2013年-世界发展银行全球_Is_Foreign_Aid_Fungible__Evidence_from_the_Education_and_Health_Sectors_52页_960kb

Summary of "Is Foreign Aid Fungible? Evidence from the Education and Health Sectors"

Core Content

This paper investigates the fungibility of foreign aid in the education and health sectors, using a new dataset that includes disaggregated aid disbursements. Fungibility refers to the extent to which aid is used for its intended purpose or whether it is reallocated by recipient governments to other sectors, leading to a reduction in public spending in the targeted area.

The study addresses a major limitation in previous research: the failure to distinguish between on-budget and off-budget aid. Off-budget aid, which is not recorded in a recipient government’s budget, can lead to biased estimates of fungibility because it is not captured in the data used for analysis.

Main Views

• Fungibility is a critical issue in assessing the effectiveness of foreign aid. It affects how aid influences public spending and development outcomes.
• On-budget aid is defined as sector program (SP) aid, which is directly channeled through the recipient's budget and is more likely to be used for its intended purpose.
• Off-budget aid is represented by technical cooperation (TC), which is used as a proxy for aid that is not recorded in the recipient’s budget.
• Previous studies often overestimated the fungibility of aid because they did not account for off-budget flows, leading to inaccurate conclusions about the fiscal response of recipient governments.
• Fungibility is limited for both on- and off-budget aid. SP aid has a one-to-one correlation with public sectoral expenditures, indicating low fungibility. TC aid, which is the largest component of total aid, does not significantly displace public spending, suggesting even lower fungibility.

Key Findings

• On-budget aid (SP): The coefficient of SP aid in the regression is approximately 1, indicating that each dollar of SP aid increases public sectoral spending by one dollar. This suggests low fungibility.
• Off-budget aid (TC): The coefficient of TC is close to zero and not significantly smaller than zero, indicating that TC does not displace public spending. This implies very limited fungibility.
• Total fungibility estimates are biased when off-budget aid is not properly accounted for, as the omission of TC leads to an overestimation of the fungibility of total aid.
• Dynamic models (e.g., system GMM) show greater volatility in the coefficient of SP aid compared to static models, due to lack of variation in SP aid over time.
• Off-budget aid is significant in many countries and varies over time, making its inappropriate treatment a major concern in assessing the true impact of aid.

Data and Methodology

• The OECD Creditor Reporting System (CRS) is used to disaggregate aid by sector, but it is incomplete.
• A novel data construction method is proposed to improve the accuracy of sectoral aid disbursements by combining CRS data with DAC Table 2a and DAC Table 5.
• The methodology involves:
  • Calculating the residual between DAC2a and CRS disbursements.
  • Allocating the residual across sectors using DAC5 data.
  • Scaling the resulting sectoral disbursements to match aggregate disbursements.

Implications

• Previous studies that used donor-based aid measures or recipient-reported data may have overestimated the fungibility of aid.
• Off-budget aid is not captured in recipient reports, leading to omitted variable bias in estimating the impact of on-budget aid.
• The correct approach requires distinguishing between on- and off-budget aid and evaluating their separate marginal effects on public spending.
• The results are robust across various specifications, highlighting the importance of proper data treatment in assessing aid effectiveness.

Conclusion

The paper concludes that foreign aid is not as fungible as previously thought, especially in the education and health sectors. By introducing a more accurate dataset and methodology, it provides evidence that off-budget aid has limited fungibility, and that previous estimates were biased due to the omission of off-budget flows. The findings underscore the need for better data collection and analysis in the study of aid effectiveness.