Fixed costs of Internet and telephony provision per capita,
varying installation cost and population density
Cost of line installation ($) Population density/km2
1 5 19 50
fixed costs of provision/capital/year
2,500 45.90 9.18 4.59 0.92
5,000 52.15 10.43 5.22 1.04
10,000 64.65 12.93 6.47 1.29
25,000 102.15 20.43 10.22 2.04
50,000 164.65 32.93 16.47 3.29
Served population 80 400 800 4000
The numbers in Table 4 suggest that the fixed costs of Internet provision are high
enough to pose a significant sustainability challenge for poor populations even at higher
population densities. With 400 people sharing the same telephone and computer, access
per person would be limited to around 11 hours a year if the terminal or telephone were
available (and used) 12 hours a day. Under these circumstances, the utility of the
Internet would be severely diminished by long waiting periods and short access times. If
a user-base per computer of 400 is considered a fair maximum, it is unlikely that the
fixed costs of Internet provision would be under 80% of total expenditures for a poor
community if strong technical support (one technician per three terminals) is considered
a necessity, or under 40% even in the absence of such support.
The approximate cost per hour of accessing different ICTs at different levels of
total usage per year can be estimated, based on a number of simplifying assumptions.17
Low values are used for the fixed cost of telephone access. At a usage rate of 2,000
17. The fixed costs per year of Internet (excluding telephone) are assumed, as above, to be $3,172. The fixed
costs of telephony per year are assumed to be $300, and of radio (a wind-up version) $21. The marginal
cost of Internet and telephony is $3 an hour for a regional connection, $1 for a local connection. There are
no marginal costs for listening to the (wind-up) radio.
Information and Communication Technologies for Direct Poverty Alleviation 149
hours per year (about 6 hours per day) the cost per hour of radio is about 1 US cent, as
compared with $1.15 per hour for local telephone and $2.69 for local Internet access.
The poor (spending under $10/year on communications) could afford perhaps 2 hours
per year at an actively used facility offering Internet access provided through a regional
(rather than local) connection, and about 40 minutes a year if the Internet connection
was rarely used (compared with about 5 hours a year in local calls or 100 hours for a
radio).18 These figures will be one reason why Internet use and access are currently even
more concentrated within and across countries than telephone access (see Heeks and
Kenny, 2002).
Overall, then, the low income of the poor combined with their location in rural
areas makes the use of the Internet, in particular, a financially disadvantageous method
for information access for many people in poverty, even if the necessary inputs are
available. Evidence from South Africa, where the telecentre movement is fairly well
developed, suggests that these potential problems do translate into significant financial
difficulties. Even in this fairly rich (upper middle-income) country, almost all Internet-
enabled telecentres rely on government, donor or NGO funding to remain viable
(Pigato, 2001).
Requirements for use
The Internet presents straightforward affordability questions when it comes to
increasing access. There are reasons to suggest that the requirements for their succesful
utilisation make radio and telephone far more suitable technologies for the poor.
Neither the telephone nor radio faces significant language barriers. For the radio,
programming is cheap enough to be produced locally and in a range of languages. For
example, in Latin America most radio (as opposed to television or Internet content) is
produced locally or nationally (http://commons.somewhere.com/rre/2000/RRE.
Radio.and.the.Intern.html). In Peru alone an estimated 180 radio stations offer
programmes in Quecha, a language spoken by only 10 million people in the whole Latin
American region (and one that is almost completely absent from the Internet)
(http://commons.somewhere.com/rre/2000/RRE.Radio.and.the.Intern.html). Telephone
and radio signals can also provide information access to the illiterate and those with no
training in ICT use.
Compare these features with those of the Internet, a tool that requires a fairly high
level of education and computer literacy to be used constructively. For example, a
recent study of Capacity Building for Electronic Communication in Africa (CABECA,
1998) found that 87% of Zimbabwean and 98% of Ethiopian Internet users had a
university degree (Ethiopia is a country where 64% of the population is illiterate).
