Options for Reducing the Impact of MDRI Netting Out on New IDA Country Allocations

Transcription

1 IDA15 MID-TERM REVIEW Options for Reducing the Impact of MDRI Netting Out on New IDA Country Allocations International Development Association IDA Resource Mobilization Department (CFPIR) October 2009

2 Abbreviations and Acronyms AfDF CPIA HIPC IDB IDA IFC IMF MDG MDRI NPV ODA PBA SDR African Development Fund Country Policy and Institutional Assessment Heavily Indebted Poor Country Inter-American Development Bank International Development Association International Finance Corporation International Monetary Fund Millennium Development Goal Multilateral Debt Relief Initiative Net Present Value Official Development Assistance Performance Based Allocation Special Drawing Rights

3 Table of Content EXECUTIVE SUMMARY... I I. INTRODUCTION... 1 II. ADVERSE IMPACT OF MDRI ON NEW IDA COUNTRY ALLOCATIONS Impact of the MDRI Netting Out under a Baseline Scenario Impact under Alternative Graduation and IDA Resources Scenarios The Case for Reforming MDRI Netting Out...8 III. SCENARIOS FOR REDUCING THE ADVERSE IMPACT Four Scenarios Considered Impact of Eliminating MDRI Netting Out...11 IV. CONCLUSIONS AND RECOMMENDATION REFERENCES ANNEX 1: ADDITIONAL REFORM SCENARIOS CONSIDERED ANNEX 2: IMPACT OF MDRI NETTING OUT ON NEW IDA COUNTRY ALLOCATION UNDER THE BASELINE SCENARIO Tables Table 1. Impact of the MDRI Netting Out on New IDA Country Allocation under a Baseline Scenario 1/... 5 Table 2. Impact of MDRI Netting Out on New IDA Country Allocation, Baseline and Alternative Scenarios 1/... 7 Table 3. Comparison of Projected Average Annual IDA Country Allocation under Baseline and No MDRI Netting Out 1/ Figures Figure 1. Asymmetric Impact of MDRI Netting Out during IDA15-20: Percent Losses are Large while Percent Gains are Modest (Loss and Gain in Percent of Gross Allocation)... 9 Figure 2. Impact of Eliminating MDRI Netting Out: Fragile States Figure 3. Impact of Eliminating MDRI Netting Out: Non-Fragile States... 14

4 EXECUTIVE SUMMARY i. For those countries benefiting from debt relief under the Multilateral Debt Relief Initiative (MDRI), IDA s gross assistance flows are reduced by the amount of annual debt service forgiven. This is done in order to help reduce moral hazard and promote equity of treatment among low income countries. The compensatory donor contributions that IDA receives in lieu of the debt service forgiven are then reallocated to all IDA-only countries using the performance based allocation (PBA) system. The MDRI debt relief mechanism thus affects new IDA allocations through a two-step process. In the first step, an eligible country s forgone debt service in any given year will be deducted from its annual PBA allocation (also known as MDRI netting out ). In the second step, compensatory donor resources will be reallocated across all IDA-only countries according to the PBA system. ii. While the principles underlying the MDRI netting out may be sound, unfortunately it leads to significant declines in new IDA allocations for some MDRI recipients over time. Projections under a set of conservative baseline assumptions about resource availability and country graduations from IDA show that new IDA allocations after MDRI would become negligible, or zero for some post HIPC completion point countries, starting from IDA17. iii. Whereas the total volume of the MDRI netting out is small relative to the total IDA resources allocated through the PBA system, its impact on country allocations is severe for a set of IDA countries. IDA s total debt service forgone during IDA15 is estimated at around 4.3 percent of the total annual resources allocated through the PBA system. Although this amount is projected to rise over time it still remains relatively small as a share of the IDA allocations. Yet, the resource allocation impact on the affected countries is severe. Projections under the base case scenario show that the 15 most affected countries will, on average, lose about 46 percent of their gross PBA allocations between IDA15 and IDA20. Moreover, the distributional impact of the MDRI netting out is asymmetric: the losses of those adversely affected by the MDRI netting out, measured as a percent of their gross annual allocations, far outweighs the percentage gains of those benefiting from it. iv. Most of the countries adversely affected by the MDRI netting out are among the poorest countries in Africa and Latin America and the Caribbean. The baseline projections show that while twenty-nine IDA-eligible countries are, overall, adversely impacted by the MDRI netting out, 15 countries will be particularly hard-hit. For these countries, new IDA allocations after MDRI netting out are projected to become negligible starting from IDA17. Alternative assumptions regarding more country graduations from IDA in particular of India and Vietnam and the continued growth of IDA resources (through higher replenishments) will not alter this trend significantly. Unless addressed, the significant declines in new IDA allocations will in turn have adverse operational implications for IDA, including possible disengagement, reduced donor coordination, and even contribute to the reversals of the modest development gains achieved in these countries over the past several decades. v. Nine of the 15 countries most impacted by the MDRI netting out are fragile states, where not only policy and institutions are weak but also additional resources are urgently needed to stem the impact of the global crisis and maintain progress towards the

