JOURNAL OF ECONOMIC DEVELOPMENT 25

Volume 30, Number 2, December 2005

EVALUATING THE IMPACT OF FOREIGN AID ON ECONOMIC

GROWTH: A CROSS-COUNTRY STUDY

SANDRINA BERTHAULT MOREIRA*

College of Business Administration Setúbal Polytechnic

One branch of the literature on aid effectiveness attempts to measure the contribution of
foreign aid to the growth of developing countries. The micro results are clear and
encouraging: foreign aid is beneficial to economic growth. However, until recently, the
macro results were inconclusive: the impact of aid on growth was positive, negative, or even
non-existent, in statistical terms. This contradiction is known as the “micro-macro paradox”.
Certain methodological and econometric flaws inherent in the assessments being carried out
up to the mid-nineties may provide an explanation for the misleading macro results.
Examining a large panel data set, I have found that foreign aid has had a positive impact on
economic growth. In light of these findings, I conclude that earlier-generation work is in
accordance with the new and recent generation of aid effectiveness studies. Thus, less
importance should be attributed to the “micro-macro paradox” as an overall appraisal of aid
effectiveness. In terms of magnitude, I have also found that aid has less effect on growth in
the short-run than in the long-run. I also conclude that the time lags in the aid-growth
relationship should not be ignored.

Keywords: Foreign Aid, Economic Growth, Panel Data, Generalised Method of

Moments
JEL classification: F35, O11, O40, C23

*
I am especially grateful for remarks and comments from an anonymous referee, Finn Tarp, Jochen
Oppenheimer, and José Passos. The unpublished work of Hansen et al. (1998) has been kindly provided by
Finn Tarp, who is acknowledged for. Earlier versions of this paper were presented at the 15th Annual Meeting
on Socio-Economics, by SASE and LEST, Aix-en-Provence, France, June 2003 and the 7th Conference on
Mathematics Applied to Economics and Management, by CEMAPRE, Lisbon, Portugal, September 2003. I
would like to thank panel participants for comments.
26 SANDRINA BERTHAULT MOREIRA

1. INTRODUCTION

Official development assistance (ODA), more commonly known as foreign aid,

consists of resource transfers from the public sector, in the form of grants and loans at
concessional financial terms, to developing countries. Many studies in the empirical
literature on the effectiveness of foreign aid have tried to assess if aid reaches its main
objective, defined as the promotion of economic development and welfare of developing
countries. When focusing on the traditional purpose of foreign aid - promotion of the
economic growth of developing countries -, one notes that the results obtained differ
according to the approach used. Studies at the micro-level, mainly using cost-benefit
analyses, support the view of those in favour of the effectiveness of foreign aid. In
contrast, the results presented in studies at the macro-level, namely cross-country
regression studies, are, to say the least, ambiguous. Mosley (1986) called this
contradiction the “micro-macro paradox”.
The “official announcement” for a move towards selectivity in allocating
development assistance gave new impetus to the discussion about aid effectiveness. As a
result, a number of econometric studies linking economic growth to foreign aid have
been published in the last few years. Compared to previous work, they break new
ground in the field. Panel data econometric tools have been employed to allow for
non-linear effects of aid on growth and the endogeneity of aid and other variables.
Furthermore, these recent studies have been inspired by the “new growth” literature,
which encompasses various modifications to the Solow-Swan neoclassical growth model
and endogenous growth models, which in turn provides a different analytical basis
compared to earlier work. The third-generation of cross-country regression studies, as it
is known, with it’s advances in theory and method, have achieved the macro results
foreseen by those in favour of the effectiveness of foreign aid and, therefore, the
“micro-macro paradox” ceases to exist. Even so, the widespread perception that a
disparity exists between micro and macro results reported by former studies is still very
much alive.
This paper concentrates on the aid-growth relationship at the macro-level. An overall
analysis of cross-country regression studies published from the late sixties to the
mid-nineties reveals that whether the dependent variable is savings, investment or
economic growth, the ODA regressor is sometimes significantly positive, sometimes
significantly negative, and sometimes even non-significant, in statistical terms. That is,
the contribution of foreign aid to the economic growth of developing countries may be
positive, negative, or even non-existent, in statistical terms. The explanation for the
inconclusive results remains unclear, but many authors have suggested theoretical and/or
methodological and econometric causes.
The underlying theory of the macro studies in focus here assumes that physical
capital accumulation is the key to economic growth. However, advances in growth
theory have come to show that the growth process relies on a complex set of
interdependent factors. In other words, a host of other factors besides physical capital
EVALUATING THE IMPACT OF FOREIGN AID ON ECONOMIC GROWTH 27

accumulation is known to affect growth. Therefore, according to many authors, the

Harrod-Domar growth model and the Chenery and Strout two-gap model are
over-simplified.
The econometric aid-growth literature has been also criticised on several grounds.
Indeed, after a careful study of twenty-nine macro studies, I have recorded a number of
methodological and econometric weaknesses that may explain the inconsistent results of
regression studies. Therefore, this paper assesses the macroeconomic impact of foreign
aid on the economic growth of developing countries, and proposes improvements to the
methodological and econometric procedures found in studies of the aid-growth
relationship. Growth regressions, based on a large sample of developing countries
covering a 29-year period, are estimated using the generalised method of moments
(GMM) suggested by Arellano and Bond (1991).
The remainder of the paper is organised as follows. Section 2 briefly presents the
theoretical basis for the cross-country regression studies on aid effectiveness reviewed in
Section 3. Section 4 describes the main features of my empirical study, namely
methodology framework, model specification, data and variables, followed by the
analysis of the estimated results in Section 5. Section 6 briefly presents the new
generation of cross-country regression studies and Section 7 this author's conclusions.

2. HARROD-DOMAR MODEL AND GAP MODELS

Empirical studies of the aid-growth relationship carried out until the mid-nineties
were influenced by the early growth theories, which asserted that the growth process
depends on the ability to surpass the constraints regarding the accumulation of physical
capital. Investment was perceived as the key to economic growth.
Traditionally, the lack of savings crucial to investment was regarded as the most
important limitation to the economic growth of developing countries. Indeed, one
characteristic of developing countries is their limited capacity to generate savings, due to
their low per capita income. The original Harrod-Domar model was expanded in the
sixties in the Chenery and Strout (1966, 1979) two-gap model. The foreign exchange
shortage was introduced as another possible growth constraint. Typically, developing
countries need to import goods and services, vital to investment and production; but
import requirements usually exceed export earnings.
Investment can be constrained either by a shortage of domestic savings (the savings
gap) or by a shortage of exports earnings (the trade gap). Therefore, foreign aid inflows
in particular, and foreign capital inflows in general, are needed to fill the prevailing gap,
so that countries can grow more rapidly than their internal resources would otherwise
allow. If these inflows do not exist, the country will experience slower growth and
inefficient employment of internal resources (labour and natural resources). The
desirable outcome is self-sustaining growth.
Following the crippling debt crisis of the 1980s, Bacha (1990) and other
28 SANDRINA BERTHAULT MOREIRA

neostructuralist authors, like Lance Taylor, introduced a third fiscal gap between
government revenue and expenditures. The three-gap model predicts that government
budget limitations rather than foreign exchange constraints or an overall savings
restriction, may be binding. If foreign aid supplements government revenue, then it will
be perceived as promoting economic growth.

