The Role of Donors in Creating Aid Volatility and How to Reduce It Final Report with Additional Programme Data Independent Research Undertaken for Save the Children, UK August 2006 Susy Ndaruhutse and Laura Brannelly CfBT Table of Contents TABLE OF CONTENTS..............................................................................................................................................................2 LIST OF ACRONYMS................................................................................................................................................................3 EXECUTIVE SUMMARY .....................................................................................................................................................4 INTRODUCTION.......................................................................................................................................................................8 OBJECTIVES OF RESEARCH .....................................................................................................................................................8 1. AID MODALITIES AND RECENT TRENDS .................................................................................................................9 2. LITERATURE REVIEW .................................................................................................................................................11 2.1 VOLATILITY IN COUNTRIES WITH IMF PROGRAMMES ............................................................................................12 2.2 TRENDS AMONGST DAC DONORS IN AID DISBURSEMENTS....................................................................................13 2.3 TRENDS AMONGST MULTILATERAL DONORS IN AID DISBURSEMENTS....................................................................14 2.4 HAS AID PREDICTABILITY BEEN INCREASING IN RECENT YEARS? ..........................................................................15 3. CONSEQUENCES OF UNPREDICTABLE AID FLOWS ...........................................................................................15 4. EMPIRICAL EVIDENCE FROM DAC DATABASE ON AID PREDICTABILITY IN DEVELOPING COUNTRIES .........................................................................................................................................................................17 5. PREDICTABILITY OF AID IN FRAGILE STATES ...................................................................................................21 6. REVIEW OF DONOR PRACTICES...............................................................................................................................22 6.1 DONOR TRENDS AND POLICIES...............................................................................................................................22 1. The Netherlands .........................................................................................................................................................23 2. Norway.......................................................................................................................................................................23 3. Sweden .......................................................................................................................................................................23 4. UK..............................................................................................................................................................................24 5. European Commission ...............................................................................................................................................24 6. World Bank ................................................................................................................................................................24 6.2 BILATERAL DONOR DATA FROM OECD DAC CREDITOR REPORTING SYSTEM DATABASE 5 ..................................25 6.4 COMMENTS ON WORLD BANK DISBURSEMENT DATA AVAILABILITY .....................................................................29 7. REVIEW OF GLOBAL FUNDS......................................................................................................................................30 7.1 HEALTH.................................................................................................................................................................30 1. GFATM ......................................................................................................................................................................30 2. PEPFAR.....................................................................................................................................................................30 3. The Bill and Melinda Gates Foundation....................................................................................................................31 7.2 EDUCATION ...........................................................................................................................................................32 Catalytic Fund Progress on Commitment and Disbursements.......................................................................................33 7.3 THE MILLENNIUM CHALLENGE ACCOUNT..............................................................................................................34 8. REASONS FOR AID VOLATILITY...............................................................................................................................35 8.1 TECHNICAL AND ADMINISTRATIVE DELAYS ...........................................................................................................36 Donor Side .....................................................................................................................................................................36 Recipient Country Side...................................................................................................................................................36 8.2 CONDITIONALITIES ................................................................................................................................................37 Donor Side .....................................................................................................................................................................37 Recipient Country Side...................................................................................................................................................37 9. RECOMMENDATIONS AND CONCLUSION .............................................................................................................39 9.1 PRIMARY RECOMMENDATIONS...............................................................................................................................40 1. Long-Term Aid Commitments and Graduated Responses..........................................................................................40 2. Public Reporting of Donor Performance and an International Aid Agreement.........................................................42 3. Applying Conditionality to Future Aid Commitments ................................................................................................43 4. Strengthening Absorption Capacity and Using Alternative Funding Channels Especially in Fragile States.............43 5. More Transparent Reporting on Aid Disbursements by All Donors ..........................................................................44 9.2 SECONDARY RECOMMENDATIONS..........................................................................................................................45 1. Discounting of Aid Disbursements.............................................................................................................................45 2. Using Macroeconomic Policy as a Buffer Stock Tool ................................................................................................45 9.3 AID EFFECTIVENESS...............................................................................................................................................46 9.4 CONCLUSION .........................................................................................................................................................47 REFERENCES ........................................................................................................................................................................49 APPENDIX: DATA ISSUES AND SOURCES ...............................................................................................................................53 2 List of Acronyms AIDS Acquired Immune Deficiency Syndrome CF Catalytic Fund CSO Civil Society Organisation DAC Development Assistance Committee DAH Development Assistance to Health DFID Department for International Development (UK) DPL Development Policy Loan EC European Commission EFA Education for All EFA-FTI Education for All Fast Track Initiative EPDF Education Program Development Fund FDA Federal Drugs Agency GBS General Budget Support