5 ii MDGs. Many fragile states are heavily dependent on commodity exports, remittances, ODA, and emergency and security assistance, all of which have come under intense pressure as a result of the current global economic crisis. At a time when the recent food and fuel price crises as well as the ongoing global financial crisis have further widened their financing gap, the MDRI netting out is set to take even more IDA resources away from these countries in the years to come. Furthermore, some of the same countries from which the MDRI netting out is projected to take away substantial resources do also receive special IDA allocations through the post-conflict and reengagement window in order to support their development needs better. Reforming the MDRI netting out system thus not only brings these opposing incentives in harmony (for these countries), but it would also be a concrete step that will bring about higher fresh IDA resources almost immediately, starting with the FY11 allocation. vi. This paper considers four scenarios for mitigating the adverse impact of the MDRI netting out: Scenario 1: Eliminating the MDRI netting out and the associated step of allocating compensatory donor resources; Scenario 2: Capping the MDRI netting out at 50 percent of gross PBA allocation; Scenario 3: Capping the MDRI netting out at 75 percent of gross PBA allocation, with delayed NPV-neutral netting out; and Scenario 4: Reallocating compensatory donors resources to MDRI-only countries. vii. These scenarios have the highest impact, in descending order, on new country allocations for the 15 most affected countries. Scenario 1 not only ensures the highest level of IDA resource flows but also helps simplify the PBA system. Scenario 2 continues to preserve the link between additional MDRI-related compensatory resources and performance, but it further complicates the management of the PBA system by adding an extra step of having to monitor whether a country s annual debt service due is larger than 50 percent of its gross PBA allocation. Scenarios 3 and 4 are administratively very complex to manage, and insufficient to avoid IDA disengagement in some countries when debt service foregone may be at their peaks. Moreover, these scenarios do not offer additional fresh resources that will allow these countries to deal with the manifold impact of the global financial crisis. viii. Recommendation. Based on the analysis, Management therefore seeks IDA Deputies endorsement to recommend eliminating the MDRI netting out and the associated step of allocating compensatory donor resources (Scenario 1) to the Executive Directors. This would be the most effective option to deal with the adverse impact of MDRI netting out and maintain a critical mass of new IDA allocation in the most affected, and mostly fragile and poorest, African and Latin American countries. Relative to the base case of continuing with the current practice, the elimination of the MDRI netting out is projected to increase new IDA allocations to the 15 most affected countries by an average of 25 percent in IDA15 and 62 percent in IDA16, slowly rising to 120 percent in IDA20. The additional IDA resources will allow these countries, in the short-term, to respond to the impact of the global crisis, and over the longterm, maintain stability and growth, and assist in the provision of key services that will allow them to make progress towards the MDGs. Eliminating the MDRI netting out will also have the added benefit of reducing administrative costs, simplifying the PBA system, and contributing to greater transparency in country allocations.

6 I. INTRODUCTION 1. In the spring of 2006, donors and shareholders approved IDA s participation in the Multilateral Debt Relief Initiative (MDRI), which provides 100 percent cancellation of eligible debt owed to IDA by countries reaching the Completion Point for the Heavily Indebted Poor Country (HIPC) Initiative. Following that, IDA began implementing MDRI in July 2006 as approved by IDA s Executive Directors and Board of Governors 1 and has, as of June 2009, provided SDR 28.6 billion (USD 43.5 billion) of debt relief to 26 HIPC countries. 2 In order to safeguard IDA s long-term capacity to finance international development, donors have agreed to provide dollar-for-dollar compensation for the debt relief provided by IDA. 2. A key element in the MDRI agreement was the principle of equity-of-treatment for all low-income countries and reduction of moral hazard associated with debt relief. 3 The equity-of-treatment principle emphasized the need to maintain equity between HIPC countries that are eligible for MDRI on the one hand and non-hipc low-income countries that do not receive debt relief because they did not accumulate unsustainable debt in the first place on the other. The principle emphasized that the latter group should also benefit from the compensatory resources donors provide to IDA for debt relief. The moral hazard concern was that debt relief should not reward, through high future aid flows, countries that had accumulated high and unsustainable debt in the past. In other words, new development assistance should be allocated based on current performance, not current debt payment obligations, in order to reinforce country incentives for pursuing responsible, pro-growth policies going forward. 3. The MDRI affects annual IDA allocations through a two-step process. 4 First, since IDA s debt cancellation is provided by relieving eligible countries of repayment obligations, it was agreed that IDA s gross assistance flows to the MDRI countries would be reduced by the amount of debt service forgiven, also referred to as MDRI netting out. Second, the compensatory donor contributions provided to IDA in lieu of debt relief would be reallocated to all IDA-only countries through IDA s existing performance-based allocation system. The rationale for netting out was that it would help allay moral hazard and ensure equity of treatment among all low income countries. 4. While MDRI netting out may be based on sound principles, unfortunately it is projected to lead to significant declines in new IDA allocations for some MDRI recipient countries over time. 5 This problem was anticipated in one of the early papers on MDRI 6, but 1 IDA (2006), Additions to Resources: Financing the Multilateral Debt Relief Initiative, Resolution No. 211 adopted on April 21, Data are as of June 30, 2009; see IDA (2009), Debt Relief Provided by IDA under the MDRI and HIPC Initiative: Update on Costs and Donor Financing as of June 30, 2009, October. 3 See: G8 Finance Ministers Conclusions on Development, London, June 2005; and Technical Note: G8 proposal for HIPC debt cancellation in The G8 Debt Relief Proposal: Assessment of Costs, Implementation Issues, and Financing Options by IDA (September, 2005). 4 See IDA (2006), IDA s Implementation of the Multilateral Debt Relief Initiative, March See IDA, 2006, IDA s Implementation of the Multilateral Debt Relief Initiative, March.

