Forthcoming as Chapter 2 in R. Grant and J. Nijman (eds.). 1998. The Global Crisis in Foreign
Aid, Syracuse: Syracuse University Press.
Multilateral Aid, Politics and Poverty
London School of Economics and Political Science
London School of Economics and Political Science
The last half-century of foreign aid programs has comprised a broad economic
experiment without historical precedent. Beginning modestly with the goal of promoting
economic development and alleviating poverty, no one could have guessed in 1944 that five
decades later industrial countries would be giving annual grants equal to 8% of recipient
countries' GNP. Nor could anyone have predicted the spectacular growth in the number and
size of bilateral and multilateral aid agencies. Gross disbursements of aid by industrialized
countries in 1994 amounted to over $100 billion dollars, nearly three times the GNP of Ireland,
or half the GNP of Australia.
And yet fifty years on most of Africa and much of Asia is still mired in chronic poverty.
The scale of global inequality remains vast: several dozen of the poorest countries register a
yearly GNP per capita1 barely equal to a few days' pay in the richest countries. Human
development indicators present an equally stark picture. Of those reporting data, more than
twenty countries have primary-school gross enrollment ratios below 75%2, and some as low as
19%, and in forty-two countries (mostly African, Arab and South Asian) more than ten percent
of children die before age five. These figures are much better than they were fifty years ago,
but it would be a mistake to conclude that past improvements imply we can be complacent with
our current efforts at poverty reduction. In Figure 1 we have extrapolated past trends in order
to project the pattern of child mortality to 2025. The top two poverty maps show the gradual
progress made in reducing child mortality over the last thirty years. But as shown in the lower
map, even if this progress continues unabated, by 2025 more than 10% of children will
continue to die before age 5 in thirty-four countries (mostly African), and this figure rises to
above 20% in ten countries.3
1World Bank (1994a)
2World Bank (1994a)
3These figures were projected from historical data compiled by the United Nations.
Logarithmic projections were made based on trends from 1950 (before which very little data is
The main theme of this chapter is that we need to do much more to address poverty in
the future. No reader should be surprised that the human costs of poverty will continue to be
large in the future. Much more worrying is the observation that fifty years of aid work and
investment have not done more to reduce deprivation to achievable levels in most of the world.
And if the trends that brought us here continue, thirty years hence the global situation will be
little better in relative terms, and much worse in absolute terms as population growth in the
poorest countries multiplies the masses of their poor. It is time to ask what five decades of
work have accomplished, and to reassess the role and effectiveness of foreign aid.
This chapter presents an overview of empirical evidence on the successes and failures
of aid programs with the aim of recommending a future role for multilateral aid agencies. We
examine the impact of aid programs that have come into operation since the first Bretton
Woods meetings, including long term grants and concessional assistance, major infrastructure
projects operated by the multilaterals, and short term aid programs to support new governments
or governments in crisis.
This evidence provides one clear message: long term aid programs have broadly
failed to achieve their poverty reduction goals. Some forms of short-term aid programs, on
the other hand, have been more successful.
The failure of long term aid programs leaves open the question of how aid can best
serve development goals in the future. This is discussed in the final section of this chapter,
which argues that improved capital markets, the rapid rise of foreign funding available for
infrastructure investment, and the general globalization of world markets sharply reduce the
need for multilateral institutions. Multilaterals should accordingly restrict the number of
countries they deal with, and gradually phase out their infrastructure and other large-scale,
capital-intensive lending programs.
An important role remains for aid agencies, however, in the provision of short term aid
available for developing countries) and 1970.
to new governments, governments in crisis, and governments embarking on serious poverty
reduction programs. We believe that foreign aid can and should be used as a creative tool to
encourage countries to embark on serious poverty reduction programs, and in the conclusions
we discuss several means to achieve this.
Multilateral agencies may also play a role promoting the development of commercial
markets for infrastructure investments. The advantage of using private markets instead of
development institutions to provide capital instead of development institutions is that loans and
projects are subjected to market discipline. But for this to function efficiently, we need to
carefully consider international laws, regulations and procedures that can best ensure efficient
private capital flows.
Thirdly, multilateral aid can provide technical assistance and support for world public
goods such as the environment. Fourthly, these institutions should continue to take advantage
of the substantial technical expertise which they have accumulated in a number of technical
areas. But they should be forced to compete in a market with other providers of such services.
This implies much smaller and less capital-intensive organizations.
The chapter is organized as follows: Section 2 outlines early arguments for and against
aid, showing the theoretical ambiguity surrounding the question of the potential impact of aid.