GCE Global Campaign for Education GFATM Global Fund to Fight Against AIDS, Tuberculosis and Malaria GNI Gross National Income HIPC Heavily Indebted Poor Country HIV Human Immunodeficiency Virus HLFH High Level Forum for Health IAA International Aid Agreement IDA International Development Association IFF International Finance Facility IMF International Monetary Fund LDC Less Developed Country LIC Low Income Country LMIC Lower Middle Income Country OLIC Other Low Income Country MCA Millennium Challenge Account MCC Millennium Challenge Corporation MDG Millennium Development Goal MIC Middle Income Country MTEF Medium Term Expenditure Framework NGO Non-Government Organisation OECD Organisation for Economic Co-operation and Development ODA Overseas Development Assistance PEM Public Expenditure Management PEPFAR President's Emergency Plan for AIDS Relief PRGF Poverty Reduction and Growth Facility PRS Poverty Reduction Strategy Sida Swedish International Development Agency SPA Strategic Partnership for Africa SWAp Sector Wide Approach TB Tuberculosis UK United Kingdom UMIC Upper Middle Income Country UN United Nations UPE Universal Primary Education US United States USAID United States Agency for International Development 3 Executive Summary The need for increased aid has been discussed extensively in recent years from grassroots level to high-level fora of international donors and politicians. Whilst it was a known fact that aid levels were decreasing in the 1990s, the new millennium and the setting of the eight Millennium Development Goals has seen increased pressure on donors to deliver larger volumes of aid in a sustainable way over at least the next ten years. Whilst many donors have subsequently made a commitment to do this, there is a concern that these commitments do not necessarily lead to increased aid being disbursed to developing countries, with aid remaining volatile. Added to this, the move by several donors towards providing more aid via budget support to assist in the sustainable financing of recurrent costs in developing countries such as salaries, makes the issue of aid predictability even more critical, as any delays in this aid arriving in country will have detrimental effects on the provision of social sector services like health and education in poorer countries. Thus the aims of this piece of research are to examine how predictable aid is, to ascertain what the potential causes and impacts are of unpredictable aid, and to provide recommendations for improving the current situation. The research starts with a literature review that shows that aid, both in project form and in programme form (budget support) remains unpredictable and volatile both within year and between years. However, programme aid is significantly more volatile than project aid, and this occurs whether or not a country's IMF programme is on-track. Bulir and Lane (2002) find that in the most heavily aid-dependent countries, aid is up to seven times more volatile than domestic fiscal revenue (p.19). The impact of uncertain aid flows undermines the public expenditure management process in developing countries and has a negative impact on both spending in the social sectors, and on economic growth. Bulir and Hamann (2003) found that one percentage point of prediction error in project aid amounts to around 0.1% of GDP, and the same prediction error in programme aid amounts to around 0.05% of GDP (p.80-1) whilst Foster and Keith (2004) noted that average shortfalls in overall aid receipts (project and programme) compared to what was committed totalled nearly 2% of GDP in a sample of 28 countries (p.38). Where significant aid is given as budget support, and there are delays in disbursing this aid, this can mean that teachers and health workers do not get paid, or that important medical supplies and school textbooks do not reach health clinics and schools on time, having adverse effects particularly on the poor and on children. Looking at individual donors using the data available in the DAC database, the Scandinavians are the most generous donors (in terms of volumes of aid as a % of GNI) but have significant room for improvement in the timely disbursement of aid commitments. For the period 2002- 2004, when data analysis was conducted, Norway was the only bilateral donor showing improvements in both the percentage of programmes that are disbursing funds in full each year, and in the total percentage of volume of committed aid actually being disbursed on time. By contrast, the Netherlands has seen improvements in the percentage of programmes that are disbursing funds in full each year, but has seen some variation in the total percentage of volume of committed aid actually being disbursed on time, though this has consistently 4 been in excess of 85%. Sweden experiences around half of the total number of programmes disbursing all their funds on time corresponding to an average of 75% of the volume of its programme commitments being disbursed in full each year but this percentage has been decreasing slightly each year thus leaving room for improvement. The UK exhibits the greatest variability in its performance across the three years. It is disbursing a far greater quantity of aid as budget support compared to the other donors, meaning that when it does disburse funds late, the negative monetary impact this has on recipient countries is significant. In 2003, only 57% of the total amount of the programme commitment was actually disbursed, with the undisbursed component totalling more than the combined budget support programme commitments due to end in 2003 of The Netherlands, Norway and Sweden. The UK does show significant improvement in 2004, disbursing over 98% of funds on time, although this only corresponded to 8 out of 11 programmes fully disbursing on time. Without data for the most recent years, it is not yet clear if this improvement is a trend or a one off. Thus there is significant room for improvement in performance for all the donors analysed. This research unearthed the fact that the World Bank does not report on disbursements to the OECD Development Assistance Committee (DAC) database unlike all other DAC donors, and has no other method for publicly reporting on its disbursement performance, which is a major weakness. In addition, the commitment data that it reports is adjusted commitments agreed by the Bank board at the time of aid disbursements rather than originally agreed commitments in project documents. For the EC, data in the DAC database was incomplete so no direct data analysis was possible though several reports criticised the EC for poor performance in various countries. For all donors, there were some gaps and inconsistencies in the data recorded in the DAC database, showing significant room for improvements in reporting between donors and the DAC so that donors can be held to account regarding timely disbursements of aid. With respect to the global funds, information was not publicly available for either the US- initiated Millennium Challenge Account (MCA) or the President's Emergency Plan for AIDS Relief (PEPFAR) though reports hinted that they have both been slow to disburse funds. By contrast the Global Fund for AIDS, TB and Malaria (GFATM) has an excellent track record disbursing at nearly 100%, and the Bill and Melinda Gates Foundation has disbursed approaching US$6 billion for global health initiatives in developing countries since it began. The Education for All Fast Track Initiative (EFA-FTI) Catalytic Fund has done less well, only disbursing around 55% of committed funds during 2005. A significant finding of this research is that aid is twice as volatile in fragile states as in other LICs, yet fragile states include some of the countries furthest away from achieving the MDGs. Thus, this is a critical group of countries where aid predictability is of utmost importance for reasons of global security as well as important developmental concerns. The research uncovered the following principle reasons for aid volatility. Table: Main Reasons for Delays in Aid Disbursements Donor Side Recipient Country Side Technical and Administrative Delays 1. Transactions costs and cumbersome administrative procedures in donor countries 2. Different parts of the donor government having responsibility for 1. Weak procurement systems 2. Lack of willingness to sign long-term aid agreements 3. Over-optimism by government planners about the levels of aid that 5 aspects of decision-making 3. The tendency for donors to make short- term rather than long-term commitments 4. Fluctuating donor budget allocations to aid 5. Exaggerated optimism by donors on how much aid can be disbursed over a given period can be disbursed in a given time period Conditionalities 1. Excessive conditionality policies attached to aid given by donors 2. No formal system to hold donors to account for slow disbursement 1. Political concerns such as human rights violations and anti-democracy issues 2. Corruption, weak governance and the lack of a transparent budget process 3. The in-country IMF programme going off-track 4. Low absorption capacity for existing aid Other 1. Exogenous shocks (cannot be blamed on either