3. OVERVIEW OF CROSS-COUNTRY REGRESSION STUDIES

ON AID EFFECTIVENESS1

Hansen and Tarp (2000) consider three generations of cross-country regression

studies. The first-generation studies offered an empirical assessment of how aid
influences domestic savings (savings regressions). According to the Harrod-Domar
equation, growth depends on investment, which is financed by savings (domestic plus
foreign). If the effect of aid on domestic savings is positive, then one may state that aid
will spur growth. Otherwise, aid will probably be detrimental to the economic growth of
developing countries. The second-generation studies assessed the link between aid and
growth, either via investment (investment regressions) or directly in reduced form
equations (growth regressions). Like the second strand of the second-generation studies,
the third-generation ones have explored the direct relationship between aid and growth.
The recent studies of the aid-growth relationship are classified as a new generation
of aid effectiveness studies, because “in our view, the third-generation studies represent
a distinct step forward in empirical cross-country work on aid effectiveness” (Hansen
and Tarp (2000), p. 114). The current generation of cross-country regression studies is
not at the core of the analysis. However, it should be stressed that their contributions
were important in shaping the empirical research in this paper, as will become clearer in
the next section (see also Section 6).
The study by Hansen and Tarp encompasses a list of twenty-nine empirical studies
of the aid-growth relationship published from the late sixties to 1998.2 An analysis of the
main characteristics of these first and second generation studies provides a general
understanding of the methodological and econometric procedures prevailing in the
literature. They are as follows:

1. Single-equation regressions for the total sample, and sub-samples selected

according to geographical region to take into account regional specificities;
2. Cross-section data with period averages, which means that the main evidence
turns out to come from the cross-sectional (between-country) variation;
3. Non-specification of time lags in the aid-growth relationship, in spite of the

1
See Moreira (2002) for a more detailed study of this subject.
2
The list of cross-country regression studies offered by Hansen and Tarp (2000) is available in the Appendix.
EVALUATING THE IMPACT OF FOREIGN AID ON ECONOMIC GROWTH 29

perception that the effect of aid on growth does not end in a single time period;
4. ODA as an exogenous variable, even though there are reasons for suspecting
correlation between aid and the error term in a given model;
5. Aid flows not identified separately from other foreign capital flows (a practice
not prevalent, though strongly criticised);
6. Control variables, even though some of them are not fully documented;
7. Little mention of diagnostic tests, which are important when evaluating the
quality of model specification;
8. The ordinary least squares (OLS) estimation method.

The results of these studies, summarised in Table 1, show inconsistent evidence of a

positive and statistically significant effect of aid on growth.

Table 1. Estimated Results from Twenty-Nine Cross-Country Regression Studies

Unit of measure: number of regression studies
Impact of Savings Investment Growth
Aid: Regressions Regressions Regressions
Positive 1 17 40
Negative 25 0 1
Non-existent 15 1 31
Total(131) 41 18 72
Source: Adapted from Moreira (2002)

However, one of the main purposes of Hansen and Tarp’s detailed survey (2000) is to
offer a re-examination of the empirical cross-country literature covering the period up to
the mid-nineties. After some theoretical and empirical considerations, these authors
conclude that regressions giving empirical support to a positive aid-growth relationship
prevail. Therefore, the “micro-macro paradox” is non-existent. This was the first paper
to draw such a startling conclusion based on a wide range of cross-country regression
studies on aid effectiveness.

4. EMPIRICAL STUDY

4.1. Methodology Framework

As can be seen in the previous section, there is evidence of methodological and

econometric shortcomings in the first and second generation studies. These and other
shortcomings were criticised in the literature and in some cases, seen as the explanation
for the disparities between micro and macro results in general, and the inconclusive
30 SANDRINA BERTHAULT MOREIRA

macro results in particular. Therefore, I propose an alternative methodological and

econometric procedure to heighten the accuracy of aid-growth studies.
From the list of cross-country regression studies supplied by Hansen and Tarp (2000),
I have focused on single-equation growth regressions, since it is the most common
practice found in the literature. Within this regression subset, I have chosen the model
specification suggested by Dowling and Hiemenz (1982) and Mosley et al. (1987, 1992).
Despite the unsophisticated empirical growth equations specified, the model
specification of each study has been remarked for the inclusion of relevant control
variables (e.g., White (1992); Durbarry et al. (1998); Hansen et al. (1998)).
Like the majority of growth regression studies, both studies mentioned above use the
reduced form equation proposed by Papanek (1973) as their basic model. This is also my
point of departure. The original derivation of the Papanek regression was based on the
Harrod-Domar growth equation and a behavioural equation in which investment
depends on its major financing components, including domestic savings as well as
various forms of foreign resource inflows (ODA, private and other official inflows).3
The first practice I have changed was that of expressing the dependent variable in
per capita terms. Real per capita GDP is the most common indicator of a population’s
standard of living. This implies a small change to the Harrod-Domar growth equation, in
order to incorporate the effects of population growth.
A non-linear relationship between aid and growth is not taken into account in any of
the 72 growth regressions selected from the literature. However, there are reasons for
expecting that “too much aid” is detrimental to economic growth. As already put
forward by Chenery and Strout (1966, 1979), the capacity of foreign aid to accelerate
economic growth is contingent upon the absorption capacity of aid recipients. The
capacity to make productive use of external resources depends on numerous factors such
as the existing infrastructure, the available skilled labour and the institutional and
administrative capacity of national and local governments. Excessively high amounts of
foreign aid raise problems of absorption capacity and are thus counterproductive. A
further effect of excessive aid is known as “Dutch disease”. The “Dutch disease”
operates through the spending effect. When part of the additional income generated by a
strong inflow (boom) of aid is spent in-country on non-traded goods and services
(education, health, welfare, construction, and other services), the result is an excessive
demand for this type of goods and services. Since imports cannot flood in to meet
demand, and since domestic supply constraints exist, the price of the non-tradable goods
and services will therefore rise in relation to the price of those tradable. This
appreciation of the real exchange rate is detrimental to external competitiveness and
economic growth. Both absorptive capacity constraints and “Dutch disease” problems
justify the possible existence of an inverted U-shaped relationship between aid and

3
Out of the 72 growth regressions identified from the 29 articles, 51 were derived from the reduced form
Papanek-type regression.
EVALUATING THE IMPACT OF FOREIGN AID ON ECONOMIC GROWTH 31

growth and, in particular, the notion of negative returns at high levels of aid inflows. I
have allowed for non-linear effects of aid on growth by including the squared aid term.4
Most growth regression studies also take no account of the time lags that most
probably exist in the aid-growth relationship. One would not expect aid to be effective in
a single time period. Instead, lags may occur between aid-financed activities and their
eventual impact on growth. The difficulty is how to allow for this time lapse
econometrically. To provide for this time lapse, I have introduced some dynamic into the
non-linear effect of aid by using an autoregressive distributed lag (ADL) relation
between aid and growth. Aid and growth are each lagged once to reveal that the current
growth value depends on the current and previous values of foreign aid. In other words,
this relationship shows that the current value of foreign aid has an effect on the current
and future values of growth.5
For macroeconomic data, reparameterisations of autoregressive distributed lag
models have proven to be quite useful. I have reparameterised the ADL (1,1) scheme
with first differences. Switching to first differences involves only linear transformations
of the variables and does not impose any restrictions. Thus, the estimated results of the
reparametrised model will be identical to those of the ADL model. Moreover, the move
from a relation in terms of the levels of the variables to a relation in terms of both levels
and first differences has substantial advantages. For instance, it usually gives a reduction
in the collinearity of the regressors, thus reducing standard errors. The outcome is the
basic model with both aid and aid squared in levels and first differences, and the
dependent variable lagged once as a regressor.
Papanek (1973) and Mosley et al. (1987, 1992) estimate single-equation growth
regressions for the total sample, and sub-samples selected according to geographical
region. As an alternative, Dowling and Hiemenz (1982) prefer to estimate
single-equation growth regressions with regional dummies. Both practices have been
widely used to point out that, ceteris paribus, growth performances in countries in those

