Econ 461: Economic Development Lecture 23: April 9 Aid effectiveness

1
Econ 461 Economic DevelopmentLecture 23 April
9Aid effectiveness
Sandra Poncet Lorch Hall, room 207 Email
sponcet_at_umich.edu Office Hours Mondays 1630-18
Wednesdays 1030-12
2
Introduction
In a very unfavorable context the World Bank
reopened the debate on aid and allocation
effectiveness, with the release of its 1998
report entitled Assessing Aid based on work
from Burnside and Dollar.
International environment change (end of the
Cold War) transformed political foundations of
bilateral aid, long considered as an instrument
for promoting the political and geostrategic
interests of donor countries.
Economic crisis and strong budgetary constraints
weighing on several donor countries, notably
European countries in the Euro zonehas, since
1992, led to a brutal fall in aid flows to
developing countries.
Problems of increasing debt in recipient
countries and successive financial crises opened
a broad debate over the reform of international
financial architecture and the role given to
multilateral institutions.
Economic foundations and the justification of
development aid were attacked by numerous very
critical studies emphasized the absence of
macro-economic effectiveness, and the failure of
conditionnalities.
3
Introduction
Literature was inconclusive until the paper by
Burnside and Dollar came out in the mid-90s
One major challenge (as always in the economic
development literature) is the issue of
endogeneity
-reverse causality aid goes where development is
difficult
Implication is tendency to under-estimate the
potential positive impact of aid on growth
-omitted variables factors both correlated with
aid and growth are not taken into account
Two key contributions of Dollar and Burnside
paper take non-random allocation seriously
seek exogenous variation in aid allow effect of
aid to differ as a function of the policy
environment
Debate has been intense since!!
4
Lecture Outline
Introduction I-Dollar and Burnside paper A-Lack
of consensus before BD B-BD approach C-BD results
and implications II-Intense polemic since A-The
implicit hypothesis of BD B-Critics C-Back to
where we were before BD or what else should be
considered? Conclusion
5
I-Dollar and Burnside paper
A-Lack of consensus before BD
Not surprising that this research has mixed
findings for several reasons
-Economic growth is not the sole objective of
foreign aid, and in some cases it is not the
objective at all.
aid given following natural disasters is aimed
at supporting immediate consumption and
humanitarian needs
aid provided to build political systems or
support democracies has growth as only a
secondary and distant objective.
much aid is given primarily for political
purposes support ward action, promote firms
interests
French aid explicitly aims at promoting French
language
-Aid effectiveness will depend on the context
Conditional/Unconditional aid Absorption
capacity of the country Resolution of the
incentives problem
6
I-Dollar and Burnside paper
A-Lack of consensus before BD
Not surprising that this research has mixed
findings for several reasons
-Typical problem with impact evaluation on
economic growth endogeneity due to omitted
variable
Geo Proximity favorable agglomeration forces and
preferential trade Eastern Europe
Successful development
Banned regime (Cuba, Zimbabwe)
Conflict, drought, nuclear threat (North Korea)
SOLUTION need to find an exogenous source of aid
to instrument
If cases 2 and 3 instrumentation will recover a
positive impact
7
I-Burnside and Dollar paper
B-BD approach
Use panel data across 6 four-year periods from
1970-73 and 1990-93 for 56 countries to run
regressions of the form
where ait is instrumented (see below) pit is a
"policy index" increasing in the quality of
policies, i is country and t represents 4-year
periods
so that ß only captures the effect of aid in a
country that has pit 0 ? gt 0 indicates that aid
is more effective in countries that pursue
good policies
8
I-Burnside and Dollar paper
B-BD approach

• To instrument aid they use measures of "political
  influence", that is
• Log population,
• Arms imports,
• Egypt dummy,
• Franc zone dummy,
• (v) Central America dummy