7 2 the paper may have, with hindsight, underestimated the adverse impact on new country allocations by assuming that the declines could be mitigated through higher IDA resources and faster country graduations. The long-term impact of the MDRI on new IDA allocations will depend on the relative magnitudes of gross PBA allocations and forgone debt service. Projections under a set of conservative baseline assumptions about resource availability and country graduations from IDA show that new IDA allocations after MDRI are set to become negligible, or zero for some post HIPC completion point countries, starting from IDA17. Most of these countries are among the poorest in Africa and in Latin America and the Caribbean. 5. Nine of the 15 countries most impacted by MDRI netting out are fragile states, where policy and institutions are weak and additional resources are critical to making progress towards the MDGs. 7 As a recent report notes, many fragile states are heavily dependent on commodity exports, remittances, ODA, and emergency and security assistance, all of which have come under intense pressure as a result of the current global economic crisis. 8 At a time when the food and fuel price crises of as well as the ongoing global financial crisis have further widened their financing gap, MDRI netting out is set to take even more IDA resources away from these countries in the years ahead. 9 Furthermore, some of the same countries from which the MDRI netting out is projected to take away substantial resources (e.g. the Republic of Congo, Togo and Haiti) also receive special IDA allocations through the post-conflict and reengagement window in order to support their development needs better. Given that the exceptional allocations are time-bound, and many countries like Haiti are already towards the end of phasing out from the exceptional window, the continued practice of MDRI netting out will mean even less resources for these countries, where IDA engagement will be even more difficult. Reforming the MDRI netting out system thus not only brings these opposing incentives in harmony, but it could also confer immediate real additional resources to these countries, which in turn are critical for supporting policy dialogue, responding to the global financial crisis, and continuing to deliver essential public services over the long-term. 6. This paper analyzes the impact of MDRI netting out on new country allocations and explores scenarios for dealing with it. The paper is organized as follows: Section 2 projects the potential medium to long-term impact of the current MDRI netting out mechanism on new IDA country allocations under a baseline and alternative scenarios. Section 3 considers four scenarios to mitigate the potential adverse impact of MDRI netting out (three of which are discussed in Annex 1), with emphasis on the impact of eliminating the MDRI netting out (Scenario 1), the option with the highest impact from the point of view of the country allocations of the most-affected countries. Section 4 offers conclusions and a recommendation Ibid. Fragile states are defined as countries with a Country Policy and Institutional Assessment (CPIA) score of less than 3.2. See Protecting Progress: The Challenge Facing Low Income Countries in the Global Recession, Background Paper Prepared for the G-20 Leaders Meeting, September Ibid. The paper estimates that low income countries will face a core financing shortfall of US$11.6 billion, of which about two-thirds are for countries in sub-saharan Africa, and about 58 percent is for fragile states.

8 3 II. ADVERSE IMPACT OF MDRI ON NEW IDA COUNTRY ALLOCATIONS 7. The long-term impact of the MDRI on a new IDA country allocation will depend on the country s relative magnitudes of annual gross PBA allocation and annual forgone debt service. For this reason, assessing the medium to long-term impact of the MDRI netting out requires projections of gross PBA allocations as well as estimates of MDRI debt relief. While the amount of debt service forgone annually depends on the debt service schedule of a country, projections of future gross PBA allocations, from which forgone debt service will be deducted, are subject to more uncertainty. 10 A country s PBA-based allocation in any given year will depend on a number of factors, including: (i) the size of IDA s overall available resource envelope; (ii) the number of IDA-eligible countries, which in turn depends on country graduations from IDA and reverse graduations back into IDA; (iii) a country s performance relative to other countries; and (iv) a country s risk of debt distress and the associated terms of eligibility for IDA resources. Under a reasonable set of assumptions, new IDA country allocations would fall below a critical minimum size, below which IDA may be unable to continue to engage effectively. Furthermore, contrary to the objectives agreed under IDA15 to simplify the PBA formula and increase transparency, the MDRI netting out will continue to complicate the management of the PBA system by adding two more steps to the allocation process, reducing transparency around country allocations, increasing staff administrative costs, and adversely impacting country dialogue. 8. This section presents the impact of the MDRI netting out on new country allocations under a baseline scenario and alternative scenarios on IDA graduation and resource envelope. The baseline scenario projects IDA country allocations under a conservative set of assumptions, while the alternative scenarios consider additional assumptions on the graduation of two large IDA-eligible countries Vietnam and India and a higher growth of the IDA resource envelope (commitment authority) through successive replenishments IMPACT OF THE MDRI NETTING OUT UNDER A BASELINE SCENARIO 9. Baseline Assumptions. The level of gross PBA allocations is projected over IDA15- IDA20 using the following core set of assumptions: The overall nominal IDA resource envelope grows, from IDA16 onwards, at 6 percent per replenishment (or about 2 percent per year), sufficient only to cover expected inflationary erosions and with no real growth; The set of IDA eligible countries remains largely unchanged, except for 4 countries which are assumed to graduate consistent with the assumptions presented in the IDA15 replenishment: Armenia and Azerbaijan in FY12 and Georgia and Bosnia-Herzegovina in FY16; See IDA, 2006, IDA s Implementation of the Multilateral Debt Relief Initiative, March. Although the impact of the global financial crisis may somewhat delay this graduation timeline, this assumption is consistent with the graduation assumptions underpinning the IDA15 replenishment, as laid out

