IDA13 IDA's Performance-Based Allocation System Current and Emerging Issues International Development Association October 2003 - i - Acronyms and Abbreviations ARPP Annual Review of Project Performance CAS Country Assistance Strategy CPIA Country Policy and Institutional Assessment GNI Gross National Income IBRD International Bank for Reconstruction and Development IDA International Development Association JSA Joint Staff Assessment LSR Lending Strategy Review OED Operations Evaluation Department PBA Performance-Based Allocation system PCPA Per Capita Per Annum PR IDA Country Performance Rating SDR Special Drawing Right - ii - Table of Contents I. Introduction..........................................................................................................................1 II. IDA Ratings Disclosure .......................................................................................................1 III. Governance in the PBA System...........................................................................................2 IV. Poverty Weighting in Allocations........................................................................................3 V. Blend Countries: Allocation Limits and Hardened Terms .................................................4 VI. Assessment and Financing for Low Income Countries Under Stress..................................5 VII. Pilot Program for Regional Projects ....................................................................................7 VIII. Summary..............................................................................................................................9 Figure 1: Alternative Post-Conflict Disbursement Profiles............................................................7 Annex I: Summary of the CPIA Criteria .....................................................................................10 Annex II: Poverty Efficient AID Allocation.................................................................................11 I. Introduction 1. Effective allocation of IDA resources is of critical importance to IDA's overall mission to help reduce poverty and stimulate economic growth. The allocation tool that IDA has in place ­ the Performance-Based Allocation (PBA) system ­ is a cornerstone of IDA's Thirteenth Replenishment arrangements. While there is broad agreement on its key features, the PBA system has from time to time been modified in light of experience, and to make sure that its results are congruent with research findings, especially with respect to aid effectiveness.1 2. The PBA system was extensively reviewed during the IDA13 replenishment negotiations.2 Several PBA-related issues were identified as topics for further discussion at the time of the IDA13 Mid-Term Review. The issues include: (i) ) increased disclosure of the IDA country performance ratings; (ii) the experience with the governance factor, introduced in 2001; (iii) the weighting of poverty in the PBA system; and (iv) allocation limits on some large blend countries. Since IDA13 was finalized, two additional issues that have a bearing on the PBA system have surfaced: how IDA can support Low-Income Countries under Stress (LICUS) and the handling of regional projects. The paper describes ongoing work on these issues, and seeks Deputies' feedback to guide consideration of possible enhancements to the IDA13 PBA framework for IDA14. II. IDA Ratings Disclosure 3. The CPIA-derived IDA Country Performance (ICP) ratings have evolved over the past decades as the key element underpinning IDA's PBA system. Because IDA's assessment system is so important in the allocation of resources, there is considerable interest in it, and growing sentiment in favor of its greater transparency and disclosure. At the start of IDA12 in FY00, IDA initiated the disclosure, in a quintile format, of the CPIA and ICP relative ratings for IDA eligible countries. Management instructed country teams of IDA eligible countries to discuss with each country's authorities their country's CPIA and ICP ratings and the resulting country IDA allocation, and posted on the external website the quintile-based rating results for the CPIA, its four clusters, the country portfolio, and the quintile-based ICP rating.3 Deputies asked Management to report at the IDA13 Mid-Term Review on readiness of the system for further public disclosure and on the timing of public disclosure of individual country ratings.4 4. Over the past six months Management has deliberated, internally and informally with the Board, on how best to move towards public disclosure of the IDA ratings. A proposal has been made to provide at this time, in addition to the quintile-based rankings, range-based information for the same seven ratings, plus the average governance score. It is foreseen that this will be followed in 2004 by an extensive consultative process with respect to full disclosure. The 1 For a review of the PBA system's recent evolution and results, see Allocating IDA Funds based on Performance - Fourth Annual Report on IDA's Country Assessment and Allocation Process, IDA/SecM2003-0130, March 26, 2003. 