4
The treatment of non-linearities in the aid-growth relation is a common feature shared by the recent
empirical cross-country work on aid effectiveness. Hadjimichael et al. (1995), Lensink and White (2001), and
other authors introduce the squared aid term in the regression to explicitly address this issue.
5
Out of the 29 macro studies I have examined, only Mosley et al. (1987, 1992) attempt to deal with the issue
of time lags in the aid-growth relation. Indeed, this has been a main concern in the research carried out by
Hudson and Mosley (see, for instance, Mosley et al. (1987, 1992); Hudson and Mosley (1994); Mosley and
Hudson (1995)). Recent work on aid effectiveness does not explicitly address the question of the lag structure
that should be used. Yet, in the studies where the endogeneity of aid is found to be relevant, the procedure has
been to include lagged aid terms. To take two examples, Hadjimichael et al. (1995, p. 51) note that “this
study attempts to address these problems [endogeneity problems] by lagging foreign aid by one period” and
Hansen and Tarp (2001, p. 552) state that “following Burnside and Dollar, we show results of instrumental
variable estimations in which all regressors involving aid are treated as endogenous. We use, however, a
different set of instruments. The main change is that we include all the aid regressors lagged one period”.
32 SANDRINA BERTHAULT MOREIRA

regions appear to differ from those of other developing countries. However, using a
panel data model with individual effects has a number of benefits, among which is that it
allows us to account for individual heterogeneity. Indeed, developing countries differ in
terms of their colonial history, their political regimes, their ideologies and religious
affiliations, their geographical locations and climatic conditions, not to mention a wide
range of other country-specific variables. Failing to take this heterogeneity into account
will inevitably bias the results, no matter how large the sample is. The empirical model I
have chosen to use is therefore a dynamic panel data model with fixed country effects.
Time dummies are also taken into account to correct for possible fixed time period
effects.6
Most growth regression studies assume that foreign aid is an exogenous variable,
even though aid is expected to be endogenous in growth regressions. On the one hand,
foreign aid may present issues of reverse causality, especially because, if aid depends on
the level of income, it will necessarily depend on economic growth. If reverse causality
is not taken into account, it can lead to serious inaccuracies in research results. Not only
are the parameter estimates inconsistent, but the magnitude and the meaning of the aid
parameter is altered as well. On the other hand, the error term in a given model may
include factors that both affect growth and are correlated with aid, thus rendering the
parameter estimates inconsistent. Consequently, I have employed Arellano and Bond’s
GMM-type estimator (1991) to deal with the issue of endogeneity in the context of panel
data models.
The GMM estimator proposed by Arellano and Bond, also known as two-step
estimation, is constructed in two phases. Firstly, first differences from the dynamic panel
data model are calculated; then, lagged levels of right-hand side variables are used as
their instruments. With a lagged dependent variable and other endogenous regressors (as
is the case with aid and aid squared), the lagged levels are dated t-2 and earlier (t indexes
time). If there are predetermined regressors, all their lagged levels are used as
instruments.7
Before addressing the model specification issue, two technical aspects must be
mentioned. First, one practical difficulty found in the estimation process is that the early
values of the instruments do not show a close correlation to their late values, though the
quality of the instruments depends on it. For this reason, I have used the common
procedure of limiting the number of lags for each variable. Second, Arellano and Bond
(1991) developed, not only a GMM estimator to apply to dynamic panel data models,

6
I have chosen fixed effects rather than random effects mainly because, when Hausman’s specification test is
employed, the fixed effects estimator is consistent whether the null hypothesis (no correlation between
individual effects and regressors) is true or not. A large loss of degrees of freedom may however arise, if the
number of countries (N) is large, since one is including N-1 dummies in the regression.
7
As done by Hansen and Tarp (2001), the additional regressors of the present empirical study (being other
official flows and time dummies an exception) are assumed to be predetermined.
EVALUATING THE IMPACT OF FOREIGN AID ON ECONOMIC GROWTH 33

but associated specification tests as well. The Sargan test evaluates if the instruments are
valid. In turn, the second-order serial correlation in the first-differenced residuals
evaluates if there is serial correlation in the residuals. The GMM estimator is consistent,
when the null hypothesis of both tests is not rejected.

4.2. Model Specification

On the basis of considerations discussed above, the empirical model which I have
estimated is of the following form:

PCGit = β1 ( Sit ) + β 2 (ODAit ) + β 3 (∆ODAit ) + β 4 (ODA2it ) + β 5 (∆ODA2it )

(1)
+ β 6 ( PFit ) + β 7 (OOFit ) + δ ( PGit ) + ρ ( PCGi ,t −1 ) + τ t + X it + wit ,

with wit = µi + ε it , where i indexes countries, t indexes time, PCGit is per capita
GDP growth rate, Sit are domestic savings relative to GDP, ODAit is official
development assistance relative to GDP, PFit are private flows relative to GDP, OOFit
are other official flows relative to GDP, PGit is the population growth rate, τ t
represents time period effects, X it represents other growth determinants, and wit
represents both country effects ( µi ) and the remainder error term which varies over
both country and time (ε it ) .
In brief, I have changed Papanek’s conventional basic model to incorporate: first, a
dynamic non-linear aid-growth relationship; second, the effects of population growth
and other growth determinants suggested by Dowling and Hiemenz (1982) and Mosley
et al. (1987, 1992); third, fixed country and time period effects. Then I have used the
Arellano and Bond’s GMM-type estimator (1991), assuming that foreign aid is an
endogenous variable.
Given the lag structure proposed in the previous subsection:
- The immediate, short-run effect of a unit change on the ODA/ GDP ratio on the
average growth value is defined, from Equation (1), as:

∂ PCG it / ∂ ODA it = ( β 2 + β 3 ) + 2 ( β 4 + β 5 )( ODAit ), (2)

where ODAit is the mean value of the ODA/ GDP ratio, with t = 1,..., or, T.
- Provided the stability condition ρ < 1 is satisfied, the total, long-run effect of a
unit change on the ODA/ GDP ratio on the average growth value is defined, from
Equation (1), as:

∂PCG / ∂ODA = β 2 /(1 − ρ ) + 2[ β 4 /(1 − ρ )](ODA), (3)

34 SANDRINA BERTHAULT MOREIRA

where ODA is the mean value of the ODA/ GDP ratio.

