Department of Economics
McAnulty College of Liberal Arts
Duquesne University
Pittsburgh, Pennsylvania
THE IMPACT OF FOREIGN AID ON ECONOMIC GROWTH
Mark T. Whitaker
Submitted to the Economics Faculty
in partial fulfillment of the requirements for the degree of
Bachelor of Arts in Economics
December 2006
Faculty Advisor Signature Page
Pinar Geylani, Ph.D.
Associate Professor of Economics
Date
Antony Davies, Ph.D.
Associate Professor of Economics
Date
2
THE IMPACT OF FOREIGN AID ON ECONOMIC GROWTH
Mark T. Whitaker, B.A.
Duquesne University, 2006
Abstract
The massive expenditures on foreign aid programs by developed nations and
international institutions, in combination with the perceived lack of results from these
disbursements, raise important questions as to the actual effectiveness of monetary
assistance to less developed countries (LDCs). In this analysis, I focus on 119 low- and
medium-development countries, and measure the impact that foreign aid has on their
growth rates of gross domestic product, using dummy variables for geography and
conflict in a geometric lag model.
The results indicate that foreign aid donations do have a positive impact on the
economic growth of the recipient nation. The effect is extremely modest, however, and
other factors such as armed conflict and geography can easily mitigate this impact, in
some cases to the extent that foreign aid becomes detrimental to economic growth.
Further analysis of the results indicate that this impact is quickly felt, with half of the
total impact of foreign aid felt in approximately six months.
Key Words: Foreign aid, economic growth, economic development
3
Table of Contents
1. Introduction ................................................................................................................... 5
2. Literature Review ......................................................................................................... 6
3. Methodology ................................................................................................................ 10
3.1 Data......................................................................................................................... 10
3.2 Model Specification ................................................................................................ 13
3.3 Expected Results ..................................................................................................... 14
4. Results and Analysis ................................................................................................... 14
5. Conclusions and Suggestions for Future Research .................................................. 19
References ........................................................................................................................ 22
Appendix A: Included Countries ................................................................................... 25
Appendix B: Autocorrelation Test Results ................................................................... 26
Appendix C: Heteroskedasticity Test Results .............................................................. 27
Appendix D: Correlation Matrix ................................................................................... 28
4
1. Introduction
Over the last half century, foreign aid has emerged as a dominant strategy for
alleviating poverty in the third world. Not coincidentally, during this time period major
international institutions, such as the United Nations, World Bank, and International
Monetary Fund gained prominence in global economic affairs.1 Yet it seems that sixty
years later, the lesser developed countries (LDCs) of the world continue to suffer from
economic hardship, raising questions of whether foreign aid is a worthwhile and effective
approach to boosting growth and development in recipient economies. Research on the
subject has attempted to draw an empirical connection between foreign aid and economic
growth. Despite these efforts, however, there is no solid consensus among scholars on
the actual effectiveness of foreign aid inflows.2
The term “foreign aid” can imply a number of different activities, ranging from
humanitarian support in the wake of natural disasters to military assistance and arms
donations.3 For the purposes of this analysis, however, I refer to the standard definition
of “official development assistance,” or aid that is aimed at increasing economic
development, and has a grant component of at least 25% of the total aid package.4 Critics
of development assistance cite a variety of reasons why it is a poor strategy for
combating global poverty. Some argue that it can breed corruption, weaken
accountability, and cause government to become excessively large.5 Nonetheless, as
Peter Hjertholm and Howard White, “Survey of Foreign Aid: History, Trends, and Allocation,”
Discussion Papers, University of Copenhagen Institute of Economics (2000): 3.
2
Raghuram G. Rajan and Arvind Subramanian, “Aid and Growth: What Does the Cross Country Evidence
Really Show?” International Monetary Fund Working Papers 05 (2005): 2.
3
H. VanBuren Cleveland, “Purposes of International Aid Programs,” Scientific Monthly 85 (1957): 77-81.
4
George W. Norton, et. al., “Impact of Foreign Assistance on Agricultural Growth,” Economic
Development and Cultural Change 40 (1992): 775-786.
5
Stephen Knack, “Aid Dependence and the Quality of Governance: Cross Country Empirical Tests,”
Southern Economic Journal 68 (2001): 310-329.
1
5
researchers Hansen and Tarp (2000) write, “it is neither analytically defensible or
empirically credible to argue from the outset that aid never works.”6 Indeed, a number of
studies have shown a positive relationship between foreign aid and economic growth,
especially in countries which have responsible economic policies regarding trade,
inflation, and other macroeconomic concerns.
The purpose of this analysis is to study the effects of foreign aid inflows on real
gross domestic product growth rates. It differs from existing research in two key ways.
First, I utilize a geometric lag model to capture the continued impact of foreign aid
inflows for years after its initial introduction into the economy.7 Second, I incorporate
several dummy variables for geography, political stability, and development to determine
their additional impact on foreign aid’s effectiveness in growing GDP.
