Ending Africa's Poverty Trap
AFRICA'S DEVELOPMENT CRISIS is unique. Not only is Africa the poorest
region in the world, but it was also the only major developing region with
negative growth in income per capita during 1980-2000 (table 1). Some
African countries grew during the 1990s, but for the most part this growth
recovered ground lost during the 1980s. Moreover, Africa's health conditions
are by far the worst on the planet. The AIDS pandemic is wreaking
havoc, as is the resurgence of malaria due to rising drug resistance and
the lack of effective public health systems. Africa's population continues
to soar, adding ecological stresses to the economic strains. Policy-based
development lending to Africa over the past twenty years, known as structural
adjustment lending, did not solve the problem. A heavy debt burden
is evidenced by the 155 Paris Club restructurings of African countries'
debt between 1980 and 2001, much more than for any other region. In
general, Africa remains mired in poverty and debt.
This paper focuses on the tropical sub-Saharan African countries with
populations of at least 2 million people in 2001. We leave out North Africa
(Algeria, Egypt, Libya, Morocco, and Tunisia,), southernmost Africa
(Botswana, Lesotho, Namibia, South Africa, and Swaziland), and a number
of very small economies (Cape Verde, Comoros, Djibouti, Equatorial
Guinea, Gabon, The Gambia, Guinea-Bissau, Mauritius, Sao Tome and
Principe, and Seychelles). Both nontropical ends of Africa are much richer
We would like to thank our colleagues in the UN Millennium Project, an advisory
project to UN Secretary-General Kofi Annan, and especially the Task Force coordinators
and members, for ample advice and consultations. We especially thank Vijay Modi for
his detailed assistance on issues regarding energy and transport infrastructure.
117
118 Brookings Papers on Economic Activity, 1:2004
Table 1. Selected Development Indicators for Major Developing Regionsa
Average
annual Under-five Average
growth Life mortality annual
GNI in GDP expectancy rate, 2001 growth in
per capita, per capita, at birth, (deaths population,
2001 1980-2000 2001 per 1,000 2001
Region (dollars)b (percent) (years) live births) (percent)
Tropical sub-Saharan 271 -1.1 46.0 172.5 2.3
Africac
South Asia 449 3.3 62.6 95.3 1.7
Latin America 3,669 0.5 70.6 32.7 1.4
East Asia and 3,710 6.4 70.2 38.3 0.8
the Pacific
Middle East and 2,207 0.9 68.4 49.8 2.0
North Africa
Source: World Bank (2003a).
a. Population-weighted averages of countries in the region.
b. GNI, gross national income.
c. Thirty-three-country sample listed in table 2; other regions are as defined by the World Bank.
than tropical Africa. They grow temperate crops, escape the worst of
malaria, enjoy (in the south) vast deposits per capita of gold and diamonds,
and (in the north) benefit from proximity to EU markets. The smallest
economies present idiosyncrasies that would distract more than inform
the analysis.
The thirty-three sub-Saharan African countries on which we focus (and
which are listed in table 2) had a combined population of 617 million in
2001, with a population-weighted average annual income of $271 per person,
or a mere 74 cents a day. Every country on the list is a low-income
country according to World Bank country classifications, and twenty-six
are among the forty-nine Least Developed Countries in the world by the
United Nations classification. Of the four countries with income per capita
of $500 or more, three (Angola, Cameroon, and Congo) are oil exporters,
and only COte d'Ivoire, which is currently in a vertiginous political and
economic collapse, is a non-oil exporter. Every country on the list has a life
expectancy at birth below sixty years, and in all but Ghana, Madagascar,
and Sudan life expectancy at birth is below fifty-five years. Child mortality
rates (deaths before the age of five per 1,000 live births) are above 100
in every country.
The standard diagnosis is that Africa is suffering from a governance
crisis. With highly visible examples of profoundly poor governance, for
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 119
Table 2. Governance Ratings and Household Consumption
in Tropical Sub-Saharan Africa
Rating Household
based on Rating final
World Bank based on consumption
governance Transparency Freedom expenditure
indicators, International House per capita, 2000
Country 2002a index, 2003a rating, 2003 (1980 =100)
Benin Good n.a. Free 98.9
Burkina Faso Good n.a. Partly free 111.0
Ghana Good Average Free 92.8
Madagascar Good Good Partly free 64.0
Malawi Good Good Partly free 111.2
Mali Good Good Free 95.3
Mauritania Good Good Partly free 104.8
Senegal Good Good Free 99.6
Cameroon Average Average Not free 102.5
Central African Rep. Average n.a. Partly free n.a.
Chad Average n.a. Not free n.a.
Congo, Rep. of Average Average n.a. 80.5
C6te d'Ivoire Average Average Not free 78.2
Eritrea Average n.a. Not free n.a.
Ethiopia Average Good Partly free n.a.
Guinea Average n.a. Not free n.a.
Kenya Average Average Partly free 100.7
Mozambique Average Good Partly free 79.4
Niger Average n.a. Partly free n.a.
Nigeria Average Average Partly free n.a.
Rwanda Average n.a. Not free 83.9
Sierra Leone Average Good Partly free 43.9
Tanzania Average Good Partly free n.a.
Togo Average n.a. Not free 112.4
Uganda Average Average Partly free n.a.
Zambia Average Good Partly free 47.0
Angola Poor Poor Not free n.a.
Burundi Poor n.a. Not free 65.0
Congo, Dem. Rep. of Poor n.a. Not free 45.1
Sudan Poor Average Not free n.a.
Zimbabwe Poor Average Not free 88.4
Liberia n.a. n.a. Not free n.a.
Somalia n.a. n.a. Not free n.a.
Sources: Kaufmann, Kraay, and Mastruzzi (2003); Radelet (2004); authors' calculations using data in Transparency International
(2004); Freedom House (2003); World Bank (2003a).
n.a., not available.
a. Determined from the residuals of a regression of countries' governance indicators or scores on income per capita (at purchasing
power parity); countries with residuals more than 1 standard deviation above or 1 standard deviation below the predicted
value are categorized as "good" or "poor," respectively, and those with residuals within 1 standard deviation as "average." For the
first column, the World Bank governance indicators are those calculated by Kaufmann, Kraay, and Mastruzzi (2003). For the second
column, Transparency International's Corruption Perceptions Index is used, which relates to the degree of corruption in the
country as perceived by business people, academics, and risk analysts and ranges between 10 (highly clean) and 0 (highly corrupt).
120 Brookings Papers on Economic Activity, 1:2004
example in Zimbabwe, and widespread war and violence, as in Angola,
the Democratic Republic of the Congo, Liberia, Sierra Leone, and Sudan,
the impression of a continent-wide governance crisis is understandable.
Yet it is wrong. Many parts of Africa are well governed even though stuck
in poverty. Governance is a problem, but Africa's development challenges
run much deeper.
Using our thirty-three-country sample, table 2 reports some common
governance indicators that make this point. The first column presents a
ranking of African governance compiled by Steven Radelet,1 who regresses
a set of widely used World Bank governance indicators due to Daniel
Kaufmann, Aart Kraay, and Massimo Mastruzzi on GDP per capita,2 and
ranks countries according to the residuals from that regression, thereby
standardizing the measurement of governance by level of income. This
procedure recognizes that poorer countries have systematically poorer
governance measures than richer countries, since good governance itself
requires real resources. Well-governed African countries are defined as
those with residuals at least 1 standard error above, and poorly governed
countries as those with residuals at least 1 standard error below, the
regression line. "Average" countries are those with residuals within
1 standard error on either side of the regression line. The table shows
that, by this ranking, many countries are well governed. Nor is there
evidence that Africa's governance, on average, is worse than elsewhere
once we control for income levels: a regression (not reported in this
paper) of several different governance indicators on the logarithm of GDP
per capita (measured at purchasing power parity) and a dummy for tropical
sub-Saharan Africa results in a statistically insignificant coefficient
for the dummy variable.
The second column of table 2 repeats the same procedure using the Corruption
Perceptions Index of Transparency International.3 All but one of
the African countries in our sample for which scores are available receive
scores of "good" (that is, low corruption) or "average" after we control for
income. The third column reminds us as well that many African countries
have become democracies in recent years, and thus are scored as "free" or
"partly free" by the well-known Freedom House ranking.4 The final col1.
Radelet (2004).
2. Kaufmann, Kraay, and Mastruzzi (2003).
3. Transparency International (2004).
4. Freedom House (2003).
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 121
Table 3. Regressions Explaining Economic Growth in Tropical Sub-Saharan
Africa with Governance Indicatorsa
Independent variable 3-1 3-2 3-3 3-4 3-5
Dummy variable for tropical -3.28 -3.06 -2.68 -3.43 -3.40
sub-Saharan Africab (-6.56) (-6.50) (-6.11) (-7.05) (-6.46)
Corruption Perceptions 0.83 ... ... ... ...
Index, 2003c (5.23)
Index of Economic Freedom, ... -0.96 ... ... ...
2001d (-2.75)
Average Kaufmann, Kraay, ... ... 1.89 ... ...
Zoido-Lobaton indicators, (5.91)
2000e
Average ICRG indicators, ... ... ... 1.56 ...
1982-971 (5.29)
Average ICRG indicators, 1982' ... ... ... ... 0.68
(3.78)
Logarithm of GDP per capita -2.07 -1.65 -1.75 -2.00 -1.82
(at purchasing power (-7.02) (-6.06) (-7.07) (-7.01) (-5.84)
parity), 1980
R 2 0.58 0.46 0.59 0.59 0.54
No. of observations 60 71 78 65 52
Sources: Authors' regressions using data from Transparency International (2004); Miles, Feulner, and O'Grady (2004); Kaufmann,
Kraay, and Zoido-Lobaton (2002); PSR Group (2004); World Bank (2003a).
a. The dependent variable is average annual growth of GDP per capita, 1980-2000. The sample consists of ninety-two countries
worldwide, excluding high-income countries and former republics of the Soviet Union. All regressions are ordinary least
squares and include a constant term (not reported). Numbers in parentheses are t statistics; all coefficients reach statistical significance
at the 1 percent level.
b. Variable equals 1 if the country is one of the thirty-three countries listed in table 2.
c. See table 2 for description.
d. The index is published by the Heritage Foundation and the Wall Street Jourtnal and ranges from I to 5, where 5 indicates the
greatest government interference in the economy and the least economic freedom.
e. Average of six World Bank governance indicators measured in units ranging from about -2.5 to 2.5, with higher values corresponding
to better governance outcomes.
f. Average of six governance indicators from the International Country Risk Guide, with values ranging from 1 to 6, with
higher values reflecting better governance.
umn, however, shows real consumption expenditure per capita in 2000, as
a percentage of its 1980 level, to point out that many of the relatively well
governed African countries have been unable to increase the material
well-being of their populations. Table 3 makes this point more systematically
in a cross-country regression, which shows that, after we control for
initial income in 1980 and the quality of governance (according to several
alternative measures), sub-Saharan African countries grew more slowly
than other developing countries, by around 3 percentage points a year.
Africa's crisis requires a better explanation than governance alone.
Our explanation is that tropical Africa, even the well-governed parts, is
stuck in a poverty trap, too poor to achieve robust, high levels of
122 Brookings Papers on Economic Activity, 1:2004
economic growth and, in many places, simply too poor to grow at all.5
More policy or governance reform, by itself, will not be sufficient to overcome
this trap. Specifically, Africa' s extreme poverty leads to low
national saving rates, which in turn lead to low or negative economic
growth rates. Low domestic saving is not offset by large inflows of private
foreign capital, for example foreign direct investment, because Africa's
poor infrastructure and weak human capital discourage such inflows.
With very low domestic saving and low rates of market-based foreign
capital inflows, there is little in Africa's current dynamics that promotes
an escape from poverty. Something new is needed.
We argue that what is needed is a "big push" in public investments to
produce a rapid "step" increase in Africa's underlying productivity, both
rural and urban. The intervention of foreign donors will be critical to
achieving this step increase. In particular, we argue that well-governed
African countries should be offered a substantial increase in official development
assistance (ODA) to enable them to achieve the Millennium
Development Goals (MDGs), the internationally agreed targets for poverty
reduction, by 2015.6 The goals are useful intermediate targets for breaking
Africa's poverty trap, because they address the key sectors in which
major productivity improvements are both needed and achievable, and
because the rich countries have repeatedly committed themselves to help5.