donor or recipient country) 2. Recipient country disagreeing with content of IMF programme but still managing its economy sufficiently (cannot be blamed on either donor or recipient country) In light of the evidence, the research concludes by outlining key recommendations on how to reduce aid volatility and make aid flows more predictable in the future, thereby giving developing countries more of a chance of reaching the MDGs. They are outlined in the table below. Table: Key Recommendations Primary Recommendations Details 1. Long-Term Aid Commitments and Graduated Responses ˇ Donors should ideally move towards medium to long term commitments of 5-10 years ˇ Donors should commit funds early enough in the year to coincide with the budget cycle and support countries operating a cash budgeting system ˇ Donors should make more accurate projections of future aid allocations 2. Public Reporting of Donor Performance and an International Aid Agreement ˇ The UN should improve its coordination role to ensure that donors report disbursement and commitment data fully and accurately 3. Applying Conditionality to Future Aid Commitments ˇ Donors should change the practice of applying conditionality to present aid commitments often leading to within-year delays in disbursements, but instead apply conditionality to the following years' aid commitments which would enable the recipient government to plan ahead more effectively 4. Strengthening Absorption Capacity and Using Alternative Funding Channels Especially in Fragile States ˇ Donors should provide a capacity building fund alongside budget support to ensure that state capacity is strengthened over the longer-term ˇ In fragile states, alternative funding channels such as through NGOs, Non-State Actors or Trust Funds may need to be considered to improve the predictability of aid, but ensuring that they do not create parallel systems but instead build state capacity 5. More Transparent Reporting on Aid Disbursements by All Donors, but Particularly the World Bank and the EC ˇ All donors should more fully report on details of programme data, such as end dates, which the EC does not report at all ˇ The World Bank should report disbursements and all donors should report more complete and accurate aid commitment and disbursement data, i.e. original commitments agreed with countries rather than what is agreed at the board/governing body just before disbursement is made Secondary Recommendations Details 1. Discounting of Aid Disbursements ˇ Where donors disburse less than what they commit, recipient country government planners should be able to discount the projected aid 6 disbursements for the following year by an appropriate amount to enable more accurate budget planning 2. Using Macroeconomic Policy as a Buffer Stock Tool ˇ The recipient country can use foreign exchange reserves as a buffer when aid disbursement is lower than expected ˇ NB the main challenge with this approach is for each country to distinguish between a temporary shortfall in aid receipts and more fundamental errors in forecasting, and to have a flexible fiscal framework, so it is not a strong recommendation overall The secondary recommendations are there to mitigate the damaging effects of unpredictable aid whilst donors adapt their practices, and are not able to make aid any more predictable, thus are very much second-best solutions. The research concludes by stating that the onus is very much upon donors to take on board the initial set of recommendations with the warning that if they do not, this is likely to severely impact the ability of developing countries to reach the MDGs by 2015. 7 Introduction The international community has committed itself to the achievement of eight Millennium Development Goals (MDGs) by 2015. Clearly, existing funding flows, both national and international, will need to be channelled and targeted more efficiently, to achieve these goals. However, even if great strides forward are made in increasing the effectiveness of external aid and national resources, they are unlikely to be of the order of an extra $50 billion per year, which is the estimated annual financing gap between 2001 and 2015 needed to achieve the MDGs stated in the Zedillo Report (UN, 2001). This figure could be achieved by doubling present levels of official development assistance (ODA). However, trends in official aid given show overall declines relative to rich country income of 30% in the last thirty years (Harford, Klein and Tilma, 2004, p.1), and more specifically a 7% decrease between 1990 and 2000 (Foster and Keith, 2003, p.49). These trends are now changing and many donors have made significant commitments to increase aid over the next few years, with several countries making those commitments public at the July 2005 G8 summit at Gleneagles, Scotland. Five donors are already meeting the 0.7% of gross national income (GNI) international benchmark: Norway, Denmark, Netherlands, Luxembourg and Sweden, whilst five other donors have a plan to meet the benchmark before 2015: Belgium, Finland, France, Spain and the UK (Oxfam, 2005, p.35). However, there is a concern that the behavioural reality of many of these same donors in recent years, has been to promise aid and make commitments but then either disburse this money late, or not in its full amount. Thus the potential positive impact of aid is undermined by the unpredictability of aid flows reaching developing countries. This unpredictability can be during the life of a project, where funds are delayed but eventually disbursed with the project having its deadline extended and less of a critical impact on end users; or more critically in the case of aid provided via budget support, where the aid tap is switched off part way through the fiscal year either because the recipient country has not fully complied with donor conditionalities, or worse still, because of bureaucratic administrative systems within the donor administration causing delays in fund disbursements. As a result, there is a concern that aid flows to developing countries remain volatile and highly unpredictable. Objectives of Research Given the reliance of many developing countries on budget support to fund a significant proportion of their recurrent expenditure, the importance of disbursing this money in a timely manner in the case of salaries and other vital inputs for education and health services, and the increasing concern that aid is not being disbursed in a predictable manner, Save the Children UK felt that more evidence and analysis of aid unpredictability and its likely impacts was required. Hence, this piece of research seeks to explore the extent of the problem of aid unpredictability: whether aid predictability has been improving in recent years, whether unpredictability is more of a problem in programme aid or project aid, which donors are the most unpredictable, how unpredictable aid is in fragile states, and whether unpredictability is more of a problem in some countries and/or regions than in others. It also seeks to examine 8 different reasons for aid unpredictability both from the donor perspective and the recipient country perspective. Section 1 begins by setting the scene and commenting on recent trends in relation to aid modalities. Section 2 carries out a review of existing literature, both academic and policy- based, from Non-Government Organisations (NGOs), bilateral and multilateral agencies, on aid predictability and its impact on Low-Income Countries (LICs) with section 3 discussing some of the consequences of unpredictable aid flows. Section 4 analyses empirical data from the Development Assistance Committee (DAC) database, which is the main comparative source of data on ODA commitments and disbursements, and section 5 looks specifically at the issue of aid predictability in fragile states. Section 6 reviews the practice of some key bilateral and multilateral donors whilst section 7 investigates the predictability of global health and education funds and the Millennium Challenge Account (MCA) to see if these vertical funds provide predictable aid flows to developing countries. Section 8 seeks to identify the main reasons for aid volatility, whilst section 9, in light of the findings, outlines some recommendations on how to reduce aid volatility and make aid flows more predictable in the future. 