Indeed, the Internet is part of a cluster of IT technologies that is widening the gap
between the incomes of those with a university degree and those without.19 This is not a
18. Because of low demand for ISPs in poor rural areas, it is likely that the connection would be regional –
although some countries cross-subsidise calls to ISPs by offering local-cost connection country-wide.
19. Skill-upgrading and growing premia for education in the US, for example, have been widespread, but
particularly rapid in computer-intensive industries (Autor et al., 1998). The poor, who cannot afford access
or training, and the majority of whom are not even literate, will suffer from growing wage inequality.
While it is possible to design Internet applications for illiterate populations, their cost-effectiveness over
150 Charles Kenny
situation that is likely to change quickly, because low expenditure on education makes
the Internet an unaffordable tool for general education or for IT technical training in the
primary schools attended by poor children (see below).
Even those of the poor who find access, who are literate, and who acquire basic
computing skills face the significant barrier to use of language. Along with training,
language skills are vital to utilise the Internet’s resources fully, because minority
languages are dramatically under-represented on the World Wide Web. Igbo (Ibo), a
language spoken by 17 million people in Nigeria, is all but completely absent from the
Internet.20 Conversely, English – a language spoken by very few of the world’s poorest
– remains overwhelmingly dominant. A 1999 survey found that 72% of sites were in
English. Japanese and German added another 12%, and French, Chinese and Spanish all
had between 1 and 2% (Nunberg, 2000).21 The effect of English dominance on the
efficacy of the web for non-English speakers can be estimated by looking at a recent
study conducted in Tokyo, Beijing, Seoul, Bangkok, Singapore and Jakarta. This study
found that English speakers were two to four times more likely to use the Internet than
the non-English speaking population.22
Poor people’s use of more advanced Internet operations, such as e-commerce, faces
an even greater number of barriers. Poor people do not have the requisite credit
facilities, they are far from logistics services that could deliver goods, and the types of
goods they would want to buy or sell have limited outlets on line. It is for this wide
range of reasons that UNCTAD’s survey of e-commerce use in least developed
countries, in the 2001 E-Commerce and Development Report, was able to find only a
few examples of e-commerce, largely serving niche markets, limited to sales of between
$2,000 and $30,000 per year, and employing a maximum of 50 people (UNCTAD,
2001).
Benefits of utilisation
There is ample evidence of the benefits of utilisation for a range of ICTs. Radio, in
addition to being the only accessible mass medium, is a trusted source of information.
In Nepal 71% of rural people surveyed used the radio as a source for information and
found it effective. Friends, family and political leaders were the only information
sources ranked as more effective, and the radio was judged far better than schools,
newspapers (used and found effective by 24% of the population), television, telephone
other technologies has yet to be demonstrated, and the range of such applications remains very limited,
suggesting that illiterates will (at the very least) garner less utility from the new technology.
20. The only documents that could be found in Igbo after an extensive (two-hour) search were translations of
the Universal Declaration of Human Rights, of something called the four spiritual laws, and of the food
pyramid, a two-page Igbo phrase book and a prayer manual. Another example is Quecha, a language
spoken by some 10 million people in Bolivia, Ecuador and Peru, which is completely absent from the
Internet (http://commons.somewhere.com/rre/2000/RRE.Radio.and.the.Intern.html). Internet search found
no web sites in Quecha, but one site, in Spanish, run by Peru’s Red Scientifica, that discussed the Quecha
language.
21. While 8% of websites could be classed as multilingual, all such sites had English as one of their
languages, compared to the next highest (31%) with French or German as one of their languages (Lavoie
and O’Neill, 1999).
22. Translation programmes are only a partial solution –being created only for languages spoken by many
millions (http://www.feedmag.com/daily/dy070799.html).
Information and Communication Technologies for Direct Poverty Alleviation 151
(used by 19%, but found effective by only 1%) and the computer (also used and found
effective by only 1%) (Pigato, 2001).
