Transcript Evaluation of the Aid

Evaluation of the Aid-Growth
Relationship
Presented by Ghassan Baliki and Emiko Nishii
Development Workshop
04.11.2010
Outline
• Empirical Framework of Rajan and Subramanian (2005)
• Potential drawbacks
• How can we understand the Aid-Growth relationship better?
• Aid Effectiveness Literature (AEL): A Meta-Study
• Main Findings and Concluding Remarks
Rajan and Subramanian (2005): What Does the
Cross-Country Evidence Really Show?
• Endogeneity issues – Aid may depend on level of income
(i.e. donors increase aid inflows based on recipients’ needs)
>> Aid can’t be exogenous with respect to growth
• Constructing instruments for Aid is necessary
• Alesina & Dollar (1998)
>> Aid is often allocated based on historical & diplomatic
reasons
Rajan and Subramanian (2005): Cont’d
Constructing IV for Aid:
1)
Find the share of donor d’s aid allocated to recipient r in year t.
2)
Use the predicted share to compute aid to GDP ratio received by
country r in year t.
Rajan and Subramanian (2005): Cont’d
•
Dependent variable: average annual growth rate of per
capita GDP ( 1960-2000, 1970-2000, 1980-2000, 19902000)
• The results suggest that with exclusion of outliers, 3 out
of 5 cases, the coefficient of Aid is negative, and
significant in none.
>> decomposition of Aid is necessary to understand the
Aid-Growth relationship better
Rajan and Subramanian (2005): Cont’d
Disaggregate Aid by:
1) Sectors (social, economic and food)
2) Timing of impact (short, and long impact)
e.g. whereas food aid should not be expected to affect long-run growth,
social & economic aid should
3) Type of donor (multilateral vs. bilateral)
i.e. multilateral aid is less ‘political’ than bilateral aid
>> the results show that no sub-categories have any
significant impact
Rajan and Subramanian (2005): Cont’d
• Non-linear & conditional effects of Aid on growth.
- Aid effectiveness depends on policy environments?
“aid effectiveness depends on the institutions that restrict
appropriation of public funds by rent seeking agents” Hodler (2007)
>> inclusion of policy measures (e.g. CPIA by the World Bank)
- Diminishing return of aid?
>> inclusion of aid squared
• Results suggest that in no case, the coefficients are significant.
>> potentially driven by endogeneity and country-specific
characteristics
Rajan and Subramanian (2005):Cont’d
- the first-difference GMM
- the system GMM
First-difference
GMM
system GMM
Total aid
- and significant
- and insignificant
Short-impact aid
- and significant
- and insignificant
Economic aid
+ and significant
+ and significant
>> the results are fragile (e.g. depends on the # of lags or
independent variables included, the results change)
Rajan and Subramanian (2005): Cont’d
• Quantitative Impact of Aid
Assumption: Mainly, Aid influences growth through increasing public
investment.
α=0.35, Y/K=0.45, and β=1 give a suggested coefficient of 0.16.
>> the coefficients for many existing literature are overestimated.
Rajan and Subramanian (2005): Cont’d
• We must pay attention to the potential importance of a previously
neglected factors. The importance of understanding ‘Aid influences
growth through which channels exactly?’: >> In this context,
investigating ‘What’s preventing aid from having a positive impact
on growth?’ may be helpful.
Related Literature:
- Lensink and Morrissey (1999) “Uncertainty of Aid Inflows and
the Aid-Growth Relationship”
- Rajan and Subramanian (2005) “What Undermines Aid’s Impact
on Growth?”
Lensink and Morrissey (1999): “Uncertainty of Aid
Inflows and the Aid-Growth Relationship”
• Aim: The paper seeks to find whether uncertainty associated with
(volatility of) the level of aid inflows affects the impact of aid on
growth.
• Potential impact of Aid on growth with the presence of
Uncertainty:
- investors may postpone/cancel investment decisions
- Aid is an important component of government revenues
>>volatility of receipts may impact on fiscal behavior, thus growth
Policies/Institutions may be conditional on aid inflows.