2. Literature Review
There are two contrasting sides to this debate: one which argues that aid has a
positive effect on economic growth, with even more impact in countries with sound
economic and trade policies; and another which contends that foreign aid causes
corruption, encourages rent-seeking behavior, and erodes bureaucratic institutions. A
renewed interest in explaining cross-country economic growth emerged in the early
1990s, with numerous studies attempting to answer the foreign aid question. To date,
however, there is no consensus among scholars as to the actual effects of foreign aid on
economic growth.
6
Qtd in Finn Tarp, Foreign Aid and Development: Lessons Learnt and Directions for the Future, New
York: Routledge, 2000.
7
Douglas C. Dacy, “Foreign Aid, Government Consumption, Saving, and Growth in Less-Developed
Countries,” The Economic Journal 85 (1975): 548.
6
There have been several prominent studies which find a causal link between
foreign aid and economic growth. Perhaps the most well-known of these was performed
by two researchers for the World Bank, Craig Burnside and David Dollar (1997). They
found that foreign aid enhances economic growth, so long as “good” fiscal policies are in
place. These policies can include maintaining small budget deficits, controlling inflation,
and being open to global trade.8 Durbarry, et. al. (1998) also found a positive association
between foreign aid and economic growth, and confirmed Burnside and Dollar’s finding
of conditionality on good economic policy. The study also concluded, however, that the
degree to which aid impacts GDP depends largely on other factors as well, such as
geography.9 Ali and Isse (2005) further confirmed the findings of Burnside and Dollar.
The study also demonstrated, though, that aid is subject to decreasing marginal returns,
indicating a threshold beyond which development assistance can become detrimental to
economic growth.10
Not all research has shown a positive relationship to exist between aid and
growth. Even before Burnside and Dollar’s monumental findings, a study by Peter
Boone (1994) found that aid-intensive African countries experienced zero per capita
economic growth in the 1970s and 80s, despite foreign aid actually increasing (as
measured by share of GDP).11 Additionally, Knack (2001) found that high levels of
Craig Burnside and David Dollar, “Aid, Policies, and Growth,” American Economic Review 90 (1997):
847-868.
9
Ramesh Durbarry, et. al., “New Evidence on the Impact of Foreign Aid on Growth,” Center for Research
in Economic Development and International Trade 8 (1998): 3.
10
Abdiweli M. Ali and Hodan S. Isse, “An Empirical Analysis of the Effect of Aid on Growth,”
International Advances in Economic Research 11 (2005): 1-11.
11
William Easterly, “Does Foreign Aid Add Up?” Foreign Policy 125 (July 2001): 94.
8
7
foreign aid can erode bureaucratic and institutional quality, triggering corruption, and
encouraging rent-seeking behavior.12
There is also evidence that the effects of foreign aid can be mitigated by other
non-economic factors. Situations of state failure, such as ethnic conflict, genocide or
politicide, and revolution can all potentially influence the extent to which aid impacts
growth. George Mason University’s Political Instability Task Force (PITF) created a
binary dataset indicating in which countries and during what years these events take
place. According to the PITF, an ethnic conflict requires the clash of two separate ethnic,
religious, or nationalistic factions, and also must meet two threshold criteria: 1,000
people must be mobilized for armed conflict, and at least 1,000 people per year must
have died as a direct result of this conflict. Similarly, revolutions are defined as episodes
of violent conflict between political groups in hopes of overthrowing the current regime,
and must meet the same threshold criteria as ethnic wars. Finally, genocide and
politicides are defined in a slightly different manner. These events occur when the group
in power carries out sustained policies that target ethnic, religious, or political rivals,
ultimately resulting in the deaths of a “substantial” portion of one of those groups.13
Easterly and Levine (1997) studied the effects of high ethnic fractionalization on
economic growth. By fractionalization, they mean the probability that two randomly
chosen people from a population will be of different ethno-linguistic backgrounds.
Easterly and Levine conclude that movement from heterogeneity to homogeneity
(decreasing fractionalization) results in better schooling, more efficient infrastructures,
12
Knack 2.
Monty G. Marshall, et. al., Political Instability Task Force, George Mason University (2006). Refer to
http://globalpolicy.gmu.edu/pitf/.
13
8
and more developed financial systems and foreign exchange markets.14 According to
their findings, then, it is entirely possible that ethnic conflict, in its attempt to move away
from ethnic diversity and towards ethnic homogeneity, will actually improve economic
growth. Despite their findings, however, the instability of the regime could still
negatively impact the degree of aid’s effectiveness.
Not a lot of attention is paid to genocide, politicide, and revolution and their
effects on growth in the literature. Moreover, there has been virtually no research
performed on this question as it concerns the effectiveness of aid. It is reasonable to
believe, though, that resources (including foreign aid) are siphoned off by the dominant
party and used for individual benefit rather than for economically efficient activities, as
intended.