9 4 A country's relative performance as compared to all other IDA eligible countries remains constant at the level of FY10; IDA country allocations are fully committed in a given year, with no front-loading or back-loading of allocations across years; Country classifications e.g., IDA-only, blend remain constant as in FY10 except for four countries: Angola, Georgia, Honduras, and Sri Lanka which are assumed to reach blend status in FY12; and post-conflict and re-engaging countries graduate from the special allocation window according to the current phasing down schedule; and A country s risk of debt distress classification (and therefore its traffic light ) is assumed to remain as in FY10 except for the countries that will reach HIPC completion point. 12 Upon reaching HIPC completion point, these countries are assumed to attain a moderate risk of debt distress ("yellow-light" under the traffic light system). 10. Impact on Country Allocations. Based on these assumptions, Annex 1 provides the impact of the MDRI netting out on new country allocations for the full list of IDA-eligible countries over IDA15-IDA20. Table 1 presents the impact of the MDRI netting out on the 15 most adversely affected countries, including those countries that have not yet reached the HIPC completion point (e.g. Chad, Cote d Ivoire Guinea, Guinea-Bissau, and Togo) but are projected to reach the HIPC completion point and thus benefit from the MDRI in the near future. 12 in The demand for IDA15 resources and the strategy for their effective use, IDA15 replenishment paper, June Revised assumptions will be formulated at the time of presenting a similar paper for the IDA16 replenishment process in Nine IDA-eligible countries (Afghanistan, Cote d Ivoire, Chad, Republic of Congo, Congo Democratic Republic, Guinea, Guinea-Bissau, Liberia and Togo) have not yet reached the HIPC completion, while five other countries (Comoros, Eritrea, Kyrgyz Republic, Somalia and Sudan) are in the pre-decision point stage. The assumptions on projected completion and decision points for these countries are consistent with Tables 2A and 2B of the 2009 report prepared by IDA and IMF entitled: Debt Relief Provided by IDA under the MDRI and HIPC Initiative: Update on Costs and Donor Financing as of June 30, 2009.

10 5 Table 1. Impact of the MDRI Netting Out on New IDA Country Allocations under a Baseline Scenario 1/ (Average Annual Allocation, SDR Million unless otherwise indicated) Region Country Item IDA15 IDA16 IDA17 IDA18 IDA19 IDA20 Average change in % of gross allocation AFR Chad Gross allocation before MDRI netting-out Projected allocation after MDRI netting-out (baseline) % LAC Guyana Gross allocation before MDRI netting-out Projected allocation after MDRI netting-out % AFR Guinea Gross allocation before MDRI netting-out Projected allocation after MDRI netting-out % AFR Mauritania Gross allocation before MDRI netting-out Projected allocation after MDRI netting-out % LAC Haiti Gross allocation before MDRI netting-out Projected allocation after MDRI netting-out % AFR Gambia, The Gross allocation before MDRI netting-out Projected allocation after MDRI netting-out % LAC Bolivia Gross allocation before MDRI netting-out Projected allocation after MDRI netting-out % LAC Honduras Gross allocation before MDRI netting-out Projected allocation after MDRI netting-out % AFR Congo, Republic of Gross allocation before MDRI netting-out Projected allocation after MDRI netting-out % AFR Central African Republic Gross allocation before MDRI netting-out Projected allocation after MDRI netting-out % AFR Togo Gross allocation before MDRI netting-out Projected allocation after MDRI netting-out % AFR Senegal Gross allocation before MDRI netting-out Projected allocation after MDRI netting-out % AFR Cote d'ivoire Gross allocation before MDRI netting-out Projected allocation after MDRI netting-out % AFR Guinea-Bissau Gross allocation before MDRI netting-out Projected allocation after MDRI netting-out % AFR Zambia Gross allocation before MDRI netting-out Projected allocation after MDRI netting-out % Total 15 Most Affected Countries Total allocation before MDRI netting-out Total allocation after MDRI netting-out Difference due to netting out Difference in % of gross allocation -20% -38% -48% -54% -59% -54% -46% All 29 Affected Countries Gross allocation before MDRI netting-out Total allocation after MDRI netting-out Difference due to netting-out Difference in % of gross allocation -8% -12% -14% -16% -20% -20% -15% 1/ Data are presented for 15 HIPC countries that are projected to be most severely affected by MDRI netting out. 2/ Refers to the average percent increase in annual allocations over IDA15 to IDA Two key results are especially worth noting. First, while the total volume of MDRI netting out is small relative to the total IDA resources allocated through the PBA system, its impact on country allocation is severe for a set of IDA countries. IDA s total debt service forgone during IDA15 is, for example, estimated to be around SDR 360 million per year, or about 4.3 percent of the total annual resources allocated through the PBA system during the same period. Although this amount is projected to rise to SDR 1.1 billion per year in IDA20, it