Financial Resources Inflows from DAC to Developing Countries
Lensink and Morrissey (1999) Cont’d
•
Dependent variable: avg. growth rate of GDP per
capita
•
Aid=level of Aid
•
Construct proxy for uncertainty
2)
calculate the standard deviation of the residuals from the
forecasting equation
1)
a forecasting equation is estimated (as a first or second-order
autoregressive process, extended with a time trend)
>> The coefficients on uncertainty are negative and significant.
>> When the uncertainty measure is included Aid becomes significant
and positive
Lensink and Morrissey (1999): Cont’d
Still some drawbacks………
• By using the cross-country approach, there are possibilities that
exogenous factors leading to a bias estimator.
• Almost any explanatory variable could be found to have a significant
effect whereas the ‘truth’ is that apparent significance is due to
common causalities or spurious regressions >> omitted variable
bias still remains.
• Is “growth” a good variable to capture the effectiveness of aid?
Rajan and Subramanian (2005): “What Undermines Aid’s
Impact on Growth?”
What’s preventing Aid from having a positive impact on growth?
- The Aid-Competitiveness Approach
• Best way to check aid-effectiveness is to compare ‘fact’ and ‘counterfact’ >> not possible.
• Instead, check whether labor-intensive industries grow relatively
slower in countries with high aid-inflow compare to non-laborintensive industries.
• This approach allows us to capture 1) within-country differential
effects, and 2) country treatment effect to understand the effect of
aid.
Rajan and Subramanian (2005): “What Undermines Aid’s
Impact on Growth?”
How Aid can influence growth through ‘competitiveness’ channel?
Under the fixed exchange rate:
• Aid spent on domestic goods pushes up the price of recourses that
are in limited supply domestically (e.g. skilled worker).
Under the flexible exchange rate:
• Aid inflows increase nominal exchange rate, thus reducing
competitiveness.
Rajan and Subramanian (2005): “What Undermines Aid’s
Impact on Growth?”
• Strong evidence consistent with aid undermining the competitiveness
of the labor-intensive or exporting sectors.
• In countries that receive more aid, labor-intensive and exportable
sectors grow slower relative to capital-intensive and non-exportable
sectors.
• Aid inflows do cause overvaluation
Are the results compelling?
• Major exports sector for all recipient counties is labor-intensive?
• on balance, whether these adverse competitiveness effects offset any
beneficial effects of aid is unclear.
AEL – Doucouliagos and Paldam
(2006, 2007a, 2008)
▫ Do the estimates of the AEL converge to something we
might term 'truth'?
▫ Can we identify the main innovations which cause
(prevent) convergence?
▫ Do biases exist while uncovering the 'truth' about aid
effectiveness ?
Three Perennial Problems
▫ 1) Priors
▫ 2) Data Mining
▫ 3)Incentives
- Innovation with skepticism
- Reliance on independent
replication
- and the Reluctance Hypothesis
Absolute Aid Ineffectiveness
Why is it Puzzling?
▫ “Why would they” vs. “If it is, it must be rational”
→ Aid fatigue
▫ Marginal Project vs. Financed Project
▫ Always via accumulation?
▫ No repay means no crowding out
Meta-Analysis
• Priors and Biases:
▫ Polishing
▫ Ideology
▫ Goodness
• Meta Analysis Methodologies:
▫ Meta-Significance Test (MST)
▫ Precision-Effect Testing (PET)
▫ Funnel Asymmetry Test (FAT)
The Three Family Models of AEL
Does Aid Cause Increasing Accumulation?
Large but probably not full crowding effect
A Large Crowding Out
Does Aid Cause Increasing Growth?
• Neoclassical Model:
• Results:
▫ Decline in variation over time and with sample
size
▫ More Extreme points
▫ Average decreasing
▫ Non-symmetrical funnel around horizontal axis
Funnel Plot, No Economic
Significance!
Aid Growth Effects: Reluctance Trends
Is the Effect of Aid on Growth Conditional?
• Good Policy Model (Burnside and Dollar, 2000):
• The Medicine Model (Lensink and White, 2001):
Conclusion of the Three Meta-Studies