Furthermore, out of respect for state sovereignty, these events are not likely to
prompt a major international response, which would perhaps eliminate local control over
resources and allow them to be used productively. Ethnic conflict, on the other hand,
typically ignores state boundaries. One study by Gurr (1993) estimated that over twothirds of identified ethnic communal groups in the world have kindred in another country.
The spread across state borders allows other states to intervene without violating state
sovereignty, which could positively impact how resources are used, and ultimately,
economic growth.
Additionally, a country’s geographic location can influence economic
performance; nations that are landlocked, for instance, are at a natural disadvantage in
global trade. Sachs and Warner (1996) write,
William Easterly and Ross Levine, “Africa’s Growth Tragedy: Policies and Ethnic Divisions,” The
Quarterly Journal of Economics 112 (1997): 1203.
14
9
Landlocked countries, in particular, face very high costs of shipping, since they
must pay road transport costs across at least on international boundary in addition
to sea freight costs. Although air shipments can help overcome many of these
problems, only certain goods can be economically shipped by air, and most
countries still import and export the majority of goods by the sea.15
A report by the UN Economic and Social Commission for Asia and the Pacific (1999)
specifically mentions the positive relationship between aid and growth in landlocked
countries, noting that they are at a disadvantage for these reasons, as well.16 Due to their
geographical position, then, landlocked countries could potentially benefit from foreign
assistance, as it may fill the gap in trade that they experience relative to countries with
easy access to international trade.
3. Methodology
3.1 Data
I direct the focus of this analysis to low- and medium-development countries as
defined by the United Nations Development Programme (UNDP) in its Human
Development Index (HDI).17 These nations were selected since they are the most likely
to be recipients of foreign aid, whereas high-development nations are the most likely to
be donors. I select the HDI as a basis for classification because in addition to income, the
index accounts for life expectancy as measured by infant mortality rates, and educational
attainment as measured by adult literacy rates and gross enrollment ratios for primary,
secondary, and tertiary schools. This provides for a more thorough understanding of a
Jeffrey Sachs and Andrew Warner, “Sources of Slow Growth in African Economies,” Harvard Institute
for International Development, Development Discussion Papers 545 (1996): 14.
16
Available at http://www.unescap.org/55/e1140e.htm.
17
Available at http://hdr.undp.org/reports/global/2005/pdf/HDR05_HDI.pdf, pages 219-222. Mediumdevelopment nations are defined as those with scores below 0.800, and low-development nations are
defined as those below 0.500.
15
10
country’s stage of development and a comprehensive measure of quality of life.18 In all,
119 countries of the 177 analyzed by the UNDP (67%) meet the development criteria and
were included in this study.19
Due to data availability issues, I restrict the range of this study to the period from
1980 to 2003. With 119 cross sections, there is a potential 2,856 observations over this
time span. After taking into account missing data for the independent variables included
in the model, 1,760 remain, or about 62%. A vast majority of the missing data is a result
of the overall lack of information regarding Sub-Saharan Africa and Soviet bloc countries
during the early 1980s. Furthermore, I aim to measure the impact of foreign aid on
average, across both time and countries. Thus, I employ pooled data analysis.
I collect the data in annual format from several sources. Most of the data come
from the United Nations Conference on Trade and Development (UNCTAD)20 and the
International Monetary Fund (IMF).21 Table 1 below lists the variables included in this
study and the source from which they were gathered:
Table 1: Data Sources
Variable
Unit
Source
Gross Domestic Product
Growth Rate
IMF
Official Development
Assistance
Millions $US
UNCTAD
Household Consumption
Growth Rate
UNCTAD
Government Expenditures
Growth Rate
UNCTAD
Exports*Petroleum
Exporter
Growth Rate
UNCTAD
Imports
Growth Rate
UNCTAD
18
Ibid. 214.
See Appendix A for a list of included countries.
20
Available at http://www.unctad.org.
21
Available at http://www.imf.org.
19
11
Agricultural Production
Growth Rate
UNCTAD
Gross Capital Formation
Growth Rate
UNCTAD
Inflation
Growth Rate
IMF
Openness to Trade22
Share of GDP
UNCTAD
Millions of BTUs
Energy Information Agency, U.S. Dept of
Energy
Energy Consumption Per
Capita
Major Petroleum Exporter
Dummy
Non-Tropics Dummy23
Foreign Direct Investment
Inflows
Ethnic Conflict Dummy
Genocide Dummy
Revolution Dummy
Landlocked Country
Dummy
Low Development Dummy
1=Yes,
0=Otherwise
1=Yes,
0=Otherwise
UNCTAD
IUCN World Conservation Union
Millions $US
UNCTAD
1=Yes,
0=Otherwise
1=Yes,
0=Otherwise
1=Yes,
0=Otherwise
1=Yes,
0=Otherwise
1=Yes,
0=Otherwise
Political Instability Task Force, University of
Maryland
Political Instability Task Force, University of
Maryland
Political Instability Task Force, University of
Maryland
UNCTAD
United Nations Development Programme
Data for household consumption, government expenditures, exports, imports,
agricultural production, and gross capital formation were only available in share of GDP
format. Since I aim to explain growth rates in GDP, however, percentage changes in the
dollar amounts of each of these variables would be more appropriate. Thus, I transform
these numbers into growth rates as well.24
22
I measure openness to trade by adding Exports (as % of GDP) and Imports (as % of GDP).