Section 3 summarizes evidence from previous research on the impact of foreign aid on growth,
investment and human development indicators. Section 4 examines the effectiveness of
public infrastructure programs and relates these to the debt crisis in Sub-Saharan Africa,
arguing that the major legacy of large aid flows in the sixties and seventies is not higher income
but rather a large debt burden. Section 5 discusses several examples of short term aid
programs that have been successful, and Section 6 concludes with a general discussion of the
future role for development assistance.
2. Early Arguments For and Against Aid
Despite the rapid growth in aid, there have never been compelling theoretical
underpinnings or empirical evidence showing that foreign aid programs actually work. John
Maynard Keynes took the view that capital shortages were a primary cause of poverty. The
World Bank was created in part due to his assessment that developing countries needed
financing for major public and private investment projects. This argument was developed
fully in the fifties and sixties by economists such as Rostow (1990) and Chenery (1966). Aid
proponents argued that developing countries have low savings because of their absolute
poverty. Since they cannot afford to save, and world capital markets are imperfect, they need
a "jump start" with foreign aid in order to take off.
But careful examination calls this argument into question. During the late nineteenth
century and in the interwar period world capital markets functioned quite well. Britain was a
major exporter of capital to the new world, where both private and public investment, most
notably railways, were financed by bond issues. This capital mobility reemerged after the
second world war. Korea financed substantial investment through foreign borrowing
beginning in the early sixties. A true explosion of international capital flows began in the
The idea that poor countries have low savings because households cannot afford to save
is also dubious. Today the richest 20% of the population earn 50% of total income in low-
income countries. It is really the high-income elite that conduct what investment and savings
go on in these countries, and they command the resources to do so. The question should then
be rephrased: Why do the high-income elite save and invest in some countries while in others
they do not?
The early critics of foreign aid programs, notably Bauer (1973) and Friedman (1958),
argued that the root cause of poverty was not capital shortage but rather government failure. If
low investment is due to a lack of opportunities to invest, say due to corruption and
distortionary economic policies, then giving aid will not lead to higher investment or growth.
Growth can only come from removing the distortions that prevent development and poverty
alleviation, which itself requires a fundamental shift in economic policy. If aid allows
governments to reduce distortions, say through reducing tax rates or strengthening the
government so it can enforce laws, then aid could still lead to growth and investment. But if
governments do not have incentives to change these policies, and if aid does not change the
basic incentives of government, then foreign aid will fail and aid funds will be wasted.
Basic Facts about Aid
(in the sample of 97 countries)
Average Aid/GNP ratio
81-90: range 0.00-0.54, sdev 0.112 71-80: range
0.00-0.35, sdev 0.035
in Base Sample:
81-90: range 0.00-0.144, sdev 0.039
71-80: range 0.00-0.146, sdev 0.045
Grant component (1980) 0.93
Restrictions on Procurement
(fraction by category):
Uses of Aid (fraction by type):
Social and Admin. Infrastructure
Sources of Aid
Bilateral (of which):
Aid as a fraction of GDP from Donor Countries
(members of DAC)
Note: Social and administrative infrastructure applies to health care, education,
technical assistance to governments, etc. Economic Infrastructure is highways,
electricity, irrigation, and other large public investment projects. Program aid is
balance of payments support.
3. The Cross-Country Impact of Aid
Given the ambiguity surrounding the theoretical case for aid, we now turn to its
empirical effects. We begin by examining the impact of the largest component of aid flows:
non-military concessional assistance, called Official Development Assistance, which includes
all non-military aid with at least a twenty-five percent grant component. Basic summary
statistics are shown in Table 1. The empirical results in this chapter summarize previous
research examining the impact of foreign aid on economic and human development indicators
using cross-country regressions.4 We present simple graphs that summarize the relation
between aid and improvements in human development indicators based on regression results
from approximately 100 developing countries.
There are two major difficulties with measuring the impact of aid in cross-country
regressions. The first is the potential simultaneity bias caused by the fact that aid is given to
relatively poor countries, so that a simple regression of, for example, infant mortality on aid
would erroneously predict that higher aid leads to higher infant mortality.5 In order to get
around this problem, we take advantage of the fact that a large portion of the variance in aid
flows is caused by political determinants. Former French colonies, for example, receive
substantially more aid than their equally poor neighbors. It is possible to test whether the
additional aid that these countries receive, when compared to other similar countries, has
positive effects on growth and human development indictors. In previous research we have
taken advantage of this insight, using appropriate instrumental variables, to test the impact of
aid and also to conduct a large number of robustness and sensitivity tests.
4There is surprisingly little research using cross-country regressions to examine the
impact of aid. There was substantial research on the impact of aid using cross-country
regressions in the 1970s. This literature abruptly ended when the problems with measurement
error and potential cause and effect biases were realized. See Papanek(1973) for references to
this literature and problems encountered. Paul Mosley has written several articles examining
the cross-country impact of aid more recently, see e.g. Mosley et. al. (1986). The regression
results reported here are discussed and justified more fully in Boone(1994a) and
5See Papnek (1973).