Unsurprisingly, the radio has shown itself to be a powerful tool for development. In
the Philippines, for example, one development programme is providing local radio
equipment and training to a number of remote villages. The project has not only
increased local business and agricultural productivity, but also resulted in the formation
of civic organisations and more constructive dialogue with local officials (UNESCO,
1996). At the cross-country level, the availability of independent radio broadcasting
services has been found to be positively and significantly correlated with a range of
development outcomes including life expectancy, lower infant mortality, schooling
outcomes and better functioning markets (Djankov et al., 2001).
The radio has a particularly important role in extension and education. A survey of
some of the 21,000 farmers enrolled in radio-backed farm forums in Zambia found that
90% of respondents thought the programmes were relevant and more than 50% credited
them with increasing their crop yields (Dodds, 1999). In another study sponsored by
UNESCO, Paul Neurath studied the effects of a Farm Radio Forum project at Poona,
India. According to Neurath (as reported in Nwaerondu and Thompson, 1987: 105):
Radio farm forum as an agent for transmission of knowledge has proved to be a success
beyond expectation. Increase in knowledge in the forum villages between pre- and post-
broadcasts was spectacular, whereas in the non-forum villages it was negligible. What
little gain there was occurred mostly in the non-forum villages with radio.
Dodds (1999) also notes that a significant percentage of health workers in Uganda
(54%) and Kenya (20%/year) have taken part in radio-backed training courses, and
there are consistent reports and surveys suggesting that these result in improved
knowledge, attitudes and practices. Looking at education more generally, Adkins’
(1999) survey of seven studies of the cost-effectiveness of educational intervention
suggests that, in terms of incremental improvement, the impact of a dollar spent on
interactive radio instruction (IRI) is nearly 70% greater than $1 spent on purchasing
textbooks and over 11 times more than $1 spent on teacher training.
Turning to telephony, a recent study found that in rural Thailand the introduction of
telephones which enabled farmers to check prices regularly had as much as doubled
farm income (ITU, 1999). In Colombia, community telephone access in Tumaco in
1994 increased trade, employment and government service delivery (ITU, 1998).
Access to telephony has improved opportunities in the rural non-farm sector, thereby
increasing incomes in rural Ecuador (Elbers and Lanjouw, 2001). Operating public call
centres is in itself a major source of rural employment and income; in the Indian state of
Punjab, for example, there were 10,000 telecentres, each generating an average of
$9,000 in revenue in 1996 (Grace et al., 2001). The strong link between telephone roll-
out and income growth has also been found in numerous cross-country studies. Cross-
country evidence further suggests that limited access to telephony within a country is a
powerful force behind growth in income inequality; those with access benefit, whilst the
incomes of those unconnected to the telephone network stagnate (Forestier et al., 2001).
The Internet is a more powerful ICT than either the radio or telephony, not least
combining the ‘broadcast’ features of the radio with the interactive features of the
telephone. As such, it has a range of capabilities that cannot be matched by the two
152 Charles Kenny
more basic ICTs. And direct access to Internet-enabled computers has also made a
difference to many poorer people in developing countries. For example, in Chile, the
national agricultural extension service created an Internet-based rural information
service for farmers’ groups, local authorities and NGOs. Transmitting price and market
information through the Internet cost 40% less than using a printed bulletin, and was
also almost instant, rather than taking 45 days (Balit, 1998: 4).
Having said that, the scale of the additional utility of the Internet to the poorest,
above and beyond that which can be garnered through more basic ICTs, has yet to be
conclusively demonstrated. For example, with regard to the transmission of crop-price
data via the Internet, both the telephone and the radio have been used for this purpose
for many years, with a significant impact on the price received by rural farmers for their
crops.
Regarding e-commerce opportunities, even if we ignore the multiple barriers to the
poor’s use of such technology, a recent study estimates that transactions savings from e-
commerce on food ingredients in the US would be of the order of 3-5%, compared with
29-39% for electronic components, for example (Goldman Sachs, 1999). In turn, this
might suggest that the poor as consumers will not see dramatically lower prices for their
largest consumption item (food), and the poor as producers (even those with access to
the Internet) will not see dramatically increased demand for their products because of
lower prices.