11 6 still remains relatively small (about 10 percent) as a proportion of resources allocated to IDA countries. Yet, the impact on the most affected countries is severe: the 15 most affected countries will, on average, lose about 46 percent of their gross PBA allocations over IDA Chad, Guyana, Guinea, and Mauritania are all set to lose 60 percent or more of their gross PBA allocations during the same period. On the other hand three countries Bangladesh, Nigeria, and Kenya together account for about 60 percent of the gains from MDRI netting out in each replenishment during IDA15-IDA20 though for each country the amount is a small proportion of their gross PBA allocations. 12. Second, most of the countries adversely impacted by the MDRI netting out are among the poorest countries in Africa and Latin America and the Caribbean (Table 1). The projections show that while twenty-nine IDA-eligible countries are, overall, adversely impacted by the MDRI netting out, new IDA allocations after MDRI netting out are projected to become negligible for some African and Latin American countries starting from IDA17. In Africa, new allocation to Chad and Guinea will fall to zero, 13 while new allocations to Mauritania, the Gambia, and the Republic of Congo will become negligible. In Latin America, new allocation to Guyana will fall to zero in IDA19; in addition, country allocations will be more than halved for Haiti, Bolivia, and Honduras. Finally, allocations for African countries such as the Central African Republic, Togo, Zambia, Cote D Ivoire, Guinea-Bissau, and Senegal will be reduced by between a third and a half of their gross PBA allocations. The significant declines in new IDA allocations will in turn have significant adverse operational implications, including possible IDA disengagement, reduced donor coordination, and even reversals of the modest development gains achieved in these countries over the past several decades IMPACT UNDER ALTERNATIVE GRADUATION AND IDA RESOURCES SCENARIOS 13. This section relaxes the assumptions underpinning the baseline scenario to examine how the impact of the MDRI netting out may be different. Under the baseline case, it is assumed that Armenia, Azerbaijan, Georgia, and Bosnia-Herzegovina will graduate from IDA in IDA16, consistent with the graduation scenario discussed during the IDA15 replenishment. However, given the longer time frame of this analysis, one alternative scenario is that two additional large IDA countries India and Vietnam may graduate in the course of IDA Given the large amount of IDA allocation that these two countries receive, their graduation would in turn have significant impact on the allocation to other IDA countries by increasing the resources available for distribution. 14. Similarly, the baseline scenario assumes no real growth in IDA resources from IDA15 onwards. Under the baseline scenario, the overall IDA envelope is assumed to grow at a nominal rate of 6 percent per replenishment after IDA15 to cover only expected inflationary erosion. However, IDA s own replenishment history indicates that donors have increased IDA resources in real terms through successive replenishments, by about 12 percent per replenishment (or 4 percent per year) during IDA1-IDA15. Another alternative scenario is to 13 Per the calculations, amounts would go below the country allocations implying that countries would have to return funds to IDA, which is not seen to be feasible.

12 7 use this historical average growth with resulting higher country allocations for IDA eligible countries. Table 2. Impact of MDRI Netting Out on New IDA Country Allocation, Baseline and Alternative Scenarios 1/ (Average Annual Allocation, SDR Million) Countries IDA15 IDA16 IDA17 IDA18 IDA19 IDA20 Chad Baseline scenario Alternative graduation Alternative IDA growth Guinea Baseline scenario Alternative graduation Alternative IDA growth Mauritania Baseline scenario Alternative graduation Alternative IDA growth Gambia, The Baseline scenario Alternative graduation Alternative IDA growth Congo, Rep. Baseline scenario Alternative graduation Alternative IDA growth Togo Baseline scenario Alternative graduation Alternative IDA growth Central African Rep. Baseline scenario Alternative graduation Alternative IDA growth Senegal Baseline scenario Alternative graduation Alternative IDA growth Cote d'ivoire Baseline scenario Alternative graduation Alternative IDA growth Guinea-Bissau Baseline scenario Alternative graduation Alternative IDA growth Zambia Baseline scenario Alternative graduation Alternative IDA growth Guyana Baseline scenario Alternative graduation Alternative IDA growth Haiti Baseline scenario Alternative graduation Alternative IDA growth Bolivia Baseline scenario Alternative graduation Alternative IDA growth Honduras Baseline scenario Alternative graduation Alternative IDA growth Total 15 Most Affected Countries Baseline Change due to alternative graduation Change due to alternative IDA growth % increase alternative graduation 0% 0% 23% 27% 30% 27% % increase alternative IDA growth 0% 8% 23% 41% 64% 75% All 29 Affected Countries Baseline Change due to alternative graduation Change due to alternative IDA growth % increase alternative graduation 0% 0% 15% 15% 16% 15% % increase alternative IDA growth 0% 7% 15% 23% 33% 41% 1/ Data are presented for 15 HIPC countries that are projected to be most severely affected by MDRI netting out. Alternative graduation scenario: Vietnam and India will graduate from IDA during IDA16 and IDA17 respectively. Alternative IDA growth scenario: IDA resources will grow by 12 percent per replenishment (or 4 percent per year), consistent with the historical trend during IDA1-IDA15.