The second problem, as shown in Table 1, is that aid is targeted to many different
expenditure categories, so by pooling different types of aid we may bias our results.6 But in
fact the allocation of aid is quite flexible. Contrary to popular belief a surprisingly small
fraction of aid is tied (30%, see Table 1), and 30% of aid is pure cash grants (not reported). In
targeted aid donors and recipients initially negotiate over how to allocate aid, after which
recipients can freely dispose of project receipts and benefits over long periods of time.
Recipients can additionally avoid expenditures on maintenance if they choose (see discussion
related to infrastructure in World Bank (1994)). The overall implication is that aid is highly
The one exception to this is for those small countries that receive very large amounts of
aid. In these countries one investment project can be 50% of GNP, meaning that aid is no
longer fungible, and that these projects will necessarily cause sharp increases in investment.
But when aid comes in smaller amounts it is once again fungible; in this section we report
results from a subsample of countries with aid/GNP ratios less than 0.15.
Figures 2 and 3 show that in our sample of 96 countries most aid goes to consumption,
and there is no significant impact on investment. These are CPR (component plus residual)
plots which show the residuals from regression equations (reported in the appendix) after
controlling for other factors that are potentially correlated with aid and the dependent variable.
All regressions are run on World Bank data using long time averages over ten years in order to
control for business cycle effects.
In Figure 2 the marginal propensity to consume from aid is insignificantly different
from one (point estimate 1.08, t-statistic 3.56), while the marginal propensity to invest is
6Cassen(1994) and Killick(1990) discuss some of the problems encountered when trying
to measure the aggregate impact of aid.
7Pack and Pack(1993), and Khilji and Zampelli(1991) found that governments were able
to, ex-post, fully redirect expenditures agreed to under aid programs to alternative uses. Pack
and Pack(1990) found that aid flows were not fungible in Indonesia, and they concluded this
was due to the large amounts of aid that Indonesia was receiving.
insignificantly different from zero (point estimate 0.03, t-statistic 3.5). This implies that over
long time periods most aid is used to raise consumption with little overall impact on
investment. In results not shown here, we divided the sample into those countries where aid
led to higher public consumption, and a remaining group where aid was not used to raise
government expenditure. This allowed us to test whether aid allocated through the private
sector had a different impact from aid allocated through government consumption. But we
found that even when government did not spend aid itself, all funds went to private
consumption. This implies that neither the private nor public sector has a strong incentive to
invest in these countries.
Given that most aid goes to consumption, it is not surprising that it has no significant
impact on growth. Figure 4 shows this. The point estimate from this regression is
insignificant, but it implies that a 10 percent increase in the annual aid/GNP ratio causes a 0.4
percent rise in the average growth rate over the full ten years. These results imply that the
factors which cause high investment and growth in developing countries do not correlate with,
nor are engendered by, foreign aid receipts. They are strong evidence that capital shortage is
not the major cause of poverty in developing countries.
Figures 5, 6 and 7 show that the poor do not benefit when a country receives higher aid
flows. Infant mortality is highly related to nutritional standards, sanitation, provision of basic
health services, housing conditions, and maternal education. Evidence from many countries
(e.g. Chile, China and Sri Lanka) shows that infant mortality improves quickly when these
basic factors improve, so infant mortality can be considered a "flash" indicator of the
conditions of the poorest groups in the population.8
Figure 5 shows that aid does not correlate with more rapid improvements in infant
mortality. Figure 6 shows that higher aid flows do not have a significant impact on
improvements in life expectancy. Figure 7 shows that aid does not correlate with
8See Flegg (1982) for empirical evidence on the determinants of infant mortality.
improvements in primary school enrollment ratios.
The coefficient estimates in Figure 5 imply that countries that received 10 percent
higher aid as a fraction of GNP had a 2 percent greater improvement in infant mortality than
countries that received no aid. This is an extremely small impact, and indicates that the
poorest fraction of the population do not receive much of the benefits of aid programs. But
there is no question that aid funds are spent. It seems likely that long term aid receipts are
actually used to increase the size of government and the civil service, or to make transfers to
relatively wealthy members of the political elite.
These results are consistent with the negative predictions of Bauer and Friedman, but it
is still possible that certain political regimes use aid more effectively than others. It is often
claimed, for example, that liberal and democratic political regimes will use aid more
effectively since they are more representative of the poor.
In other research we have tried to test whether liberal political regimes use aid more
effectively. We conducted similar regressions to those reported in Figures 2 through 7, but in
this case we allowed for different impacts of aid according to the degree of political liberties in
the recipient countries. We used indexes of political liberties calculated by Raymond Gastil,
and as an alternative we used the categorization of countries into liberal democracies and
"other" regimes as determined by Derbyshire and Derbyshire(1992).