2 Additions to IDA Resources: Thirteenth Replenishment, September 17, 2002 IDA/SecM2002-0488. 3 See http://siteresources.worldbank.org/IDA/Resources/CPIA2002.pdf. 4 Additions to IDA Resources: Thirteenth Replenishment, para 18, op.cit. - 2 - proposal is described in more detail in a paper5 which will be discussed by the Bank Executive Board on October 28, 2003. III. Governance in the PBA System 5. To help steer scarce IDA funds away from less effective use, and to raise the profile of governance as a key determinant of the capacity to use development finance effectively, the already substantial focus on governance in the PBA system6 was further strengthened in IDA13 with the introduction of the governance factor. The latter was designed to address a concern about its predecessor, the governance discount, to allow a more continuous stress on governance across the full performance range. 6. The governance factor is based on the country's average rating for seven governance criteria that are part of the PBA system.7 The weighted average of the country's CPIA and portfolio management performance ratings is multiplied by this factor, resulting in an increase or decrease of the rating, depending on the degree to which the country's governance rating is satisfactory or unsatisfactory. As a result, the governance criteria are effectively double-counted in the final ratings. 7. The governance factor has proven to be an effective means to highlight the importance of governance across the rating spectrum. One important index of this is the increasing frequency with which significant governance triggers are included in CAS programs. At the same time, it is important to look at the cumulative effect on countries with relatively poorer governance. It is found that a one-point drop in just one of the CPIA's seven governance criteria results in a 7.5% drop in the overall IDA rating, and thus a 15% drop in the country's allocation. 8. Management is exploring an adjustment that would simplify the governance factor and would smooth out the discontinuity described above. One option would be to eliminate the exponent currently applied to the factor. In substance, governance would remain dominant, but the impact of marginal changes in countries' governance ratings would be slightly reduced. Management will consider this possible modification further in the context of the PBA system's overall treatment of poorly performing countries (some thoughts on this are presented below in Section VI) and may submit a proposal to the IDA Deputies in the context of the IDA14 replenishment negotiations. 5 Disclosing IDA Country Performance Ratings IDA/R2003-0187, October 9, 2003. 6 In IDA12 a governance discount was applied to the weighted average of the CPIA and portfolio performance ratings in cases where governance was particularly weak, reducing the allocation to the dozen or so countries in this situation by as much as 50%. 7 The seven criteria include six governance-related CPIA criteria (4 and 16-20, see Annex 1), plus the procurement practices criterion included in the ARPP rating (the governance discount was based on the same seven criteria). The factor is calculated by dividing the average rating on these seven criteria by 3.5 (the mid point of 1-6) and applying an exponent of 1.5 to this ratio: governance factor = (average governance rating / 3.5)1. . 5 - 3 - IV. Poverty Weighting in Allocations 9. In line with IDA's principal mission to reduce poverty in the world's poorest countries, a country's per capita income must be below $865 in order to be considered eligible for IDA assistance.8 The formula that is used to allocate IDA resources within this group of countries gives the highest weight to the IDA Country Performance Rating.9 Half as much weight is given to population, and one sixteenth as much to the degree of poverty, using per capita income as a proxy. Given IDA's overarching objective of poverty alleviation, Deputies have asked whether the PBA system's modest bias in favor of the poorest among these poor countries is adequate. Accordingly, Management has reviewed the impact of the current poverty weighting, actual borrowing patterns in these countries, as well as the most recent findings of the Bank's aid effectiveness research in order to determine whether additional weight should be given to poverty in the PBA formula. 10. IDA's current PBA system has a significant pro-poor bias, as it ensures that countries with low per capita incomes receive significantly higher per capita allocations than those that are relatively better off. At the same level of performance, a country with a per capita income of $100 receives a "poverty bonus" of about 30% more on a per capita basis than a country with a per capita income of $800. In addition, it is worth noting that as a share of the economy IDA's aid is typically much higher in the poorest countries, even without the poverty bonus. In a country with a per capita income of $100, the average annual per capita allocation of around $7.5 would constitute 7.5% of gross national income, while in a country with per capita income of $800 it would be less than 1%. 