1. Aid Modalities and Recent Trends There is a spectrum along which different donors can choose to position themselves as regards giving financial assistance to a country. There are two main types of support ­ (i) project support, where the donor gives funds for a specific period defining what the funds are to be used for and how the funds are to be managed, but has a restricted role in policy dialogue with the government; and (ii) budget support (either general budget support or support to specific sectors such as education or health), where broad policy agreements are made between government and donors, but where government uses its own procedures to manage the funds and decide how money is spent. There are also various hybrid examples that exhibit only some of the features of these two main types. This research will look at the two main modalities to see if one is more predictable than the other. The new millennium has seen changes in the way donors are financing aid. The move has been away from traditional projects with separate project management units and parallel financial systems, towards sector and general budget support (also known as programme aid), with donors giving money directly to governments to manage. Not all donors have yet embraced this paradigm shift, but it is certainly becoming more popular as a way of distributing aid both in general terms, and more specifically to the education and health sectors. Programme aid is not new ­ the International Monetary Fund (IMF) and World Bank followed this approach during the structural adjustment period in the 1980s and 1990s. But Booth (2004, p.1) argues that there has been a philosophical change in the way budget support is viewed: in the structural adjustment period, programme aid was used to bridge specific financial gaps or increase recipient countries' commitments to World Bank and IMF policy reforms, whereas now, it is used to rebuild countries' capacities to develop and implement policy for themselves, and is the preferred approach for many bilateral donors. However, it still makes up a low proportion of overall ODA funding as it is only targeted to countries considered to have a low to medium fiduciary risk and a reasonably sound macroeconomic and policy environment. 9 A recent study (Foster, 2004, p.15) of 14 countries receiving General Budget Support (GBS) shows the following modalities being used for the disbursement of every $1 of aid: Direct donor spending (TA and direct payments) not recorded in balance of payments1 $0.30 Recorded in Balance of Payments, but not reported as part of Government spending2 $0.20 Aid earmarked to specific projects $0.30 Provided as Budget Support $0.20 If the highest users of budget support (Uganda, Tanzania and Rwanda) are included, then budget support accounts for around 40% of aid flows. The use of budget support, often coupled with Sector Wide Approaches (SWAps), reflects the desire of donors to improve the efficiency and effectiveness of aid and enables donors to distribute larger volumes of aid more rapidly. Donors also argue that it moves the responsibility for planning and prioritising to the aid-recipient governments, thereby strengthening ownership. Thus two of the main advantages of budget support are: (i) it avoids creating the multiple parallel systems, thereby reducing transactions costs and strengthening government financial systems and ownership; and (ii) it is available to finance recurrent costs such as teachers' salaries; the area of the budget that is often under high stress and which projects are unable to finance. Projects traditionally only finance one-off expenditures and capital costs like building roads, hospitals and schools, and providing some basic start-up equipment. Once these are in place, the main challenge to governments is to fund the on-going cost of running education and health care systems, including paying the salaries of existing and additional new teachers and health workers, which often take up between 90 and 95% of sector budgets in developing countries. This means that once the basic infrastructure is in place, the major future need for developing countries is for recurrent financing of staff, maintenance and quality inputs such as textbooks and medicines, as the government expands access to education and healthcare to meet the MDGs and reduces overcrowding of services through providing a higher number of trained personnel. Given that projects cannot provide this kind of support, budget support rather than project support will be the main vehicle for scaling up aid. Donors have committed themselves to providing increased levels of aid funding in a more predictable manner and more aligned with recipient countries' systems, through both the Monterrey Consensus and the Paris Declaration on Aid Effectiveness. Box 1 gives details of some of the main commitments in these two documents. Box 1: The Monterrey Consensus and the Paris Declaration The Monterrey Consensus The Monterrey Consensus, which was agreed at the Monterrey Conference on Financing for Development in March 2002, lays out a framework of mutual accountability where developing countries accept responsibility for their own development whilst developed countries commit to supporting developing countries in doing this and accounting for their support. Thus whilst aid is an important tool, it should be used to support development 1 This is money paid directly by the donor to companies or individuals providing goods and consultancy services to the recipient country; thus the money is not passing through the balance of payments of the recipient country. 2 This includes debt repayments that are not counted as government spending as the donor is directly repaying the debt. 10 efforts already being made within developing countries, and should complement rather than replace domestic resources. This may require improvements in public administration, public financial management and governance structures within many developing countries. The Monterrey Consensus reinforced agreements for bilateral donors to raise ODA to 0.7% of GNI with between 0.15% and 0.20% specifically targeting Less Developed Countries (LDCs) (FTI Secretariat, 2006b, p.9). Since the Monterrey Conference, ODA has been growing in real terms and several donors have finally pledged to increase ODA as a percentage of GNI to reach the 0.7% UN recommended target that was originally agreed in 1970. The real challenge is whether these pledges have led to actual commitments and then whether these commitments are being disbursed in full. UK and France seem to be on-track, whilst Germany, Italy and Japan still have a long way to go (OECD, 2004, p.23-24). At the Monterrey Summit, all European governments formally agreed that debt cancellation should be additional to ODA rather than counted as part of it, yet this is still not the practice of most bilateral donors (Korach and Wilks, 2006, p.17). The Paris Declaration on Aid Effectiveness The Paris Declaration was agreed by more than 60 aid donors (bilateral, multilateral and civil society) and aid recipient countries at a High Level Forum in March 2005. It sets out a statement of resolve, partnership commitments, and indicators of progress regarding the issues of ownership, alignment, harmonisation, managing for results and mutual accountability. In relation to aid predictability, the Declaration states its commitment "to taking concrete and effective action to address the remaining challenges, including...failure to provide more predictable and multi-year commitments on aid flows to committed partner countries" and for donors to commit to "provide reliable indicative commitments of aid over a multi-year framework and disburse aid in a timely and predictable fashion according to agreed schedules". In addition, donors are committed to "provide timely, transparent and comprehensive information on aid flows so as to enable partner authorities to present comprehensive budget reports to their legislatures and citizens". The main indicator of progress by 2010 related to aid predictability is for donors to halve the proportion of aid not disbursed within the fiscal year for which it was scheduled (Paris Declaration, 2005). The Paris Declaration has been criticised for being reliant on weak and watered down indicators. 2. Literature Review There have been several key studies in the last five years looking at the area of aid predictability from various angles in different groups of developing countries. Bulir and Hamann have undertaken the main work in several different studies. In addition, Odedokun and a few large international NGOs have undertaken research looking at the different reasons for aid volatility. Overall evidence from these studies shows that aid is volatile, procyclical and unpredictable leading to it being less effective than it should be and often having a negative impact on economic growth, thus leaving recipient countries with challenges to short-term fiscal 11 management. High aid volatility can also increase exchange rate variability. Specific highlights from the different studies are summarised below. 