The clear implication of these results was that all political regimes use aid in a similar
manner. Most aid goes to consumption, having little impact on investment or growth and no
significant impact on basic human development indicators. This was true regardless of how
democratic or liberal the political regime was.
It should not be surprising that liberal democracies do not use aid more effectively than
other regimes. Even in democracy the poor are relatively weak actors in the political process
precisely because they lack education and good health. When governments receive additional
funds it is only natural that they allocate them to their strongest political supporters.
Lack of improvement in human development indicators, and the continued high levels
of infant mortality in aid recipient countries, are clear signs that poverty reflects government
failure. Caldwell (1986) and Drèze and Sen (1989) show that countries can radically improve
basic human development indicators when they choose to. Examples from Cuba, Sri Lanka,
China and the Indian State of Kerala show that concerted efforts on the part of government to
introduce good public health programs lead to dramatic improvements in infant mortality and
life expectancy. These programs are not costly. The World Bank (1993) estimates that a
health program costing 3.1% of GNP in low income countries, and 0.9% of GNP in middle
income countries would be sufficient to bring life expectancy and infant mortality indicators
near to OECD standards. This argument is neatly illustrated in Figure 5. The two outliers in
the bottom left corner are Costa Rica and Chile, two countries famous for their effective health
4. Is Public Infrastructure Oversold?
Public infrastructure investment is the bread and butter of multilateral aid programs -
approximately 20% of Official Development Assistance is used for public infrastructure
(included in the data described in the previous section). Overall, infrastructure commitments
are 24% of total development assistance; the IBRD and IDA alone provided 13.6 billion dollars
of infrastructure investment in 1991, or more than half the total infrastructure investment
provided by all aid programs. Has this been successful? and is it really needed?
There are two arguments for donor provision of public infrastructure investment.
First, as explained above, capital shortage may be the cause of poverty. Second, it is argued
that cash-strapped governments cannot afford public infrastructure, and poor infrastructure is
an important bottleneck in the development process.
But it is extremely difficult to empirically measure the potential benefits of public
infrastructure. Aschauer (1989) estimates that public capital in the United States has a 60
percent annual return, and Uchimura and Gao (1993) estimate that investments in Taiwanese
and Korean infrastructure earn returns of 77 percent and 51 percent respectively. But these
and other similar studies are based on growth regressions where simultaneity bias may
overestimate the results. It could be that a common factor is driving both growth and
investment, or that public infrastructure spending is responding to higher growth. In either
case such growth regressions will sharply overestimate the returns to public infrastructure.
The World Bank provides direct measures of the efficiency of the projects it finances.
It examines rates of return on public infrastructure projects across regions and sectors for
projects where the financial returns should reflect social marginal products (some transport,
agriculture, industry, etc.). Table 2 shows the Bank's estimated returns on these projects
ranging from 8 to 21 percent. Returns vary systematically across regions, with Sub-Saharan
African projects earning the lowest returns.9
These figures suggest that individual projects earn lower rates of return than one would
expect for capital-poor, developing regions of the world. Government regulators in North
America and Europe typically concede rates of return around 15% to public utilities, and many
technological firms demand returns of 25% or more to finance projects. The Bank's estimates
suggest either that development agencies should look much harder for worthwhile projects, or
that the policy environment in these countries is so generally distorted that good projects
cannot be generated.
These numbers are probably not good proxies for the overall impact of aid on
macroeconomic performance. First, World Bank estimated returns are generally biased
upward at the time projects are begun, and over time are reduced as projects are implemented.
World Bank data shows that returns on public investment projects that were initially estimated
to average 24.4% were subsequently revised downwards to 15.8%. This suggests a Bank
tendency to overestimate project benefits prior to project approval.
More importantly though, the satisfactory performance of individual projects may not
reflect the broad macroeconomic impact of aid. Recipient governments can use foreign aid to
9See Killick(1991) for an overview of other literature on this issue and description of
some of the problems in measuring these returns.
Re-estimated Rates of Return on World Bank Infrastructure Projects
for Evaluated Projects by Region
Eastern Africa 15.0 9.5
Western Africa 27.0 8.4
East Asia and Pacific 20.6 20.6
European, Mediterranean and North
Latin America 16.2 14.8
Source: World Bank (1988) page 9.
reduce their own public investment programs, allowing them to increase expenditures in other
areas. In cases where aid is not an increment to government spending but rather a substitute, it
may well achieve nothing. Governments can also fail over time to maintain public
infrastructure projects. The evidence from the previous section suggests these factors may be
empirically large enough to swamp the benefits of investment projects.