4.3. Description of Data and Variables

The empirical model presented above is estimated examining 48 developing

countries covering the period 1970 to 1998. I have used six sub-period averages instead
of yearly data. 8 The presence of missing values produced a total sample of 170
observations (unbalanced panel data). The main data source is the World Bank (2001)
and the OECD-DAC (1999, 2000). In Appendix II, I present the list of sample countries
according to income group and geographical region, summary statistics for the main
variables, and the correlation matrix.9 Appendix III shows the list of variables and
sources.10
A few words must be said regarding the intuitive sign for each independent variable.
As sources of physical capital accumulation, domestic savings, official development
assistance, private inflows, and other official inflows (all as a percentage of the GDP)
are expected to have a positive impact on investment and therefore on economic growth.
The quadratic term of the ODA/ GDP ratio is expected to be negatively related to
growth. As has been pointed out, very high aid inflows (measured in relation to the
GDP) are counterproductive. The population growth rate is also expected to have a
negative effect on the growth rate of real per capita GDP.
Dowling and Hiemenz (1982) added four policy variables to Papanek’s basic model.
First, they expressed the degree of openness as the ratio of exports plus imports as a
proportion of GDP. Theory and evidence suggest that trade liberalisation raises
economic growth. Second, the role of governments in domestic resource mobilisation is
measured by central government tax revenues as a percentage of the GDP. The sign of
this variable is a priori ambiguous, because higher taxes can raise public savings and
therefore contribute to domestic resource mobilisation. However, it can also reduce

8
In the estimates, average data over sub-periods of five years are used, except for the last period, which
refers to four years.
9
Care should be taken in analyzing the correlation matrix presented in Appendix II, since it is computed by
ignoring the panel structure of the data. Finding a negative correlation between aid and either economic
growth or domestic savings is somewhat odd. It suggests - but clearly does not prove - that as the
ODA / GDP ratio increases either the per capita GDP growth rate or the domestic savings relative to GDP
decreases. However, part of the correlation may be spurious, reflecting the effects of third factors (e.g.,
traditional growth determinants and unobserved country effects).
10
The developing countries listed in Appendix II were selected from The 2001 World Development
Indicators (World Bank (2001)). Countries that did not have foreign aid data for at least half of the sample
period were excluded. The econometric package also removed countries that had missing values in the six
sub-periods.
EVALUATING THE IMPACT OF FOREIGN AID ON ECONOMIC GROWTH 35

private savings and discourage private capital formation. Third, the share of the public
sector in economic activities is measured by total government expenditure in GDP. The
inefficiency usually associated with public enterprises and the oversized bureaucracies
suggest a negative sign for this variable. Fourth and last, we have M2 over GDP as a
proxy for financial development. Financial development stimulates economic growth by
enlarging the services provided by financial intermediaries such as savings mobilisation,
project evaluation, and risk management. The size of financial intermediaries,
traditionally measured by the ratio of M2 to GDP, is assumed to be positively associated
with the provision of financial services.
In Mosley et al. (1987) and a follow-up paper that extended the database to the
1980s (Mosley et al. (1992)), the authors proposed an alternative expanded version of
the Papanek-type regression. The additional independent variables chosen were changes
in export values and in the literacy rate. Faster export growth is expected to contribute to
economic growth by increasing the supply of foreign exchange and thus the capacity to
import raw materials and equipment essential to rapid and sustained growth. We also
have the growth of adult literacy rates to factor in the positive role for changes in the
stock of human capital on economic growth.

5. REGRESSION RESULTS

The GMM results using Equation (1) are displayed in Table 2.11 Regression [1]
presents the estimated results of my basic model. The high correlation between tax ratio
and government spending (correlation coefficient equal to 0.84 - see Appendix II)
suggests that multicollinearity is a problem. So, as Dowling and Hiemenz (1982), I have
put the policy variables into separate regressions, named Regressions [2] and [3].
Regression [4] enters the additional variables suggested by Mosley et al. (1987, 1992).
From regression [1], we see that the stability condition ( − 0.124 < 1) is met;
otherwise, we wouldn’t have been able to determine the short and long-run effect of
foreign aid on economic growth. Furthermore, the explanatory variables are all correctly
signed and statistically significant. The result concerning the key explanatory variable
therefore suggests that foreign aid contributes to economic growth (as long as the aid to
GDP ratio is not excessively high).

11
I have used the DPD package for OX, available at http://www.nuff.ox.ac.uk/Users/Doornik/.
36 SANDRINA BERTHAULT MOREIRA

Table 2. Growth Regressions Using Panel Data with Fixed Effects

Dependent variable Growth rate of real per capita GDP
Sample / Period 48 countries / 1970-74, 75-79, 80-84, 85-89, 90-94, and 95-98
Estimation method GMM-type estimator of Arellano and Bond (1991)
Regression [1] [2] [3] [4]
Lagged dependent variable -0.124*** -0.101** -0.140** -0.212***
(-3.08) (-2.30) (-2.47) (-3.78)
Domestic savings 0.129*** 0.074* 0.113*** 0.078**
(5.00) (1.84) (2.90) (1.97)
Foreign aid 0.434*** 0.540*** 0.595** 0.655***
(3.57) (3.21) (2.34) (3.77)
Foreign aid (first differences) -0.400*** -0.327* -0.397* -0.480***
(-3.72) (-1.93) (-1.75) (-2.92)
Square of foreign aid -0.009** -0.017*** -0.019** -0.014***
(-2.33) (-2.81) (-2.38) (-2.94)
Square of foreign aid (first differences) 0.008*** 0.012** 0.015** 0.010***
(2.79) (2.15) (2.26) (2.67)
Private flows 0.103*** 0.076* 0.077* 0.051*
(6.70) (1.73) (1.89) (1.94)
Other official flows 0.291*** 0.036 -0.085 0.130
(5.76) (0.41) (-0.99) (1.15)
Population growth (lagged) -0.853*** -0.298 -0.040 0.541
(-3.70) (-0.45) (-0.06) (1.65)
Openness 0.074***
(4.54)
Government spending -0.105*
(-1.95)
Financial depth 0.035*
(1.74)
Tax ratio 0.109
(1.63)
Export growth 0.178***
(6.39)
Literacy growth 3.179***
(2.73)
Number of observations 176 134 134 161
Sargan test 0.560 0.966 0.941 0.996
Serial correlation test 0.052 0.143 0.277 0.635
Wald test - significance of all regressors 0.000 0.000 0.000 0.000
Wald tset - significance of time dummies 0.000 0.000 0.000 0.000
Notes: a) *, ** and *** indicate that the estimated parameter is statistically significant at the 10%, 5% and
1% level, respectively.
b) t-values are shown in parenthesis. Heteroskedasticity-consistent standard deviations.
c) Regressions with time dummies.
d) The p-value for the tests.
EVALUATING THE IMPACT OF FOREIGN AID ON ECONOMIC GROWTH 37

