Professional Documents
Culture Documents
Easterly Etalaer04
Easterly Etalaer04
net/publication/4732180
CITATIONS READS
670 4,570
3 authors:
David Roodman
Bill & Melinda Gates Foundation
43 PUBLICATIONS 9,903 CITATIONS
SEE PROFILE
All content following this page was uploaded by Ross Levine on 17 May 2014.
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=aea.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
American Economic Association is collaborating with JSTOR to digitize, preserve and extend access to The
American Economic Review.
http://www.jstor.org
Aid, Policies, and Growth:Comment
By WILLIAM ROSSLEVINE,ANDDAVIDROODMAN*
EASTERLY,
variable enters insignificantly (we will show stage least squaresas shown in columns (3) and
these results below). (4) of Table 1. We use the same set of instruments
We performthe same exercisewith BD regres- as BD. We areagainableto reproducetheirresults
sion 8 for the sample of low-income countries with our data set (see Table 2 below).
(also following them in omitting outliers). BD The aid*policy term is insignificant in their
note that low-income countriesmight be a pre- regression8 when we simply add all the datafor
ferred sample to detect the effects of aid, and low-income countries that we can collect for
indeed theiraid*policyinteractiontermis signif- 1970-1993 and the data for 1994-1997 [col-
icant in both OLS and two-stage least squares umn (4)]. The coefficient not only becomes
(2SLS) in theirregression8. In orderto check the insignificant,but changes sign. Oursampleis 52
robustnessof the estimates of the instrumental observationslarger than the BD sample for re-
variablesestimates, we do the exercise in two- gression 8.
VOL.94 NO. 3 EASTERLYET AL.: AID, POLICIES,AND GROWTH:COMMENT 777
ELR data, full sample, outliers included, 0.05 0.07 0.00 -0.06
1970-1993 (0.82) (0.86) (0.03) (-0.52)
Observations 300 300 205 205
ELR data, full sample, outliers included, 0.05 0.06 -0.01 -0.08
1970-1997 (0.81) (0.79) (-0.06) (-0.73)
Observations 356 356 244 244
Notes: ELR data refers to the data set constructedfor this paper as describedin the text. All
regressions omit outliers, either in the original Burnside and Dollar results as described in
their paper, or in the ELR results using the Hadi method, except where otherwise noted.
T-statisticsare in parentheses.The numberof observationsis given below the t-statistics.
* Significant at the 5-percent level.
** Significant at the 1-percentlevel.
The fragile results on aid effectiveness re- sion, for using their countries but the whole
main evident when varying the sample. For period sample or for their sample period but all
brevity, Table 2 shows only the aid*policy co- countries, and for samples excluding outliers
efficients, t-statistics, and number of observa- and for samples including outliers. Not only
tions for OLS and 2SLS for regressions 5 and 8 does significance vanish, but the magnitude of
for various combinations of sample periods, the coefficient changes greatly across the differ-
country samples, and when including and ex- ent permutations.
cluding outliers. We reproduce statistical signif- The only significant coefficient out of our
icance when restricting our data to the various permutations was for OLS for regres-
Burnside-Dollar sample period and sample of sion 8 (the low-income sample) using the
countries, though the coefficient sizes are larger Burnside-Dollar countries for the full sample
when using the new data. The significance of period. Since this is one significant coefficient
the relationship between growth and the at the 5-percent level out of 20 permutations, we
aid*policy interaction term vanishes, however, do not think this provides strong support for the
if we relax either the sample period constraint or robustness of the Burnside-Dollar results.