4.1 Can Centralized Allocation of Infrastructure be Efficient?
While the overall effectiveness of infrastructure programs is unclear, there is still an
important question as to how best to allocate assistance for infrastructure investment. The
founders of the World Bank, to take the most important example, anticipated that the Bank
would primarily provide guarantees to private investors but not enter into substantial direct
lending. But over time it became clear that the pool of such private projects was too small to
justify large-scale bank activities. Nor was there sufficient capacity in developing countries to
measure infrastructure needs and feasibility. The Bank thus evolved to become almost
entirely a direct lender.
But is centralized allocation of funds through large development institutions the best
means to select and choose infrastructure projects? Internal investment criteria and discipline
in the World Bank, as in other development agencies, can never be expected to mimic market
conditions. Mosley (1991) reports that loan approvals and disbursements are at the center of
the Bank's internal incentive structure. The impact of these loans cannot be assessed for long
periods of time, and it is impossible for any bureaucracy to single out blame for failed
programs years after a project was begun. It is thus understandable that the project selection
process often fails to weed out projects with poor prospects.
Neither do the normal forces of market mechanisms provide external discipline to the
operations of the World Bank and other multilaterals. They earn very high credit ratings
because their bonds are guaranteed by all member countries. Recipient countries are required
to pledge that the World Bank is senior to all other creditors, and the IMF and regional
development banks have similar clauses. This is reinforced by the fact that commercial banks
look to multilateral activity for signs of a country's or government's creditworthiness. The
result is that governments service multilateral debt before attending to other creditors.
Multilaterals can thus safely lend to developing countries without facing the risk of losses that
other commercial lenders face.
In addition to this, politicians, motivated by the possibility of private or political gain
and the reality of electoral cycles or political instability, will be willing to enter into projects
even when the social returns are low or negative. In a system where neither markets nor
borrowers play a discriminating role in project selection, Bank staff are left alone to cope with
the entire burden of quality control and monitoring in Bank lending. Such a system presents a
risk that loan programs can provide a mechanism for an explosion of debt and inefficient public
infrastructure investment in developing countries through easy access to multilateral credit
4.2 The Dangers of Inefficient Investment
The build-up of official debt to Sub-Saharan African countries with little effect on
growth, the most striking example of the failure of international development programs, is
precisely such a case. From the late fifties and early sixties it was already apparent that aid
projects were not generating sufficient government revenues to allow recipients to repay their
loans. This led to the creation of the International Development Agency (IDA), sister agency
to the IBRD, which was made responsible for providing long term lending with a ten year grace
period and concessional interest rates to developing countries.
But during the 1960s and 1970s substantial bilateral assistance continued at
non-concessional rates. By 1969 past principal and debt service was beginning to mount.
The Pearson (1969) report warned that there was an impending debt crisis in Sub-Saharan
Africa and called for donor countries to increase the portion of concessional lending to
But the warnings of these reports were never fully heeded, and today official debt
makes up three quarters of Sub-Saharan Africa's external obligations. Without official aid
programs African countries would not have been able to build up the severe levels of
indebtedness they have.
This failure of foreign aid to cause growth is generally consistent with the findings and
arguments made above in sections three and four. A large part of aid was probably used for
consumption, most likely via permitting governments to increase transfers and/or the size of
government. Further, as shown in table 2, and despite a probable upward bias, the World
Bank's estimated returns on investment projects in African countries were generally low. This
puts into question the ability of donors to screen projects with adequate rigor, and leads us to
conclude that the fact that market mechanisms are not guiding the process is an important
reason why lending to Africa got out of hand.
5. When does aid succeed?
There are, of course, numerous examples of successful aid programs. Many specific
aid projects have worked well, bringing significant benefits to their host countries and target
populations. Famine relief and immunization drives are prime examples of inexpensive
programs where relatively small amounts of assistance may be decisive.
Sachs (1994) presents a clear case in favor of aid to assist countries attempting financial
stabilization, or entering into large-scale liberalization programs. He shows that during the
last century virtually every country that successfully stabilized received foreign assistance.
Amongst many others these include Austria (1922), Germany (1924 and 1946), Bolivia (1986),
Poland (1990), and Estonia (1992).
Such short term aid also has clear theoretical backing for its potential usefulness.
During high inflation the tax base is often eroded as enterprises delay payments (known as the
Tanzi effect). Seignorage revenues also fall as people stop using domestic money. To get
out of the trap of high inflation and low revenues the government needs to muster the political
strength to make tough budget decisions, and raise credibility so that households and
enterprises will once again trust the domestic currency. Here foreign assistance provides
bridge financing while budget measures take hold and governments gain political strength.
Additionally, international support usually raises government's credibility.