The Internet is also not a particularly efficient tool for the utilisation of unskilled
labour (the type of labour that most of the poor have to offer). It might have some use as
an information tool for connecting unskilled workers with employers who have a
demand for them, but, as a direct source of income-generating activities, the Internet is a
tool for skilled (at least computer-literate) employees, as we have seen.
Finally, direct provision of government services to the poor through the Internet
would be both impractical (excluding those who are illiterate, and lack access, language
or computing skills) and unsustainable. Taking the example of education services, the
discretionary budget in primary schools in poorer developing countries (that left over
after teachers’ salaries have been paid) can be as low as about $5 per student per year.
For comparison, estimates from a recent World Bank project in Turkey suggest that
costs per student for IT classrooms being set up in that country, excluding training,
housing and recurrent costs, are approximately $141. Even if the resources for Internet-
enhanced educational services could be found, evidence on the cost-effectiveness of the
technology is far weaker than that for interactive radio instruction, for example (Grace
and Kenny, 2001).
In contrast to the case of radio and telephone, then, the poor face multiple barriers
both to Internet utilisation and to benefiting from that use. They have little to spend on
communications, live in areas where the Internet is costly and complex to provide, have
low rates of education and speak languages ill-represented on the World Wide Web.
The multiple factors behind the low utility of the Internet compared with more
traditional ICTs is also suggested by two recent studies that have found low Internet
usage rates even in rural areas with access to the technology. Figure 1 is based on a
survey of two villages in Uganda carried out by Samuel Kyabwe and Richard Kibombo
(see http://www.idrc.ca/telecentre/evaluation/nn/22_Buw.html) at the time when
Internet-enabled telecentres were being set up in the villages. It suggests that even
Information and Communication Technologies for Direct Poverty Alleviation 153
villages with Internet access may see usage rates as low as 5% – compared with close to
30% telephone and almost 100% radio usage.23
If providing Internet access would be very expensive, and yet providing that access
might be of limited additional benefit beyond access to radio and telephony, this must
raise questions as to the advisability of embarking on large-scale Internet direct access
programmes as a tool of poverty relief, especially as compared with access programmes
focusing on the more ‘suitable’ technologies of radio and telephony.
Nonetheless, there is a potentially important role for the Internet to support poverty
alleviation efforts indirectly, through the more efficient functioning of governments and
expanded export opportunities, for example. The Internet is a very powerful tool for
information transfer; switching from fax to email can reduce the costs of sending 10,000
pages of text from Mozambique to the US over the course of a year by 83% (Africa
Internet Forum, 1999). The CARD project in Andhra Pradesh, India is but one example
of networked computer use in developing country government operations that help the
poor. The Computer-aided Administration of Registration Department system has
significantly reduced the time needed to register property and reduced corruption in the
process, making the security of registration much more accessible to the poor.
Figure 1: ICT use in Nabweru and Buwama, Uganda
Furthermore, traditional ICTs can act as a sustainable intermediary for the poor to
access the power of the Internet indirectly. Rural radio, in particular, can benefit from
the presence of the Internet. In Kothmale, Sri Lanka, a joint project between UNESCO,
the Ministry of Posts, Telecommunications and the Media, the Sri Lanka Broadcasting
23. A recent study of a pilot programme of the Ministry of Environment, Natural Resources, and Fisheries in
Mexico also found limited demand for Internet use. Of 23 telecentres set up in rural areas around the
country, only 5 remained functional after two years (Robinson, 2000). These findings are echoed by
surveys in India, Nepal, Tanzania and Uganda reported in Pigato (2001) that suggest low Internet use
amongst small-scale entrepreneurs with access because available content is not relevant, and a general low
level of trust in the use of modern ICTs as a tool of information exchange.