11. An analysis of actual usage of IDA allocations shows that the take up has been lower at the lower end of the income scale. While this outcome can be attributed to a variety of factors, many of which lie beyond the country's control, it does suggest that, until these impediments are identified and addressed, a greater pro-poor bias may not result in a more effective allocation of IDA funds. 12. Another factor to consider is that IDA countries are assessed in a manner which is development-neutral, with no advantage for the better off countries. The system is designed so that countries at different stages of development have an equal possibility to score well, and hence an equal chance at obtaining a given level of IDA support. 13. IDA's current allocation formula is based on the research finding that, for two countries with the same population and CPIA score, aid relative to income should be higher for the poorer country. More recent findings raise some questions on this analytical base by suggesting that "poverty efficient" aid allocation would require aid per GNI actually to increase slowly with per capita income, starting from the level of the very poorest country, and only beginning to decline 8 Some exceptions are made to the income criterion, in particular for small island states. 9 Country Allocation = f (Performance2.0, Population1.0, GNI/capita­0.125 ). - 4 - after a per capita income of about $800 (at PPP).10 This would suggest an intuitively inequitable formula that would involve decreasing the weight given to income per capita as countries move up the income spectrum. Modifying the PBA formula to match such a nuanced relationship between income and aid effectiveness would introduce additional complexity into the formula at a time when transparency of the system is at a premium. Considering the points discussed in this Section, Management sees on balance no clear argument for proposing a change of the poverty weight in the PBA system at this time, but intends to keep new findings under review. 14. Lastly, Management has considered the possibility of improving the targeting on poverty by substituting, for the aggregate population figure now used in the allocation formula, the number of people living below the poverty line. This could help better direct IDA resources to those countries where the true extent of poverty is masked by uneven income distribution. To do this with the accuracy and cross-country comparability provided by population and gross national income figures would, however, require far better poverty data than are presently available in most IDA countries. Until reliable data on the extent of poverty in each IDA country are available, IDA will need to rely on total population and income figures as the best available proxy for the scale of poverty. V. Blend Countries: Allocation Limits and Hardened Terms 15. The limiting ("capping") of allocations to some blend countries constitutes an important exception to the application of the PBA system.11 Deputies have asked IDA Management to review current policy vis-ą-vis the blends, including with respect to their allocations, and with respect to hardened terms for these countries. 16. Historically IDA has set limits on its allocations to blend countries, in light of their access to IBRD funding. Normally IDA reduces the blend country's performance-based allocation by 25% of its projected borrowing from IBRD. In addition, some further limits were put in place to avoid an excessive concentration of IDA funding in a few very populous recipient countries. Accordingly, IDA currently caps the performance-based three-year allocations of: India (SDR 2 billion), Pakistan (SDR 1.3 billion), and Indonesia (SDR 600 million). Without a cap on the allocations of these three countries, they would have received 71% of available IDA resources in the FY04-06 IDA allocation exercise, reducing by nearly two thirds the allocations to all other IDA countries. 17. The capped blend allocations constitute an exception to the PBA system, and it should be noted that these countries are under-funded vis-ą-vis the PBA allocations, leaving important opportunities for alleviating poverty unfunded.12 Thus this exception is of a different nature than 10 Collier and Dollar have calculated an updated "poverty efficient" allocation of aid based on updated data for poverty and for policy performance, as measured by the CPIA, which shows some increase at the lowest income levels and then a tapering off (see chart in Annex II). 11 The other exception are the exceptional allocations to eligible post-conflict countries. 