Foreign assistance has also played a key role in encouraging shifts in economic policies
and political structures during revolutionary periods or economic crises. United States aid to
Taiwan played a decisive role in ensuring nationwide land reform in the early 1950s (Cheng
(1961)). The Taiwanese also followed US advice to introduce export-oriented policies in the
1960s when US AID threatened to cut aid to the Chiang Kai-Shek regime.
In Korea the government maintained closed, distorted markets throughout the 1950s.
The decisive factor which caused the Korean government to embark upon market liberalizing
reforms was the US administration's threat to end aid in the early 1960s. The American
commitment to reduce aid to Korea provided the political impetus for the government to
stabilize its macroeconomic policies, open the economy and promote exports.
Another common example of aid providing decisive political support was the Marshall
plan to assist reconstruction in postwar Europe. This was also largely short-term political aid
aimed at boosting liberal regimes and supporting market liberalization.
These experiences suggest that conditional aid programs, as proposed by the IMF for
example, could also be effective means to bring about good policies. But the empirical
evidence on the success of IMF conditionality at introducing greater macroeconomic stability
even in the short run is mixed.10 Most of the countries which enter into IMF-sponsored
programs do not achieve long term stability or growth, and hence must return to the Fund some
years later to negotiate another program. The underlying factors that cause instability are
apparently not affected by the short term conditional programs that the IMF prescribes.
Nor do World Bank conditional aid programs have clear positive or long-lasting
impacts. Mosley (1991) summarizes results from several case studies. It is not evident,
firstly, that World Bank conditionality is strongly enforced, nor that it is even enforceable
given the range of commitments made. Bank conditionality generally involves several dozen
specific conditions, but in practice countries can implement only a fraction of these and still not
risk loan suspension. The real reason behind this may in turn reflect the incentive structure of
the Bank - if it measures its own performance by success in approving and disbursing loans,
then it is not really in the interest of Bank staff to suspend loans due to noncompliance of
conditionality. There is reason to think that other multilaterals' conditional programs operate
The most recent examples of success are seen in the formerly socialist countries. Here
IMF conditionality has played a fundamental role in promoting financial stability and
liberalization. The IMF helped design and finance the Polish stabilization and liberalization
program which became the model for Russia, Czechoslovakia, Ukraine, and most other CIS
countries. The IMF has also played a critical role training staff in the new central banks in
these countries, and providing technical assistance to ministries. Many of the price and trade
liberalization programs in CIS countries are timed in conjunction with specific requirements by
10See Khan(1992) for a survey of the effectiveness of conditionality.
the IMF, showing the decisive role it can play in bringing about change.
The lesson here is that foreign aid programs can be successful during and after periods
of political change. These are times when (new) interest groups are still forming and the role
of government is being redefined. Very often the government has the mandate to carry out
rapid reform, or is in a position where a few key reformers can introduce changes that
eventually become self-sustaining. But once interest groups are entrenched, and governments
settle into more stable political patterns, the potential for major reforms is greatly diminished.
In these cases even conditional foreign aid is a weak instrument when used to affect changes in
Whither Multilateral Institutions? Whither Foreign Aid?
The evidence from the last fifty years points to three broad conclusions. First, poverty
and the lack of economic development are not driven by capital shortages. Aid programs have
transferred large capital flows to developing countries, but investment has not risen
significantly in response and consumption has increased instead. This suggests that a lack of
investment opportunities is the main reason for low investment and savings. Poor investment
possibilities could in turn be driven by the lack of clear property rights, political instability, and
poor economic policies and a lack of credibility on the part of the government.
Secondly, we must attribute to government failure the lack of improvement in human
development indicators in many countries. This failure may be caused by political deadlock,
or by political regimes that simply choose to spend resources on uses other than the provision
of basic services to the poor. A small number of the poorest nations have shown that
concerted efforts by government can lead to sharp improvements in development indicators.
Amongst the remaining poor countries today, the missing ingredient appears to be political will
rather than financial capital.
Finally, both examples and sound theoretical reasons support the position that short
term aid can be effective during revolutionary periods when governments have the opportunity
and/or a mandate to carry out major reforms. During these periods aid can strengthen
governments that choose to introduce major liberalization programs or enfranchise the poor in
the political system. And reforms that lead to better education and health for the poor may
well result in their political voice being strengthened, and so become self-sustaining.
What role does this leave for multilateral institutions? The world economy has
changed radically since the end of the second world war. Economists have changed their
views toward the role of the public sector in the economy: throughout the world governments
are privatizing activities that used to be considered well within the public domain. The private
sector is held to be generally more efficient, and doesn't face the incentive problems discussed
above that both governments and aid agencies face.
This narrowing of the public domain should also be applied to aid programs. In recent
years there has been a spectacular rise in foreign investment for public infrastructure in
developing countries. Today foreign investors can travel easily and communicate
instantaneously with most regions of the world, and their resources have followed. Examples
include toll roads in Mexico, dams in Argentina, telecommunications facilities in Mongolia
and energy and power plants in China. Poor countries can and should seek to harness these
financial flows by adjusting their macroeconomic policies to promote investment.