0
10
20
30
40
50
60
70
80
90
100
Internet
email
fax
video cassettte
telephone
TV
postal services
audiocassette
radio
Percentage of Respondants who Use
154 Charles Kenny
Corporation, and the Sri Lanka Telecommunication Regulatory Commission uses radio
as an interface between rural people and the Internet. A daily one-hour live radio
programme in which an announcer and a panel of resource persons browse the Internet
at the request of listeners, has proved capable of overcoming linguistic barriers in using
the Internet for non-English speakers. The radio station adds value to the information by
interpreting it for the local context, by broadcasting it in vernacular languages and by
providing a platform for feedback through local discussion and networks of local
correspondents.
The Internet can also act as a distribution network among independent broadcasters.
The Panos Institute’s Banque de Programmes On Line (www.oneworld.org/
panos_audio/), located in Mali, has correspondents in 20 francophone African countries,
and Latin America’s Agencia Informativa Pulsar is a similar initiative for Spanish-
language programming (http://commons.somewhere.com/rre/2000/RRE.Radio.and.
the.Intern.html, www.pulsar.org.ec). Both these projects provide radio content
accessible by community stations worldwide for broadcast.
Conclusion and policy recommendations
Radio is a powerful, sustainable technology for meeting many of the information needs
of the poor. A policy that promotes access to as wide a range of radio (and television)
broadcast options as possible is clearly important for the development of opportunities
for the poor. Opportunities for private, competitive provision of radio content will
expand choices and development impact. Governments should therefore legalise the
private provision of national and local radio and issue spectrum licences for
broadcasters. In Colombia, for example, over 1,000 new licences were issued to
community stations in 1995, dramatically increasing listener choice and information
flow. Even in poorer countries, opening the airwaves increases choice and information
flow. There are 19 independent rural or community radio stations in Mali alone;
Mauritania now boasts at least four such stations; and South Africa has over 80 (Vogt,
2001).
In addition to opening up to private and community provision, there also remains a
significant role for government in the broadcast sector, especially in the delivery of
public service content in areas such as education, health and disaster preparedness.
Furthermore, there might be a role in providing access to receivers. Radio sets are
already fairly ubiquitous in developing countries. Nonetheless, there might be an
argument for providing wind-up sets to schools and community centres to ensure wider
access amongst the poorest people. Donors, including the World Bank and the UK’s
Department for International Development, have supported carefully designed
programmes of providing communities with radios.
The case of telephones is slightly different. Because of the historical concentration
of access amongst wealthy urban populations, we have seen that telephone roll-out has
traditionally been a force for divergence in incomes both between rich and poor
countries and within poor countries. However, technological change, policy reform and
innovative universal access programmes such as that in Chile have made the goal of
extending telephone access to the majority of the poor an increasingly feasible idea.
It is clear that the first step remains a programme of reform towards well-regulated
private, competitive markets, which have repeatedly delivered expanded network access
Information and Communication Technologies for Direct Poverty Alleviation 155
at lower cost (see Kenny, 2001, for a review of the cross-country evidence). To extend
access beyond the market, subsidy auctions to provide lowest-cost, privately-provided
public access in unserved areas (the model used to achieve universal access in Chile)
have proved affordable and sustainable (Wellenius, 1997). It is hoped that this will turn
telecommunications roll-out from a force for divergence into a force for convergence.
It will be a while before the same can be said of the Internet. Again, this does not
mean that the technology is irrelevant to developing countries. It will have a range of
uses in production, trade and the provision of government services that should increase
incomes and improve the quality of life of the poorest. Via intermediary technologies
including radio and telephony, the Internet might also have a significant impact on
information flows directly to and from the poorest people.
However, at least until technological advance has made Internet access less
expensive and more straightforward for the illiterate and minority-language speaker to
use – and until education has become more widespread – the use of traditional
computers hooked up to the Internet as a tool for poverty alleviation should probably
not involve programmes for universal access. Even the cost of universal access to
simple telephony would be a large burden on government and aid budgets. If we assume
that service can be provided to the great bulk of the poor who currently lack access at
the same cost as in Chile ($10 per head), this cost is equal to half of the annual per
capita expenditure on health in low-income countries. Adding the Internet to universal
access goals would greatly increase that cost. As we have seen, such programmes are
likely to be complex and very expensive, and quite possibly of limited benefit. Instead,
the Internet should be used as an indirect supporting tool in efforts to improve
information and communications flows that do benefit poor people.