I define “non-tropic” as a country with less than 50% of land mass lying between the Tropic of Cancer
and the Tropic of Capricorn.
24
Growth rates were calculated by multiplying the share of GDP times real GDP values, which resulted in
the real dollar value of each variable. The percentage change was then calculated for inclusion in this
analysis.
23
12
3.2 Model Specification
I assume that inflows of foreign aid will continue to impact the economy for years
after its initial introduction, but at a decreasing rate. It would therefore be unsuitable to
use an ordinary least squares model, since it would only take into account aid inflows in
the year they were received and disregard the continued impact that foreign aid has on the
economy in the years after its introduction. To effectively capture this rationale, I use a
geometric lag model which incorporates an infinite number of lags for each variable, but
weights each lag in a geometrically declining fashion. The general form of this type of
model is:




Yi ,t    1 X i ,t  X i ,t 1  2 X i ,t 2 ...   2 Z i ,t  Z i ,t 1  2 Z i ,t 2 ...  ...  
(1)
Note that in the model a weight is attached to each lag (λ), a value between zero and one
that diminishes geometrically as time passes. Mathematically, this model is the same
as:25
Yi ,t   1     Yi ,t 1  1 X i ,t   2 Z i ,t  ...  
(2)
This simpler form, however, shows the dependent variable Y on the right side of the
equation. Since Y is already shown to have an error component in (1), this simplification
introduces a stochastic regressor into the model, requiring two-stage least squares (TSLS)
regression. In order to ensure the instruments required for TSLS are non-stochastic, I lag
25
Equation (2) is derived by lagging (1) one period on both sides of the equation and subtracting from (1).
13
each one period. Thus, to the observer at time t, values for instruments at t-1 are fixed.
In other words, these instruments are stochastic but predetermined.
3.3 Expected Results
I expect to find a positive relationship between foreign aid and economic growth
on average, as indicated by most prior research on this subject. I further anticipate,
however, that aid will have a detrimental effect on low-development countries since they
lack efficient infrastructures and institutions which might make foreign aid donations
more effective. I expect ethnic conflict, genocide and revolution to negatively influence
the effectiveness of foreign aid, but leave open the possibility that ethnic conflict could
positively influence aid’s impact based on Easterly’s study. Furthermore, I expect
landlocked countries to experience additional positive gains from foreign aid, since they
are at a trade disadvantage.
4. Results and Analysis
The results of the TSLS regression are shown below in Table 2:
14
Table 2: TSLS Regression Results
Parameter
Estimate
Std. Error
t-Statistic
Prob.
Constant Term
GDP(-1) [Lambda]
Household Consumption
Government Expenditures
Exports*Petroleum Exporter
Imports
Agricultural Production
Gross Capital Formation
Inflation
Openness to Trade
Energy Consumption
Energy Cons.*Low Dev.
Less than Half of Land in Tropics (1=Yes)
Foreign Direct Investment
Foreign Aid
Foreign Aid*Ethnic Conflict
Foreign Aid*Genocide*Low Dev.
Foreign Aid*Revolution
Foreign Aid*Landlocked
Foreign Aid*Landlocked*Low Dev.
0.091
0.233
6.307
4.505
9.825
-3.746
10.976
7.262
-0.001
0.020
-0.013
-0.052
0.742
0.000
0.001
0.001
-0.017
-0.001
0.002
-0.003
0.400
0.087
2.241
1.305
1.866
0.963
1.992
0.834
0.000
0.005
0.004
0.014
0.326
0.000
0.000
0.000
0.009
0.000
0.001
0.001
0.228
2.692
2.814
3.452
5.266
-3.891
5.510
8.703
-2.282
4.301
-3.212
-3.822
2.275
2.124
3.233
2.202
-1.948
-2.731
1.847
-2.320
0.820
0.007
0.005
0.001
0.000
0.000
0.000
0.000
0.023
0.000
0.001
0.000
0.023
0.034
0.001
0.028
0.052
0.006
0.065
0.021
R-squared
Adjusted R-squared
0.415
0.408
S.E. of regression
Durbin-Watson stat
4.535
2.069
The model can be written as in general terms as follows:
GDPi ,t  0.091  0.233GDPi ,t 1  0.001ODAi ,t   1ODAi ,t * DUMMYi ,t   2 Z i ,t
(3)
Where:
GDP = Gross Domestic Product Growth Rate (for country i at time t)
ODA = Official Development Assistance (for country i at time t)
DUMMY = Vector for Dummy Variables (for country i at time t)
Z = Vector for All Other Variables (for country i at time t)
The results of the regression indicate that approximately 42% of the variation in
GDP growth rates is explained by the variables included in the model, as evidenced by
15
the R-squared value. Further, each coefficient estimate is significant at the 0.05 level,
with the exception of a few borderline cases and the constant term. These coefficients are
also consistent with my expectations, however the coefficient for the ethnic conflict
dummy did turn out to be in harmony with Easterly’s study of ethnic fractionalization.