This means that multilaterals should gradually phase out their major infrastructure
programs. They can instead promote more rapid growth of foreign private investment flows,
and assist countries in designing policies and legal reforms to best promote domestic
investment and access to funds. They could also piggyback on these projects, providing
partial financing or guarantees against political instability, nationalization, etc. that such
international agencies could enforce.
Market discipline will then be the guiding factor that determines the choice of
investment projects, leaving multilaterals in a passive role in the project selection process.
This discipline will pressure governments to maintain an economic climate conducive to
investment in the long term, thus providing a more credible, clear and binding conditionality
than the current programs of the World Bank and other donors. It will also reduce, though
certainly not eliminate, the possibility of another major debt crisis in these countries.
Several of the multilateral agencies have built up considerable technical and research
expertise. They have every reason to continue to take advantage of this. But technical
assistance should also be subject to more competition and openness. The Bank could sharply
reduce its size, and instead move to a system where it contracts more services from private
consulting firms or experts.
Finally, the evidence on aid programs suggests that long term aid can be sharply
reduced and possibly replaced by short and medium term programs more directed at poverty
reduction. The clearest examples of the successful use of foreign aid have been during periods
of revolution and crisis. Aid can be helpful during stabilization, as famine relief, and to
provide the political support new governments and reformers often need when they push
through major economic and political reform. To achieve this, we should bring much greater
competition into the aid allocation process. We should target aid to countries that introduce
major poverty eradication campaigns, and countries that take measures to liberalize their
economies. This aid should be highly conditional, and we should be prepared to stop foreign
aid if reforms slow or are reversed. When government failure is a major cause of poverty, aid
must be used as a tool to change government incentives. In this way we can support reformist
factions within existing governments, and give greater incentives for all governments to
embark on programs to help the poor.
Aschauer, David A. “Is Public Expenditure Productive ?” Journal of Monetary Economics 23
Bauer, P.T. Dissent on Development. London: Weidenfield and Nicholson, 1971.
Boone, Peter. “The Impact of Foreign Aid on Savings and Growth.” Center for Economic
Performance Discussion Paper No. 627, London School of Economics (1994a):
Boone, Peter. “Politics and the Effectiveness of Foreign Aid.” Center for Economic
Performance, Discussion Paper No. 628, London School of Economics (1994b):
Brandt Commission. Common-Crisis: North South Cooperation for World Recovery. London:
Caldwell, John C. “Routes to Low Mortality in Poor Countries.” Population and Development
Review 12 (2 1986): 171-212.
Cassen, Robert et. al. Does Aid Work ? Second ed., Oxford: Clarendon Press, 1994.
Chenery, Hollis and A.M. Strout. “Foreign Assistance and Economic Development.”
American Economic Review 56 (1966):
Cheng, Chen. Land Reform in Taiwan. Taipei: China Publishing Company, 1961.
Conteh-Morgan, Earl. American Foreign Aid and Global Power Projection: The Geopolitics
of Resource Allocation. England: Dartmouth, 1990.
Dreze, Jean and Amartya Sen. Hunger and Public Action. Oxford: Oxford University Press,
Friedman, Milton. “Foreign Economic Aid.” Yale Review 47 (4 1958): 500-516.
Griffin, Keith. “Foreign Capital, Domestic Savings and Economic Development.” Oxford
Bulletin of Economics and Statistics 32 (February 1970): 99-112.
Khilji, Nasir M and Ernest M Zampelli. “The Fungibility of U.S. Assistance to Developing
Countries and the Impact on Recipient Expenditures: A Case Study of Pakistan.” World
Development; 19(8), August 1991, pages 1095-105 (1991).
Killick, Tony. The Development Effectiveness of Aid to Africa. World Bank Working Paper
Series No. 646, Washington, D.C.: World Bank, 1990.
Levy, Victor. “Aid and Growth in Sub-Saharan Africa: The Recent Experience.” European
Economic Review 32 (9 1988): 1777-95.
Maizels, Alfred and Machiko K. Nissanke. “Motivations for Aid to Developing Countries.”
World Development 12 (9 1984): 879-900.
McKinlay, R. D. and R. Little. “A Foreign Policy Model of U.S. Bilateral Aid Allocation.”
World Politics 30 (1 1977): 58-86.
McKinlay, R.D. and R. Little. “A Foreign Policy Model of the Distribution of British Bilateral
Aid: 1960-70.” British Journal of Political Science 8 (1978a): 313-322.
