Burnside - Dollar (2000)

Question 1

Why?

Answer 1

Investigate whether aid is affected by policy conditions within a country

Question 2

What is the empirical issue? How do we solve for these?

Answer 2

• We may have reverse causality and Omitted Variable Bias issues with aid and growth.
• Use 2SLS with and without interactions of a policy index

Question 3

What is the policy index? How is it constructed?

Answer 3

• weighted average of the degree of trade openness, budget surplus and inflation
• weightings chosen by running regressions on GDP per capita and using coefficients

Question 4

What was the methodology?

Answer 4

• run regression without any interaction of policy index and use excluded variables such as population growth, the Egypt dummy (commonly used in aid regressions) as instruments for changes in aid receipts
  (since aid suffers from reverse causality etc). These are exogenous to the equation for growth. Found to have a significant impact
• Run the regression with a policy index interaction. This will show how policy impacts the instrumented changes in aid receipts on growth.

Question 5

What were the main results?

Answer 5

• Good policy interacted with aid is positively correlated with growth, even when dropping middle income countries from the sample
• Diminishing returns to the level of aid (AID/GDP)^2*Policy term has a negative coefficient and is significant
• Extending the sample to 1993 shows the growth spurt we witnessed was the effect of aid after the ‘lost decade’ (decade of low growth in Africa)
• Bilateral aid (makes up 2/3 of foreign aid) features no allocation in favor of countries with good policy which leads to higher government consumption and no real lasting impacts on growth

Question 6

Conclusion

Answer 6

Aid is effective in an appropriate policy environment