13 8 15. Table 2 shows the impact of MDRI netting out on country allocations under these alternative scenarios. The results show that, even under these scenarios, the impact of MDRI netting out on new country allocations will continue to be severe for the most affected countries. The graduation of India and Vietnam from IDA does moderate the impact of MDRI netting out on new country allocations as more IDA resources will become available to the remaining IDA eligible countries, particularly in Africa and Latin America. Relative to the baseline, total annual net allocation to the 15 most severely affected countries will increase by SDR 62 million in IDA17, slowly rising to about SDR 71 million by IDA20. Similarly, under the historical IDA resource growth scenario (of 12 percent), total annual net allocation to the 15 most severely affected countries will increase by SDR 26 million in IDA16, slowly increasing to SDR 199 million in IDA20. However, these increases are not sufficient to mitigate the sharp reductions in IDA allocations due to the MDRI netting out. For example, the final allocations to Chad, Guinea, and Guyana remain at zero for at least one replenishment period even under the alternative graduation scenario; and zero or negligible under the real growth scenario. These results therefore confirm that further scenarios would need to be considered to deal with the adverse impact of MDRI on new IDA country allocations THE CASE FOR REFORMING MDRI NETTING OUT 16. The case for reforming MDRI netting out rests on four principal arguments. First, the distributional impact of MDRI netting out is asymmetric: the losses of those adversely affected, measured as a percent of their gross annual PBA allocation, is far larger than the gains of those benefiting from MDRI netting out, also measured as a proportion of their gross annual allocation (Figure 1). 14 While the 15 most affected countries will, on average, lose about 46 percent of their gross PBA allocations over IDA15-20, and Chad, Guyana, Guinea, and Mauritania are all set to lose between percent of their gross PBA allocations, the top beneficiaries will gain only about percent their gross annual allocation. Furthermore, the gains are concentrated in a few countries. Three countries Bangladesh, Nigeria, and Kenya are projected to capture about 60 percent of the gains from MDRI netting out (distributed through the compensatory allocation) in each replenishment over IDA15-IDA The top ten beneficiaries, as proportion of their gross PBA allocations, are the Republic of Yemen, Lao PDR, Tajikistan, Nepal, Cambodia, Kyrgyz Republic, Lesotho, Bangladesh, Nigeria and Kenya.

14 9 Figure 1. Asymmetric Impact of MDRI Netting Out during IDA15-20: Percent Losses are Large while Percent Gains are Modest (Loss and Gain in Percent of Gross Allocation) Chad Guyana Guinea Mauritania Haiti Yemen Lao PDR (In percent) Tajikistan Nepal Cambodia 17. Second, most of the countries affected by the MDRI netting out are fragile states that need additional development aid. Indeed, nine of the 15 most affected countries are fragile states, with significant resource needs to make progress towards MDGs. 15 The significant declines in new IDA allocations could in turn have significant adverse operational implications, including possible IDA disengagement, reduced donor coordination, and even reversals of the modest development gains achieved in these countries over the past several decades. Conversely, reforming the MDRI netting out will ensure that fresh additional IDA resources will flow to these countries starting from the FY11 allocation cycle, which will allow them to respond to the impact of the global crisis, maintain stability and growth, and in some cases, support countries as they attempt to prevent a slide back into conflict and instability. 18. Third, alternative assumptions around the possibility of more country graduations from IDA in particular of India and Vietnam and a higher growth of IDA resources (consistent with its historical growth 12 percent per replenishment) will not alter the projected impact significantly. 19. Finally, the MDRI netting out has added considerable complexity and administrative costs to the PBA system, and reforming it will contribute to the simplification and transparency of the PBA system. The MDRI netting out has added two additional steps (one for deducting debt service forgone and another for distributing compensatory resources) to the PBA system, requiring additional staff time and resources both in terms of administering the PBA model, but more importantly, in terms of explaining to the Bank country teams and government authorities why annual country allocations have changed. Indeed, it has since its implementation in 2007 compounded the problem of transparency around annual country allocations and generated requests from various stakeholders Executive Directors, government authorities, and Bank regional management and resource management officers seeking explanations on movements in annual country and regional allocations. For Bank staff managing the PBA system, this has meant a disproportionate amount of time being devoted to MDRI-related questions and issues. Reforming the system 15 Exceptions are Mauritania, Senegal, and Zambia in Africa; and Guyana, Bolivia, and Honduras in Latin America.