2.1 Volatility in Countries with IMF Programmes Bulir and Hamann (2001) undertook a survey of 37 countries with IMF programmes in 1998, which was then drawn upon further in Bulir and Lane (2002). These two studies provide the following results: Bulir and Hamann (2001): ˇ Aid is much more volatile than domestic revenues and volatility is more severe as countries become more dependent on aid (p.30) ˇ Quarterly disbursements deviate from quarterly commitments by around 50%. In a sample of 23 countries, only 2 countries received programme aid with quarterly disbursements differing from commitments by less than 20% (p.27) ˇ Grant disbursements are lower than projections by 13%, with loan disbursements being 40% lower than projections (p.27) ˇ 24 out of 28 countries receiving programme aid saw disbursements fall short of projections by 42%, whilst the other 4 countries received disbursements in excess of projections by an average of 14% (p.27) Bulir and Lane (2002): ˇ For the 33 most heavily aid-dependent countries (i.e. aid to revenue ratios of more than 50%), aid is up to seven times more volatile than domestic fiscal revenue (p.19) ˇ Total aid disbursements in countries with IMF-supported programmes were around 20% less than what was projected at the beginning of the period (p.22) ˇ For those countries with an interruption in the IMF programme, aid disbursements were over 80% below commitments (p.22) Project Aid ˇ Average project aid disbursements were 10% lower than projections (p.22) ˇ There was little impact from interruptions to the IMF programme on project aid disbursements (p.22) Programme Aid ˇ Average programme aid disbursements were 32% and 25% smaller respectively in all countries and in countries with no interruptions in the IMF programme (p.22) From this analysis, we can see that interruptions in the IMF programme have negative effects on programme aid disbursements but no significant effect on project aid disbursements. What is more concerning is the general unpredictability of programme aid leading to consistently lower than projected disbursements and within year fluctuations in aid flows whether or not the IMF programme is on track. Another concern is that both recipient countries and the IMF tend to systematically overestimate aid disbursements. Building on this study, Bulir and Hamann undertook a further study (2003) looking at the difference between aid disbursements and projected aid commitments made by both the IMF one year in advance and the Ministry of Finance national budget planning team at the beginning of the fiscal year. Their study, along with others they refer to (Gemmell and McGillavray (1998) and Pallage and Robe (2001a)) finds empirical evidence to show that: 12 (i) Aid is more volatile than tax revenues (ii) Aid volatility increases with the degree of aid dependency (measure by the aid-to- revenue ratio) (iii) Countries with high volatility of tax revenues also experience high volatility in aid revenues (p.66) In addition, they cite Gemmell and McGillivray (1998) (p.65) who find that shortfalls in aid due to lower disbursements than commitments are often followed by cuts in government expenditure and sometimes by increases in taxation or both measures combined; and Collier (1999) (p.65) who is the only study that finds aid to Sub-Saharan Africa to be less volatile than tax revenues, and countercyclical. The general empirical results from Bulir and Hamann (2003) based on 72 countries with data from 1975 to 1997 are as follows: Project Aid ˇ For project aid, average disbursement was lower than the IMF projections by 5.1% and lower than authorities' budget projections by 15% (p.80) ˇ One percentage point of prediction error amounts to around 0.1% of GDP meaning that overestimating project aid disbursements is likely to have a negative impact on projected GDP growth (p.80) ˇ Interruptions in IMF programmes have a limited impact on disbursements of project aid (p.80) Programme Aid ˇ Programme aid shortfalls are larger than project aid shortfalls when comparing IMF projections with actual disbursements (p.81) ˇ Both IMF and original projections overestimated disbursements of programme aid by over 30%, with each percentage point of prediction error amounting to around 0.05% of GDP (p.81) ˇ Countries with off-track IMF programmes received only one-third of projected programme aid yet those countries where IMF programmes were on-track with no interruptions still received only 75% of projected programme aid (p.81) These results are consistent with their first study, showing programme aid to be more volatile than project aid and underlining again that even where IMF programmes are on-track, this does not guarantee timely programme aid disbursements. Also this study shows that projected commitments are over-optimistic compared to actual disbursements. 2.2 Trends Amongst DAC Donors in Aid Disbursements Odedokun (2003) undertook a study to explore country-specific factors that are potential determinants of donors withholding or delaying aid disbursements, based on annual data from 1970 to 2000 for the 22 DAC donors. He found the following results: ˇ G7 plus Denmark, Norway, The Netherlands and Sweden account for more than 80% of the total aid volume for all 22 DAC members ˇ Annual disbursements fell short of commitments in around 57% of the data points for all donors 13 ˇ Over the three decades, only 4.7% of commitments remained undisbursed, but this hides short-term yearly fluctuations and individual programmes or projects which were much more volatile ˇ The dollar volume of disbursements over the thirty year period was only 86% of the dollar value of commitments in monetary terms due to the delays in disbursements and the real value of the dollar commitments deteriorating over time ˇ Being a G7 member country has a negative effect on the disbursement-commitment ratio ˇ The more generous donors (defined as having a high net aid disbursement as a percentage of GDP) tend to disburse less of their aid commitments ˇ Donors giving grants tend to disburse less of their commitments than donors giving loans ˇ Donors giving a greater proportion of their aid to the poorest countries seem to disburse less of their aid than other donors. The author explains that this may be due to the lack of strong lobby groups in or for those countries who would hold donors to account ˇ A higher proportion of procurement-tied aid results in a greater proportion of committed aid being disbursed ˇ Large-sized donor governments (with large government spending to GDP ratios in their own countries) disburse a higher percentage of their aid commitments ˇ The more checks and balances there are in the donor political system, the lower the proportion of aid disbursed ˇ Conditionalities seem to matter sometimes and be overlooked at others, meaning that donors sometimes behave erratically rather than failure to meet conditionality leading automatically to donors disbursing less than they committed 2.3 Trends Amongst Multilateral Donors in Aid Disbursements Further evidence for the last conclusion from Odedokun's study can be found for multilateral donors as well as bilateral ones. Odedokun provides evidence to show donors being pressurised to disburse aid even in light of some of the failings in the recipient country to meet conditionalities. Ravi Kanbur, in his roles as IMF and World Bank resident representatives in several African countries stated that he received a lot of pressure to disburse aid from parties with a vested interest in seeing IMF and World Bank aid disbursed. In Ghana, both private sector representatives and other bilateral donors with aid linked to the World Bank's programme, put pressure on him to get World Bank funds released when Ghana had violated the conditionality of the World Bank structural adjustment credit. He also refers to this behaviour occurring in former Zaire and Senegal in the 1980s and 1990s when the US and France were putting a lot of pressure on the multilaterals to release aid even when these countries failed to meet their adjustment conditionalities. A third example relates to debt servicing, and ensuring that aid flows in so that debt servicing can flow out, an example of which is the Côte d'Ivoire (Odedokun, 2003, p.146-7). These examples show that aid predictability is not based around macroeconomic policy objectives alone, but is also strongly influenced by politics and the role and influence of the business community. Another study by Mosley and Abrar (2005, p.23) points out that compliance with loan conditionality is not necessarily a precursor for more reliable aid disbursements. They compare four countries (Uganda, Ethiopia, Zambia and Malawi) that were at a similar level (50-60%) of compliance with loan conditionalities in the 1990s. In Uganda, compliance increased over time, but aid volatility remained the same; Ethiopia where compliance increased and aid volatility decreased; Zambia where compliance did not change much and 14 aid volatility remained high; and Malawi, where compliance improved yet aid volatility became worse. 2.4 Has Aid Predictability Been Increasing in Recent Years? A more recent study by Bulir and Hamann (2005) shows that aid became slightly more predictable from the 1970s to the 1980s, but this trend stopped during the 1990s and then reversed. Aid disbursements fell short of commitments by over 40% on average over the whole period, and during 1999-2001 donors promised 50% more than they disbursed (p.10). This study confirms their two earlier studies showing that aid has been significantly more volatile than domestic revenue, and remains unpredictable and countercyclical, thus not protecting countries against GDP shocks (p.7). The authors points out that aid often exceeds 20% of GNI in recipient countries meaning that predictability is very important as is whether aid coincides with positive income shocks (procyclical aid) or negative income shocks (countercyclical aid) (Bulir and Hamann, 2005, p.5). The study concludes by stating that the combined efforts made since 2000 to better coordinate donors and harmonise aid, improve the design of aid programmes, and improve policy implementation in recipient countries do not show any improvements in aid delivery over the last five years (p.1). They comment that aid volatility has not been decreasing even in light of the Heavily Indebted Poor Country (HIPC) debt relief initiative, Poverty Reduction Strategies (PRS) and Poverty Reduction and Growth Facilities (PRGFs). In their sample of 76 countries over 1975-2003, aid commitments continue to remain poor predictors of aid disbursements, and this is a particular problem in the poorest countries (p.5). 