The Durbin-Watson statistic fails to conclusively determine the presence of serial
correlation. Further analysis of the residuals, however, indicates that it is not a
statistically significant problem.26 The model was also tested for the presence of
heteroskedasticity, both across time and cross sections using the Breusch-Pagan Test.
The results of this test fail to show statistically significant evidence of
heteroskedasticity.27 Multicollinearity was investigated using a correlation matrix of the
regressors, but no major evidence of this anomaly was detected, either.28
The results provide insight as to foreign aid’s effectiveness in a number of ways.
Most obvious is that it is has a positive, though modest effect on economic growth,
significant at the 0.01 level. Increasing foreign aid by $1 million US will result in an
increase in GDP of approximately 0.001%, ceteris paribus. According to the data, the
average annual amount of official development assistance received over all years and
countries is approximately $570 million US. In this case, aid is estimated to increase
growth in GDP by approximately 0.6%.
As shown in Table 3, however, this impact can be greatly diminished by other
factors, in some cases to the point where aid actually becomes detrimental to growth.
Using the baseline case of a country with no ethnic conflict, revolution, or genocide,
which is not landlocked, and does not suffer from low development, I estimate the
26
See Appendix for additional information regarding serial correlation tests.
See Appendix for methodology and results of the Breusch-Pagan Test.
28
See Appendix for correlation matrix.
27
16
additional impacts of any of those circumstances on economic growth. Those factors
with N/A listed under “Impact” were not statistically significant at the 0.05 level.29
Table 3: Factors Influencing Aid Effectiveness
Factor
Impact
Ethnic Conflict
Ethnic Conflict in Low Development Countries
Genocide/Politicide
Genocide/Politicide in Low Development Countries
Revolution
Revolution in Low Development Countries
Landlocked Country
Landlocked Country with Low Development
0.001
N/A
N/A
-0.017
-0.001
N/A
0.002
-0.003
Overall Impact of Aid +
Additional Factor(s) on GDP
0.002
N/A
N/A
-0.016
0.000
N/A
0.003
-0.002
The model indicates that foreign assistance actually becomes detrimental to
growth in situations where there is genocide or politicide in low development nations, as
predicted. I attribute this to the fact that resources are typically controlled by the
dominant party in genocidal conflicts, and it is likely that aid dollars are siphoned off and
used for their own benefit instead of productive and efficient activities. Revolutionary
conflict eliminates entirely the impact aid has on the economy, resulting a net effect of
about zero. I argue that this is the case because the institutions required to effectively
utilize foreign assistance are in jeopardy during a major transfer of power, reducing their
ability to act efficiently and distribute aid dollars according to the country’s best interests.
Interestingly, ethnic conflict actually increases the effectiveness of aid. This finding is
consistent with Easterly’s study of ethnic fractionalization and its impact on economic
growth.
29
Other factors were tested but failed to show statistical significance, including dummies for Low
Development, Sub-Saharan Africa, Openness to Trade, Afrotropic Climate, Tropical Geography, and Major
Petroleum Exporters.
17
In landlocked countries, aid is particularly effective, tripling the extent to which it
impacts economic growth. As Sachs and Warner pointed out, landlocked countries are
limited in their ability to engage in global trade. Thus, it seems reasonable that foreign
aid positively impacts growth in these areas since their capacity to engage in trade is
restricted. However, in low-development countries that are landlocked, this relationship
no longer holds. This indicates that whatever benefits aid has in landlocked countries is
reversed in low-development countries, possibly due to poor institutional quality,
corruption, or other factors.
As for other variables besides foreign aid, the model shows the effect of foreign
direct investment (FDI) on economic growth is surprisingly small; an increase of only
0.00003% in GDP for every $1 million US invested. In contrast, foreign aid boosts GDP
by 0.001% with the same amount of money. This indicates that foreign aid has a
substantially greater impact on growth than foreign direct investment, all else equal.
According to the model, being open to trade seems to be a much more effective strategy
in growing the economy, even more so than foreign aid and FDI. It is important to note,
however, that since openness to trade is measured as a share of GDP, the impact is not
directly comparable that of foreign aid or FDI, since economies included in this study
vary greatly in size.
To quantify how quickly foreign aid impacts the economic growth of a country, I
calculate the median lag as outlined by Davies and Quinlivian (2006).30 This measure
estimates how quickly half of the impact of foreign assistance is felt, and is calculated as
follows:
Antony Davies and Gary Quinlivan, “A Panel Data Analysis of the Impact of Trade on Human
Development,” Journal of Socioeconomics (2006).
