T=
he
Increasing Selectivity of Foreign Aid.
Dollar and Levin, WD =
34(12):
2034-2046
The “Monterrey
Consensus” emerged from the UN conference in 2002.
Developing countries =
own
institutions and policies are key to sustained
development.
Sound policies and go=
od
governance at all levels are necessary to ensure ODA effectiveness.
Case studies suggest =
this is
true –
Project level studies=
(like
the sowing and reaping paper we will look at).
M=
oney if
fungible, so the project outcomes are not conclusive. T=
hat is,
if the project did not fund it, would the private sector / government have
funded it?
Thus, look at a third=
kind of
evidence, macroeconomic indicators.
That is the spirit of the Burnside and Dollar paper we just looked a=
t.
Leave aside the quest=
ion of
whether aid effectiveness is conditional on ‘good governance’.<=
span
style=3D'mso-spacerun:yes'>
Here, has the behavio=
r of
donors changed since they signed on to
I=
ndicators
of poverty and institutional quality. How are=
aid
flows conditioned on each of these?
U=
sing the
International Country Risk Guide as a measure of the economic institutions.=
U=
sing the
Freedom House democracy index as a measure of political institutions.
Other variables (colo=
nial relationships,
commercial ties, distance).
L=
ooking at
multilateral as well as bilateral.
Fitting in to what we=
have
been looking at with the recent articles – growing agreement that
institutions matter.
This has given rise t=
o the
idea of institutional selectivity.
However, at the same time, there is agreement that aid should go whe=
re
the need is the greatest. The=
se can
be at odds with each other.
L=
ooking at
OECD DAC data, net of emergency aid.
U=
sing WDI
indicators for some of the data.
E=
quation on
page 4 of the in press version, 2037 of printed version.
Tobit estimation (whe=
n there
is a probability mass at zero).
Log, so add one so it=
does
not go to negative infinity!
Groups: bilateral tha=
t had
colonies, bilateral that did not have colonies, multilateral.
Rows are different ty=
pes of
donors, columns are different regressors.&=
nbsp;
The freedom house ind=
ex used
has higher numbers meaning more authoritarian (so it is also possible to st=
ate
that aid has a negative relationship with authoritarianism). Clear relations=
hip
between increased freedom and increased aid.
Not as sharp a findin=
g for
economic institutions.
The colonial dummy te=
nds to
be positive.
Trade variables are n=
ot all
that relevant for most (with the exceptions of
Tend to give in their=
area,
as negative relationship distance to aid for most.
In section three, the=
same
type of analysis, but now add in earlier periods. 1984-89, 1990-94, 1995-99, 2000-03=
.
Table two focuses on =
the key
variables of democracy, rule of law, and poverty over time. Looking at elasticities – ho=
w does
aid respond to a change in the variable?
Democracy varies in
magnitude, but not in sign over time.
Rule of law changes s=
ign, and
the significance is being driven by the multilaterals.
P=
overty
consistent in sign.
C=
onclusion
in section four.
=
Alesina and Weder, AER 92(4=
):
1126-1137. 2002
Do corrupt governments
receive less foreign aid?
Paper has three core
questions (p. 1126)
First, do less corrupt
governments receive more aid or debt relief?
Second, do various do=
nors
differ in their willingness to discriminate amongst corrupt governments?
Third, does foreign a=
id
reduce corruption?
They contend that one=
strand of
literature they engage looks at the determinants and effectiveness of forei=
gn
aid. The results to date:
=
1) <=
/span>Foreign aid is most often used for largely waste=
ful
public consumption.
=
2) <=
/span>Countries following good policies are helped by
foreign assistance, but the probability of adopting good policies is not
influenced by the aid flow.
=
3) <=
/span>Donor countries disburse aid as a function of
strategic considerations rather than needs.
Another theme in the
literature is the impact of corruption.&nb=
sp;
I=
ssue of
reverse causality yet again.=
Does the fact that poorer countries
receive more aid mean aid causes poverty?&=
nbsp;
Likewise with institutions.
However, with corrupt=
ion it
is a bit easier.