12 For a further discussion of the potential for scaled up lending in the large Asian IDA eligible countries, see Supporting Sound Policies with Adequate and Appropriate Financing: Implementing the Monterrey Consensus at the Country Level, DC2003-0016 Development Committee Meeting September 22, 2003. - 5 - the extra post-conflict allocations which temporarily provide funding above PBA norms to help consolidate a promising opportunity for peace. 18. The IDA13 arrangements provided that countries with per capita income levels above the IDA operational cut off (currently $865) that continue to be IDA eligible, should access IDA resources on hardened terms.13 During IDA13, hardened terms have so far been applied to Albania, Bolivia, Bosnia & Herzegovina, and Serbia & Montenegro (Albania is still an IDA-only country, the others are blend countries). For these countries the per capita income had exceeded the operational cut off for more than two consecutive years. Hardened terms have not yet been applied to Djibouti and Honduras, where the per capita income exceeded the operational cut off for the first time in FY03.14 The hardened terms are a step on the path towards blend status and ultimately graduation from IDA to IBRD. For this reason, IDA programs in countries now receiving hardened terms are focused on strengthening creditworthiness ­ first as a step towards blend status, and subsequently to prepare for graduation from IDA. 19. Deputies requested that, at the time of the Mid-Term Review, the possibility be revisited of hardening the terms for "notional" blends15. At this time only three countries fall in this category: Nigeria, Papua New Guinea, and Zimbabwe, with per capita incomes ranging from $290 - $580. They would not appear to be appropriate cases for applying hardened terms which were intended to introduce somewhat higher pricing of IDA as countries' incomes rise and as their capacity to repay IDA increases. For this reason, Management recommends continuing the current practice of limiting the application of hardened terms to countries that are above the IDA operational cut off of $865. VI. Assessment and Financing for Low Income Countries Under Stress 20. The needs of Low Income Countries Under Stress (LICUS) have been looked at systematically by a Bank-wide LICUS Task Force. The task force's principal focus has been on improving analytical work, enhancing the selectivity and realism of reform programs, increasing capacity-building efforts and improving internal support through human resources and procedural reforms. The task force has recommended that IDA take a more finely tuned approach to these countries, though without compromising its overall performance-based allocation process. 21. LICUS countries vary across a spectrum, from more stable countries with modest IDA allocations to post-conflict countries and those in non-accrual status with the Bank. As part of 13 Regular IDA terms are 40 years (35 years for blend countries) with 10 years grace period, an administrative charge of ¾% per annum and a commitment charge of 0 to 0.50%; the hardened terms involve a maturity of 20 years. The hardened terms provision does not apply to small islands that have exceptional access to IDA. 14 Djibouti and Honduras also exceeded the cut off in FY04; if this continues into FY05 they will be subject to hardened terms at that time. 15 "Notional" blends are borrowers that have a capacity or history of market-based borrowing and a per capita income below the IDA eligibility threshold, and which are currently unable to borrow from IBRD due to marginal or deteriorated creditworthiness. - 6 - the implementation of the LICUS initiative, the Bank has studied the application of the PBA system in these countries. In this context a number of ideas are under discussion: · Improving Data Availability. Few sources of reliable internal or external data are available in these countries. The Bank will continue to accelerate analytical work in LICUS countries, and will provide additional guidelines for country teams on more diverse external sources, including on the political and security environment. · Amending the Assessment Scale. The current CPIA rating system allows for little differentiation at the bottom of the performance spectrum, where the lowest rating is "unsatisfactory".16 In the context of being able to differentiate LICUS countries that are attempting to take basic reform steps ("zero generation reforms") one possibility might be to define explicit ratings that depict different levels of low performance. This change would also involve explicitly defining outstanding performance at the top of the scale and is under consideration for the CPIA exercise in 2005. · Post-Conflict Country Assessments. Countries that are eligible for IDA post-conflict allocations17 are annually rated on a special set of Post-Conflict Progress Indicators (PCPI).18 IDA Management is planning to improve the accuracy of these ratings by providing specific characterizations for all the rating levels, which could reduce the subjective element in the rating process. 