The concept of a low-level poverty trap is, of course, a long-standing hypothesis in
the theories of economic growth and development. The earliest mathematical formalization
was by Nelson (1956). The theoretical possibility of poverty traps in the neoclassical
growth model is covered briefly in the economic growth textbook by Barro and Sala-iMartin
(1998), which also discusses briefly the possible case for large-scale development
assistance to overcome such traps. The connection of a low-level trap to subsistence consumption
needs is spelled out in Ben-David (1998), and connections to agriculture and education
are described in the World Economic and Social Survey 2000. Two recent empirical
studies claiming that such poverty traps exist in very poor countries are UNCTAD (2002)
and Bloom, Canning, and Sevilla (2003).
6. Adopted by the world's leaders in September 2000, the UN Millennium Declaration
committed their nations to stronger global efforts to reduce poverty, improve health, and
promote peace, human rights, and environmental sustainability. The MDGs that emerged
from the declaration are specific, measurable targets, summarized by the following eight
goals: eradicate extreme poverty and hunger; achieve universal primary education; promote
gender equality and empower women; reduce child mortality; improve maternal
health; combat HIV/AIDS, malaria, and other diseases; ensure environmental sustainability;
and develop a global partnership for development. More information is available at
www. unmillenniumproject.org and www.un.org/millenniumgoals.
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 123
ing Africa achieve these goals, with more funding if necessary.7 However,
the rich countries have not yet delivered on that promise.
We begin by outlining our theory of Africa's poverty trap. This is followed
by a discussion of the structural conditions and history that have
led the continent into such a trap. Following that, we identify how a big
push in key investments could enable Africa to meet the MDGs, and how
that, in turn, would help to extricate Africa from its current development
trap. We then estimate the financial costs required to meet the MDGs, and
we suggest how those costs could be allocated between domestic resources
in Africa (both public and private) and increased ODA. Finally, we propose
a new framework for donor-African relations to underpin a big push
designed to meet the MDGs.
The Theory of Africa's Poverty Trap
Consider a standard neoclassical growth model in which output per
capita q is produced by a production function Af(k), where A is total factor
productivity and k is the capital-labor ratio. For the moment we take A
to be constant. The national saving rate is s, and d is the rate of capital
depreciation, with n denoting the rate of population growth. The rate of
capital accumulation dkldt is given by
(1) dkldt = sAf(k) - (n + d)k.
A change in the capital-labor ratio is called capital deepening. The term
(n + d)k is called capital widening and is equal to the amount of saving
per capita that is needed to hold the capital-labor ratio constant in the face
of population growth and depreciation. Equation 1 says that capital deepening
equals saving per capita minus capital widening.
7. Perhaps the clearest commitment is in the 2002 Monterrey Consensus of the International
Conference on Financing for Development, where in paragraph 42 all countries
(including the United States, the European Union, and other donors) adopted the following
position: "[W]e urge developed countries that have not done so to make concrete
efforts towards the target of 0.7 per cent of gross national product (GNP) as ODA to
developing countries and 0.15 to 0.20 per cent of GNP of developed countries to least
developed countries . . ." (pp. 9-10). ODA was 0.23 percent of the combined gross
national income of donor countries (members of the OECD's Development Assistance
Committee) in 2002, a shortfall of approximately $120 billion a year on a combined donor
GNP of around $25 trillion.
124 Brookings Papers on Economic Activity, 1:2004
Figure 1. Standard Neoclassical Growth Model
Output
units
(n+d)k
Af(k)
sAf(k)
Capital
kE
Source: Authors' model described in the text.
The economy grows in per capita terms as long as saving per capita
exceeds capital widening. If saving is lower than capital widening, the
economy experiences a decline in output per capita. The standard neoclassical
model is typically presented as if the economy necessarily grows
when k is very low, in the famous diagram reproduced in figure 1. In the
figure, sAf(k) is assumed to be very steep at the origin and, in particular,
steeper than the (n + d )k ray from the origin. In that case, when k is low,
dkldt is positive, since it is equal to the vertical distance between the
sAf(k) curve and the (n + d )k ray. Indeed, starting at a very low capitallabor
ratio, k and q rise asymptotically to a unique, positive steady-state
equilibrium at kE and qE = Af(kE).
Actually, figure 1 is a special case, and a not very plausible one at that.
When k is very low, two other things tend to be true. First, the marginal
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 125
productivity of capital also tends to be very low (instead of nearly infinite,
as the standard theory assumes), because a minimum threshold of capital
is needed before modern production processes can be started. Factory production
requires, for example, a basic infrastructure of electricity, roads,
and a functioning port, as well as a literate and numerate labor force.
When these basic conditions are not present, small increments of k may
have little effect. However, once the basic infrastructure and human capital
are in place, the marginal productivity of capital may indeed become
very high in a low-income country. The capital threshold can be seen in
figure 2: small increments of k below a threshold kT might do little to raise
production, and the sAf(k) curve might be very flat near the origin. It then
becomes steep in a middle range before flattening out once again at high
levels of k. In short, there is an early period of increasing returns to scale
in capital before the more traditional constant or decreasing returns to
scale set in.
Figure 2. Growth Model with Minimum Capital Stock Threshold
Output
units
, ~~~~~~~~~~Af(k)
@ / @ ~~~~~~~~(n+d)k
sA f(k)
-A>y> 1' Capital
kT kE
Source: Authors' model described in the text.
126 Brookings Papers on Economic Activity, 1:2004
Figure 3. Saving Trap
Output
units
4 / , , ~~~~~~~~~~~~~~~(n+d)k
/ ~~~~~~~~~~~~~~~~sAf(k)
Capital
kT kE
Source: Authors' model described in the text.
Second, the saving rate can become very low or even negative when
k is very low, because impoverished households do not save, but rather
must use all (or more than all) of their current income in the struggle just
to stay alive. Once basic needs are met, poor households may save quite a
lot of income, but not before. The saving trap leads to a picture like figure
3, in which sAf(k) is less steep than (n + d )k when k is very low, and then
turns steeper in an intermediate range. Note that sAf(k) looks as it does in
figure 2 even without assuming a threshold level of capital. The key
implication of figures 2 and 3 is that dkldt is negative when k < kT. When
an economy begins with very low capital, both the capital-labor ratio
and output per capita tend to decline over time. The very poor indeed get
poorer, pushed into more extreme poverty by the lack of capital accu-
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 127
mulation coupled with population growth. Only when an economy has a
capital-labor ratio above a minimum threshold kT does it tend to achieve
economic growth and converge to the steady-state kE and qE.
To formalize these ideas using the simplest version of the neoclassical
model, we can use the Harrod-Domar or AK model, in which q = Ak. In
this case, equation 1 becomes
(2) dk/dt = sAk - (n + d)k.
The proportionate growth rate of the economy is given by
(3) y = (llq)dqldt = sA - (n + d).
The economy grows if sA > (n + d) and declines if sA < (n + d). As is well
known, the AK model does not have a steady-state equilibrium. The economy
either grows or declines at a constant rate.
Let us now see how a poverty trap arises in the AK model, beginning
with a capital threshold. Suppose that A is low (AL) until a threshold level
of infrastructure capital kT (the minimum capital needed for roads, ports,
electric power, and human capital) is reached, and then becomes high
(AH > AL). If it is the case that sAL < n + d while sAH > n + d, the economy
can get stuck in a poverty trap. Specifically, if the capital stock starts
below kT, the economy shrinks at the rate sAL - (n + d) < 0, whereas if k
starts above kT, the economy grows at the rate sAH - (n + d) > 0. The picture
is similar to figure 2, with a relatively flat sAf(k) curve near the origin,
but with a vertical segment at k = kTas the economy transits from AL
to AH, and then a relatively steep segment for k > kT.
Consider alternatively the case of the saving trap, in which the saving
rate is very low or negative when k is very low. As an illustration, suppose
that households save a constant fraction cs not of total income but of discretionary
income, which we define as income after deducting the costs of
meeting basic needs. Poor countries calculate basic needs according to an
estimate of the minimum caloric intake necessary for a healthy and productive
life, plus ancillary essential needs of clothing, shelter, cooking
fuel, and a few other basic amenities. Let these minimum needs be represented
by the poverty line qP = Akp. Discretionary income is then qd = q -
qP if q > qP and qd = 0 otherwise. We assume that saving per capita sq is
equal to Gyqd. The saving function is therefore
128 Brookings Papers on Economic Activity, 1:2004
(4) s = 0 for q < qP (or k < k )
= (i I k /k) for q > qP (or k > kP)
This kind of saving function gives rise to a poverty trap. Specifically, in
the AK model, the capital stock must be greater than or equal to kT =
aAkPI [6A - (n + d)] to achieve positive economic growth. The growth
rate is easily calculated to be as follows:
(5) y = -(n+d), for k < kP
y = caA(1 - kP/k) - (n + d) < 0, for kP < k < kT
7 = (5A(1 - kP/k) - (n + d) > 0, for k > kT .
There is a long theoretical tradition and strong empirical evidence to support
the proposition that the saving rate is very low at low income levels
and rises as income rises above the poverty threshold. Simon Kuznets
remarked, in his classic study on Modern Economic Growth: Rate, Structure,
and Spread, that "It is hardly surprising that with the much lower
per capita product in the underdeveloped countries, their gross or net capital
formation proportions-which represent gross or net national saving
rates-are distinctly lower than those of industrial countries."8 Norman
Loayza, Klaus Schmidt-Hebbel, and Luis Serven find that, in developing
countries, a doubling of private income raises the long-run private saving
rate by 10 percentage points.9 Many other studies have shown that saving
rates rise at low levels of income.'0 In an interesting unpublished survey
of rural Ugandan households, only 23.8 percent indicated that they had
undertaken any saving, and "the most common reason given for not saving
was low income (85.4 percent), although a number of non-saving respondents
also referred to the poor access to financial institutions.""11
A third factor that can push an economy into a poverty trap is very
rapid population growth at low levels of k. Very poor people do not save
much, if anything, in the form of reproducible capital or even human cap8.
Kuznets (1966, p. 427).
9. Loayza, Schmidt-Hebbel, and Serv6n (2000).
10. For example, Ogaki, Ostry, and Reinhart (1996).
11. Musinguzi and Smith (2000, p. 9).
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 129
Figure 4. Demographic Trap
Output
units
r ~~~~Af(k)
(n+d)k
Capital
kT kE
Source: Authors' model described in the text.
ital, but they do save in the form of progeny. Since n is a function of k, it
is quite possible that sA < n + d when k is low and n is high, whereas sA >
n + d when k is high and hence n is low. This is diagrammed in figure 4.
The highest fertility rates in the world are observed among the world's
poorest people. This initially seems paradoxical, since the poorest people
have the least chance to actually care for and invest in such large families.
But ample research on fertility choices in low-income settings has
explained this paradoxical situation very well. First, children are net economic
assets from an early age in rural areas where most poor people live
(or, in a less extreme version, they are a much smaller net cost than children
in urban areas). They collect fuel wood and water and tend to the animals
and other household chores. Second, the children of the poor are
more likely than other children to die, and so risk-averse poor parents have
large numbers of children to compensate, and indeed a much larger number
than needed simply to break even in an actuarial sense. Third, parents
130 Brookings Papers on Economic Activity, 1:2004
view children as old-age security in the absence of a well-functioning and
reliable private or public old-age financing scheme. Fourth, poor families
lack access to contraceptives and family planning services, so there is
considerable unmet desire to limit childbearing: about 17 percent of married
women in developing countries would prefer to avoid a pregnancy
but are not using any form of family planning.12
These three factors-capital thresholds, saving traps, and demographic
traps-all interact. It is quite possible that no one of them would be sufficient
by itself to cause a poverty trap, but that they do so in combination.
For example, a country with substandard infrastructure may find itself with
a low level of productivity AL, which by itself would not create a trap, but
does so when combined with the low saving rate and high population
growth rate caused by the low productivity.
In any case, positive growth in income per capita is more likely if there
is rapid technological change. When A is rising, the growth equation
(equation 3) becomes
(6) y = sA - (n + d) + (1/A) (dA/dt).
We shall see, alas, that Africa suffers not only from low A, low s, and high
n, but also from very slow growth in total factor productivity. The slow
rise or stagnation of A is mainly for structural rather than governance reasons,
as we discuss later.
Why Has Africa Suffered the Most Persistent Poverty Trap?
The simple theory outlined above explains how a country or region can
get stuck in a poverty trap. It does not explain, however, why sub-Saharan
Africa has been the region most prone to getting stuck. We propose five
structural reasons that have made sub-Saharan Africa the most vulnerable
region in the world to a persistent poverty trap:
-very high transport costs and small market size
-low-productivity agriculture
-a very high disease burden
12. Ross and Winfrey (2002), who estimate the corresponding figure to be highest in
sub-Saharan Africa, at 24 percent.