Corruption is hard to
measure. They use seven indic=
ators
of corruption from six different sources.&=
nbsp;
International Country=
Risk
Guide (used as a source for the rule of law index in Dollar and Levin and
generated by a group in
World Development Rep=
ort /
S=
tandard
and Poors.
B=
usiness
International (EIU).
World Competitiveness
Yearbook form the Institute for Management Devel=
opment
in
T=
ransparency
International.
Table 1 on 1129 summa=
rizes
the data and the sources.
Table two illustrates=
the correlation
across the corruption indicators.
Though not perfect, all positive, mostly large correlations. 22 of 28 are above 0.5.
Other regressors:
Level of income
Size
Trade openness
Index of political ri=
ghts
(here high is more democratic)
Years as a colony
Frequency of voting w=
ith
donor in UN
Indicators for
S=
everal
time periods, from 1970 to 1995.
T=
able three
and four. Three give general pattern, four
presents results for the corruption index for different types of measures.<=
span
style=3D'mso-spacerun:yes'> Four is a table of t-stats, all sh=
owing
not significant (only one significant at the 10% level).
No evidence that less=
corrupt
countries receive more aid (or that more corrupt get more / less for that
matter).
S=
ome
information on debt relief in section B.
D=
ebt relief
not influenced by corruption. See table 5 on page 11=
33 and
six for different corruption measures.
Table 7 looks at indi=
vidual
donors.
First column gives th=
e most interesting
pattern of the results (though how about those colony results for
[If democracy is as
described, I think the discussion on pages 1134-5 is incorrect. The
Table 8 tries to get =
at the
dynamics of corruption and aid.
Change in corruption is the dependent variable now. Since this is coded such that high=
er
numbers mean less corruption, a positive coefficient is a corruption reduct=
ion,
a negative a corruption increase.
Sowing and Reaping, D=
ollar
and Levin, WBPRWP 3524, February 2005
The macro study resul=
ts are
clearly under debate.
Look at micro evidenc=
e with
this study.
P=
roject level
indicators, economic rate of return and success / failure variables from the
Operations and Evaluations department.
This paper also fits =
into the
‘institutions rule’ debate.&nb=
sp;
World Bank funded pro=
jects,
nearly 100 different countries during the 1990s.
OED reports to the
shareholders, and has a basic ‘thumbs up / thumbs down’ on whet=
her
the project met the original objectives.&n=
bsp;
All countries that ha= d at least 3 WB funded projects. <= o:p>
Variation:
Rule of law index from
ICRG. Higher scores better.
F= reedom house with 1 being highest level of freedom, 7 being lowest.<= span style=3D'font-size:20.0pt'> <= o:p>
Also as a check use t=
he
composite measure of KKZ.
Log of per capital GD=
P in
1990.
Overall level of aid =
relative
to GDP – increasing returns, or diminishing returns arguments could be
made.
Geography – sha=
re of
territory in the tropics / regional dummies.
Equation on page 7 / =
table on
page 17.
Political freedom and=
rule of
law significant and as predicted.
Initial income and aid are not significant.
‘The most robus=
t result
in these OLS regressions is that rule of law has a strong positive relation=
ship
to project success, and that geography does matter in that Sub-Saharan Afri=
can
countries have lower success rates, ceteris paribus.’ (p. 8).
Reverse causation:
Successful projects b=
uild
strong institutions?
Visibly successful co=
untries
with good institutions are more likely to get better evaluations / halo eff=
ect
/ benefit of doubt.
Aid may be that those=
getting
more aid need more help and are more likely to have worse outcomes.
Instruments: our by now old friends: share of the population speaking E=
nglish
/ European language, distance from the equator, level of population, and
interactions to get 8 instruments.
Table 2 is a robustne=
ss check
using more countries and the KKZ institutional composite indicator.
Section 3: different kinds of things being
funded. Some are policy orien=
ted,
some are investment oriented.
Differentiate between
adjustment and investment.
Law and democracy imp=
act
adjustment operations, but for investment loans, the coefficient on democra=
cy
is not significant.
Democracy seems more
important for the policy kinds of aid, rule of law for the investment kind =
of
aid.
&=
#8216;Our
results make it very unlikely that aid works the same everywhere’ (p.
13).