The hypothesis is that (i)-(v) influence growth
only through their effect on aid.
It is however debatable
As a measure of policy, they use an index that
combines (1) trade openness, (2) inflation, (3)
fiscal surplus/gdp
9
I-Burnside and Dollar paper
C-BD results and implications
1-BD results
Findings indicate that instruments are good
predictors of aid allocation aid is indeed
correlated to income per capita aid has no
effect on growth on average aid has no effect
on growth when the policy index is low, but has a
positive effect when the index is high enough (?
gt 0 and ß 0)
10
Source Burnside and Dollar (2000) AER
11
It is possible to compute the marginal impact of
aid on growth increase in growth for an
additional increase of GDP in aid
1-BD results
Source Overseas Development Institute ODI
12
I-Burnside and Dollar paper
C-BD results and implications
2-BD results
Findings indicate that -instruments are good
predictors of aid allocation -aid is indeed
correlated to income per capita -aid has no
effect on growth on average -aid has no effect
on growth when the policy index is low, but has a
positive effect when the index is high enough (?
gt 0 and ß 0)
They also find that bilateral aid (but not
multilateral aid) is fungible as it increases
government consumption (and not investment) thus
conditionality fails
13
I-Burnside and Dollar paper
C-BD results and implications
3-BD results implications
Three main implications Aid works if good
policies, if no good policies then ideas.This is
the concept of Selectivity
It is possible to derive an optimal allocation
formula (Collier Dollar) to maximize the
effectiveness of aid best results per dollar of
aid
Possibility to evaluate the good practices of
aid policies by individual multilateral
donors. Collier-Dollar provides a striking
measure of effectiveness the number of people
lifted out of poverty by an extra 1 million of
aid. By this measure, there are huge differences
between donors, e.g. about 350 or higher for
Denmark, Norway, the Netherlands and IDA, just
over 200 for the US and 150 for the EU
Commission.
The share of aid going to countries with good or
very good policies declined after 1998 for all
donors except for Norway, the DAC average falling
from 55.5 in 1993-8 to 50.1 in 1999-2002.
WHY???
14
II-Intense polemic since
DBs simple credo Aid works if good policies, if
no good policies then idea did not convince
Several questions emerged in the literature
Is DBs theoretical framework reliable?
Are DBs empirical results reliable? Good
methodology and robustness
Is good policy the only conditioning factor? Is
it the single right criterion to which aid should
be allocated?
15
II-Intense polemic since
A-Disagreement on BDs implicit assumption
The assumption (on which the World Bank policy
conclusion of targeting aid at countries with
good policies critically depends) that aid is
ineffective in promoting policy reform
-The view that conditionality is ineffective is
not held universally.
-Aid may improve the results policies and
projects may be better designed and implemented
as a result of donor engagement.
-Aid may help to compensate losers of reform
and thus increase the support for the reforms
Empirical evidence on relation from Aid to Policy
is mixed -no correlation between aid and policy
-some evidence of a dynamic mechanism aid
effectiveness is all the more important that
initial quality of institutions is low
16
II-Intense polemic since
B-Critics of BD results
1-Lack of robustness of the results to time
extension and more countries
Easterly, Levine and Roodman (2004) extended the
dataset to include 4 more years and 8 more
countries. They find that the link between
aid and growth disappears in the larger
sample Both unconditionally and
conditionally Casts doubts on the external
validity of BDs findings highly sensitive to
sample selection
17
II-Intense polemic since
B-Critics of BD results
2-Critics of the methodology
Critics of the indicator of economic policy
-why only 3 components -critics of the
weighting
Critics of the PPP measurement of GDP
Critics of the empirical estimations -sensitiv
ity of outliers -pooling of all aid flows
should consider only aid for which an impact is
expected within a 4-year window -why restrict
the potential conditional effectiveness
relationship only to policy?
18
II-Intense polemic since
B-Critics of BD results
2-Critics of the methodology
Non-linearity captured by the interactive term
AP in fact vanishes when other non-linearity is
allowed for
So aid effectiveness depends on the economic
vulnerability of a country (potential bad
luck) on the existence of a post-conflict
situation (Collier-Hoeffler, 2004) on the
capacity of aid to allow for an improvement of
policies when at first they are wrong (so aid
improves policies) on factors more or less
precisely grouped under the term of capacity
absorption and which condition the rhythm of the
reduction of the aids marginal effectiveness up
to the point where it becomes negative, etc.
19
I-Stylized facts about Aid
C-Back to where we were before BD or what else
should be considered?
1-Back to where we were State to the art
instrumentation method of aid by Rajan et
Subramanian 2005
20
C-Back to where we were before BD or what else
should be considered?
2-Additional areas of necessary
investigation Potential solution of the
conundrum will come from Decomposition of aid
into -Urgency aid -Short term projects -Long term
projects (training)
Look at long period / short period of time aid
may need less or more than 4 years to have pay
Potential detrimental effects Dutch disease,
weakening of institutions
Look at other left-hand side variables since aid
as other objectives than just growth -poverty
alleviation -reduction of migration -improving
health care -reduction of conflicts
21
Conclusion
Two different questions in the very important
policy issue of aid 1-Is the way we empirically
test the proper way? Much remains to be done to
assess its effectiveness and optimal design
Interesting results come from micro-level
studies but incapacity to generalize the results
It would seem logical that aid effectiveness
very much depends on the context and that
possibly aid would not be the most effectiveness
tool in all cases
-for small countries (like Uganda or Mozambique)
which have recently implemented reforms creating
favorable environments for the development of the
private sector. Priority is that they lack
access to capital markets because of a poor
dissemination of information regarding the
reforms under way.
-for countries in pre-conflict situations. While
aid has important effects, other actions may
better prevent conflicts regulating the diamond
trade, for example, to make financing rebellions
more difficult, sending United Nations
peace-keeping missions to prevent an explosion of
violence, etc Lecture 24
22
Conclusion
Two different questions in the very important
policy issue of aid 2-If no unconditional
effectiveness, how to improve?
As Easterly argues a major impediment to
effectiveness is the lack of feedback and
accountability no learning on past failures
The Planners approach of pouring massive
resources is not as effective (despite the
goodwill) as it does not take into account the
incentives
The Searchers approach of trial and
experimentation works better but requires more
time flexible approach and thus is incoherent
with the pre-determined setting schedule and
massive disbursement of international donors
Illustration of getting the bed nets to the
poor -hand out free nets often out of stock
and 70 are not used -sold by nurses (little
commission) to pregnant women always in stock
and 100 use.