15 10 will help reduce the administrative costs, simplify the PBA system, and promote greater transparency on country allocations. 20. These benefits should be weighed against the original rationale of the MDRI netting out, namely the equity-of-treatment and moral hazard principles that were built into the MDRI agreements. It could be argued that the elimination of netting out may weaken the equity-of-treatment principle as non-mdri low-income countries currently benefiting from compensatory resources will cease to benefit, and there would be no redistribution of compensatory resources according to country performance. However, it must also be recognized that, so long as the netting out is maintained, the asymmetric adverse impact will continue. III. SCENARIOS FOR REDUCING THE ADVERSE IMPACT 21. In exploring available scenarios, it is important to note that net annual IDA allocation consists of three principal components: the gross PBA allocation; the annual debt service forgone that is deducted from the gross PBA allocation (i.e. the MDRI netting out); and, receipts from reallocation of donor compensatory resources. The gross PBA allocation is determined by a country s performance (which in turn is measured by its CPIA and IDA portfolio rating). In general, the gross PBA allocations of the most affected countries will only increase in line with improvements in policy and institutional performance, which generally improves only gradually. The scenarios would therefore have to invariably seek to affect either the treatment of debt service forgone in IDA allocation (i.e. the MDRI netting out itself) or the reallocation of compensatory resources FOUR SCENARIOS CONSIDERED 22. Accordingly, this paper considers four reform scenarios: Scenario 1: Eliminating the MDRI netting out; Scenario 2: Capping the MDRI netting out at 50 percent of gross PBA allocation; Scenario 3: Capping the MDRI netting out at 75 percent of gross PBA allocation, with delayed NPV-neutral netting out; and, Scenario 4: Reallocating compensatory donors resources to MDRI-only countries. 23. This paper has explored the four scenarios in detail, all of which have some positive effects; however, given the need for an allocation of meaningful size for the countries most affected, and the administrative complexity and costs associated with each scenario, Scenario 1 is the only feasible option that effectively meets the twin objectives of maximum allocation and simplicity and transparency of the PBA system. Eliminating MDRI netting out (Scenario 1) has the most desirable impact on the most-affected countries. Overall, these scenarios have a mitigating impact, in descending order, in terms of their effects on country allocations of the 15 most affected countries. Scenario 1 not only ensures the highest level of IDA resource flows to the most affected countries but also helps simplify the PBA system. Scenario 2 (capping MDRI netting out at 50 percent of gross PBA allocation)

16 11 would enhance IDA resource allocation towards those countries most affected by MDRInetting out, while continuing to preserve the link between additional MDRI-related compensatory resources and performance, but it complicates the management of the PBA system even further by adding one step. Scenarios 3 and 4 are not only administratively complex to manage, but also insufficient to avoid IDA disengagement in some countries when debt service foregone may be at their peaks. Annex 1 explores in detail the impact on country allocations of Scenarios 2, 3 and 4, while the remainder of this section focuses on the impact of Scenario IMPACT OF ELIMINATING MDRI NETTING OUT 24. Table 3 compares, for the most affected countries, new IDA country allocations under the baseline (where netting out continues) to the case where the MDRI netting out is eliminated (Scenario 1). Figures 2 and 3 provide a visual illustration of the same impact for the 15 most affected countries, grouped into fragile and non-fragile states. 25. Impact on country allocations. Relative to the base case of continuing with the current practice of netting out, the elimination of MDRI netting out is projected to increase new IDA allocations to the 15 most affected countries by 25 percent in IDA15 and 62 percent in IDA16, slowly rising to 120 percent in IDA20. New IDA resource flows to Chad, Guyana, Guinea and Haiti will more than double during IDA16, with similarly significant projected increases for the other countries as well. New IDA allocation would at least more than double starting from IDA17 for most of the affected countries, and no country s allocation would become close to zero or negative (Table 3). At the regional level, this scenario will increase, compared to the baseline, IDA s resource transfers to Africa and Latin America. The elimination of the netting out will come at the cost of equity-of-treatment since non-mdri low-income countries currently benefiting from compensatory resources will cease to benefit, and there will no longer be a redistribution of compensatory resources according to country performance.

17 12 Table 3. Comparison of Projected Average Annual IDA Country Allocation under Baseline and No MDRI Netting Out 1/ (Average Annual Allocation, SDR Million) Region Country Item IDA15 IDA16 IDA17 IDA18 IDA19 IDA20 AFR Chad Baseline Scenario 1: No MDRI netting out % increase 38% 418% 2143% 385% 377% 434% LAC Guyana Baseline Scenario 1: No MDRI netting out % increase 108% 158% 209% 222% 1401% 798% AFR Guinea Baseline Scenario 1: No MDRI netting out % increase 23% 163% 388% 2199% 19687% 1130% AFR Mauritania Baseline Scenario 1: No MDRI netting out % increase 80% 94% 93% 146% 1554% 588% LAC Haiti Baseline Scenario 1: No MDRI netting out % increase 40% 186% 221% 202% 137% 102% AFR Gambia, The Baseline Scenario 1: No MDRI netting out % increase 79% 74% 114% 244% 194% 152% LAC Bolivia Baseline Scenario 1: No MDRI netting out % increase 49% 64% 191% 210% 201% 121% LAC Honduras Baseline Scenario 1: No MDRI netting out % increase 34% 98% 123% 144% 157% 128% AFR Congo, Republic of Baseline Scenario 1: No MDRI netting out % increase 8% 52% 80% 74% 294% 477% AFR Central African Republic Baseline Scenario 1: No MDRI netting out % increase 33% 120% 97% 83% 62% 52% AFR Togo Baseline Scenario 1: No MDRI netting out % increase -2% 120% 113% 109% 80% 65% AFR Senegal Baseline Scenario 1: No MDRI netting out % increase 42% 73% 68% 69% 58% 48% AFR Cote d'ivoire Baseline Scenario 1: No MDRI netting out % increase -2% 18% 73% 120% 96% 80% AFR Guinea-Bissau Baseline Scenario 1: No MDRI netting out % increase 17% 50% 46% 51% 48% 114% AFR Zambia Baseline Scenario 1: No MDRI netting out % increase 21% 23% 23% 25% 135% 144% Total 15 Most Affected Countries Baseline Scenario 1: No MDRI netting out Change from the baseline % increase 25% 62% 93% 117% 144% 120% All 29 Affected Countries Baseline Scenario 1: No MDRI netting out Change from the baseline % increase 9% 14% 16% 18% 25% 25% 1/ Data are presented for 15 HIPC countries that are projected to be most severely affected by MDRI netting out.