30
18
Median Lag =
ln 0.5
= 0.477
ln  
(4)
A median lag of 0.477 indicates that in approximately 5.7 months, half of the
entire impact of foreign aid on GDP growth will be realized. Half of the remaining
impact is then felt in another 5.7 months, and so on, as the cumulative impact of the aid
asymptotically approaches 100%. This phenomenon is illustrated in Chart 1 below.
Graph 1: Cumulative Impact of Foreign Aid on Growth
Cumulative Impact on GDP
Growth (% of Total)
Cumulative Impact on GDP Growth
100%
80%
60%
40%
20%
0%
0
5
10
15
20
25
30
35
40
Time (Months)
The median lag indicates that aid can quickly impact an economy, but for a relatively
short amount of time. After only two years of circulation in the recipient economy, over
95% of the total impact of foreign aid is experienced.
5. Conclusions and Suggestions for Future Research
The purpose of this analysis was to determine the effects of development
assistance on economic growth. The model developed in this paper provides evidence
supporting the contention that foreign aid positively impacts economic growth in the
19
developing world. Therefore, it is not in the interest of developed countries and
international bodies to discontinue aid programs. Moreover, as Gunning (2004) points
out, it would be extremely difficult for a donor country to stop aid since it would be seen
by both the domestic and foreign populations as punishing an already poor country.31
The model also shows, however, that the effects of aid on economic growth are
modest, and “buying” economic growth through foreign aid would be incredibly
inefficient and expensive. For instance, using foreign aid alone to increase GDP by 1%
in a country would require a foreign aid package of approximately $1 billion US. With
almost 120 countries identified as low- and medium-development, spurring economic
growth in developing world to desirable levels would be an enormous expenditure. This
also assumes that the negative effects of conflict and geography shown to be significant
in the model do not apply, and ignores the potential problems of aid dependence,
corruption, and bureaucratic erosion that research has associated with high levels of
foreign aid.
The aforementioned studies by Burnside and Dollar (1997) and others have
shown aid to be more effective in sound economic policy environments. Thus, donor
governments and multilateral institutions should continue to push economic reforms and
trade liberalization on recipient governments. Not only will this improve the
effectiveness of foreign aid according to these studies, but it will also result in less aid
being required.
The armed conflict dummies indicate, with the exception of ethnic conflict, that
state failure and political instability reverse the positive effect of aid, even making it
Jan Willem Gunning, “Why Give Aid,” Presented to the 2 nd Annual AFD-EUDN Conference
Development Aid: Why and How, Paris, 25 November 2004.
31
20
detrimental to economic growth in some cases. Therefore, donor governments should be
aware of the political situations in recipient countries, and work with international bodies
to ensure as much stability as possible. Further, since geography is essentially fixed,
foreign aid donations to landlocked countries should be designed to facilitate
improvements in transportation infrastructures, which increase their capacity to engage in
trade.
Future research should further explore the role of sound economic policies and
good governance in aid effectiveness. Scholars should also explore other ways of
quantifying climate, tropical geography, and governance to provide for additional testing
of potential impacts on the effectiveness of foreign aid. Finally, future study of foreign
aid should also investigate its effects on economic development, instead of growth.
Doing so will shed light on the question of whether aid actually improves the quality of
life in lesser developed countries.
21
References
Ali, Abdiweli M., and Hodan S. Isse. "An Empirical Analysis of the Effect of Aid on Growth."
International Advances in Economic Research 11 (2005): 1-11. JSTOR. Gumberg
Library, Pittsburgh, PA. 2 Oct. 2006.
Burnside, Craig, and David Dollar. "Aid, Policies, and Growth." American Economic Review 90
(1997): 847-868. 12 Sept. 2006.
Cleveland, H. Vanburen. "Purposes of International Aid Programs." Scientific Monthly 85
(1957): 77-81. JSTOR. Gumberg Library, Pittsburgh, PA. 14 Nov. 2006.
Dacy, Douglas C. "Foreign Aid, Government Consumption, Saving, and Growth in LessDeveloped Countries." The Economic Journal 85 (1975): 548-562. JSTOR. Gumberg
Library, Pittsburgh, PA. 05 Nov. 2006.
Davies, Antony, and Gary Quinlivan. "A Panel Data Analysis of the Impact of Trade on Human
Development." Journal of Socioeconomics (2006). 20 Nov. 2006
<http://www.business.duq.edu/faculty/davies>.
Durbarry, Ramesh, Norman Gemmell, and David Greenaway. "New Evidence on the Impact of
Foreign Aid on Growth." Center for Research in Economic Development and
International T Rade 98 (1998). 04 Nov. 2006
<http://www.nottingham.ac.uk/economics/credit/research/papers/cp.98.8.pdf>.
Easterly, William, and Ross Levine. "Africa's Growth Tragedy: Policies and Ethnic Divisions."
The Quarterly Journal of Economics 112 (1997): 1203-1250. JSTOR. Gumberg Library,
Pittsburgh, PA. 20 Oct. 2006.