References
Adkins, D. (1999) ‘Cost and Finance’in A. Dock and J. Helwig (eds), Interactive Radio
Instruction: Impact, Sustainability and Future Directions. Washington, DC: World
Bank.
Africa Internet Forum (1999) ‘Economic Toolkit for African Policy Makers’. Available
at http://www.worldbank.org/infodev/projects/finafcon.htm.
Autor, D. Katz, L. and Kreuger, A. (1998) ‘Computing Inequality: Have Computers
Changed the Labor Market?’, Quarterly Journal of Economics 113 (4): 1169-1214.
Balit, Silvia (1998) ‘Listening to Farmers: Communication for Participation and Change
in Latin America’. Available at http://www.fao.org/WAICENT/FAOINFO/
SSTDEV/CDdirect/Cdan0018.htm.
CABECA (1998) (Menou, M.J.) Connectivity in Africa: Use, Benefits and Constraints
of Electronic Communication – Synthesis Report. Part 2: Overview of the Findings
of the Project. Study carried out under the CABECA project of Padis as part of the
IDRC-sponsored research programme on the Impact of Information on
Development. Addis Ababa: UNECA/PADIS, May.
Datt, G. and Gunewardena, D. (1997) Some Aspects of Poverty in Sri Lanka: 1985-90.
World Bank Policy Research Working Paper 1738. Washington, DC: World Bank.
de Melo, J. (2000) Telecommunications and the Poor. Internal World Bank Report.
Washington, DC: World Bank.
156 Charles Kenny
Djankov, S., McLiesh, C., Nenova, T. and Shleifer, A. (2001) ‘Who Owns the Media’,
Background paper for World Bank World Development Report 2002. Washington,
DC: World Bank.
Dodds, T. (1999) ‘Non-Formal and Adult Basic Education through Open and Distance
Learning in Africa’. University of Namibia: Centre for External Studies (mimeo).
Easterly, W. and Levine, R. (1997) ‘Africa’s Growth Tragedy: Policies and Ethnic
Divisions’,Quarterly Journal of Economics 112: 1203-50.
Elbers, C. and Lanjouw, P. (2001) ‘Intersectoral Transfer, Growth, and Inequality in
Rural Ecuador’,World Development 29 (3): 481-96.
Elwan, A. (1999) ‘Poverty and Disability: A Survey of the Literature’, Background
Paper for World Bank World Development Report 2000/1. Washington, DC: World
Bank.
Forestier, Emmanuel, Grace, Jeremy and Kenny, Charles (2001) ‘Can Information and
Telecommunications Technologies be Pro-Poor?’Paper presented at World Bank
Economists’Forum, 3-4 May.
Goldman Sachs (1999) ‘B2B –Just How Big is the Opportunity?’. Available at
http://www.gs.com/hightech/research.
Grace, Jeremy, Kenny, Charles and Zhen-wei Qiang, Christine (2001) ‘ICTs and Broad-
Based Development’. Washington, DC: World Bank (mimeo).
Grace, Jeremy and Kenny, Charles (2001) ‘A Short Review of Information and
Communications Technologies and Basic Education in LDCs’. Washington, DC:
WorldBank(mimeo).
Harrell, M., Parillon, C. and Franklin, R. (1989) ‘Nutritional Classification Study of
Peru: Who and Where are the Poor?’,Food Policy 14 (4): 313-29.
Heeks, R. and Kenny, C. (2002) ‘Is The Internet a Technology of Convergence or
Divergence?’Washington, DC: World Bank (mimeo).
Intven, H. (2000) Telecommunications Regulation Handbook. Washington, DC: World
Bank.
ITU (1998) ‘Rural Telecommunications in Colombia –Lessons Learned’. Paper given
at World Telecommunication Development Conference (WTDC-98), Valletta,
Malta, 23 March –1 April.
ITU (International Telecommunications Union) (1999) ‘General Principles:
Particularities of Rural Environments and Their Implications for
Telecommunications’, in Handbook on New Developments in Rural
Telecommunications. Geneva: ITU.