From Regressions [2], [3], and [4], we see that the introduction of other growth
factors does not alter the expected sign; nor does it change the statistical significance of
the domestic savings parameter or of the private flows parameter. In addition, the
magnitude of these parameters only presents small variations. The same goes for the
parameters associated with a dynamic, non-linear aid-growth relationship. On the other
hand, the results for other official flows (share of GDP) and population growth rate are
not very stable. Thus, with the exception of OOF and PG, the basic model parameters are
robust in that they show little sensitivity to small changes in the basic model
specification.12
The variables added to the basic model have the intuitive signs. The respective
parameters are statistically significant, except for the tax ratio parameter. We could
however question the appropriateness of the proxies used for the additional growth
factors. For instance, Barro and Lee (2000) argue that the average number of years of
education in the total population and other educational measures suggested by these
authors have advantages over others commonly used in cross-country studies such as
school enrolment ratios and literacy rates. To take another example, there is a battery of
alternative trade openness measures that have been employed in the empirical growth
literature, with no “best” measure emerging. More reliable measures for the control
variables used in my regressions may indeed exist. This is an empirical matter well
worth exploring, but doing so is beyond the scope of the present paper.
I have used four tests to evaluate the quality of model specification. As can be seen
from the table above (Table 2), the null hypothesis of valid instruments has not been
rejected nor has the null hypothesis of no second-order serial correlation in the
first-differenced residuals. In addition, I have rejected the null hypothesis of regressor
parameters that are simultaneously null, and the null hypothesis of time dummy
parameters that are simultaneously null.13
As a further check on the reliability of the estimation method chosen and, hence,
underlying assumptions, I have replicated the regressions in Table 2 using OLS. The
presence of a lagged dependent variable among the regressors is a major drawback when
using least squares, because it renders the OLS estimator biased and inconsistent. Even

12
There is empirical evidence in favour of the conditional convergence hypothesis, when the negative
relationship between the growth rate and initial level of real per capita GDP is found after controlling the
differences in structural characteristics among countries that generate cross-country differences in long-run
income levels. It is useful to note that finding a statistically significant and negative sign on the lagged
dependent variable does not corroborate the conditional convergence hypothesis. The lagged dependent
variable means lagged growth, not lagged income.
13
Sargan is a test of over-identifying restrictions, distributed as Chi-Squared ( χ 2 ) under the null
hypothesis of instrument validity. In Regression [1], for example, χ 2 [40] = 38.02 , which is far smaller than
the 95 percent critical value of 55.47 (i.e., p-value = 0.560 > significance level = 0.05 ). One can therefore
conclude that the null hypothesis of the Sargan test is not rejected.
38 SANDRINA BERTHAULT MOREIRA

so, this estimation method proceeds by essentially treating the variables included in the
regression as exogenous and the country-specific effects as homogeneous among
different individuals. If these assumptions do hold, there should be no substantial
differences between the OLS and the GMM results. As shown in Appendix IV, this is
not so. Economic growth does not respond to foreign aid in any of the four OLS
regressions; the parameters associated with a dynamic, non-linear aid-growth
relationship are highly insignificant. The same goes for some of the control variables,
namely openness, government spending, financial depth, and literacy growth. One can
therefore conclude that both the issue of endogeneity and that of country heterogeneity
should be taken into account when evaluating the impact of aid on growth.
The GMM results presented above (Table 2) allow one to calculate the
macroeconomic contribution of foreign aid to the economic growth of developing
countries, namely its short and long-run effect. The information reported in Table 3 is
obtained by using the regression parameters from Table 2 and the mean value of aid for
the sample countries. For the short-run results, the mean aid value was computed from
aid figures of the last sub-period, whereas for the long-run results it was computed from
averages across all sub-periods. As Table 3 indicates, for developing countries as a
whole, an increase in the ODA/ GDP ratio of one percentage point leads to a per capita
growth rate increase of 0.16 percentage points, approximately. However, the total impact
on per capita growth of a one percentage point increase in the ODA/ GDP ratio
oscillates between 0.34 and 0.43 percentage points, depending on the regression.

Table 3. Aid Effectiveness Results

Regression [2] [3] [4]
Short-run impact of aid 0.167 0.164 0.141
Long-run impact of aid 0.337 0.360 0.427

In sum, aid effectiveness results displayed in Table 3 may not be sizeable in terms of
magnitude, but they do show that the immediate effect of aid on growth is positive and
lower than its long-run effect.

6. NEW GENERATION OF CROSS-COUNTRY REGRESSION STUDIES

The Assessing Aid report (World Bank (1998)) explicitly laid the foundations for a
move towards selectivity, giving aid to developing countries with a proven track record.
The arguments put forward for aid selectivity were based on a number of background
papers. The one by Burnside and Dollar (1997), which was later published in the
American Economic Review (Burnside and Dollar (2000)), caught researchers’ attention
and gave new stimulus to the discussion on the effectiveness of aid. In view of that, a
EVALUATING THE IMPACT OF FOREIGN AID ON ECONOMIC GROWTH 39

number of cross-country regression studies linking growth to aid have been published
since the World Bank’s report Assessing Aid.
Overall, the new generation of aid-growth econometric studies share a common set
of characteristics. First of all, they examine the variation in growth rates between
countries within specified time periods by using panel data with sub-period averages.
Some studies analyse the aid-growth relationship using both the total sample of
developing countries and the sub-sample of lower income countries. Second, the
overwhelming majority of studies introduce time dummies in the regressions to correct
for the world business cycle. Many authors also use regional dummies, even though
some of them prefer to take individual heterogeneity into account by including country
specific effects. Third, as is standard in the empirical “new growth” literature, both the
initial level of per capita income (which captures the conditional convergence effect) and
a number of economic, political, and institutional factors are included in growth
regressions. Fourth, a non-linear relationship between aid and growth is taken into
account by using quadratic terms and/or interaction terms. The squared aid term allows
for diminishing returns to aid. The interaction term between aid and a given variable
addresses the hypothesis that aid effectiveness is conditional on that variable. Finally,
most growth regression studies assume that foreign aid is an endogenous variable and
only a few consider the possible endogeneity of other explanatory variables.
The new empirical work on aid effectiveness presents evidence on the relationship
between foreign aid and economic growth that should make one trust in aid as a
growth-enhancing factor. Indeed, a fair conclusion from the recent empirical evidence on
aid and growth is that foreign aid appears to promote economic growth, but it's impact
differs across countries depending on the conditions they face. Aid appears to be more
effective in the following circumstances: post conflict situations (Collier and Hoeffler
(2002)); structurally vulnerable countries, including those undergoing trade shocks
(Collier and Dehn (2001); Guillaumont and Chauvet (2001); Chauvet and Guillaumont
(2002)); countries outside the tropical areas (Dalgaard et al. (2002)); politically stable
regimes (Chauvet and Guillaumont (2002)); and democratic countries (Svensson (1999)).
Moreover, aid also seems to be subject to diminishing returns, as the squared aid term is
found consistently negative in a “new growth” framework (e.g., Hadjimichael et al.
(1995); Hansen and Tarp (2000, 2001); Dalgaard and Hansen (2001); Hudson and
Mosley (2001); Collier and Dollar (2002); Collier and Hoeffler (2002); Dalgaard et al.
(2002)).
There is a controversy as to whether the beneficial impact of aid on growth is
dependent on good policies being in place or if it takes place irrespective of policy. The
former is put forward by Burnside and Dollar (1997, 2000) and later supported by
Collier and Dehn (2001), Chauvet and Guillaumont (2002), Collier and Dollar (2002),
and Collier and Hoeffler (2002). The latter is advanced by a set of studies that
demonstrate the sensitivity of the Burnside and Dollar's key result to variations in
sample (including selection of outliers), model specification (regressors and instruments),
policy index and estimation method (e.g., Hansen and Tarp (2000, 2001); Dalgaard and
40 SANDRINA BERTHAULT MOREIRA

Hansen (2001); Guillaumont and Chauvet (2001); Hudson and Mosley (2001); Lensink
and White (2001); Lu and Ram (2001)). Nevertheless, the fact that policies do or at least
potentially matter concerning the effectiveness of aid appears undisputed.