3. Consequences of Unpredictable Aid Flows Whilst domestic revenues may fluctuate somewhat, particularly in a country heavily dependent on agriculture, unpredictable aid flows undermine the planning and public expenditure management (PEM) process in developing countries and can cause difficulty in managing the economy which can in turn lead to further aid interruptions pushing countries into a vicious cycle. If it is not really known what the likely aid flows are going to be in the next year, then it is difficult for the recipient government to plan the next budget, and with the move towards governments preparing 3-5 year medium term expenditure frameworks (MTEFs), this makes forward planning very difficult, thus undermining the meaningful discussion of resource allocation in the national budget process, and undermining the implementation of sector strategic planning. This means that social sectors are not sure of their allocations when budget support makes up a significant percentage of government resources. A World Bank study of the period 1975-95 for the fifty most aid-dependent countries calculated that aid constituted 53.8% of central government expenditure (van de Walle 1998 cited in Quartey, 2005, p.5). Foster and Keith (2004) point out that in 1990-95, aid to Sub- Saharan Africa averaged 50% of public expenditure and 71% of gross investment, compared to 20% and 31% respectively in South Asia (p.92). A more recent study examining dependency on ODA from 2000 to 2002, found that there was a large variation between countries from a small 2.2% of GDP in Bangladesh to a very significant 36% of GDP in Mozambique (World Bank, 2005, p.92). For those countries at the high end of dependency, 15 aid volatility is likely to have a significant negative impact as governments have very few choices they can make to mitigate for lower than expected aid disbursements. This can lead them to a place of budgetary instability and is a particular problem where budget support flows make up a significant proportion of the government's expenditure plans, which may lead government to have to delay or even cut important areas of social spending. Where planning is over-optimistic, as various pieces of research that this study has explored show to be the case, this will have a serious impact on the operations of the social sectors. Salaries of teachers and health workers may well end up being withheld for several months or delayed, undermining motivation and the quality of services they provide; other important recurrent inputs which impact on the services provided such as textbooks and life-saving medicines, may not be available due to the lack of funds, meaning that the quality of education suffers, and some patients end up dying due to a lack of available drugs; if money for fuel and maintenance is not released, then remote health workers may lack the resources to visit rural areas and provide care and treatment to the sick. Delays in aid disbursements in Ghana in recent years as shown in Table 1, have forced government to have to cut back on development spending or led to unplanned domestic financing or non-concessional borrowing. Table 1: Aid Commitments and Disbursements in Ghana (billions of Cedis) Year Disbursements Commitments Shortfall (%) 1999 1,275.0 1,498.1 17.5% 2000 2,385.5 2,978.9 19.9% 2001 3,739.4 3,784.6 1.2% 2002 2,868.6 4,706.3 39.1% Source: Government of Ghana's Annual Budget Statements, 1998-2003 cited in Quartey, 2005, p.13 Overoptimistic projections of aid disbursements are likely to feed through into overoptimistic projections of economic growth. Bulir and Hamann (2003) found that one percentage point of prediction error in project aid amounts to around 0.1% of GDP, and the same prediction error in programme aid amounts to around 0.05% of GDP (p.80-1). Therefore a 10% overestimation of project aid and a 30% overestimation of programme aid could lead to a 2.5% overestimation of GDP, which is sizeable. As a specific example, Foster and Keith (2004) note: "Average shortfalls in aid receipts relative to the budget were equivalent to nearly 2% of GDP in a sample of 28 countries, with no less than 24 of them suffering shortfalls. Moreover, the shortfalls were greatest on programme aid, the untied funds of most importance for macro and budget management. Even countries that met policy conditions experienced large shortfalls." (Foster and Keith, 2004, p.38) Any similar delays or shortfalls in disbursing money on more traditional projects generally have less of an immediate negative impact, as the majority of these projects are financing capital costs rather than recurrent ones. In addition, project aid flows are generally based on multiple-year disbursement schedules which may have some delays but are unlikely to be totally interrupted, whereas programme aid is usually only disbursed if the IMF programme is on track. This is supported by a report by the Strategic Partnership for Africa (SPA) that found that African partners generally rank the predictability of project aid higher than other aid modalities directly supporting sector plans. This is followed by sector budget support, 16 which was ranked higher than project assistance in two cases, and as good as project aid in one case (SPA Secretariat, 2005, p.3). Whilst humanitarian aid and emergency assistance are both volatile by nature as external shocks are often difficult to predict, more regular on-going development assistance should be more predictable to enable recipient countries to plan over at least a medium-term horizon with the assurance that the commitments donors made are reliable and are likely to be delivered. This is particularly important in highly aid-dependent countries that have very little ability to protect themselves against the adverse shocks of aid volatility, due to liquidity constraints. 4. Empirical Evidence from DAC Database on Aid Predictability in Developing Countries Graph 1 below shows overall ODA commitments and gross disbursements (i.e. total disbursements made which does not take into account recipient country debt repayments) in 4-year averages between 1964 and 2004. It also shows these commitments and disbursements disaggregated into loans and grants. Whilst loans and grants were given in nearly equal proportions in the 1960s and early 1970s, the late 1970s saw a move away from loans towards grants, and this has grown so that between 2000 and 2004, a far greater proportion of ODA is provided through grants rather than loans. This is mostly due to the 1980s debt crisis and the decision by nearly all bilateral donors to give all bilateral aid in grant rather than loan form. As a general trend, we see that disbursements are usually lower than commitments except in the early years, which may be due to incomplete data availability or because commitments may be for periods of more than one year, whereas disbursements only show the level of funds disbursed in that year. Since the early 1980s, we can see that grants are disbursing on average at a higher percentage than loans, a view supported by the empirical results found by Bulir and Hamann (2001). Graph 1: Total Grants and Loans Committed and Disbursed (Gross) 1964-2004 0 100000 200000 300000 400000 500000 600000 1964- 67 1968- 71 1972- 75 1976- 79 1980- 83 1984- 87 1988- 91 1992- 95 1996- 99 2000- 04 US$millions(2004) Grants: Commitments Grants: Gross Disbursements Loans: Commitments Loans: Gross Disbursements ODA: Commitments ODA: Gross Disbursements 17 Graph 2 below breaks down overall net ODA disbursements (i.e. after debt repayments have DC Less Developed Country y ry he 2005 DAC Development Cooperation Report points out that excluding amounts nspecified by region, 62.6% of bilateral aid is disbursed directly to LDCs and OLICs, with a proportion of net ODA disbursed (no gross isbursement data was available) to each main region of the world in 4-year averages been made ­ gross data was not available) by four different groupings of country: L OLIC Other Low Income Countr LMIC Lower Middle Income Count UMIC Upper Middle Income Country Graph 2: Destination of Net ODA by Country Income Level 0 10000 20000 30000 40000 50000 60000 70000 80000 90000 100000 1960- 63 1964- 67 1968- 71 1972- 75 1976- 79 1980- 83 1984- 87 1988- 91 1992- 95 1996- 99 2000- 04 US$millions(2004) LDC OLIC LMIC UMIC T u further 32.7% going to LMICs and 4.6% going to UMICs (OECD, 2006, p.217). This could be seen as a slight distortion in the allocation of aid according to need. A study undertaken by the Education for All Fast Track Initiative (EFA-FTI) Secretariat found that Middle Income Countries (MICs) received around two-fifths of global ODA in 2004, and DAC donor funding to LICs varied from a low of 12% from Greece to a high of 96% from Portugal. In addition, five DAC donors (the US, Japan, the UK, France and Germany) contribute 65% of aid to LICs (FTI Secretariat, 2006b, p.10-11). Graph 3 on the following page shows the d between 1960 and 2004. Over the last two decades, around one-third of ODA has consistently been given to Sub-Saharan Africa, the biggest regional aid recipient. ODA to Europe has been increasing slightly over the last fifteen years, largely as a result of the expansion of the European Union, and increased aid to Eastern Europe. Overall aid to middle-income regions remains small or declining. 