22. Post-Conflict Country Allocations. Eligible post-conflict countries receive special allocations for up to three years, after which they phase down during years four and five to their regular performance-based norm. This approach, formalized during IDA13, needs of course to be continually evaluated in the light of knowledge and research. Collier has looked at the time profile of post-conflict absorptive capacity19, and argues from this research that it may be desirable to have somewhat lower early allocations, but phase special post-conflict assistance over a longer period of time. As an illustration, stretching out the post-conflict allocation over seven years (e.g. four years of post conflict allocations plus three years of phase down to the performance-based norm) would leave the overall amount of extra post-conflict allocations20 unchanged. Figure 1 below shows how such stretching out would affect the cumulative disbursement profile for a typical post-conflict country.21 16 If such rating persists for three consecutive years, the unsatisfactory ("2") score becomes a "1". 17 Currently nine countries are eligible for special post-conflict allocations: Afghanistan, Angola, Burundi, Congo Republic, Cote D'Ivoire, Democratic Republic of the Congo, Eritrea, Sierra Leone, Timor Leste (Sri Lanka does not receive special allocations, in view of its substantial regular performance-based allocation). 18 See Additions to IDA Resources: Thirteenth Replenishment, Annex 2: IDA's Approach for Allocating Resources to Post-Conflict Countries. 19 Breaking the Conflict Trap: Civil War and Development Policy, World Bank 2003 20 The nine countries that are currently eligible for post-conflict allocations are projected to receive in FY04 close to SDR 0.5 billion in allocations above their regular CPIA and ARPP-based norms. 21 The graph assumes a norm of SDR 50 million and a typical 4 year disbursement profile. In the current scenario it assumes allocations of SDR 100 million in years 1-3; and in the alternative scenario allocations of SDR 85 million in years 1-4. - 7 - FIGURE 1 - Alternative Post-Conflict Disbursement Profiles 100 90 80 70 60 amount d 50 se 40 Disbur 30 3 years P-C Allocations with 2 years phase down 20 4 years P-C Allocations with 3 years phase down 10 0 1 2 3 4 5 6 7 8 9 10 11 12 Year of Disbursements 23. Work on these LICUS issues is ongoing. Management hopes to bring specific recommendations for supporting LICUS countries to the IDA14 negotiations. VII. Pilot Program for Regional Projects 24. The increasing importance of regional integration for growth in IDA countries has been emphasized by borrowers and donors alike. The IDA13 replenishment agreement calls for steps to spur trade and regional integration and to help address regional challenges of the greatest relevance to IDA countries22. In response, staff has been exploring mechanisms for increasing IDA support for regional integration, especially in regions such as Africa where history, geography, shared borders, large numbers of landlocked countries, and/or a predominance of small economies makes it productive for donors and governments to think and act regionally. 25. Regional integration is generally recognized as a strong potential engine of growth and poverty reduction in poor countries. Integration also offers the potential to create more effective and integrated markets ­ both regionally and internationally. These factors, plus many other cross-border developmental issues, provide strong arguments for broad and proactive "deep integration" to increase the ability of IDA's clients to compete internationally and to make the most of regional market. Regional integration can include convergence of macroeconomic policies including fiscal, monetary, and exchange rate policies; harmonizing legal and regulatory frameworks; and improving scale and competition through the integration of infrastructure and markets for goods, finance, labor, and energy. 22 See Additions to IDA Resources: Thirteenth Replenishment, Op Cit. para.54. - 8 - 26. IDA, as a major source of financing across the regions is well placed to play a catalytic role in helping countries capture the benefits of regional integration and cooperation. In order to provide more reliable and effective support for the regional integration agenda during the IDA13 period, a pilot envelope has been set up for regional projects of up to $450 million per annum during FY04-0523. To ground this pilot in IDA's PBA system, individual country allocations will cover one third of the project cost attributable to an individual country wherever feasible. In recognition of the significant positive externalities of regional projects, the remaining two thirds of the credit amount will be from general IDA resources. This approach acknowledges the need for additional funding for regional projects - which has also been the rationale for a set aside in the African24 and Asian Development Funds - while maintaining an important link with IDA's PBA system. The size of the additional funding represents less than 5% of available IDA commitments for FY04-FY05. Management will report on the experience with this pilot to inform upcoming IDA14 deliberations once there is a sufficient number of projects on which to assess progress and performance. 