7. CONCLUSION

This paper has sought to evaluate the macroeconomic impact of foreign aid on the
economic growth of developing countries. Cross-country growth regressions carried out
until the mid-nineties were at the core of the analysis. In an attempt to achieve greater
accuracy and improve upon existing procedures, which were viewed as possible causes
of the ambiguous macro results underlying the “micro-macro paradox”, I have proposed
a methodological and econometric procedure that differs from the most prevalent one
used in the literature. I have done this while using two expanded versions of the
well-known reduced form Papanek-type regression, in which aid's contribution to
growth is assumed to be roughly the same for all developing countries. The results
achieved are in line with the micro results, and the common macro result from
cross-country regression studies published in the last few years, i.e., foreign aid is
beneficial to the economic growth of developing countries. Given this, one may then
state that the method rather than the theoretical basis is the main problem inherent in the
assessments being carried out up to the mid-nineties. Moreover, there is empirical
evidence to assert that the “micro-macro paradox” should be given less importance as an
overall appraisal of the effectiveness of foreign aid. It should be stressed, however, that
the new generation of cross-country regression studies have gone beyond earlier work
(including the present one) to address the conditions that must be place for aid to be
(more) effective.
The present empirical results also suggest that non-linearity (negative effects of high
aid inflows) and time lags in the aid-growth relationship, country heterogeneity, and
endogeneity of foreign aid should be factored in when assessing the impact of foreign
aid on the economic growth of developing countries. The recent cross-country studies on
aid effectiveness do employ econometric tools to account for at least, one of these
factors. However, the squared aid term is mainly to find evidence for decreasing returns
to aid rather than a fixed capacity constraint (a threshold) and, even more importantly,
the issue of time lags between aid-financed activities and their eventual impact on
growth has been neglected.
The empirical study described in the present paper shows that the immediate and
overall impact of aid on growth differ in terms of magnitude. This provides support to
assert that the time lags in the aid-growth relationship should not be ignored.
Nonetheless, the issue of time lags in the aid-growth relationship remains a Gordian knot
in the empirical cross-country work on aid effectiveness. Indeed, the required lag
structure will change according to the recipient country and the type of aid allocated.
Programme aid is expected to have a more rapid impact than project aid and this, in turn,
EVALUATING THE IMPACT OF FOREIGN AID ON ECONOMIC GROWTH 41

is expected to have a more rapid impact than technical cooperation aimed at raising the
level of human skills. This suggests that future research should focus on in-depth,
country-specific, case studies.

APPENDIX I. List of Cross-Country Regression Studies Reviewed in Section 3

Ahmed, N. (1971), “A Note on the Haavelmo Hypothesis,” Review of Economics and

Statistics, 53, 413-414.
Areskoug, K. (1969), External Public Borrowing: Its Role in Economic Development,
New York: Praeger Publishers.
Bornschier, V., C. Chase-Dunn, and R. Rubinson (1978), “Cross-National Evidence of
the Effects of Foreign Investment and Aid on Economic Growth and Inequality: A
Survey of Findings and a Reanalysis,” American Journal of Sociology, 84, 651-683.
Dowling, J.M., and U. Hiemenz (1982), “Aid, Savings and Growth in the Asian
Region,” Asian Development Bank Economic Office Report Series No. 3.
Durbarry, R., N. Gemmell, and D. Greenaway (1998), “New Evidence on the Impact of
Foreign Aid and Economic Growth,” Centre for Research in Economic Development
and International Trade Research Paper No. 98/8.
Feyzioglu, T., V. Swaroop, and M. Zhu (1998), “A Panel Data Analysis of the
Fungibility of Foreign Aid,” World Bank Economic Review, 12, 29-58.
Griffin, K.B. (1970), “Foreign Capital, Domestic Savings and Economic Development,”
Bulletin of the Oxford University Institute of Economics and Statistics, 32, 99-112.
Griffin, K.B., and J.L. Enos (1970), “Foreign Assistance: Objectives and Consequences,”
Economic Development and Cultural Change, 18, 313-327.
Gupta, K.L. (1970), “Foreign Capital and Domestic Savings: A Test of Haavelmo’s
Hypothesis with Cross-Country Data: A Comment,” Review of Economics and
Statistics, 52, 214-216.
_____ (1975), “Foreign Capital Inflows, Dependency Burden, and Savings Rates in
Developing Countries: A Simultaneous Equation Model,” Kyklos, 28, 358-374.
Gupta, K.L., and M.A. Islam (1983), Foreign Capital, Savings and Growth: An
International Cross-Section Study, Dordrecht: D. Reidel Publishing Company.
Haveli, N. (1976), “The Effects on Investment and Consumption of Import Surpluses of
Developing Countries,” Economic Journal, 86, 853-858.
Heller, P.S. (1975), “A Model of Public Fiscal Behavior in Developing Countries: Aid,
Investment, and Taxation,” American Economic Review, 65, 429-445.
Khan, H.A., and E. Hoshino (1992), “Impact of Foreign Aid on the Fiscal Behavior of
LDC Governments,” World Development, 20, 1481-1488.
Levy, V. (1987), “Does Concessionary Aid Lead to Higher Investment Rates in
Low-Income Countries?” Review of Economics and Statistics, 69, 152-156.
_____ (1988), “Aid and Growth in Sub-Saharan Africa: The Recent Experience,”
European Economic Review, 32, 1777-1795.
42 SANDRINA BERTHAULT MOREIRA

Massell, B.F., S.R. Pearson, and J.B. Fitch (1972), “Foreign Exchange and Economic
Development: An Empirical Study of Selected Latin American Countries,” Review
of Economics and Statistics, 54, 208-212.
Mosley, P., J. Hudson, and S. Horrell (1987), “Aid, the Public Sector and the Market in
Less Developed Countries,” Economic Journal, 97, 616-641.
_____ (1992), “Aid, the Public Sector and the Market in Less Developed Countries: A
Return to the Scene of the Crime,” Journal of International Development, 4,
139-150.
Over, A.M. (1975), “An Example of the Simultaneous-Equation Problem: A Note on
Foreign Assistance: Objectives and Consequences,” Economic Development and
Cultural Change, 23, 751-756.
Papanek, G.F. (1973), “Aid, Foreign Private Investment, Savings, and Growth in Less
Developed Countries,” Journal of Political Economy, 81, 120-130.
Rahman, A. (1968), “Foreign Capital and Domestic Savings: A Test of Haavelmo’s
Hypothesis with Cross-Country Data,” Review of Economics and Statistics, 50,
137-138.
Singh, R.D. (1985), “State Intervention, Foreign Economic Aid, Savings, Growth in
LDCs: Some Recent Evidence,” Kyklos, 38, 216-232.
Snyder, D.W. (1990), “Foreign Aid and Domestic Savings: A Spurious Correlation?”
Economic Development and Cultural Change, 39, 175-181.
_____ (1993), “Donor Bias Towards Small Countries: An Overlooked Factor in the
Analysis of Foreign Aid and Economic Growth,” Applied Economics, 25, 481-488.
Stoneman, C. (1975), “Foreign Capital and Economic Growth,” World Development, 3,
11-26.
Voivodas, C.S. (1973), “Exports, Foreign Capital Inflow and Economic Growth,”
Journal of International Economics, 3, 337-349.
Weisskopf, T.E. (1972), “The Impact of Foreign Capital Inflow on Domestic Savings in
Underdeveloped Countries,” Journal of International Economics, 2, 25-38.
White, H. (1992), “What Do We Know About Aid’s Macroeconomic Impact? An
Overview of the Aid Effectiveness Debate,” Journal of International Development, 4,
121-137.
EVALUATING THE IMPACT OF FOREIGN AID ON ECONOMIC GROWTH 43
44 SANDRINA BERTHAULT MOREIRA