18 Graph 4 gives a comparison of investment project aid and programme aid (budget support) for the period 1995-2004, when data for both types of support was available. As noted above for Graph 1, commitments may be for periods of more than one year, whereas disbursements (gross disbursement data is used) only show the level of funds disbursed in a given year, meaning that we might expect to see a time lag in the disbursement data. This time lag might be evident in the short-term, however over the life span of a typical programme (one to three years) we would expect commitments and disbursements to average out and equalise. In addition, aid commitments have been increasing in recent years. However, apart from two or three exceptional years, even allowing for disbursement lags and multiple year commitments, the graphs seem to imply that both types of aid are not disbursing in full and are being quite volatile, which would lead to difficulties in recipient country planning schedules, particularly for programme aid that is more critical if it is not disbursed in full. Further analysis is undertaken in section 6 to establish a clearer picture of how much of a problem volatility actually is at individual programme level. Graph 3: Proportion of Net ODA Disbursed to Each Region 1960-2004 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 1960- 63 1964- 67 1968- 71 1972- 75 1976- 79 1980- 83 1984- 87 1988- 91 1992- 95 1996- 99 2000- 04 Oceania Middle East Europe C&S Asia Far East Asia S America N&C America Sub-Saharan Africa North Africa Graph 4: Investment Project Aid and Programme Aid Commitments and Gross Disbursements, 1995-2004 2000 4000 6000 8000 10000 12000 14000 16000 US$000s(2004) 19 0 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 Invesment Project Aid Commitments Investment Project Aid Gross Disbursements Programme Aid Commitments Programme Aid Gross Disbursements Graph 5 shows the top ten most aid dependent countries in terms of ODA (the data used is stated as total ODA and it is not clear whether this is commitment data, net or gross disbursement data3 ) as a percentage of GNI per capita. Several of these countries are island economies with small populations, so a significant amount of aid then corresponds to a very high share of ODA as a percentage of GNI per capita. Graph 5: Average ODA Disbursed as % GNI per capita, 2002-2004 0 10 20 30 40 50 60 70 Sao Tome & Principe Congo Dem.Rep. (Zaire) Timor-Leste Guinea-Bissau Micronesia Marshall Islands Mozambique Palestinian adm.areas Eritrea Burundi Country ODA as % of GNI per capita Graph 6 on the following page gives an indication of the best and worst performers amongst the DAC donors in relation to the percentage of total gross ODA (budget support and projects) that they are disbursing, though as in Graphs 1 and 4 above, there may be a time lag between commitments and disbursements that partly explains poor performance. However, assuming that time lags would not be significant over a three-year period and do not fully explain poor performance, it would appear that the Scandinavians are some of the best performing donors, along with the Portugal, Spain, the UK, France, Luxembourg and Belgium, disbursing between 100% and 115% on average of the commitments made over the five-year period. By contrast, Korea only disbursed about 82%, and the US and the EC 87% of total aid commitments, with Italy, Finland, Austria, Canada, New Zealand, Japan and Germany doing only slightly better. It is not clear what the specific reasons were for the good performance by certain donors and the underperformance by others or whether apparent underperformance by some donors is partly due to time lags on multiple year commitments. Thus section 6 looks at individual donors in more detail to try and ascertain a more complete picture about aid volatility based on specific evidence for individual donor projects and programmes rather than overall possibly spurious conclusions from overall trend data. 3 Table 2a in the database 20 Graph 6: Average % of ODA Disbursed (Gross) 2000-04 (Left: 10 Worst Performers; Right: 10 Best Performers) 0 20 40 60 80 100 120 140 Korea UnitedStates EC Italy Finland Austria Canada NewZealand Japan Germany Australia Switzerland Greece Ireland Portugal Spain UnitedKingdom France Luxembourg Norway Belgium Denmark Sweden Netherlands Average%Disbursed 5. Predictability of Aid in Fragile States By definition, fragile states suffer from very low or weak capacity and are generally much further off achieving the MDGs than other countries. Thus the duration of any aid flows needs to be longer-term than in many higher capacity developing countries in order to build stronger and more sustainable institutions to enable these countries to make progress towards the MDGs. This means that any aid volatility in fragile states is more acute than in more stable countries. In addition, in fragile states, donors are more likely to use technical assistance and off-budget assistance via NGOs and Civil Society Organisations (CSOs) to avoid corruption and help build local government institutions. This means that they may be bypassing the opportunity to build the capacity and transparency of government systems. Over 80% of gross ODA to fragile states is grant-funded compared to only 57% in LICs (Levin and Dollar, 2003, p.55-56). Fragile states, like all developing countries, can be split into two main groups: (i) donor darlings who receive higher aid flows than poverty and policy would predict; and (ii) donor orphans who receive lower aid flows than poverty and policy would predict. The majority of donor orphans are in francophone Africa and donors have a tendency to give a higher proportion of aid to smaller, better-managed countries usually in post-conflict settings (Levin and Dollar, 2005, p.22). Levin and Dollar (2003, p.22-23) undertook a study on aid volatility in fragile states between 1992 and 2002 and they find that aid volatility is much higher in these countries than it is in LICs, and a little higher than MICs. They also find that aid to fragile states comes in spurts, with one year a particular country receiving substantial aid flows, and the next year, donors moving on to another country. Splitting the group into two groups: donor darlings and donor orphans, Levin and Dollar find that aid per capita volatility amongst darlings is very close to the level amongst LICs, but that aid per capita volatility amongst orphans is very high and accounts for the overall high levels of aid volatility for fragile states (p.25). For LICs and fragile states, Levin and Dollar found that over the 10-year period of the study, donors 21 disbursed a total 86% and 80% of committed funds respectively, whereas in MICs it was around 100% (p.38). The authors conclude that aid flows to fragile states have been twice as volatile as aid flows to LICs. Fragile states received 58% less bilateral aid and 34% less multilateral aid (overall 43%) than what they should do according to their population, poverty, policy and institutional performance levels (cited in McGillavray, 2006, p.12). By contrast, Sheelagh Stewart made the comment at a recent DFID presentation at a British Angola Forum event (25th May 2006) that aid absorption is two times the average levels of LICs in post-conflict states for the first decade after a conflict, meaning that certain fragile states may have a much greater capacity to absorb additional aid than is currently assumed. 