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 131
-adverse geopolitics
-very slow diffusion of technology from abroad.
In combination, these produce the confluence of low A, low s, and high n
that characterizes Africa.
High Transport Costs and Small Market Size
To a remarkable extent, Africans live in the interior of the continent
and face enormous transport costs in shipping goods between coastal
ports and the places where they live and work. Moreover, the Sahara
effectively cuts sub-Saharan Africa off from high-volume overland trade
with its major high-income trading partner, Europe, and this fact increases
the high costs of transport. Problems of isolation are compounded by
small market size. Small developing countries with little access to global
trade tend to grow more slowly than countries with large internal markets,
or than countries that have easy access to trade (such as Singapore).'3
There are several reasons why Africans tend to live away from the
coast: the soils are often better and rainfall is more plentiful in the interior
highland regions; the burden of malaria is intrinsically lower there; and
centuries of slave trade made it dangerous for Africans to live near the
coast. This problem would not be so severe if the rivers from the interior
to the coast of Africa were navigable by oceangoing vessels, but they are
not. Economists are gradually returning to the wisdom of Adam Smith,
who recognized, in The Wealth of Nations, the long-standing low levels
of development in Africa and the fundamental role of transport costs in
impeding prosperity in the region:
There are in Africa none of those great inlets, such as the Baltic and Adriatic
seas in Europe, the Mediterranean and Euxine [Black] seas in both Europe and
Asia, and the gulphs of Arabia, Persia, India, Bengal, and Siam, in Asia, to
carry maritime commerce into the interior parts of that great continent; and the
great rivers of Africa are at too great a distance from one another to give occasion
to any considerable inland navigation.'4
Recent evidence underscores the extremely high transport costs in subSaharan
Africa and their severe impact on trade. Figure 5 compares the
much higher transport costs in sub-Saharan Africa relative to those in
13. Sachs (forthcoming).
14. Smith (1776/1976, Book I, Chapter iii, para. 8).
132 Brookings Papers on Economic Activity, 1:2004
Figure 5. Transport Costs in Typical Sub-Saharan African and Asian Countries
Dollars per ton per kilometer
OE Typical sub-Saharan African countries
800 _U* Typical Asian countries
700
600 NEMM
500
400
300
200 i
Local Regional National Intemnational
Source: Starkey and others (2002).
Asia; Nuno Limao and Anthony J. Venables estimate that halving transport
costs could increase the volume of transport by a factor of five.'5
Before high-intensity modem trade can get started, Africa needs an extensive
road system both from the coast to the interior and within the interior,
where the highest population concentrations are found. These roads, however,
are very expensive to build and maintain.
In terms of our simple model with AL and AH, we can say that Africa
could have the same AH as other regions, but it must invest in very costly
transport infrastructure to be on a par with other developing regions
that have lower transport costs. These transport costs therefore cause
Africa to be stuck with AL, which in turn contributes to the low-level
poverty trap.
Low-Productivity Agriculture
Africa gets no break on food productivity. Most Africans live in the subhumid
or arid tropics, with few rivers to provide irrigation and a lack of the
15. Limao and Venables (1999). Other relevant evidence includes that provided by
Rizet and Hine (1993), who found that long-distance freight tariffs in francophone West
Africa were over five times higher than comparable tariffs in India, Pakistan, and Vietnam
between 1986 and 1988. More recently, Hine and Ellis (2001) have found that long-distance
freight rates in Tanzania are three times higher than in Indonesia.
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 133
large alluvial plains typical of much of South and East Asia that permit
cheap irrigation. As a result, Africa has the lowest share of irrigated cropland
of any major region of the developing world.'6 Moreover, rainfall in
much of the continent is erratic, vulnerable to high seasonal and year-toyear
fluctuations, and the land is subject to high rates of evapotranspiration,
due also to high temperatures. Indeed, of all major regions, Africa
loses the largest share of its precipitation to evapotranspiration."7 In addition,
rainfall in the Sahel (the semiarid region just south of the Sahara) has
been persistently low during the past thirty years, in a phenomenon that
may be linked to long-term climate change and to rising sea-surface temperatures
in the Indian Ocean. High transport costs also mean that African
farmers can afford little fertilizer, since by the time they cover the cost of
bringing fertilizer to their farms and farm output to market, fertilizers are
no longer cost effective. Consequently, farmers are farming on soils
increasingly depleted of nutrients, and without modern methods of water
management, in increasingly densely populated communities that are therefore
too impoverished to finance roads and water infrastructure that could
dramatically raise farm yields. Finally, as discussed below, the new seed
varieties that sparked the Green Revolution in Asia and Latin America are
poorly suited to African farming conditions.
Very High Disease Burden
Africa carries a disease burden that is unique in the world in its severity.
In recent years the most prominent disease in the region has been HIV/
AIDS. Tropical sub-Saharan Africa's HIV prevalence in 2001 was 7.3 percent;
in every other world region it is below 1 percent. Africa is also home
to numerous endemic tropical diseases, especially vector-borne diseases.
Among these, malaria is by far the most consequential. Of the 1 million to
3 million malaria-related deaths every year, it is estimated that 90 percent
occur in sub-Saharan Africa, the great majority of them among children.'8
Many casual observers mistakenly assume that because the United States
and Europe once were burdened by malaria and got rid of it, Africa's
ongoing malaria crisis is a symptom of its poverty and weak institutions
rather than a deep cause. In other words, many ask how malaria could have
16. World Bank (2003a).
17. Global Environmental Monitoring and Research Center (1995).
18. World Health Organization and United Nations Children's Fund (2003).
134 Brookings Papers on Economic Activity, 1:2004
played any more of a causal and intractable role in Africa's development
than it did in the southern United States, southern Europe, and other
regions that have since eliminated the disease. The answer lies in the
disease's ecology.
Africa's malaria is much less tractable than that in other regions for a
combination of climatic and biological reasons. The continent' s temperatures,
mosquito species, and humidity give Africa the world's highest
malaria burden, as captured in the malaria ecology index developed by
Anthony Kiszewski, Jeffrey Sachs, and others; this index is a measure
of the strength of malarial transmission based solely on ecological factors.'9
Unlike in other parts of the world, Africa's malaria mosquitoes
are almost exclusively human-biters, and this fact enhances the chain of
human-to-human transmission. Africa is also the only major world region
where the most malignant human malaria species, Plasmodium falciparum,
predominates.20 The combination of high year-round temperatures,
adequate precipitation for mosquito breeding, and human-biting
anopheles mosquitoes has made Africa the global epicenter of malaria
from time immemorial. The genetic code of African populations provides
evidence: the sickle-cell hemoglobin allele, which partially protects individuals
against malaria when inherited heterozygously, is extraordinarily
prevalent in Africa compared with other regions of the world.2'
Malaria contributes to a classic poverty trap. With enough investment,
even Africa's high malaria morbidity and mortality could be controlled,
although not eliminated, with current technologies. But that would
require substantially more money than Africa can afford. Thus Africa is
too poor to control malaria, and meanwhile malaria reduces productivity,
frustrates foreign investment, and (by contributing to very high child mortality
rates) delays or stops the demographic transition, thereby helping to
keep Africa poor.
To highlight Africa's unique challenges in terms of high transport
costs, low agricultural productivity, and high disease burden, table 4 presents
difference-of-means tests for a series of geographic variables and
results of simple regressions with agricultural productivity, infrastructure,
19. Kiszewski and others (2004).
20. World Health Organization (1992).
21. The Hardy-Weinberg equation can be used to link the high prevalence of the hemoglobin
S allele and the historically high death rates from malaria in Africa.
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 135
Table 4. Selected Geographic and Health Indicators for
Tropical Sub-Saharan Africa
Percent except where stated otherwise
Tropical Rest of Level of
sub-Saharan developing statistical
Indicator Africa a world average b significanceC
Population distribution
Share of population 24.9 66.3 1 percent
within 100 kilometers of coast
Share of population in 62.0 34.9 1 percent
tropical ecozones
Share of population in subhumid and 81.9 38.7 1 percent
arid ecozones
Share of population living at low 45.2 26.5 1 percent
densityd
Agriculture
Irrigated land as share of total 0.5 10.6 1 percent
agricultural land
Cereal yield (kilograms per hectare) 1,102.0 2,364.6 1 percent
Fertilizer consumption (100 grams per 95.0 1,606.2 1 percent
hectare of arable land)
Infrastructure
Paved roads (kilometers per 1,000 <0.1 4.2 Not
population) significant
Traditional fuel use as 76.3 29.2 1 percent
share of total energy usee
Health
Index of malaria ecology' 13.0 2.5 1 percent
Infant mortality rate 107.6 43.7 5 percent
(deaths per 1,000 live births)
Under-five mortality rate 176.9 58.1 1 percent
(deaths per 1,000 live births)
Total fertility rate (number of children)9 5.4 3.4 5 percent
Sources: World Bank (2003a); Center for International Earth Science Information Network, Columbia University (2002);
Kiszewski and others (2004); and authors' regressions.
a. Thirty-three countries listed in table 2. Data are averages not weighted by population.
b. Former republics of the Soviet Union are excluded. Data are averages not weighted by population.
c. For the population distribution variables, statistical significance of the difference in the two averages; for other variables,
statistical significance of the dummy variable in the following regression:
dependent variable = a + b, (log of 2001 GDP per capita at PPP) + b2 (dummy variable for tropical sub-Saharan Africa).
A tobit regression is used when the dependent variable is a percentage; otherwise ordinary least squares is used, with the dependent
variable expressed as a natural logarithm.
d. Percent of the population living in areas with density below 50 persons per square kilometer.
e. Estimated consumption of fuel wood, charcoal, bagasse (sugar cane waste), and animal and vegetable wastes.
f. From Kiszewski and others (2004). Index measures the stability of malaria transmission based on ecological factors and the
biological properties of the anopheline (mosquito) species of that region. The variable is measured at a highly disaggregated level
and averaged to the country level using population weighting.
g. Number of children that a woman would have borne if she survived her childbearing years and bore children in accordance
with age-specific fertility rates.
136 Brookings Papers on Economic Activity, 1:2004
and health measures as dependent variables. The geographic variables
show a statistically significant difference of means between sub-Saharan
Africa and the rest of the developing world, and the agriculture, infrastructure,
and health variables are significantly worse in Africa even after
controlling for income per capita.
Adverse Geopolitics
On top of the structural challenges already mentioned, Africa suffered
brutally at the hands of European powers for almost five centuries, and
the record with Arab powers was little better. A massive slave trade
helped undermine the formation of nation-states and may have depopulated
Africa's coastal regions. In the nineteenth century the slave trade
was replaced by direct colonial rule and a century of exploitation by European
imperial powers. At the end of the colonial period, most of Africa
was left with tragically little infrastructure and education, far less than in
postcolonial Asia.22 In contrast to much of Asia, Africa's borders were
drawn up without regard to ethnic and political units, with little infrastructure
connecting them. In addition, differences in colonial languages
and customs further fragmented many parts of Africa, particularly in West
Africa. Independence, ironically, was followed by the continued manipulation
of African countries as proxies in the cold war and continued depredations
by rich countries in exploiting Africa's natural resources. No doubt
many African leaders played into these abuses and exacerbated them
through corrupt and authoritarian rule, but if it is true that these leaders
hanged themselves and their fellow citizens, the rich countries often provided
the rope.
22. It is worth quoting at length an excellent paper by Torero and Chowdhury (2004,
p. 4): "Unlike Asia and Latin America, Africa inherited a highly dispersed and unevenly
distributed infrastructure from its colonial past. There was little, if any, improvement of
infrastructure during the colonial era, and, according to [Baland and] Platteau (1996), 'in
some important respects, it can even be said that colonial policy reinforced the handicaps of
SSA' (p. 200). The limited infrastructure that was built during that era was driven by the
objective of connecting natural resources to export markets. For example, Platteau noted
that 'two-thirds of the African railways built in the colonial period connected mines to a
coastal harbor' (p. 200). The rest of the continent was virtually ignored and, according to
Boserup (1981), 'only the Union of South Africa with mass immigration of Europeans had
more than six meters of railways per square kilometer in 1970 and six countries had no railways
at all' (p. 148)."