McKinlay, R.D. and R. Little. “The French Aid Relationship: A Foreign Policy Model of the
Distribution of French Bilateral Aid: 1964-1970.” Development and Change 9 (1978b):
McKinlay, R.D. and R. Little. “The U.S. Aid Relationship: A test of the Recipient Need and
Donor Interest Models.” Political Studies 27 (2 1979): 236-250.
Mosley, Paul. Aid and Power: The World Bank and Policy Based Lending. Vol. 1 and 2.
London and New York: Routledge, 1991.
Mosley, Paul, John Huson, and Sara Horrell. “Aid, the Public Sector and the Market in Less
Developed Countries.” Economic Journal 97 (September 1987): 616-641.
Organization for Economic Cooperation and Development. Development Cooperation Report.
Paris: OECD, 1992.
Pack, Howard and Janet Rothenberg Pack. “Is Foreign Aid Fungible? The Case of Indonesia.”
Economic Journal 100 (399 1990): 188-94.
Pack, Howard and Janet Rothenberg Pack. “Foreign Aid and the Question of Fungibility.”
Review of Economics and Statistics 75 (2 1993): 258-65.
Papanek, G. F. “The Effect of Aid and other Resource Transfers on Savings and Growth in
Less Developed Countries.” Economic Journal 82 (327 1972): 934 951.
Papanek, Gustav F. “Aid, Foreign Private Investment, Savings and Growth in Less Developed
Countries.” Journal of Political Economy 81 (1 1973): 121-130.
Pearson, Lester B. Partners in Development: Report of the Commission on International
Development. New York: Praeger, 1969.
Rostow, Walt Whitman. The stages of economic growth: A non-communist manifesto. Third
ed., New York and Melbourne: Cambridge University Press, 1990.
Sachs, Jeffrey. “Russia's Struggle with Stabilization: Conceptual Issues and Evidence.” 1994.
Uchimira, Kazuko and Hong Gao. “The Importance of Infrastructure on Economic
Development.” World Bank Latin America and Caribbean Regional Office, Washington D.C.
Weisskopf, Thomas E. “The Impact of Foreign Capital Inflow on Domestic Savings in
Underdeveloped Countries.” Journal of international Economics (1972): 25-38.
World Bank. Evaluation Results for 1988: Issues in World Bank Lending over Two Decades.
Washington, D.C.: World Bank, 1988.
World Bank. World Development Report 1993. Washington D.C.: World Bank, 1993.
World Bank. World Development Report 1994. Washington D.C.: World Bank, 1994a.
World Bank. World Tables. World Bank, Washington D.C.: World Bank, (1994b): 750.
Regression Results and Techniques
The regression results reported here are fully described in Boone(1994a) and (1994b).
In these papers Boone presents results using alternative instruments, subsamples, and
regression techniques. The main findings are robust when we use fixed effects, generalized
least squares, and when we instrument aid using alternative measures of political factors that
determine aid flows.
In this paper we have presented results from regressions where we create an instrument
for aid by first regressing the aid/GNP ratio of recipients on a range of factors that may be
correlated with aid and with the regression residual. For example, income per capita is
correlated with both the consumption/GNP ratio and the aid/GNP ratio. By regressing aid on
income per capita, and then taking the residuals from this equation, we can purge the
correlation between aid and income per capita.
This procedure can be implemented by simply including as right hand side variables
those factors that are correlated both with aid and the regression residual. In earlier work cited
above we include alternative sets of right hand side variables. The empirical results here are
robust once we control for income per capita on the right hand side. But since aid may also be
correlated with terms of trade shocks, debt rescheduling, income and population growth rates,
and regional dummies, we also include these in the regressions here.
Table A1 reports basic regressions for the main variables in this study. As stated
above, the econometric results reported in Table A1 are qualitatively similar when we directly
instrument aid flows, and when we use alternative regression techniques and right hand side
The data on foreign aid was provided by the OECD, and all remaining data is from the
World Bank. LGNPCAP is the log of GNP relative to OECD countries, and TERMS OF
TRADE is the cumulative loss of income measured as a proportion of GNP over the decade.
The terms of trade loss/gain is calculated by the World Bank(1993).
DEBT RESCHEDULING is a dummy set to one in 1981 to 1990 if the country entered into
Paris Club negotiations during that period.
26 Download full-text
Regressions Showing the Impact of Aid on Consumption,
Investment, Infant Mortality, Life Expectancy and Schooling
(panel data using base sample with decade-averaged data 1971-80,81-90)
I II V VI VII
Per capita GNP growth
Population growth rate
Terms of Trade
log(Dependent variable at start of
Level of Dependent variable at start
R2 0.615 0.463 0.963 0.980 0.869
SEE 0.063 0.053 0.145 0.025 0.148
N 123 123 123 123 105
1. OLS estimates, t-statistics in parentheses, standard errors are adjusted for a random individual specific component.