APPENDIX III. List of Variables and Sources

Variable name Description Source
Per capita GDP Growth rate of real per capita GDP World Bank (2001)
Domestic savings Gross domestic savings World Bank (2001)
(share of GDP)
Foreign aid Net official development OECD-DAC (1999, 2000)
assistance(share of GDP) World Bank (2001)
Private flows Net private capital flows World Bank (2001)
(share of GDP)
Other official flows Net other official flows OECD-DAC (1999, 2000)
(share of GDP) World Bank (2001)
Population growth Population growth rate World Bank (2001)
Openness Exports plus imports World Bank (2001)
(share of GDP)
Government spending Total government expenditure World Bank (2001)
(share of GDP)
Financial depth M2 (share of GDP) World Bank (2001)
Tax ratio Central government tax revenues World Bank (2001)
(share of GDP)
Export growth Exports growth rate World Bank (2001)
Literacy growth Growth in adult literacy rate World Bank (2001)
EVALUATING THE IMPACT OF FOREIGN AID ON ECONOMIC GROWTH 45

APPENDIX IV. Growth Regressions Using Panel Data with Regional Dummies
Dependent variable Growth rate of real per capita GDP
Sample / Period 48 countries / 1970-74, 75-79, 80-84, 85-89, 90-94, and 95-98
Estimation method OLS (Ordinary Least Squares)
Regression [5] [6] [7] [8]
Lagged dependent variable 0.238*** 0.242*** 0.276*** 0.148**
(3.48) (2.79) (3.20) (2.31)
Domestic savings 0.047** 0.073** 0.052* 0.036**
(2.50) (2.30) (1.73) (2.13)
Foreign aid -0.094 -0.062 -0.055 -0.049
(-0.96) (-0.43) (-0.41) (-0.54)
Foreign aid (first differences) 0.017 0.041 0.032 -0.012
(0.11) (0.18) (0.15) (-0.08)
Square of foreign aid 0.005 0.004 0.003 0.002
(1.53) (0.48) (0.47) (0.81)
Square of foreign aid (first differences) -0.004 -0.003 -0.003 -0.002
(-0.84) (-0.33) (-0.31) (-0.46)
Private flows 0.138** 0.171** 0.132* 0.082
(2.18) (2.23) (1.75) (1.47)
Other official flows -0.087 -0.256 -0.438 -0.040
(-0.47) (-0.94) (-1.63) (-0.24)
Population growth (lagged) -0.863*** -0.784** -0.610* -0.657**
(-3.18) (-2.48) (-1.95) (-2.56)
Openness 0.012
(1.31)
Government spending -0.030
(-1.00)
Financial depth -0.001
(-0.08)
Tax ratio 0.072*
(1.83)
Export growth 0.216***
(8.01)
Literacy growth -0.404
(-1.56)
Number of observations 224 182 182 210
Notes: a) *, ** and *** indicate that the estimated parameter is statistically significant at the 10%, 5% and
1% level, respectively.
b) t-values are shown in parenthesis. Heteroskedasticity-consistent standard deviations.
c) Regressions with time dummies and regional dummies for Sub-Saharan Africa, East Asia and Latin
America.
46 SANDRINA BERTHAULT MOREIRA

REFERENCES

Arellano, M., and S. Bond (1991), “Some Tests of Specification for Panel Data: Monte
Carlo Evidence and an Application to Employment Equations,” Review of Economic
Studies, 58, 277-297.
Bacha, E.L. (1990), “A Three-Gap Model of Foreign Transfers and the GDP Growth
Rate in Developing Countries,” Journal of Development Economics, 32, 279-296.
Barro, R.J., and J.W. Lee (2000), “International Data on Educational Attainment:
Updates and Implications,” Center for International Development Working Paper No.
42.
Burnside, C., and D. Dollar (1997), “Aid, Policies, and Growth,” World Bank Policy
Research Working Paper No. 1777.
_____ (2000), “Aid, Policies, and Growth,” American Economic Review, 90, 847-868.
Chauvet, L., and P. Guillaumont (2002), “Aid and Growth Revisited: Policy, Economic
Vulnerability and Political Instability,” Paper presented at the Annual Bank
Conference on Development Economics: Towards Pro-poor Policies, June 24-26,
Oslo.
Chenery, H.B., and A.M. Strout (1966), “Foreign Assistance and Economic
Development,” American Economic Review, 56, 679-733.
_____ (1979), “Foreign Assistance and Economic Development,” in Structural Change
and Development Policy, ed. by H.B. Chenery HB, New York: Oxford University
Press, 382-455.
Collier, P., and J. Dehn (2001), “Aid, Shocks, and Growth,” World Bank Policy
Research Working Paper No. 2688.
Collier, P., and D. Dollar (2002), “Aid Allocation and Poverty Reduction,” European
Economic Review, 46,1475-1500.
Collie, P., and A. Hoeffler (2002), “Aid, Policy, and Growth in Post-Conflict Societies,”
World Bank Policy Research Working Paper No. 2902.
Dalgaard, C.J., and H. Hansen (2001), “On Aid, Growth and Good Policies,” Journal of
Development Studies, 37, 17-41.
Dalgaard, C.J., H. Hansen, and F. Tarp (2002), “On the Empirics of Foreign Aid and
Growth,” Centre for Research in Economic Development and International Trade
Research Paper No. 02/08.
Dowling, J.M., and U. Hiemenz (1982), “Aid, Savings and Growth in the Asian
Region,” Asian Development Bank Economic Office Report Series No. 3.
Durbarry, R., N. Gemmell, and D. Greenaway (1998), “New Evidence on the Impact of
Foreign Aid and Economic Growth,” Centre for Research in Economic Development
and International Trade Research Paper No. 98/8.
Guillaumont, P., and L. Chauvet (2001), “Aid and Performance: A Reassessment,”
Journal of Development Studies, 37, 66-92.
Hadjimichael, M.T., D. Ghura, M. Mühleisen, R. Nord, and E.M. Uçer (1995),
“Sub-Saharan Africa: Growth, Savings, and Investment, 1986-93,” International
EVALUATING THE IMPACT OF FOREIGN AID ON ECONOMIC GROWTH 47

Monetary Fund Occasional Paper No. 118.