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 137
Even when the major powers were not exploiting Africa, they were
rarely helping it, or at least not very much. Africa is not a strategic focus
of either the United States or Europe, in contradistinction, for example,
to the United States' relationships with East Asia, the Middle East, or
Latin America, and Western Europe's relationships with Eastern
Europe, the Middle East, and North Africa. Sub-Saharan Africa is too far
away and too negligible a market to elicit a priority concern. The result is
that whereas Europe, Asia, Latin America, and the Middle East have all,
at times, been the object of U.S. financial largesse, Africa has not. Africa
has also ranked as a low priority in both trade and debt negotiations. Following
the 1982 developing country debt crisis, Latin American countries
received debt cancellation well before Africa did, and trade
discrimination against Africa has persisted for decades despite repeated
promises to remove limits on Africa's access to U.S. and European markets.
Even new initiatives to expand access to the U.S. market, such as
the African Growth and Opportunity Act, remain heavily constrained by
rules-of-origin and other limits to African exports.
Slow Diffusion of Technology from Abroad
Africa has been the world's great laggard in technological advance,
notably in the areas of agriculture and health. For instance, most of the developing
world experienced a Green Revolution-a surge in crop yields in
the 1970s through the 1990s as a result of scientific breeding that produced
high-yield varieties (HYVs), combined with increased use of fertilizers
and irrigation. Africa's uptake of HYVs was the lowest in the
developing world. The reasons are very clear. The Green Revolution HYVs
were designed for the conditions of Latin America and Asia and did not
easily transfer to the agronomic and economic conditions of Africa's rainfed,
fertilizer-scarce, subhumid and arid tropics. Whereas HYV research
focused mainly on wheat, maize, and paddy rice, Africa produces maize
but little wheat and paddy rice. It depends much more on sorghum, millet,
and tubers (cassava, coco yams, and sweet potatoes). As discussed
above, high transport costs and the combination of unfavorable geohydrology
and topography hindered increases in the use of fertilizers and
irrigation. The absence of a Green Revolution in Africa had clear implications
for food productivity, as shown in table 5. Sub-Saharan Africa has
by far the lowest cereal yield per hectare of any major region and has
Table
5.
Agricultural
Technology
and
Productivity
by
Developing
Region
Average
annual
Contribution
of
Average
annual
growth
in
food
Share
of
area
planted
to
crop
genetic
Cereal
yield
growth
in
production
modern
varieties
(percent)
improvement
(kilograms
cereal
yield,
per
capita,
to
yield
growth,
per
hectare)
1980-2000
1980-2000
Developing
region
1970
1980
1990
1998
1960-98
2000
(percent)
(percent)
Sub-Saharan
Africaa
1
4
13
27
0.280
1,111.6
0.7
-0.01
Asia
13
43
63
82
0.884
3,662.4
2.3
2.3
Latin
America
8
23
39
52
0.658
2,809.2
1.9
0.9
Middle
East
and
North
Africa
4
13
29
58
0.688
2,659.9
1.2
1.0
Sources:
Evenson
and
Gollin
(2003);
World
Bank
(2003a);
FAOSTAT
data
base
(available
at
apps.fao.org/default.jsp).
a.
All
countries
in
sub-Saharan
Africa,
except
in
last
three
columns,
which
refer
to
the
thirty-three
countries
listed
in
table
2.
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 139
experienced the slowest gain in yields during the last two decades. Indeed, it
is the only major region to have recorded a (slight) decline in food production
per capita during 1980-2000.
In highlighting these points, we are certainly not arguing for any
geographical determinism. Indeed, the whole point of the paper is the
opposite: Africa's structural impediments can be overcome, if they are
compensated by an intensive investment program that directly confronts
the continent's high transport costs, low agricultural productivity, heavy
disease burden, colonial legacy of poor educational attainment, and the
like. We are arguing, however, that Africa's structural problems help to
account for its current trap. Each of the factors mentioned has made it
much harder for Africa to lift itself up by its bootstraps. Rural and inland
Africa, plagued by disease, huge transport costs, and low agricultural yields,
was unable to generate enough of an economic surplus above survival
levels that could be invested sufficiently to overcome these conditions.
Poor governance no doubt complicated the task-many natural resource
rents were squandered-but the poor governance itself was as much symptom
as cause.23 To break out of the poverty trap, Africa needs help, and it
has not yet gotten it in the form and to the extent that is necessary.
The delay in getting this help, moreover, has greatly complicated the
task that lies ahead. Because the world has waited so long to help Africa
break free of its structural constraints, the situation is vastly more complicated
today than even a generation ago. First, there is AIDS, a pandemic
unrivaled in ferocity in modern history, which the rich countries have
allowed to run almost unabated in Africa since its recognition twentythree
years ago. Second, the massive population growth in rural areas has
led to a sharp fall in the size of the average household farm. In turn, the
increased population pressure has not led to agricultural intensification as
Ester Boserup predicted,24 but has caused massive environmental degra23.
Even in clear cases of squandered wealth, such as Nigeria with its oil earnings, the
extent of the loss is often exaggerated. Nigeria is not, contrary to popular view, a rich country
that squandered vast wealth. Rather, it is a poor country that squandered a modest inheritance.
A simple calculation explains why. Nigeria produces around 2 million barrels of oil
a day, at a price of roughly $25 per barrel in recent years, or $50 million a day. Perhaps
60 percent of this, or $30 million a day, stays with Nigeria after production sharing with
foreign companies and covering the costs of extraction. Given that there are approximately
130 million Nigerians, oil revenue would be around 23 cents per Nigerian per day, or
around $84 a year. Nigeria, in short, is not oil rich on a per capita basis.
24. Boserup (1965).
140 Brookings Papers on Economic Activity, 1:2004
Table 6. Decomposition of Growth by Region, 1980-90 and 1990-2000
Contribution to growth in output
Growth in output
per worker (percentage points)
per worker Physical capital Education Totalfactor
Region and period (percent a year) per worker per worker productivity
Africaa
1980-90 -1 .1 -0. 1 0.4 -1.4
1990-2000 -0.2 -0.1 0.4 -0.5
Latin America
1980-90 -1.8 0.0 0.5 -2.3
1990-2000 0.9 0.2 0.3 0.4
Middle East
1980-90 1.1 0.6 0.5 0.1
1990-2000 0.8 0.3 0.5 0.0
East Asia
1980-90 4.4 2.4 0.6 1.3
1990-2000 3.4 2.3 0.5 0.5
China
1980-90 6.8 2.1 0.4 4.2
1990-2000 8.8 3.2 0.3 5.1
Industrial countries
1980-90 1.8 0.7 0.2 0.9
1990-2000 1.5 0.8 0.2 0.5
Source: Bosworth and Collins (2003).
a. Nineteen African countries with populations above 1 million for which data were available.
dation instead. In Boserup' s theory, higher population density should have
led to intensive cultivation, with large investments in fertilizers, water
management, seed breeding, and the like. Instead soil nutrients have been
depleted, with hardly any replacement by chemical fertilizers or nitrogenfixing
plants, and forests have been cut down, yet food production remains
at subsistence levels, with almost no modern inputs.
Some Growth Accounting
Africa's poverty trap is evident empirically through simple growth
accounting. We look at this in two ways. The first is the standard Solow
decomposition of growth, as recently implemented on a cross-country
basis by Barry Bosworth and Susan Collins.25 The second is the AK
25. Bosworth and Collins (2003).
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 141
Table 7. Economic Growth Predicted from National Saving, Population Growth,
and Capital Consumption, by Developing Region, 1980-2001
Percenta
Gross Consumption
national offixed
saving as capital as Annual
share of share of growth in output
gross gross per capita
national Growth in national
Region incomea populationa incomea Predicted Actual
Tropical sub-Saharan 11.1 2.7 9.9 -2.3 -0.4
Africab
South Asia 20.0 2.0 8.7 1.8 3.3
Latin America 18.7 1.8 9.8 1.2 0.4
East Asia and 35.1 1.3 9.6 7.2 6.4
Pacific
Middle East and 23.5 2.4 9.2 2.3 1.0
North Africa
Source: Authors' calculations using data from World Bank (2003a).
a. Annual average across countries and years, weighted by population.
b. Countries listed in table 2, except Dem. Rep. of Congo and Liberia, for which relevant data are unavailable.
model approach, which assumes a fixed incremental capital-labor ratio
and which does not require us to construct a capital stock. Both methods
give the same result: Africa has not been saving enough to keep up with
the requirements of capital widening. The Bosworth-Collins decomposition
is shown in table 6. We see that Africa is the only region to have
experienced two consecutive decades of declining output per worker.
According to Bosworth and Collins, Africa's shortfall is the result of both
a declining capital-worker ratio and an outright decline in total factor productivity
(which could reflect, for example, the effects of environmental
degradation, not measured in the capital stock, on crop yields).26
The implementation of the AK model for the period 1980-2001 is
shown in table 7. There we assume that A = 1/3, or an incremental capitaloutput
ratio of 3, which is a typical estimate in the development literature.27
According to the AK model, and ignoring growth in total factor
productivity, we should expect the growth rate to equal sA - (n + d). With
26. Bosworth and Collins (2003).
27. Note that A is the ratio of Y to K, so that 1/A is the ratio of K to Y. Kuznets (1966),
for example, assumes KIY = 3 in his discussion of saving, investment, and growth. The
unweighted average value of KIY in the Bosworth-Collins data for African countries in
2000 is around 2.5.
142 Brookings Papers on Economic Activity, 1:2004
Figure 6. Growth in Gross National Income by Developing Region,
Actual and Predicted, 1980-2001
Predicted (percent a year)
East Asia and Pacific
6
4
Middle East and North Africa
2 South Asia
Latin America
0
-2 t | Tropical Sub-Saharan Africa y = 1. 1 128x - 0.3421
R 2=0.815
I I I I I I I I I I
-2 0 2 4 6
Actual (percent a year)
Source: World Bank (2003a) and authors' calculations using model described in the text.
A = 1/3, s equal to the national saving rate, n equal to the population
growth rate, and d equal to capital consumption as a percentage of GNP
divided by 3 (to give an estimate of capital consumption as a share of the
capital stock), we see that Africa is indeed predicted by the simple AK
model to have a negative growth rate. The combination of Africa's low
domestic saving rate and high population growth rate is just as we
expected. In fact, the AK model does remarkably well in accounting for
the variation across the five regions in growth in income per capita, as
shown by figure 6.
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 143
The situation is even worse than it looks, however, because the national
income accounts data almost surely, and substantially, overestimate Africa' s
true saving rate. To a significant extent, Africa is living off of its natural
capital but counting the resource depletion as income. Africa has been
cutting down its rainforests to make room for new farmland and to provide
fuel wood and timber, but the deforestation is not counted as a loss
of natural capital. Farmers have been depleting the soils of nutrients
by growing crops without fertilizers, but the loss of soil nutrients is not
counted as a loss of capital. Countries depleting their mineral reserves
(for example, Zambia' s copper) are counted as earning income rather than
converting one form of capital, mineral deposits, to another, financial
assets. Sooner or later these forms of resource depletion will have to come
to an end, and African real consumption will fall, perhaps significantly.
Put differently, Africa's productive capital stock per person is falling even
faster than was suspected once we take into account, even imperfectly, the
decline of a number of forms of natural capital.
There is no readily accepted practical methodology for correcting measured
national saving rates for resource depletion effects. We therefore
use the World Bank's valiant, but very preliminary, attempt, adding in
one calculation of our own.28 In the World Bank's corrected saving measures,
the saving rate taken from the national accounts is augmented by
expenditure on education (which is counted as consumption in the
national accounts but should be counted as investment in human capital),
and then reduced according to estimates of the economic costs of deforestation,
energy depletion, and mineral depletion. This corrected saving
rate is shown in the right-hand column of table 8. We then make one more
correction for tropical sub-Saharan Africa, by measuring the estimated
annual loss to Africa's soil of three macronutrients-nitrogen, phosphorus,
and potassium-which are currently being depleted with each harvest.