27. The following criteria must be met for projects to qualify as "regional projects" under the pilot program. Regional projects are operations (i) that involve three or more countries, all of which need to participate for the project's objectives to be achievable (i.e. the project would not make sense without the participation of all countries); (ii) whose benefits (either economic or social) spill over country boundaries (i.e. that generate significant positive externalities or mitigate negative ones); (iii) where there is clear evidence of country or regional (e.g. ECOWAS, SADC) ownership which demonstrates commitment of the majority of participating countries; (iv) that provide a platform for a high level of policy harmonization between countries (this is key to the success of a regional initiative); and, importantly (v) are part of a well developed and broadly supported regional strategy. The Southern Africa Power Market is an example of such a project, and will be discussed by the Executive Directors in November 2003. 23 See Pilot Program for Regional Projects, September 30, 2003,IDA/SecM2003-0532, Information Note provided to the World Bank Executive Board. 24 In the African Development Bank, there was a set aside of up to 10 percent for regional economic integration and co-operation interventions. - 9 - VIII. Summary 28. IDA's PBA system will continue to be refined as new insights emerge and technical solutions become available. A key issue is how the performance-based system can be responsive to the particular problems of countries at the bottom end of the performance scale, including a re- examination of the governance factor, and some assessment and allocation amendments that could be helpful to LICUS and post-conflict countries. 29. Currently IDA makes three tightly defined exceptions to the performance-based allocations: (i) the temporary extra allocations to eligible post-conflict countries; (ii) the basic allocation of SDR1 million per annum that all countries receive independent of their performance; and (iii) the $300 million per annum that IDA has set aside to catalyze valuable regional initiatives. Taking into account these special arrangements, over 85% of IDA's allocations continue to be based on performance.25 30. Since IDA12 key elements of the PBA system have been described on IDA's external website. While it moves towards full disclosure of its assessments, IDA will be proactive in enhancing the transparency of how the PBA system works by holding regional workshops for interested recipient countries.26 In addition, an assessment and allocation workshop for bilateral and multilateral development partners is planned for the second quarter of 2004. The enhanced transparency of IDA's assessment and allocation system should permit a more in-depth collaboration of all development partners over the years to come. 25 As pointed out in Para 14, the capped allocations to the three large blend countries do not strictly follow the performance-based allocation formula, but fall within the performance-based allocations. 26 Workshops for African IDA borrowing countries were held on October 9-10 in Addis Ababa, and on October 15-16 in Dakar. - 10 - ANNEX I - SUMMARY OF THE CPIA CRITERIA Box 1 - CPIA Criteria A. Economic Management 1. Management of Inflation and Macroeconomic Imbalances 2. Fiscal Policy 3. Management of External Debt 4. Management and Sustainability of the Development Program B. Structural Policies 5. Trade Policy and Foreign Exchange Regime 6 Financial Stability and Depth 7 Banking Sector Efficiency and Resource Mobilization 8. Competitive Environment for the Private Sector 9. Factor and Product Markets 10. Policies and Institutions for Environmental Sustainability C. Policies for Social Inclusion/Equity 11. Gender 12. Equity of Public Resource Use 13. Building Human Resources 14. Social Protection and Labor 15. Monitoring and Analysis of Poverty Outcomes and Impacts D. Public Sector Management and Institutions 16. Property Rights and Rule-based Governance 17. Quality of Budgetary and Financial Management 18. Efficiency of Revenue Mobilization 19. Quality of Public Administration 20. Transparency, Accountability and Corruption in the Public Sector - 11 - ANNEX II ­ POVERTY EFFICIENT AID ALLOCATION 1. Holding policy and population constant, the relationship of the "poverty efficient" allocation of aid with log per capita GDP is as follows: 20.4 20.2 20 Aid 19.8 Optimal of 19.6 Log 19.4 19.2 19 6 6.5 7 7.5 Log of GDP per Capita, PPP, 2000 The poverty- efficient allocation of aid assigns more aid relative to GDP to poorer countries, other things equal. However there are limits to how much financing countries with a small GDP can absorb. Based on our estimates of the diminishing returns to Aid/GDP, the optimal relationship is traced out in this graph.