the country selection constraint for either re- We tried all of these same exercises for the
gression 5 or 8 (i.e., the whole sample and only other aid*policy regressions that BD report in
the low-income sample). The significance van- the paper. Burnside and Dollar found the
ishes for both OLS and 2SLS in either regres- aid*policy term to be significant and positive
778 THEAMERICANECONOMICREVIEW JUNE 2004
GAB3
11.967 -
CMR4
SYR3
NWGY5
BWA4
ML16
BWA5
._
o SL-PA' BWA6
ARG5
ZMB6 HND7
BRA7
0 Me7
0
0
CMR7
GABtC4
-10.7188 ETH5
I I I I
-8.01298 9.7002
Aid x policy
- SYR3
8.2642 NGA2
GAB2DOlQR4 ARG7
PRY4
EGY3 TGQaptjVMr BWA4
SYR4 X
BWA5
ECU2 SYWEN2 BWA6
UGA6 8*6,ftq UGA8
UGA7 CH. PNG7
ARG3 PNG7
cd
BRA5
0. :MB6 GMB5TG03
0 GMB4 j4 GHAER6ARG5
ARG5 MLIJOR5
' I
NIC8
CI NIC5 M
"
JOR7 ML17
cr- IL3
o
PNG6
v RS^L5PNG I8
ZAR7 GA IOR6
jMR
109R7
SLE8
NIC4
IRN4
-10.528 -
I~ I
-6.03525 5.0871
Aid x policy
Notes: These partial scatterplotsare from regressions 1 and 3 in Table 1. The top graph
representsBurnside-Dollaroriginalresults;the bottomgraphshows resultsusing the new data
set. The partial scatterplotinvolves the two-dimensional representationof the relationship
between growth and aid*policy controllingfor the other regressors.Thus, we regress growth
against all of the regressors listed in Table 1 except aid*policy and collect these growth
residuals.Then we regress aid*policy againstthe same regressorsand collect these aid*policy
residuals.The figuresplot the growthresidualsagainstthe aid*policy residualsalong with the
regressionline. Point labels are three-letterISO countrycodes followed by a digit for the time
period (2 = 1970-1973; 3 = 1974-1977, etc.).
when they did NOT exclude outliers but added results with our data set using their sample
another term aid2*policy (which was significant period and sample of countries(Table 3). When
and negative). Their results were significant in we try these specifications with our expanded
OLS for the whole sample and the low-income data set, the previous pattern holds: the
sample, but not in 2SLS, so we reportonly the aid*policy interaction term is not robust to the
OLS results. We are able to reproduce their use of new data, including various permuta-
VOL.94 NO. 3 EASTERLYET AL.: AID, POLICIES,AND GROWTH:COMMENT 779
4/OLS 7/OLS
Burnside and Dollar original aid*policy 0.20* 0.27*
(2.07) (2.03)
aid2*policy -0.02* -0.02*
(-2.22) (-2.45)
Observations 275 189
Notes: ELR data refers to data set constructedfor this paper as described in text. T-statistics
are in parentheses;Observationsare below t-statistics.
* Significant at the 5-percent level.
** Significant at the 1-percentlevel.
tions of period and country selection. In our data all the way up to the cross-section for the
full sample and in some of the other permu- full sample). These exercises (available upon
tations, the coefficients on the aid*policy and request) did not change our conclusion about
aid2*policy reverse sign from the BD results. the fragility of the aid*policy term-the
Thus, the result of our paper is as follows: aid*policy term is not robust to alternative
adding new data creates new doubts about the equally plausible definitions of aid and policy,
BD conclusion. When we extend the sample or to alternativeperiod lengths.
forward to 1997, we no longer find that aid
promotes growth in good policy environments. II. Conclusions
Similarly, when we expand the BD data by
using the full set of data available over the This paper reduces the confidence that one
original BD period, we no longer find that aid can have in the conclusion that aid promotes
promotes growth in good policy environments. growth in countries with sound policies. The
Our findings regardingthe fragility of the aid- paper does not argue that aid is ineffective. We
policy-growthnexus is unaffectedby excluding make a much more limited claim. We simply
or including outliers. note that addingadditionaldata to the BD study
We also experimentedwith alternativedefi- of aid effectiveness raises new doubts aboutthe
nitions of "aid"and "good policies," as well as effectiveness of aid and suggests that econo-
trying different period lengths (from annual mists and policy makersshouldbe less sanguine
780 THEAMERICANECONOMICREVIEW JUNE 2004
about concluding that foreign aid will boost of DevelopmentStudies, August 2001, 37(6),
growth in countries with good policies. We be- pp. 66-92.
lieve that BD should be a seminal paper that Hansen, Henrik and Tarp, Finn. "Aid Effec-
stimulatesadditionalwork on aid effectiveness, tiveness Disputed." Journal of Interna-
but not yet the final answeron this criticalissue. tional Development, April 2000, 12(3), pp.