Sooner rather than later, this loss of soil nutrients will have to be
made up by the addition of nutrients through chemical or biological
means. If we assume that the soil nutrients are replaced through inputs of
chemical fertilizers purchased at market value in Africa, the annual cost
of replacing the soil nutrients is around 2 percent of Africa's gross
national income (GNI). We accordingly lower the measured saving rate
by that amount. In sum, whereas Africa' s measured national saving rate is
28. Hamilton and Clemens (1999).
144 Brookings Papers on Economic Activity, 1:2004
Table 8. Gross National Saving Rates by Developing Region, 1980-2001
Percent of gross national income
Adjusted for human
capital investment and
Region Unadjusted resource depletiona
Tropical sub-Saharan Africab 11.1 3.0 (1 .0)c
South Asia 20.0 18.7
Latin America 18.7 15.7
East Asia and the Pacific 35.1 29.3
Middle East and North Africa 23.5 9.1
Sources: World Bank (2003a) and authors' calculations using data from Henao and Baanante (1999) and African Agricultural
Market Information Network data base, accessed at www.afamin.net/regionalenglish/reg-mis-en.asp on March 8, 2004.
a. Gross national saving plus education expenditure, minus energy, mineral, and net forest depletion.
b. Thirty-three-country sample listed in table 2. Averages are weighted by population.
c. When soil nutrient depletion is calculated with the 1999 sub-Saharan Africa nutrient balance midpoint of 60-100 kilograms
of nutrients (nitrogen, phosphorus, potassium) per hectare, the result is the first figure shown. When nutrient depletion indicators
and fertilizer prices are used to calculate soil depletion (around 2 percent of GDP), adjusted gross saving is reduced to 1.5 percent
of gross national income, and adjusted net saving is the figure in parentheses.
around 11.1 percent of GNI, the actual saving rate net of resource depletion
might be only around 1.0 percent of GNI. Whatever the precise number,
the conclusion is that Africa is failing to maintain its capital-labor
ratio.
A Big Push to End Africa's Poverty Trap
If Africa is caught below kT, and therefore is stuck in chronic low or
negative growth, the main policy implication is to raise k above kT, which
in turn would unleash self-sustaining growth by raising A and s and lowering
n. We propose to increase the capital stock in one step, as it were,
through a large, well-targeted infusion of foreign assistance. In other words,
we are arguing not for endless flows of increased aid, and not for aid as
simple charity, but rather for increased aid as an exit strategy from the
poverty trap. For those who fear that aid increases dependency, our response
is that aid that is ambitious enough would actually end Africa's
dependency. Moreover, we see no other likely successful strategy for ending
Africa's poverty trap.
In the short run we think of aid as augmenting national saving by a
flow of external saving, so that the growth equation becomes (s + a)A -
(n + d), where a is official development assistance as a share of national
income of the recipient economy. In our proposal this flow of aid is targeted
to a particular set of priority investments, and specifically public
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 145
sector investments, so that the aid cannot be used for consumption. The
recipient country would commit to use the aid not to replace but rather
to augment existing spending in the targeted areas. We believe that, with
some care in designing the donor-recipient process, such a commitment
can be enforceable through improved monitoring of budget processes and
expenditures, perhaps with the help of local nongovernmental organizations.
The results of debt relief under the Heavily Indebted Poor Countries
initiative, in which the fiscal resources made available by reduced debt
servicing were similarly targeted at predetermined areas of public spending
(especially social spending), have been highly encouraging. In fact,
the fear that increased aid would simply allow the recipient country to
decrease its own efforts is not a major concern in our view, because the
rates of domestic public investment in the targeted areas are already so
low that there is in fact little to substitute for. Of course, a reasonable
level of governance in the recipient country is a necessary precondition
for this process to work.
If aid is to end the poverty trap, it must be large enough and maintained
long enough to raise k above kT. Small or short-lived flows of aid will
accomplish little, since they, too, will be overwhelmed by rapid population
growth, so that (s + a)A - (n + d) will remain negative. This basic
point is important and widely misunderstood. It is often said that past aid
to Africa has little to show for it. In fact, there has been too little aid to
make a difference. Moreover, the aid has been very poorly targeted mainly
toward the donors' foreign policy clients (such as President Mobuto Sese
Seko of Zaire) rather than development purposes.
To use a simple analogy, consider a large forest fire with a tendency to
spread. Suppose that only one fireman is initially sent to fight the blaze,
and he is overwhelmed. The critics might then say, "Why should we send
more firemen? We've already seen that they are ineffective!" In dealing
with the control of any epidemic process, a proper scale of intervention is
needed. Similarly, aid has to be large enough to ensure that (s + a)A is in
fact raised above n + d. The effectiveness of aid has been demonstrated in
several well-governed African countries. For example, Mozambique has
effectively used comparatively generous aid receipts of roughly $50 per
capita to finance infrastructure investments, the provision of social services,
and the demobilization of the armed forces.29 In Tanzania progress
29. Foster (2002).
146 Brookings Papers on Economic Activity, 1:2004
in improving human and physical capital stalled after donors reduced their
high aid disbursements during the 1980s and early 1990s.30
Any aid strategy must confront many complexities, especially regarding
the conditions under which aid is provided. Unconstrained aid flows would
probably be consumed rather than invested. The strategy needs to be
designed to ensure that the aid is properly invested, and there must be a
credible mechanism for enforcing the strategy over a relatively long period.
We first take up the design of the investment program and then, in the final
section of the paper, consider the recipient-donor enforcement mechanisms.
Our specific recommendation is to use the Millennium Development
Goals for 2015 as the intermediate targets for the investment program. In
other words, we argue that aid flows should be designed to enable otherwise
well-governed African countries to meet the MDGs. There are both
pragmatic and substantive reasons for this choice. The pragmatic reason
is that the world has already promised to do just this. The MDGs are the
only international goals that can motivate a complicated international
machinery of intergovernmental cooperation and resource transfers. But
there are deeper reasons as well. The MDGs in fact constitute a very reasonable
balance of concerns and are the product of a decade or more of
international debate and deliberation on economic development. They
reflect the fact, for example, that the escape from poverty requires investments
in both human capital and physical capital, and that poverty itself is
multidimensional, involving not only income but also lack of food security,
health, education, gender equality, environmental management, and
access to basic amenities. The MDGs are also flexible enough to support
a well-constructed growth strategy. For example, they call for halving the
rate of income poverty by 2015, but they do not specify exactly how this
is to be done. That leaves open the possibility for sound analysis of the
underlying growth dynamics and allows investment targets to be introduced
in several areas (such as family planning, energy systems, and rural
roads) that the MDGs do not explicitly mention, but that are nonetheless
key inputs to a sound public investment strategy to end the poverty trap.
One final preliminary thought is important. The MDGs are an intermediate
target, not a final goal. After all, even if met, they will still leave
in place an enormous amount of extreme poverty. Indeed, it is quite likely
that, whatever is achieved in terms of the MDGs, many African coun30.
Foster and Keith (2003).
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 147
tries will continue to require large-scale aid flows for another decade or
more beyond 2015, since they will not have reached the threshold for selfsustaining
growth. We are not arguing that the MDGs are sufficient for
escaping poverty. We are arguing that they are a sensible, internationally
agreed way station on the road to eliminating extreme poverty and ending
Africa's poverty trap.
Designing a Strategy to Meet the MDGs
Ending the poverty trap in Africa and meeting the MDGs will require
a comprehensive strategy for public investment in conjunction with improved
governance. This section lays out such an investment strategy
focusing on specific interventions-defined broadly as the provision of
goods, services, and infrastructure-which we have grouped into seven
areas in appendix A. A detailed background paper of the UN Millennium
Project offers more information on both the general strategy and the various
sets of interventions.3'
Most of the interventions are expected to have effects on several goals.
For example, improved education leads to many benefits, including reductions
in income poverty, hunger, and child mortality. Conversely, meeting
each goal (poverty, hunger, education, health, and the others) requires
multiple interventions. This point is stressed in appendix B, which summarizes
how each goal is affected by multiple sets of interventions. The
combination of appendixes A and B approximates an "MDG production
function," describing the set of inputs needed to achieve the desired set
of outputs. Of course, this is a purely qualitative depiction at this stage;
we provide a quantification of the inputs in the later analysis.
In designing the recommended package, we have given careful consideration
to the issue of gender equity, that is, to an improved economic and
social status for Africa's women. To a great extent, women are the farmers,
caregivers, and child raisers of Africa. They perform an incredible
amount of physical labor every day merely to keep their households alive.
Fetching fuel wood and water, caring for the farm animals, plowing the
fields, and caring for the sick and dying in the AIDS pandemic are all
backbreaking tasks disproportionately borne by Africa's women. Many
31. UN Millennium Project (2004).
148 Brookings Papers on Economic Activity, 1:2004
Table 9. Poverty, Agricultural Employment, and Urbanization in
Tropical Sub-Saharan Africa
Percent
Share of
Share of total Share of
poor employ- population
Poverty ratea living ment in in urban
in rural agriculture, areas
Country National Rural Urban areas a,b 1990 1990
Angola n.a. n.a. n.a. n.a. 74.5 33.5
Benin 33.0 n.a. n.a. n.a. 63.5 41.5
Burkina Faso 45.3 51.0 16.5 94.3 92.4 16.2
Burundi 36.2 n.a. n.a. n.a. n.a. 8.6
Cameroon n.a. n.a. n.a. n.a. 69.7 48.0
Central n.a. n.a. n.a. n.a. 80.2 40.8
African Rep.
Chad 64.0 67.0 63.0 80.1 83.2 23.5
Congo, Dem. n.a. n.a. n.a. n.a. 67.8 n.a.
Rep. of
Congo, Rep. of n.a. n.a. n.a. n.a. 48.7 64.5
C6te d'Ivoire 36.8 n.a. n.a. n.a. 60.0 43.2
Eritrea 53.0 n.a. n.a. n.a. 80.5 18.4
Ethiopia 44.2 45.0 37.0 86.3 n.a. 15.2
Ghana 31.4 34.3 26.7 70.1 n.a. 35.8
Guinea 40.0 n.a. n.a. n.a. 87.2 27.1
Kenya 42.0 46.4 29.3 74.8 19.1 32.3
Liberia n.a. n.a. n.a. n.a. 72.3 44.3
Madagascar 71.3 76.7 52.1 76.5 78.2 28.8
Malawi 65.3 66.5 54.9 87.2 n.a. 14.4
Mali n.a. n.a. n.a. n.a. 85.8 29.5
Mauritania 46.3 61.2 25.4 57.7 55.2 56.3
Mozambique 69.4 71.3 62.0 71.1 82.7 30.8
Niger 63.0 66.0 52.0 83.7 7.8 20.1
Nigeria 34.1 36.4 30.4 60.7 n.a. 43.1
Rwanda 51.2 n.a. n.a. n.a. 91.7 6.1
Senegal 33.4 40.4 n.a. 64.6 76.7 46.7
Sierra Leone n.a. n.a. n.a. n.a. 67.4 35.9
Somalia n.a. n.a. n.a. n.a. 75.3 27.1
Sudan n.a. n.a. n.a. n.a. 69.5 35.1
Tanzania 41.6 49.7 24.4 82.3 n.a. 31.1
Togo n.a. n.a. n.a. n.a. 65.5 32.8
Uganda 55.0 n.a. n.a. n.a. n.a. 13.8
Zambia 72.9 83.1 56.0 68.9 74.7 39.5
Zimbabwe 34.9 48.0 7.9 90.0 24.3 34.6
All tropical 48.4 56.2 38.4 76.6 67.5 31.8
sub-Saharan
Africa
Source: World Bank (2003a) and authors' calculations.
a. Data are for the most recent year between 1990 and 2001 for which data are available.
b. Calculated as rural poverty rate x (100 - urbanization rate)/national poverty rate.
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 149
of the recommended investments, such as improved access to water supply,
modern cooking fuels, enhanced transport services, and improved soil
nutrients, will have a special benefit for women.
The first area of interventions focuses on raising rural productivity,
because three-quarters of Africa's poor live in rural areas, and an overwhelming
percentage are employed in agriculture, as shown in table 9. In
particular, the investments in farm productivity will increase rural incomes
and reduce chronic hunger, which in Africa is predominantly caused by
insufficient agricultural productivity. Additional interventions for roads,
power, improved cooking fuels and stoves, and water and sanitation will
provide a basis for higher productive efficiency.
The second area addresses Africa' s extraordinary disease burden, widespread
micronutrient deficiencies, and extremely high fertility rates by
focusing on health, nutrition, and family planning. This package of interventions
includes the medical interventions needed to improve child and
maternal health; prevent the transmission of and provide treatment for
HIV/AIDS, tuberculosis, and malaria; improve nutrition; and provide family
planning services. We include an entire package of family planning
and reproductive health services, aimed at enabling a significant reduction
in Africa's very high total fertility rates. We believe that access to family
planning, contraceptive services, girls' education, women's empowerment,
and off-farm employment opportunities for young women can lead to a
dramatic reduction in the total fertility rate in just a few years.