Hansen, H., and F. Tarp (2000), “Aid Effectiveness Disputed,” in Foreign Aid and
Development: Lessons Learnt and Directions for the Future, eds. by F. Tarp, and P.
Hjertholm, London: Routledge, 103-128. Also published in Journal of International
Development, 12, 375-398.
_____ (2001), “Aid and Growth Regressions,” Journal of Development Economics, 64,
547-570.
Hansen, H., F. Tarp, and S. Vikkelso (1998), “Aid Effectiveness: A Survey of 30 Years
of Literature on Aid-Savings, Aid-Investment, and Aid-Growth Linkages,”
Development Economics Research Group.
Hudson, J., and P. Mosley (1994), “Aid, Policy Reform and Endogenous Growth in Less
Developed Countries,” University of Reading Discussion Paper No. 13.
_____ (2001), “Aid, Policies and Growth: In Search of the Holy Grail,” Journal of
International Development, 13, 1023-1038.
Lensink, R., and H. White (2001), “Are There Negative Returns to Aid?” Journal of
Development Studies, 37, 42-65.
Lu, S., and R. Ram (2001), “Foreign Aid, Government Policies and Economic Growth:
Further Evidence from Cross-country Panel Data for 1970 to 1993,” International
Economics, 54, 15-29.
Moreira, S.B. (2002), “Ajuda Pública ao Desenvolvimento e Crescimento Económico:
Estudos Cross-Country da Relação Ajuda-Crescimento,” Masters Dissertation,
Technical University of Lisbon.
Mosley, P. (1986), “Aid Effectiveness: The Micro-Macro Paradox,” IDS Bulletin, 17,
22-27.
Mosley, P., and J. Hudson (1995), “Aid Effectiveness: A Study of the Effectiveness of
Overseas Aid in the Main Countries Receiving ODA Assistance,” University of
Reading and University of Bath.
Mosley, P., J. Hudson, and S. Horrell (1987), “Aid, the Public Sector and the Market in
Less Developed Countries,” Economic Journal, 97, 616-641.
_____ (1992), “Aid, the Public Sector and the Market in Less Developed Countries: A
Return to the Scene of the Crime,” Journal of International Development, 4,
139-150.
OECD-DAC (1999), Geographical Distribution of Financial Flows to Aid Recipients,
Disbursements, Commitments, Country Indicators 1960-1997, CD-ROM, Paris:
Organisation for Economic Co-operation and Development.
_____ (2000), Geographical Distribution of Financial Flows to Aid Recipients,
Disbursements, Commitments, Country Indicators 1994-1998, Paris: Organisation
for Economic Co-operation and Development.
Papanek, G.F. (1973), “Aid, Foreign Private Investment, Savings, and Growth in Less
Developed Countries,” Journal of Political Economy, 81, 120-130.
Svensson, J. (1999), “Aid, Growth and Democracy,” Economics and Politics, 11,
275-297.
48 SANDRINA BERTHAULT MOREIRA

White, H. (1992), “The Macroeconomic Impact of Development Aid: A Critical

Survey,” The Journal of Development Studies, 28, 163-240.
World Bank (1998), Assessing Aid: What Works, What Doesn’t, and Why, New York:
Oxford University Press.
_____ (2001), The 2001 World Development Indicators, CD-ROM, version 4.2,
Washington, D.C.: Development Data Group.

Mailing Address: Dra. Sandrina Berthault Moreira Escola Superior de Ciências

Empresariais Instituto Politécnico de Setúbal Campus do IPS - Estefanilha 2914 - 503
Setúbal, Portugal. Tel: +351 265 709 437, Fax: +351 265 709 314. E-mail:
[email protected], [email protected]

Manuscript received October, 2004; final revision received February, 2005.

 APPENDIX II. Sample Countries According to Income Group and Geographical Region Summary Statistics
and the Correlation Matrix
Sample of countries classified by:
Income group 48 countries
Bangladesh, Cameroon, Congo Democratic Republic, Ghana, India, Indonesia, Kenya, Lesotho, Liberia, Malawi, Nepal, Nicaragua,
Low income countries (17 countries)
Pakistan, Senegal, Sierra Leone, Zambia and Zimbabwe.
Bolivia, China, Colombia, Costa Rica, Dominican Republic, Ecuador, Egypt, Fiji, Guyana, Jordan, Papua New Guinea, Paraguay,
Lower middle income countries (20 countries)
Peru, Philippines, Sri Lanka, Swaziland, Syrian, Thailand, Tunisia and Turkey.
Upper middle countries (11 countries) Argentina, Botswana, Brazil, Chile, Malaysia, Mauritius, Mexico, Panama, South Korea, Uruguay and Venezuela.
Geographical region 48 countries
Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, Guyana, Mexico, Nicaragua, Panama,
Latin America and Caribbean (16 countries)
Paraguay, Peru, Uruguay and Venezuela.
Middle East and North Africa (4 countries) Egypt, Jordan, Syrian and Tunisia.
Botswana, Cameroon, Congo Democratic Republic, Ghana, Kenya, Lesotho, Liberia, Malawi, Mauritius, Senegal, Sierra Leone,
Sub-Saharan Africa (14 countries)
Swaziland, Zambia and Zimbabwe.
Bangladesh, China, Fiji, India, Indonesia, Malaysia, Nepal, Pakistan, Papua New Guinea, Philippines, South Korea, Sri Lanka,
Asia, Europe and Pacific (14 countries)
Thailand and Turkey.

Series in levels PCG S ODA DODA PF OOF PG X&M G M2 T XG LG

Summary statistics (No. of observations) (282) (277) (283) (235) (280) (277) (288) (279) (199) (274) (199) (262) (235)
Mean 1.80 17.10 4.93 0.26 2.60 0.66 2.32 62.30 22.93 31.07 15.98 6.42 1.33
Standard deviation 3.28 13.75 6.31 3.65 3.14 1.04 0.76 37.01 9.82 18.31 6.23 7.04 1.09
Minimum -11.48 -68.08 0.07 -11.92 -2.75 -2.84 0.15 5.71 0.00 6.71 0.00 -15.25 -0.89
Maximum 15.81 49.19 34.89 25.85 30.42 9.36 5.69 231.97 76.28 116.51 38.53 37.32 6.97
Correlation matrix (*)
Per capita GDP (growth; %): PCG 1.00
Domestic savings (% GDP): S 0.31 1.00
Foreign aid (% GDP): ODA -0.17 -0.57 1.00
Foreign aid (first differences; % GDP): DODA -0.28 -0.16 0.18 1.00
Private flows (% GDP): PF 0.27 0.30 -0.22 -0.25 1.00
Other official flows (% GDP): OOF -0.12 -0.01 0.15 0.12 0.12 1.00
Population (growth; %): PG -0.21 -0.23 0.40 0.14 -0.14 0.11 1.00
Openness (% GDP): X&M 0.04 0.06 0.41 -0.17 0.25 0.16 0.07 1.00
Government spending (% GDP): G -0.03 -0.08 0.39 -0.09 0.11 0.23 0.07 0.62 1.00
Financial depth (% GDP): M2 0.24 0.13 0.02 -0.23 0.17 0.05 -0.20 0.45 0.47 1.00
Tax ratio (% GDP): T 0.10 0.10 0.15 -0.13 0.21 0.21 -0.05 0.67 0.84 0.45 1.00
Exports (growth; %): XG 0.55 0.14 -0.18 -0.10 0.16 -0.11 -0.13 -0.09 -0.19 0.06 -0.06 1.00
Literacy (growth; %): LG -0.08 -0.26 0.49 0.05 -0.27 0.12 0.59 -0.09 0.11 -0.13 -0.08 -0.12 1.00
* Spearman correlations; cases with missing values for one or both of a pair of variables for a given correlation coefficient were excluded from the analysis.