We hope that furtherresearch will continue to 375-98.
explore pressing macroeconomicand microeco- . "Aid and GrowthRegressions."Jour-
nomic questions surroundingforeign aid, such nal of Development Economics, April 2001,
as whether aid can foment reforms in policies 64(2), pp. 547-70.
and institutions that in turn foster economic Knack,Stephenand Keefer,Philip. "Institutions
growth, whether some foreign aid delivery and Economic Performance:Cross-Country
mechanisms work better than others, and what Tests Using Alternative Institutional Mea-
is the political economy of aid in both the donor sures." Economics and Politics, November
and the recipient. 1995, 7(3), pp. 207-27.
Lensink,Robertand White,Howard."Are There
REFERENCES NegativeReturnsto Aid?"Journalof Develop-
ment Studies,August 2001, 37(6), pp. 42-65.
Beattie,Alan. "A Measure of Good Intentions." Mosley,Paul; Hudson,John and Horrell,Sara. "Aid,
Financial Times, March 11, 2002, p. 23. the Public Sector and the Market in Less
Boone, Peter. "Aid and Growth."Mimeo, Lon- Developed Countries." Economic Journal,
don School of Economics, 1994. September 1987, 97(387), pp. 616-41.
. "Politics and the Effectiveness of For- Papanek, Gustav F. "The Effect of Aid and
eign Aid." European Economic Review, Other Resource Transfers on Savings and
1996, 40(2), pp. 289-329. Growth in Less Developed Countries."Eco-
Burnside,Craig and Dollar, David. "Aid, Poli- nomic Journal, September1972, 82(327), pp.
cies, and Growth."American Economic Re- 934-50.
view, September2000, 90(4), pp. 847-68. The Economist. "Help in the Right Places."
Bush, GeorgeW. Speech at Inter-AmericanDe- March 16, 2002.
velopment Bank, Washington, DC, March U.K. Department for International Development.
16, 2002. Eliminating world poverty: Making global-
Cassen, Robert. Does aid work? Report to an isation work for the poor. White Paper on
intergovernmentaltaskforce. New York:Ox- InternationalDevelopment presented to Par-
ford University Press, 1986. liament by the Secretaryof State for Interna-
Collier,Paul and Dehn, Jan. "Aid, Shocks, and tional Development by Command of Her
Growth." World Bank (Washington, DC) Majesty, London, December 2000.
Working Paper No. 2688, October 2001. WashingtonPost. "Does Aid Help?"February9,
Collier,Paul and Dollar,David. "Aid Allocation 2002, p. A26.
and PovertyReduction."EuropeanEconomic White House. Fact sheet on Millennium Chal-
Review, September2002, 45(1), pp. 1-26. lenge Account,Washington,DC, 2002; www.
Dalgaard,Carl-Johanand Hansen,Henrik. "On whitehouse.gov.
Aid, Growth and Good Policies." Journal of WorldBank.Adjustmentin Africa: Reforms,re-
Development Studies, August 2001, 37(6), sults, and the road ahead. New York:Oxford
pp. 17-41. University Press, 1994.
Easterly, William and Levine, Ross. "Africa's . Globalization, growth and poverty:
Growth Tragedy: Politics and Ethnic Divi- Building an inclusive world economy. Lon-
sions." QuarterlyJournal of Economics, No- don: Oxford University Press, 2001.
vember 1997, 112(4), pp. 1203-50. . A case for aid: Building a consensus
Guillaumont,Patrick and Chauvet, Lisa. "Aid for development assistance. Washington,
and Performance:A Reassessment."Journal DC: World Bank, 2002.