The third area of interventions aims for universal completion of primary
education. In designing this package of interventions, particular attention
needs to be paid to increasing girls' completion rates through additional
demand-side interventions such as incentive payments to poor households
to encourage them to keep their daughters in school. We also argue that
secondary school enrollment needs to be increased, because the returns to
education now and especially in the future will increasingly depend on secondary
education. Moreover, large numbers of secondary school graduates
will be required to help deliver the other MDG intervention packages; for
example, secondary school graduates will be needed as community health
workers and agricultural extension workers. Targeted adult literacy programs
that are designed to raise educational attainment among the adult
population will complete the investments in human capital.
The fourth area of interventions addresses challenges specific to urban
areas. Throughout sub-Saharan Africa the large cities have not yet gener-
150 Brookings Papers on Economic Activity, 1:2004
ated internationally competitive manufacturing or service-based industries.
In conjunction with abject rural poverty, which fuels rural-urban
migration, the lack of urban jobs has led to extremely high levels of urban
poverty, which are rising across much of the continent. An MDG-based
urban strategy focuses on urban infrastructure and services (electricity,
transport, water and sanitation, waste disposal, and so forth) and slum
upgrading. Successful urban development and the creation of viable
export industries across Africa are contingent on improving access to rich
countries' markets, particularly for apparel and light manufacturing, and
the flexibility to use targeted industrial policies as needed.
The fifth area of interventions focuses on the mobilization of science
and technology. Currently, tropical sub-Saharan Africa produces roughly
twenty times fewer patents per capita than the average for the rest of the
developing world, and the region has only 18 scientists and engineers per
million population, compared with 69 in South Asia, 76 in the Middle
East, 273 in Latin America, and 903 in East Asia. We stress the need for
increased investment in science advising, higher education, and research
and development targeted at Africa's specific ecological challenges in
food production, disease, nutrition, construction, energy, and other areas.
The sixth area focuses specifically on interventions for achieving gender
equity that are not included as part of the other intervention areas. This
includes ensuring full reproductive health rights and access to services, as
well as the guarantee of equal property rights and access to work, backed
up by affirmative action to increase political representation. Of particular
concern in many parts of sub-Saharan Africa are persistently high levels
of violence against women and girls; this violence needs to be confronted
with legislative and administrative changes.
The seventh area of interventions focuses on regional integration as a
sine qua non for economic growth. The fifteen of the forty-nine countries
in sub-Saharan Africa that are landlocked have little chance to develop unless
they have ready access to the coast with efficient, low-cost infrastructure.
Moreover, from a global perspective, the African countries taken individually
are very small markets. Regional integration will raise the interest
of potential foreign investors by increasing the scope of the market that
accompanies an operating presence in Africa. Regional integration is also
important in achieving scale economies in infrastructure networks, such
as power grids, large-scale electricity generation, road transport, rails, and
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 151
telecommunications, and in eliciting increased R&D on problems that are
specific to Africa's ecology but that extend beyond any single country,
such as public health, energy systems, and agronomy.
As an overarching matter affecting each of these seven categories of
interventions, we note that Africa will require large numbers of teachers,
nurses, doctors, agricultural extension officers, infrastructure specialists,
scientists, and so forth-experts who are in very short supply across the
continent. Hence each of the seven investment areas relies on a substantial
scaling up of specialized human resources. In addition to expanding formal
higher education, therefore, African countries need to focus on vocational
training and short-term training courses. Already several countries
have demonstrated that community health workers or agricultural extension
officers who have undergone one-year training courses can produce
substantial improvements in the communities where they work. For example,
Ethiopia is currently training several thousand community health workers
who will be able to carry out malaria control, offer simple treatment,
and help improve households' hygiene. If extended to other areas, this
"barefoot doctor" approach could rapidly improve rural life at moderate
cost. An additional advantage from the countries' perspective is that these
vocationally trained experts are less likely to migrate abroad than their
more highly trained colleagues.
Specific Interventions to Achieve the MDGs
A full discussion of these proposed interventions is far beyond the
scope of this paper but is available from the UN Millennium Project.32
For purposes of illustration we describe here only two categories of
interventions: investments in rural productivity and in reducing child
mortality. Our aim is to show that specific investments exist that could
make enormous changes in productivity at rather low cost.
INCREASING RURAL PRODUCTIVITY. As already mentioned, sub-Saharan
Africa is the world's only major region where food output per capita has
been stagnant since 1980.33 Reversing this trend lies at the heart of any
strategy to achieve the MDGs in Africa. One key factor in this stagnation
is the declining availability of arable land per capita, which has shrunk
32. UN Millennium Project (2004).
33. World Bank (2003a).
152 Brookings Papers on Economic Activity, 1:2004
Table 10. Land under Cultivation in Selected Sub-Saharan
African Countries, 1960-99a
Hectares per capita
Country 1960-69 1970-79 1980-89 1990-99
Ethiopia 0.508 0.450 0.363 0.252
Kenya 0.459 0.350 0.280 0.229
Mozambique 0.389 0.367 0.298 0.249
Rwanda 0.215 0.211 0.197 0.161
Zambia 1.367 1.073 0.896 0.779
Zimbabwe 0.726 0.664 0.583 0.525
Source: Jayne and others (2001).
a. Data are period averages.
dramatically from 0.38 to 0.25 hectare over the past twenty years,34 mainly
because of population growth. Table 10 shows the tremendous decreases
in land under crop cultivation relative to population in agriculture in
several African countries since 1960.35
Overcoming the fall in average farm size and availability per capita of
arable land requires increased food productivity coupled with a reduction
in the growth rate of the rural population. In general, five sets of constraints
must be addressed to increase African food productivity. First, soil
fertility in Africa is low and has been declining. This is due to several factors:
lack of available soil nutrients (such as nitrogen, phosphorus, sulfur,
magnesium, and zinc), insufficient soil organic matter, poor physical structure
of the soil, poor water-holding capacity, high risk of erosion, limited
rooting depth, and lack of biodiversity.36 Second, smallholder farmers in
Africa lack access to improved seeds that are suited to their soils. Third,
limited water availability and increasingly variable precipitation patterns
pose major challenges for smallholder water management and irrigation.37
Fourth, knowledge concerning new technologies that can increase crop
yields is extremely limited. Fifth, women typically conduct the majority of
African farm work, but they frequently have limited access to farm prop34.
Centro Internacional de Agricultura Tropical, Tropical Soil Biology and Fertility,
and International Center for Research in Agroforestry (2002).
35. Jayne and others (2001).
36. Donovan and Casey (1998).
37. According to data from the Food and Agriculture Organisation (FAO), sub-Saharan
Africa includes eleven of the sixteen nations of the world that receive less than 1,000 cubic
meters of water per hectare-a situation described as "absolute water scarcity" (FAO,
1995a).
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 153
erty rights, face barriers to generating and controlling their own incomes,
and have limited access to improved agricultural technologies.
These constraints on raising rural productivity can be overcome
through a series of well-targeted interventions. Soil nutrient deficiencies
can be addressed through chemical fertilizers that help to increase crop
yields and through agroforestry techniques that use nitrogen-fixing
trees.38 One approach is to interplant such trees in young maize fields and
allow them to grow as fallows during dry seasons, accumulating 100 to
200 kilograms of nitrogen per hectare in six months to two years. In the
subhumid regions of eastern and southern Africa, this technology can
double or more the yields of maize, a staple crop.39 It can be supplemented
with investments in improved seeds adapted to African farming
environments. Suitable seeds can raise yields, reduce the time to harvest,
improve germination, reduce labor requirements, reduce vulnerability to
diseases, and improve cooking, processing, and storage qualities.
To improve crop irrigation in the arid and subhumid regions where
most Africans live, it is generally acknowledged that small-scale and
technically simple water management systems are most appropriate
because of their low cost and ease of operation and maintenance. Options
include water harvesting through the use of embankments, check dams
and small impoundment dams, shallow wells, drip irrigation, and treadle
or diesel pumps. Experience in Burkina Faso, Kenya, and Sudan shows
that rain harvested from one hectare for supplementary irrigation of
another can triple or even quadruple production.40
These physical interventions need to be augmented by agricultural extension
services that provide farmers with technical support and training.4'
Although debate continues over the most appropriate model for extension
services, there is consensus on the need for increased human resources
and research programs, and particularly on the need to improve the technical
qualifications of extension workers.42
38. Such trees include leguminous trees of the genera Sesbania, Tephrosia, Crotalaria,
and Glyricidia.
39. Sanchez (2002).
40. FAO (1996).
41. Birkhaeuser, Evenson, and Feder (1991).
42. Sasakawa Africa Association (2000). The Green Revolution relied mainly on the
training and visitation (T&V) model, whereby extension served as a transmission belt
between research centers and farmers through cadres of field-based, professional trainers.
However, this model had its limitations, including poor linkages of the research undertaken
154 Brookings Papers on Economic Activity, 1:2004
A central theme throughout the rural productivity strategy is the need for
special attention to the critical role that women play in African agriculture.
Africa' s women contribute 60 to 80 percent of the labor used to produce
food both for household consumption and for sale, yet they face enormous
cultural and legal inequities.43 Special interventions are needed to reach
them through extension services and to provide them with access to technologies
and skills that they can use to reduce the time spent in the field
and increase production. Christopher Udry, John Hoddinott, and Lawrence
Haddad show that merely allocating factors of production evenly across
plots controlled by women and men could increase plot output by 10 to 15
percent.44
Interventions outside the formal agricultural sector are also crucial for
farm productivity. Not surprisingly, education is important. The International
Food Policy Research Institute (IFPRI) estimates, for example,
that an extra year of primary education provided to women farmers in
Kenya would increase maize yields by 24 percent.45 Transport infrastructure
and services are also critical, since well-maintained feeder,
district, and national roads are crucial in order to bring products to markets
and to lower the costs of agricultural inputs, such as fertilizer and
improved seeds. In one indication of the severity of rural farmer isolation,
John Hine and John Rutter estimate that, for 51 percent of villages in
Ghana and 60 percent of villages in Malawi, the walking distance to the
nearest pickup point for motorized transport services was more than
2 kilometers; it was over 10 kilometers for 10 percent of Ghanaian villages
and 19 percent of Malawian villages.46 D. S. Spencer calculates
that, for tropical Africa to reach a road density (measured in kilometers
of roads per square kilometer of land) equivalent to that of India at the
with the needs of small farmers, inability to adapt to local conditions, and inadequate practical
knowledge on the part of the trainers; all of these shortcomings were compounded by
budgetary constraints and lack of long-term fiscal support. Increasingly, hybrid solutions
have been proposed that combine the best of the T&V model with principles of demanddriven
and farmer-led knowledge creation and management systems. However, basic principles
of increased investment in hiring and training extension workers (especially women
workers to reach women farmers), close links to research, and continual and timely training
remain critical.
43. FAO (1995b).
44. Udry, Hoddinott, and Haddad (1995).
45. IFPRI (2000).
46. Hine and Rutter (2000).
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 155
onset of the Green Revolution in 1950, the length of the road network
would have to be increased by a factor of SiX.47
Often forgotten in the rural productivity equation are modern energy
services, and in particular safe thermal energy for cooking as well as electricity
for illumination, water pumps, and refrigeration. In addition to providing
health benefits through reduced indoor air pollution, improved
cooking stoves and fuels help to preserve biomass and dung for use as fertilizer.
Subsidized access to safe fuel sources such as liquid petroleum gas
also allows women to spend less time collecting biomass fuel and cooking
and more time tending to farm production and opportunities to generate
private income.
REDUCING CHILD MORTALITY. Table 1 showed how child mortality
rates in Africa are by far the worst in the world; a total of 4 million
African children under five die each year. Remarkably, five conditions
account for approximately 95 percent of these fatalities: one quarter result
from neonatal disorders (infections, birth asphyxia and injuries, and complications
of prematurity) that occur in the first twenty-eight days of life;
22 percent from malaria; 21 percent from pneumonia; 20 percent from
diarrheal diseases; and 8 percent from HIV/AIDS.48 Although these estimates
are imprecise because of the difficulty of attributing cause of death
and the dubious quality of the underlying data, the basic message is that
controlling a limited number of preventable and treatable conditions could
bring about a dramatic reduction in child mortality.
Underlying these proximate causes of death are contributing factors
that render children susceptible. Malnutrition, for example, is a contributing
factor in at least half of all child deaths,49 because malnourished children
cannot resist common infections as readily as adequately nourished
children and, when infected, have a much higher risk of complications and
death. Even mildly underweight children are estimated to have a twofold
higher risk of death than those that are adequately nourished.50 Meanwhile,
poor indoor air quality and lack of access to drinking water and
sanitation predispose children to pneumonia and diarrhea.