James, J. (2000) ‘Bridging the Digital Divide with Low-Cost Technologies’. Tilburg
University (mimeo).
Kenny, C (2001) ‘Prioritizing Countries for Assistance to Overcome the Digital
Divide’,Communications and Strategies 41: 17-36.
Lavoie,B.andO’Neill, E. (1999) ‘How World Wide is the Web?’,Annual Review of
OCLC Research. Available at http://www.oclc.org/oclc/research/publications/
review99.
Nunberg, Peter (2000) ‘Will the Internet Always Speak English?’,The American
Prospect, 10 March p.27.
Nwaerondu, Ndubuisi Goodluck and Thompson, Gordon (1987) ‘The Use of
Educational Radio in Developing Countries: Lessons from the Past’,Journal of
Distance Education 2 (2): 43-54.
Information and Communication Technologies for Direct Poverty Alleviation 157
Pigato, M. (2001) ‘Information and Communication Technology, Poverty and
Development in Sub-Saharan Africa and South Asia’. Washington, DC: World
Bank (mimeo).
Proenza, F. (2002) ‘Telecenter Sustainability –Myths and Opportunities’,Journal of
Development Communications 13 (1) (forthcoming).
Robinson, Scott S. (2000) ‘Telecentres in Mexico: Learning the Hard Way’. Paper
presented at the ‘Partnerships and Participation in Telecommunications for Rural
Development: Exploring What Works and Why’conference at University of
Guelph, Guelph, Ontario, Canada, 26 and 27 October. Available at
http://www.devmedia.org/documents/robinson.htm.
Shakeel, H., Best, M., Miller, B. and Weber, S. (2001) ‘Comparing Urban and Rural
Telecenters’Cost’,The Electronic Journal on Information Systems in Developing
Countries 4 (2): 1-13.
UNCTAD (2001) E-Commerce and Development Report. New York: United Nations.
UNESCO (1996) ‘Information and Communication Technologies in Development: A
UNESCO Perspective’. Submission to the UNCSTD Working Group on
Information Technology for Development and the ITU Development Study Group
1, Inter-Agency Project on Universal Access to Basic Communication and
Information Services, CII-96/WS/6, September. (Available at http://www.itu.int/
acc/rtc/unesco.htm.)
Vakis, Renos and Lindert, K. (2000) Poverty in Indigenous Populations in Panama: A
Study Using LSMS Data. World Bank LCSHD Paper Series. Washington, DC:
World Bank, January.
van de Walle, D. and Gunewardena, D. (1999) Sources of Ethnic Inequality in Viet
Nam. World Bank Poverty Research Working Paper No.2297 Washington, DC:
World Bank.
Vogt, E. (2001) ‘Community Media and the Digital Divide –The Radio Model’Bonn:
FES (mimeo).
Wellenius, B. (1997) ‘Extending Telecommunications Service to Rural Areas –The
Chilean Experience’,World Bank Viewpoint Note No.105. Washington, DC: World
Bank.
World Bank (1990) World Development Report, 1990: Poverty. New York: Oxford
University Press.
World Bank (1995a) Brazil: A Poverty Assessment. Washington, DC: World Bank.
World Bank (1995b) Ghana: Poverty Past, Present and Future. Washington, DC:
World Bank.
World Bank (1995c) Vietnam: Poverty Assessment and Strategy. Washington, DC:
World Bank.
World Bank (1996) Tanzania: The Challenge of Reforms, Growth Income and Welfare.
Washington, DC: World Bank.
World Bank (1997a) India: Achievements and Challenges in Reducing Poverty.
Washington, DC: World Bank.
World Bank (1997b) Chile: Poverty and Income Distribution in a High-Growth
Economy, 1987-95. Washington, DC: World Bank.
World Bank (1999) Bangladesh: From Counting the Poor to Making the Poor Count.
Washington, DC: World Bank.
World Bank (2000) World Development Indicators. Washington, DC: World Bank.