18 13 Figure 2. Impact of Eliminating MDRI Netting Out: Fragile States Chad Guinea Republic of Congo O1 B O1 B O1 B 10.0 Central African Republic Togo Cote d Ivoire O O B 20.0 B O1 B Guinea Bissau Guyana Haiti O1 B O1 B O1 B Note: B= Base case scenario (with netting out); O1 = Scenario 1 (with no MDRI netting out)

19 14 Figure 3. Impact of Eliminating MDRI Netting Out: Non-Fragile States The Gambia Mauritania Zambia O1 B O1 B O1 B Senegal Bolivia Honduras O1 B O O1 Note: B= Base case scenario (with netting out); O1 = Scenario 1 (with no MDRI netting out) 26. Impact on countries currently benefiting from the MDRI netting out. Several countries benefit from the current practice of MDRI netting out. Among the biggest beneficiaries, in absolute levels of allocation, are Bangladesh, Nigeria, Ethiopia and Kenya, which together account for nearly 70 percent of the total MDRI netting out during IDA15. Other main beneficiaries, as percent of their gross PBA allocation, include the Republic of Yemen, Lao PDR, Tajikistan, Nepal, Cambodia, Kyrgyz Republic, Rwanda, Burkina Faso, Mongolia and Lesotho. These countries will no longer receive these allocations if the netting out were to be eliminated. However, if the Management s proposal to eliminate the MDRI netting out is endorsed by IDA Deputies and approved by the Executive Directors, Management will, for the countries adversely affected by this change, take into account the impact of eliminating MDRI netting out in determining the amount of allocations they would receive through the Crisis Response Window as well as in determining regional reallocations in order to ensure that they are not worse off during the IDA15 period. IV. CONCLUSIONS AND RECOMMENDATION 27. If the current practice of MDRI netting out continues, a number of countries including Chad, Guinea, Guyana, Haiti, Mauritania, the Republic of Congo, and the Gambia will receive, starting from as early as IDA17, little or no new IDA allocations. Alternative assumptions around country graduations from IDA in particular of India and Vietnam and continued growth of IDA resources (through replenishment) will not alter this trend significantly.

20 The decline in new IDA allocations will in turn have significant consequences for the countries affected. To the extent that IDA s level of funding in these countries, especially in Sub-Saharan Africa, falls below a critical mass, IDA will find it increasingly difficult to play its platform role, 16 as a result of which donor coordination and aid effectiveness in these countries may be significantly compromised. In cases where new IDA allocations fall to zero or negative, IDA will be unable to support these countries with fresh resources, thereby risking reversals of the modest development and poverty reduction gains that have been achieved in the past. 29. This paper has explored four scenarios. Scenario 1 (eliminating MDRI netting out) not only ensures the highest level of IDA resource flows but also helps simplify the PBA system. Scenario 2 (capping MDRI netting out at 50 percent of gross PBA allocation) would have the second largest impact in terms of IDA resource allocation towards those countries most affected by MDRI-netting out, while continuing to preserve the link between additional MDRI-related compensatory resources and performance, but it complicates the management of the PBA system even further by adding an extra step of having to monitor whether a country s annual debt service due is larger than 50 percent of its gross PBA allocation or not. Scenarios 3 (capping the MDRI netting out at 75 percent of gross PBA allocation, with delayed NPVneutral netting out) and Scenario 4 (reallocating compensatory donors resources to MDRI-only countries) are, respectively, administratively too complex to manage, and insufficient to avoid IDA disengagement in some countries when debt service foregone may be at their peaks. More importantly, these scenarios do not offer any additional fresh resources that will allow these countries to deal with the manifold impact of the global financial crisis. 30. Recommendations. Based on the analysis, Management therefore seeks IDA Deputies endorsement to recommend eliminating the MDRI netting out to the Executive Directors as the most effective option to deal with the adverse impact of MDRI netting out and maintain a critical mass of new IDA allocation in the most affected, and mostly fragile and poorest, African and Latin American countries. Relative to the base case of continuing with the current practice of netting out, the elimination of MDRI netting out is projected to increase new IDA allocations to the 15 most affected countries by an average of 25 percent in IDA15 and 62 percent in IDA16, and a further 120 percent by IDA20. New IDA resource flows to Chad, Guyana, Guinea and Haiti will increase in the remaining period of IDA15, and more than double during IDA16, with similarly significant projected increases for the other countries. By ensuring the highest level of fresh IDA resource flows to these countries almost immediately, this option will help mitigate the risks of IDA disengagement, reduced donor coordination, and even reversals of hard-won development gains achieved over many years in these countries. The additional IDA resources will in turn allow these countries to respond to the impact of the global crisis in the short term, and maintain stability and growth, and assist in the provision of key services that will allow them to make progress towards the MDGs over the long term. Eliminating the MDRI netting out will also have the added benefit of reducing staff administrative and transaction costs, simplifying the PBA system, and enhancing transparency on country allocations. 16 See IDA The Role of IDA in the Global Aid Architecture: Supporting the Country-Based Development Model, IDA Resource Mobilization Department (CFPIR), June.