Because the causes of child death are both direct and indirect, interventions
focused on saving children's lives need to both treat children
47. Spencer (1994).
48. Black, Morris, and Bryce (2003).
49. Black, Morris, and Bryce (2003).
50. Black, Morris, and Bryce (2003).
156 Brookings Papers on Economic Activity, 1:2004
who are sick and prevent children from becoming sick. Treatment interventions
typically fall within the realm of the formal health sector, but preventive
interventions necessarily span other sectors as well. Prominent
among health sector interventions to improve child health outcomes is an
integrated neonatal package that includes clean delivery, newborn resuscitation,
prevention of hypothermia, antibiotics for infection, tetanus toxoid,
breastfeeding education, and community hygiene education.5' Such a
package may reduce neonatal mortality by up to 60 percent.52 Immunizations
are another simple intervention with well-documented benefits in
the early stages of life: vaccinations against major diseases such as polio,
diphtheria, pertussis, tetanus, measles, hepatitis B, and yellow fever have
all been shown to reduce child mortality. Immunizing all children against
measles alone could save 2.3 million children in Africa within the current
decade.53
For children beyond infancy, a broad set of interventions has been established
under the UNICEF-WHO protocol known as the Integrated Management
of Childhood Illness (IMCI). This includes a series of preventive and
curative interventions that enable health workers to treat symptoms that
could have multiple underlying causes. Specific IMCI interventions include
insecticide-treated nets (ITNs) to prevent malaria; complementary feeding;54
vaccinations; micronutrient supplementation; case management of
acute respiratory infections, diarrhea, measles, malaria, and malnutrition;
treatment with iron; and antihelminthic treatment.55 This protocol has been
applied successfully in dozens of developing countries. For example, early
results from Tanzania show a significantly reduced hospital admission
rate for children in IMCI districts.56
Interventions to prevent and treat malaria are part of IMCI but deserve
special elaboration in the African context. On the prevention side, ITNs
provide both a physical and a chemical barrier to the mosquitoes that
transmit malaria, with benefits extending beyond the individual user as
51. Marsh and others (2002).
52. Bang and others (1999).
53. World Health Organization, United Nations Children's Fund, and Global Alliance
for Vaccines and Immunizations (2003).
54. Complementary feeding involves adding solid foods to the diet of breastfed infants
over six months of age to prevent nutritional deficiencies.
55. The IMCI protocol is available at World Health Organization (2001).
56. Armstrong Schellenberg and others (2004).
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 157
the chemical halo emitted by the net repels and shortens the lives of mosquitoes
in the vicinity. The impact of comprehensive ITN programs can
be substantial. A review evaluating twenty-two trials concluded that
approximately 6 lives could be saved for every 1,000 children protected
with ITNs.57 In Tanzania, C. A. Maxwell and colleagues found that ITNs
were linked to 55 to 75 percent reductions in malarial morbidity for
children aged six months to two years.58 On the treatment side, the recent
rise in malarial resistance to chloroquine and sulfadoxine-pyrimethamine
in Africa has accelerated the need to make the transition to more effective
therapies such as artemisinin-based combination treatment.59 This
treatment can enhance clinical efficacy and delay the development of
resistance.60
Of course, interventions outside of the formal health sector are also
crucial for reducing child mortality, as described in appendix B. Targeted
nutritional interventions can help enormously. For example, a UN study
demonstrated that vitamin A supplementation resulted in a 23 percent
average reduction in mortality rates among infants and children between
six months and five years old.6' Substituting clean cooking fuels for
domestic biomass fuels has been estimated to reduce acute respiratory
infections in several developing countries by approximately 30 percent.62
D. Leipziger and colleagues, in a study of forty-three developing countries,
estimated that improved access to water explains 37 percent of the
difference in child mortality between the poorest and richest quintiles
within countries.63 Simple hand washing with soap and water reduces
shigella and diarrheal disease transmission by 35 percent.64
Maternal education similarly contributes to child health. Each additional
year of a mother' s education corresponds to an estimated 7 to 9 percent
decline in her under-five child's mortality risk, although the precise
57. Lengeler (2004).
58. Maxwell and others (2002).
59. World Health Organization and United Nations Children's Fund (2003).
60. World Health Organization (2001).
61. Beaton and others (1993).
62. Reductions in the range of 21 to 44 percent were found for acute lower respiratory
infections and 24 to 64 percent for all acute respiratory infections (Ezzati and Kammen,
2002).
63. Leipziger and others (2003).
64. Kotloff and others (1999); Khan (1982).
158 Brookings Papers on Economic Activity, 1:2004
mechanisms of this relationship are not yet entirely clear.65 Access to contraception
can also have a strong impact on child mortality rates.
These interventions must be provided in an integrated manner, but one
must be careful not to think of their effects as purely additive. A critical
lesson from past experience with development programs in the health sector
is that successful scale-up and utilization of a broad range of diseasespecific
health interventions requires a functioning health system, including
an adequate complement of health workers; good management capacity
for supervision and oversight; ongoing monitoring, evaluation, and quality
assurance; and sustained outreach and education for communities,
among other features.
In a prominent recent study, G. Jones and colleagues evaluated the
impact of a broad set of health sector and non-health sector interventions
on child survival and concluded that 63 percent of annual child deaths
could be prevented by providing universal (defined as 99 percent) access
to an integrated set of interventions.66 Although these authors did not
evaluate the role of a full health management system, they did include
preventive medical interventions (such as immunization, micronutrient
supplementation, and clean delivery), curative interventions (such as oral
rehydration therapy through salt tablets, antibiotics for pneumonia and
sepsis, and antimalarial medications), promotion of breastfeeding, ITNs,
complementary feeding, and improved access to water, sanitation, and
hygiene education.67
MDG Needs Assessments for Ghana, Tanzania, and Uganda
To be aligned with the MDGs, the full intervention package outlined in
appendix A must be converted into a country-level investment plan, one
that works backward from the outcome targets to identify needed input
targets and then identifies infrastructure, human, and financial resource
requirements to meet the targets. We dub this a "needs assessment"
65. Cleland and Van Ginneken (1988).
66. Jones and others (2003).
67. Importantly, Jones and others (2003) point out that their estimate of lives saved
through known interventions may be conservative, since they evaluate only those interventions
for which good cause-and-effect evidence exists and that are "feasible" for developing
countries to implement (for example, they exclude emergency obstetric care). They do
not assess other underlying factors, such as maternal education, that have a strong although
indirect impact on child survival.
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 159
approach to the MDGs.68 Here we estimate the costs of interventions for
three countries: Ghana, Tanzania, and Uganda. We chose these countries
because all have high rates of extreme poverty, are making insufficient
progress toward achieving the MDGs, and are well governed for
their level of income per capita, as indicated by their "good" and "average"
governance ratings in table 2. These are countries where governance
is not the binding constraint.
To highlight the extent of poverty in Ghana, Tanzania, and Uganda,
table 11 summarizes available data for some key MDG indicators. Although
the underlying data are often imprecise and incomplete (a reflection of the
extreme poverty in these countries), they highlight each country's uneven
progress across the MDGs. Persistently high rates of hunger and child
mortality are particularly notable. Indeed, the fact that Ghana's child mortality
rate of 100 per 1,000 live births in 2001 is the lowest in the group
reflects how extreme poverty is in these countries. Human capital accumulation
meanwhile remains slow in Ghana and Tanzania in particular,
as reflected in their low primary enrollment rates, and more than a third
of the population in all three countries was living on less than $1 a day in
the latest available surveys.
To meet the MDGs systematically in each of these countries, we assume
that the interventions are scaled up linearly through 2015. As much as
possible, the intervention package is also adjusted according to the age
and gender structure of the population and to the rural-urban divide. To
account for population growth and other changes in a country's demographic
profile, the UN Population Division's 2002 revision of medium
population forecasts is used.69 Where more recent population data are
available at the country level, we use those.
For each intervention, such as the provision of ITNs or the hiring of
new teachers, we specify quantitative input targets that are consistent with
the MDGs. We then translate the interventions into budget lines, estimating
financial costs per year. We use local unit costs wherever possible and
regional proxies where these are not available. In some cases, unit costs
are assumed to increase as coverage rates increase. For example, teachers'
68. The UN Millennium Project needs assessment (UN Millennium Project, 2004)
draws on several works including Bruns, Mingat, and Rakotomalala (2003), World Bank
(2002), Delamonica, Mehrotra, and Vandermoortele (2001), FAO (2002), Commission on
Macroeconomics and Health (2001), and Global Water Partnership (2000).
69. United Nations (2003).
Table
11.
Progress
toward
Achieving
the
Millennium
Development
Goals
in
Ghana,
Tanzania,
and
Ugandaa
Percent
except
where
stated
otherwise
Ghana
Tanzania
Uganda
Most
MDG
Most
MDG
Most
MDG
Indicator
Earliest
recent
statusb
Earliest
recent
statusb
Earliest
recent
statusb
Share
of
population
31.4
n.a.
...
51.1
n.a.
...
55.0
n.a.
below
national
(1992)
(1991)
(1993)
poverty
line
Prevalence
of
child
27.3
24.9
Off
track
28.9
29.4
Off
track
25.5
23.0
Off
track
malnutritionc
(1994)
(1999)
(1992)
(1999)
(1995)
(2000)
Primary
net
n.a.
58.3
.
..
51.4
46.7
Off
track
n.a.
109.5
...
enrollment
rated
(2000)
(1990)
(2000)
(2000)
Ratio
of
girls
to
boys
in
n.a.
88.2
...
96.8
98.9
On
track
n.a.
88.9
...
primary
and
(2000)
(1990)
(2000)
(2000)
secondary
educatione
Under-five
mortality
126
100
Off
track
163
165
Off
track
165
124
Off
track
rate
(per
1,000)
(1990)
(2001)
(1990)
(2001)
(1990)
(2001)
Share
of
population
53
73
On
track
38
68
On
track
45
52
Off
track
with
access
to
(1990)
(2000)
(1990)
(2000)
(1990)
(2000)
improved
water
supply
Share
of
population
61
72
On
track
84
90
On
track
n.a.
79
...
with
access
to
(1990)
(2000)
(1990)
(2000)
(2000)
improved
sanitation
Source:
World
Bank
(2003a).
a.
Baseline
year
for
all
MDGs
is
1990;
where
data
for
1990
are
unavailable,
data
for
the
closest
available
baseline
year
are
used.
b.
Assumes
continuation
of
1990-2000
trend.
Target
date
for
attainment
of
goals
is
2015
except
where
noted
otherwise.
c.
As
measured
by
weight
for
age.
d.
Persons
attending
primary
school
divided
by
the
age-appropriate
population;
the
rate
can
thus
exceed
100
percent
if
persons
older
than
primary
school
age
are
attending
school.
e.
Target
date
for
attainment
of
goal
is
2005.
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 161
salaries are increased to improve retention rates. Importantly, we estimate
capital as well as operating costs, and total costs instead of incremental
costs. This implicitly assumes that all investments and service deliveries
follow best practice, regardless of current efficacy of delivery, yielding a
downward bias in our cost estimates. Since the marginal costs of a major
scale-up are unknown-they could be increasing or decreasing when
making investments in rural communities-we use average unit costs.
Next we identify key interventions that, upon implementation, would
reduce resource requirements for other interventions. The resulting cost savings
can occur through reductions in the population in need or reductions in
unit costs for other interventions. Quantifying these links is often difficult,
especially for the interventions that have numerous indirect effects, such as
girls' education. And even where clear data exist, the magnitude of the
impact may not be known for different settings or for different delivery
mechanisms. Nonetheless, our estimates include the effect of several synergies
that are likely to result in particularly large cost savings. These are a
reduction in diarrhea morbidity through expanded access to improved water
supply and sanitation, including improved hygiene behavior; a reduction in
acute respiratory infections through increased use of improved cooking
fuels; a reduction in HIV/AIDS incidence through increased condom use;
and a reduction in malaria incidence through increased use of ITNs.
As a last step in the needs assessment, we develop a basic financing
strategy for each country, distinguishing among domestic government
resources, out-of-pocket household expenditure, and ODA. We assume
that domestically financed government spending on MDG-related interventions
will rise by 4 percentage points of GDP between 2005 and 2015,
in addition to the increase in domestic resource availability resulting from
GDP growth alone. We also assume that, as a result of this package of
interventions, GDP per capita rises between 2.2 and 3.3 percent a year on
average during 2005-15, which is approximately the growth rate needed
to reduce income poverty by half at an average elasticity of headcount
poverty with respect to income per capita of -1.4.7?