Easterly, William. "Does Foreign Aid Add Up?" Foreign Policy 125 (125): 94. JSTOR.
Gumberg Library, Pittsburgh, PA. 28 Oct. 2006.
22
Emerging Issues and Developments At the Regional Level: Least Developed, Landlocked, and
Island Developing Countries. Report of the Special Body on Least Developed and
Landlocked Developing Countries on Its Fourth Session, 22 Apr. 1999, United Nations
Economic and Social Commission for Asia and the Pacific. 01 Dec. 2006.
Gunning, Jan W. "Why Give Aid." Development Aid: Why and How? 2nd AFD-EUDN
Conference, Paris. 25 Nov. 2004. 28 Nov. 2006.
"Handbook of Statistics." 2006. United Nations Conference on Trade and Development. 10 Oct.
2006 <http://www.unctad.org>.
Hjertholm, Peter, and Howard White. "Survey of Foreign Aid: History, Trends, and Allocation."
University of Copenhagen Institute of Economics Discussion Papers 2000 (2000). 12
Sept. 2006.
Human Development Report. 2005. United Nations Development Programme. 11 Sept. 2006
<http://hdr.undp.org/reports/global/2005/pdf/HDR05_HDI.pdf>.
Knack, Stephen. "Aid Dependence and the Quality of Governance: Cross-Country Empirical
Tests." Southern Economic Journal 68 (2001): 310-329. JSTOR. Gumberg Library,
Pittsburgh, PA. 10 Sept. 2006.
Marshall, Monty G., comp. Political Instability Task Force. Vers. Phase IV. 1994. George Mason
University. 10 Oct. 2006 <http://globalpolicy.gmu.edu/pitf/>.
"Official Energy Statistics From the U.S. Government." Energy Information Agency. 2006. U.S.
Department of Energy. 10 Oct. 2006 <http://www.eia.doe.gov>.
Rajan, Raghuram G., and Arvind Subramanian. "Aid and Growth: What Does the Cross-Country
Evidence Really Show?" International Monetary Fund Working Papers 05 (2005). 12
Sept. 2006 <http://www.adc.org.au/Resources>.
23
Sachs, Jeffery, and Andrew Warner. "Sources of Slow Growth in African Economies." Harvard
Institute for International Development, Development Discussion Papers 545 (1996). 14
Nov. 2006 <http://www.earth.columbia.edu/about/director/documents/jrnafec1297.pdf>.
Tarp, Finn, ed. Foreign Aid and Development: Lessons Learnt and Directions for the Future.
New York: Routledge, 2000. Google Books. 18 Nov. 2006 <http://books.google.com>.
"World Economic Outlook Databases." Sept. 2006. International Monetary Fund. 10 Oct. 2006
<http://www.imf.org>.
24
Appendix A: Included Countries
Medium-Development Countries
Albania
Algeria
Antigua and Barbuda
Armenia
Azerbaijan
Bangladesh
Belarus
Belize
Bhutan
Bolivia
Bosnia and
Herzegovina
Botswana
Brazil
Cambodia
Cape Verde
China
Colombia
Comoros
Congo, The
Dominica
Dominican Republic
Ecuador
Egypt
El Salvador
Equatorial Guinea
Fiji
Gabon
Georgia
Ghana
Grenada
Guatemala
Guyana
Honduras
India
Indonesia
Iran
Jamaica
Jordan
Kazakhstan
Kyrgyzstan
Laos
Lebanon
Libya
Macedonia
Malaysia
Maldives
Mauritius
Moldova
Mongolia
Morocco
Myanmar
Namibia
Nepal
Nicaragua
Oman
Pakistan
Papua New Guinea
Paraguay
Peru
Philippines
Romania
Russia
St Lucia
St Vincent-Grenadines
Samoa
Sao Tome and Principe
Saudi Arabia
Solomon Islands
South Africa
Sri Lanka
Sudan
Suriname
Syria
Tajikistan
Thailand
Timor-Leste
Togo
Tunisia
Turkey
Turkmenistan
Uganda
Ukraine
Uzbekistan
Vanuatu
Venezuela
Vietnam
Zimbabwe
Low-Development Countries
Angola
Benin
Burkina Faso
Burundi
Cameroon
Central African Republic
Chad
Congo, The
Côte d’Ivoire
Djibouti
Eritrea
Ethiopia
Gambia
Guinea
Guinea-Bissau
Haiti
Kenya
Lesotho
Madagascar
Malawi
Mali
Mauritania
Mozambique
Niger
Nigeria
Rwanda
Senegal
Swaziland
Tanzania
Yemen
Zambia
25
Appendix B: Autocorrelation Test Results
The regression returns a Durbin-Watson Statistic of 2.069. Critical Values for a sample
of over 2,000 observations with 20 coefficients (including the constant term) are:
DL = 1.907
DU = 1.946
4- DL = 2.093
4- DU = 2.054
Since the Durbin-Watson statistic returned by the regression falls between 2.054 and
2.093, there is inconclusive evidence of autocorrelation at the 0.05 level. However,
further analysis of the residuals indicate that there is no statistically significant evidence
of autocorrelation or partial correlation, as shown in the table below:
Lag
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
AC
-0.029
0.033
0.011
0.057
0.020
0.027
0.014
-0.010
-0.032
0.035
-0.002
0.041
0.017
0.021
0.008
0.036
-0.009
0.006
0.007
-0.006
-0.002
-0.024
0.012
-0.041
-0.026
0.015
0.005
0.037
0.014
0.017
0.030
0.009
0.001
0.006
0.013
-0.002
PAC
-0.029
0.032
0.013
0.057
0.023
0.025
0.013
-0.015
-0.037
0.030
0.000
0.041
0.022
0.018
0.007
0.030
-0.014
-0.002
0.005
-0.011
0.000
-0.026
0.010
-0.040
-0.029
0.014
0.009
0.040
0.020
0.016
0.029
0.005
-0.009
0.003
0.010
0.001
Q-Stat
1.498
3.367
3.594
9.394
10.107
11.428
11.788
11.966
13.737
15.880
15.886
18.939
19.433
20.205
20.306
22.609
22.762
22.831
22.921
22.993
22.997
24.047
24.309
27.382
28.627
29.016
29.062
31.449
31.802
32.335
33.933
34.064
34.065
34.131
34.452
34.459
Prob
0.221
0.186
0.309
0.052
0.072
0.076
0.108
0.153
0.132
0.103
0.145
0.090
0.110
0.124
0.160
0.125
0.157
0.197
0.241
0.289
0.344
0.345
0.387
0.287
0.280
0.310
0.358
0.298
0.329
0.352
0.328
0.369
0.416
0.461
0.494
0.542
26
Appendix C: Heteroskedasticity Test Results
In accordance with the Breusch-Pagan test, I capture the residuals from the regression,
square them, and regress them on a time dummy variable and a cross-section dummy
variable. Significance of either coefficient at the 0.05 level would indicate an
inconsistent variance of the error terms, and the presence of the heteroskedasticity.
Dependent Variable: RESIDSQUARED
Method: Least Squares
Date: 12/07/06 Time: 12:58
Sample (adjusted): 29 3092
Included observations: 1760 after adjustments
Variable
C
TIMEDUMMY
COUNTRYDUMMY
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
Durbin-Watson stat
Coefficient
12.671
0.034
0.128
0.001792
0.000656
97.38902
16664478
-10554.37
1.710709
Std. Error
6.924
0.368
0.072
t-Statistic
1.830
0.092
1.774
Mean dependent var
S.D. dependent var
Akaike info criterion
Schwarz criterion
F-statistic
Prob(F-statistic)
Prob.
0.067
0.927
0.076
20.33291
97.42097
11.99701
12.00634
1.577169
0.206851
As shown in the regression results above, neither the time dummy nor cross-section
dummy were significant at the 0.05 level, indicating homoskedasticity.
27
Appendix D: Correlation Matrix
X1
X2
X3
X4
X5
X6
X7
X8
X9
X10
X11
X12
X13
X14
X15
X16
X17
X1
X2
0.028
X3
0.017
-0.123
X4
-0.007
-0.148
-0.044
X5
0.020
0.076
0.058
0.050
X6
0.001
0.181
0.017
-0.125
0.030
X7
0.027
0.043
0.156
-0.053
0.277
0.185
X8
-0.004
0.033
-0.051
0.011
0.137
-0.030
0.072
X9
-0.346
0.047
0.074
0.007
0.086
-0.006
0.063
-0.043
X10
-0.176
0.000
0.029
0.086
0.013
-0.009
-0.005
-0.017
0.193
X11
-0.068
0.002
-0.020
0.007
-0.009
0.001
-0.012
-0.010
0.149
0.014
X12
0.355
0.010
0.004
0.030
0.003
0.011
-0.013
-0.025
-0.163
0.318
-0.085
X13
0.298
0.028
0.021
-0.003
0.018
0.017
0.012
-0.009
-0.081
0.094
-0.041
0.263
X14
0.627
0.009
0.015
0.008
0.007
0.002
0.002
-0.012
-0.225
-0.102
-0.074
0.162
0.257
X15
0.000
0.034
-0.021
0.105
0.116
-0.066
-0.041
0.031
0.065
-0.043
0.082
-0.026
0.000
X16
0.336
0.003
0.010
0.010
0.007
-0.016
-0.005
0.041
-0.104
-0.042
-0.033
0.165
0.006
0.164
0.081
X17
0.124
0.022
-0.008
-0.033
0.022
0.012
0.032
-0.013
-0.186
-0.188
0.013
-0.087
-0.057
-0.014
0.025
-0.061
X18
0.104
0.036
-0.006
-0.026
0.020
0.031
0.064
-0.022
-0.145
-0.180
0.065
-0.085
-0.047
-0.036
0.045
-0.061
0.059
0.797
X18
28
30