70. Several studies have estimated these elasticities for low-income countries. Bruno,
Ravallion, and Squire (1998) estimate a global average of -2, whereas Besley and Burgess
(2003) cite a low value of -0.49 for sub-Saharan Africa. In the case of Uganda, the
Economic Policy Research Center estimates a value of -1.39 (Okidi, Okwi, and Ddumba,
2000); Tanzania's elasticity for basic needs poverty has been estimated at -1.45
(TAKWIMU, 2000).
162 Brookings Papers on Economic Activity, 1:2004
Household contributions are estimated based on the incentive effects
of user fees and overall ability to pay. Since user fees can play a critical
role in preventing poor people from accessing basic services, we assume
no user fees for primary school education, adult literacy programs,
improving gender equality, basic health care, nutritional interventions,
and roads.7' In some cases, however, user fees may be necessary to avoid
wasting scarce resources after basic minimum needs have been ensured.
"Lifeline tariffs," which ensure the free provision of water (or other basic
needs) up to the minimum daily requirement for personal hygiene and
cooking, but then charge for consumption above the minimum, have been
used successfully in many parts of the world to avoid wastage while
improving access to clean drinking water. We assume that households
bear some of the cost of agricultural interventions, secondary school education,
energy provision, water supply, and sanitation. To calculate
households' ability to pay, we divide the population of each country into
three segments using income shares derived from national household budget
surveys as reported by the World Bank.72 It is assumed that the poorest
segment of the population, that living below the national poverty line,
will not contribute to the cost of meeting the MDGs. The middle segment
and the top segment (with income above twice the poverty line) are
assumed to contribute larger shares of operating and capital costs.73
A core result of our needs assessment is that the investment program
required to meet the MDGs is beyond the financing means of the poorest
countries, even when one assumes that governments will devote an additional
4 percent of GNP to meeting the MDGs alongside roughly 3 percent
annual growth in income per capita and that households contribute to
the increased investments. We estimate an "MDG financing gap" as the
total cost of the package of interventions minus the contributions by the
71. "Experience has taught repeatedly that user fees end up excluding the poor from
essential health services, while at the same time recovering only a tiny fraction of costs"
(Commission on Macroeconomics and Health, 2001, p. 61). Even modest user fees have
been shown to have a strong impact on reducing effective access to these basic services,
particularly for women, young girls, and other vulnerable groups. See World Education
Forum (2000). The Education for All Dakar Goal 2 reads as follows: "Ensuring that by
2015 all children, particularly girls, children in difficult circumstances, and those belonging
to ethnic minorities, have access to and complete free and compulsory primary education of
good quality" (available at www.unesco.org/education/efa/ed_for_all/dakfram_eng.shtml).
72. World Bank (2003a).
73. Details are outlined in UN Millennium Project (2004).
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 163
domestic government and households. This is the amount that donors
would have to bear in order to finance the MDG intervention package.
Table 12 presents the total estimated resource requirements for the three
countries, based on an eleven-year scale-up period from 2005 through
2015. The results suggest that each of the three countries will be able to
finance between 41 and 48 percent of the total investment needs through
domestic sources. This translates into an average total need for external
development assistance equivalent to $47 (for Uganda) to $60 (for Tanzania)
per capita per year. Points of interest for each country are drawn from
the UN Millennium Project and summarized below:74
-Ghana's total investment needs for meeting the MDGs average
$2.0 billion a year, or $82.8 annually per capita. Of this total,
$1.2 billion is estimated for required average annual external assistance.
This need remains significantly above the $653 million in net ODA that
donors disbursed to Ghana in 2001 for MDG- and non-MDG-related
activities.
-Of the three countries in our sample, Tanzania faces the biggest
challenges in meeting the MDGs. We estimate that the country will
require the highest per capita levels of average investment and external
finance to meet the MDGs, at $100.4 and $60.2, respectively. These figures
are equivalent to an aggregate investment of $4.2 billion a year and
imply external finance of $2.5 billion a year. More than a third of these
investments are required to achieve the health MDGs, because of the
country's high prevalence of HIV/AIDS and malaria.
-Uganda is widely lauded for its progress in education and its HIV/
AIDS control program over the past decade. However, the country still
remains off track in several areas, including hunger, health, and water.
We project that the country will need to spend an average of $3.0 billion
a year to meet the MDGs, or $89.7 per capita. More than half of these
investments will need to be financed externally, requiring an average
of $1.6 billion, or $46.7 per capita, annually. An important gap in our
Uganda study is an estimate of the transport and communications infrastructure
required to provide the landlocked country with improved access
to world markets. The Northern Corridor highway to Mombasa,
Kenya, is in critical need of upgrading, as is the rail infrastructure.
74. UN Millennium Project (2004).
164 Brookings Papers on Economic Activity, 1:2004
Table 12. Estimated Total Cost and Projected Sources of Funding to Meet
Millennium Development Goals in Ghana, Tanzania, and Uganda, 2005-15a
Ghana Tanzania Uganda
Cost (orfunding Per Per Per
source) and Percent capita Percent capita Percent capita
intervention area of GDP (dollars) of GDP (dollars) of GDP (dollars)
Total CoStb
Hunger 1.9 5.9 3.2 11.5 1.7 5.9
Education 5.6 17.7 3.3 12.1 4.1 13.9
Gender equality 0.7 2.1 0.6 2.4 0.7 2.3
Health 7.9 24.8 9.6 34.8 9.4 32.2
Water supply 2.8 8.8 1.6 5.9 1.1 3.9
and sanitation
Energy 4.5 14.3 4.0 14.4 3.8 13.0
Roads 2.9 9.2 5.3 19.4 5.4 18.4
Total 26.3 82.8 27.7 100.4 26.3 89.7
Financing sources:
Household contribution
Hunger 0.3 0.9 0.5 1.8 0.4 1.2
Education 0.3 0.9 0.2 0.7 0.6 1.9
Gender equality 0.0 0.0 0.0 0.0 0.0 0.0
Health 0.0 0.0 0.0 0.0 0.0 0.0
Water supply 1.2 3.9 0.7 2.7 0.5 1.7
and sanitation
Energy 1.4 4.5 1.3 4.8 1.2 4.0
Roads 0.0 0.0 0.0 0.0 0.0 0.0
Total 3.3 10.2 2.8 10.0 2.6 8.9
Domestically financed government expenditurec
Hunger 0.5 1.7 1.0 3.5 0.7 2.3
Education 1.6 5.0 1.0 3.6 1.5 5.3
Gender equality 0.2 0.6 0.2 0.7 0.3 0.9
Health 2.2 7.0 2.9 10.5 3.6 12.2
Water supply 0.8 2.5 0.5 1.8 0.4 1.5
and sanitation
Energy 1.3 4.0 1.2 4.3 1.4 4.9
Roads 0.8 2.6 1.6 5.8 2.1 7.0
Total 7.5 23.5 8.3 30.2 10.0 34.1
Required total external budget support
Hunger 1.1 3.3 1.7 6.2 0.7 2.4
Education 3.7 11.8 2.1 7.8 2.0 6.7
Gender equality 0.5 1.5 0.5 1.6 0.4 1.4
Health 5.7 17.8 6.7 24.3 5.9 20.0
Water supply 0.8 2.4 0.4 1.5 0.2 0.7
and sanitation
Energy 1.8 5.7 1.4 5.2 1.2 4.1
Roads 2.1 6.6 3.7 13.6 3.4 11.4
Total 15.6 49.1 16.6 60.2 13.7 46.7
Source: UN Millennium Project (2004).
a. Data are averages per year. Costs for three intervention areas (environment, improving the lives of slum dwellers, and
science and technology) are still to be determined. Data may not sum to totals because of rounding.
b. Sum of household contributions, domestically financed government expenditure, and required total external budget support.
c. Government expenditure on the MDGs financed solely through domestic revenue generation; on a pro forma basis, expenditures
are allocated to budget line items based on their relative share of funding from all sources.
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 165
In addition to estimating the costs of meeting the MDGs, the bottomup
needs assessments yield detailed estimates of each country's human resource
and infrastructure requirements. For example, our analysis suggests
that Tanzania will need to increase the number of primary and secondary
school teachers from approximately 134,000 in 2005 to 215,000 in 2015.
We emphasize that these results are preliminary and likely to underestimate
the true costs of achieving the MDGs in each country, since they
assume an optimal intervention package and exclude a number of expensive
interventions, such as ports, railroads, infrastructure for water storage
and transport, urban infrastructure and slum upgrading, expanding higher
education, and information and communications technologies.75 The estimates
nonetheless provide an important first approximation of the combination
of resources and inputs required to achieve the MDGs.
Implications for Development Assistance
The external financing requirements presented in table 12 are not total
ODA needs. Table 12 includes only MDG-oriented investments and excludes
programs that are not directly linked to the MDGs but that might
be indirectly important, such as support for the judicial system or other
institutions that underpin democratic governance. To calculate the three
countries' total ODA needs, one needs to add other important forms of
ODA to directly MDG-focused ODA. One can then identify the total
incremental ODA need for each country.
Table 13 presents estimated current ODA and incremental total ODA
needs for Ghana, Tanzania, and Uganda. The first column shows aid levels
per capita in 2001. The second indicates the estimated share of that
ODA that is focused on MDG-related activities other than those addressing
income poverty, using a decomposition accounting methodology
developed by Brian Hammond of the OECD Development Assistance
Committee (DAC). The actual results produced through that methodology
were in the range of 33 to 59 percent, but because it is difficult to allocate
aid flows precisely to each MDG, and because poor aid quality (for example,
due to lack of donor coordination) often limits resource efficacy, we
assume a figure toward the low side of that range, namely, 40 percent. The
third column of the table reports the estimate of non-MDG aid per capita.
75. Appendix A lists all the interventions that have so far not been included in the
needs assessment.
Table
13.
Estimated
Official
Development
Assistance
Required
for
Ghana,
Tanzania,
and
Uganda
to
Achieve
the
Millennium
Development
Goals
Dollars
except
where
stated
otherwise
Implied
Implied
total
Estimated
current
ODA
Estimated
Implied
ODA
per
capita
share
of
per
capita
ODA
per
Implied
current
Estimated
required
as
share
ODA
per
current
ODA
going
to
capita
total
ODA
shortfall
in
GDP
per
of
estimated
capita,
going
to
MDGs
non-MDG
required
to
per
capita
ODA
per
capita
over
GDP
per
Country
2001
(percent)a
areas
meet
MDGsb
required
capitac
2004-15d
capita
(percent)
Ghana
33
40
19.8
49.1
68.9
35.9
313
22
Tanzania
37
40
22.2
60.2
82.4
45.4
362
23
Uganda
35
40
21.0
46.7
67.7
32.7
340
20
Sources:
OECD,
International
Development
Statistics,
and
authors'
calculations.
a.
See
Simon
(2003)
for
an
application
of
Brian
Hammond's
estimation
methodology
to
these
countries.
b.
UN
Millennium
Project
estimate.
See
UN
Millennium
Project
(2004)
for
details.
c.
Implied
total
ODA
per
capita
required
minus
actual
ODA
per
capita
in
2001.
d.
Average
for
the
period.
Sachs, McArthur, Schmidt-Traub, Kruk, Bahadur, Faye, and McCord 167
We assume perfect donor delivery mechanisms and so do not distinguish
between currently efficient and inefficient aid dollars. This biases upward
the estimate of current MDG-related aid and biases downward the implied
aid gap. The fourth column of the table repeats from table 12 the estimated
total external finance gaps per capita. The fifth column then
assumes that current non-MDG aid spending will remain constant as
MDG-based investments are scaled up, so that the total ODA need per
capita is the sum of MDG-targeted ODA and complementary ODA. The
sixth column presents the implied shortfall in ODA per capita, indicating
that roughly a doubling of ODA is needed in each country.
The seventh column of table 13 then presents the average GDP per
capita that is assumed by our financing model, which in turn is used to
calculate the final column, total ODA required as a share of GDP in each
country. The results, which range from 20 to 23 percent of GDP, present a
lower bound for estimates of the aid required for these three countries to
achieve the MDGs. These are lower bounds because the costing model
excludes several "big ticket" items listed above and neglects the inherent
overhead costs of managing aid delivery, which are likely in the neighborhood
of 15 to 20 percent of total aid, or 3 to 4 percentage points of
GDP. Overall, the results thus suggest that these countries will require
average ODA equivalent to 20 to 30 percent of GDP through 2015.