6. Review of Donor Practices 6.1 Donor Trends and Policies The practices of four bilateral (The Netherlands, Norway, Sweden and the UK) and two multilateral donors (the European Commission and the World Bank) are commented on below in more detail. These donors have been selected, as they are some of the most generous donors and the ones that give the highest percentage of their aid via budget support, where predictability is more critical. Graph 7 below shows net disbursements (data on gross disbursements was not available) of the four bilateral donors as a percentage of GNI, supporting earlier evidence that The Netherlands, Norway and Sweden are all at least equalling or actually exceeding the UN-recommended 0.7% GNI target with their disbursements. Comparatively the UK is disbursing below 0.4% of its GNI as ODA, even though its aid disbursements are increasing. Graph 7: Net Disbursements as a % of GNI 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 2002 2003 2004 2002 2003 2004 2002 2003 2004 2002 2003 2004 Netherlands Norway Sweden UK To assess more clearly the predictability of the budget support component, the practices of the four aforementioned bilateral donors and two multilateral donors are explored in more 22 detail. Table 2 below shows budget support as a percentage of total bilateral support for the four aforementioned bilateral donors. Table 2: Direct Budget Support as % of Total Bilateral Support for Various Donors, 2000-2003 Country 2000 2001 2002 2003 Norway 2.0 3.2 4.7 4.9 Sweden 5.8 4.4 4.6 5.2 Netherlands - 8.5 11.3 20.4 UK1 24.0 24.0 18.0 22.0 1 Includes general and sector budget support. Source: Norad, 2005, p.5 1. The Netherlands The Netherlands is a significant provider of ODA. It currently supports 36 partner countries with bilateral support, with significant additional contributions to the EC, the World Bank and the UN agencies. At least 50% of its total development budget is mandated for Africa, and 15% for education (Minbuza, 2004). The Netherlands, like the UK, has provided around one-fifth of its total ODA via budget support since 2003, usually over three year commitment cycles. 2. Norway Norway has expressed a clear intent to increase the proportion of Norwegian aid provided via budget support and has seen this proportion increasing gradually in recent years. Whilst it was still just under 5% in 2003, for Tanzania, Mozambique, Uganda and Malawi, between 20 and 23% of total Norwegian support was given via direct budget support in that same year (Norad, 2005, p.4). 3. Sweden The Swedish International Development Agency (Sida) has bilateral development cooperation with 120 countries and gives general budget support to around 12% of these countries. Between 1998 and 2003, 84% of budget support went to African countries, 11% to countries in Latin America, 4% to countries in Eastern Europe and 1% to Asian countries. 67% of all budget support went to Mozambique, Tanzania and Uganda (Narea and Christensen, 2004, p.15). 11% of total bilateral aid from Sweden was given via programme aid between 1990 and 2002. In the early years, the main form was import support, but by the end of the period, general and sector budget support dominated (Narea and Christensen, 2004, p.11). In 2003, general budget support accounted for 5.3% of Sida's bilateral cooperation (Narea and Christensen, 2004, p.15). Sida plans and provides budget support in 2-3 year contracts, but these contracts are renewable each year. 23 4. UK Since the late 1990s, the UK has expressed a desire to move away from projects towards providing budget support in order to scale-up aid for the MDGs and provide more long-term and predictable recurrent financing to recipient countries. In 2003, UK's Department for International Development (DFID) launched a General Budget Support Evaluability Study to assess how well budget support has been working, and in its latest policy paper on Poverty Reduction Budget Support, states that it will continue to assess the most appropriate aid modalities on a country-by-country basis, and that budget support will only be an option where there is not a high level of fiduciary risk that is unlikely to be reduced (DFID, 2004, p.4). Budget support has nonetheless constituted around one-fifth to one-quarter of total UK ODA in recent years. DFID generally commits funding over a three-year time horizon according to individualised Country Assistance Plans. However, a recent press release by The Treasury states that the UK is planning to enter into 10-year agreements with countries to finance 10-year education plans, and that education ODA is set to increase dramatically over the next ten years (HM Treasury, 2006). This bold announcement, if it is carried through to action, could lead the way forward to longer-term aid commitments than the current 3-year average. DFID is also considering moving in the direction of providing two-tranche budget support similar to the European Commission (see paragraph below for more details). 5. European Commission The European Commission (EC) has developed budget support programmes with some countries which are usually 3-year programmes with one tranche release each year. Generally EC budget support is untargeted and has two components: a fixed tranche (on average 65%) which is released as an "all-or-nothing" payment dependent on whether or not the recipient country has met broad macroeconomic conditions; and a variable tranche (on average 35%) which is dependent on performance in public financial management, education and health sectors, so that if performance is only 60% of what was projected, then only 60% of funds will be released (European Commission, 2005, p.2). 6. World Bank The World Bank has two main types of financial aid instruments ­ specific investment loans/grants (projects) and development policy based loans/grants (budget support). These can be sectoral, multisectoral, or economic in nature. Poverty Reduction Support Credits (PRSCs) were introduced in 2001 as economic development policy based loans and Uganda was the first country to receive a PRSC loan. A PRSC programme usually involves two or three operations or individual PRSCs that together support the country's medium-term poverty reduction strategy. Many PRSCs release one tranche of funds each year, though some may have two tranches. 24 6.2 Bilateral Donor Data from OECD DAC Creditor Reporting System Database 5 Graph 8 below shows the number of budget support programmes due to complete in a given year and what proportion of the total funds originally committed for the programme had been disbursed by the programme's completion date4 . It was not possible to compare disbursements made in a given year with the original commitment for that year, as the commitment data was for a variety of one-year and multi-year programmes and did not specify a disbursement schedule for each tranche of the commitment. The data for many programmes was rejected due to inconsistencies or non-availability of data so this does not reflect 100% of all budget support programmes over this period. For example, for Sweden and Norway the proportion rejected due to lack of available closing date information was under 10%, but for the UK just under a third had to be rejected and for The Netherlands 41%5 . There was no data available for the closing date of any EC programmes so the analysis was unable to include the EC, though evidence from other studies on the predictability of EC aid is provided in the next section. There is no consistent trend between the four donors, although it is noteworthy that each donor failed to disburse any funds on one programme within the 3-year time period. The only country indicating a trend of improvement over the three years is Norway who consistently improved upon the number of programmes where all funds were disbursed by the expected completion date of the programme. In comparison the UK and Sweden have shown much more variable results. Sweden has disbursed 100% of funds by the completion date on at least half of the programmes each year. The UK disbursed 100% of funds by the completion date on at least half of the programmes in 2004, but less than this for programmes ending in 2002 and 2003. The results for The Netherlands are less variable, with Dutch aid being fully disbursed on nearly three-quarters of all programmes on average across the three years. 4 In order to compare commitments and disbursements on budget support programmes, disbursement data from the CRS 5 database was used. This database contained data on commitments and disbursements in the donor countries' currency, as well as the value of the disbursement in US$ in 2004 prices. The data was sorted by programme number and transaction number in order to track where disbursements had taken place over more than one year on programmes so that the total value and proportion of funds disbursed could be calculated. The stated closing date of each programme was noted and the programmes grouped into those ending in 2002, 2003 and 2004. Any disbursement data relating to programmes listed as closing before 2002 or after 2004 were rejected, as were any disbursements where the closing date was unavailable. The year of the disbursement was then compared to the stated `closing date' to make a judgement as to whether or not the disbursement was `on time' or `late'. The stated closing date is an indication of the programme's planned closing date and was interpreted as such. If the closing date was part way through the year, it was assumed that any disbursements took place before this date, as only the year of the disbursement was available, not the month or exact date. It should be noted that this categorisation of `on time' or late' is only in relation to the completion date, but gives no indication as to whether payments were completed in line with any agreed payment schedule during the life cycle of the programme. For example, if a programme spanned over 2 years and there were two disbursements there is no way of knowing from the data whether these disbursements took place at regular, agreed intervals. A previous approach to the analysis tried to compare the data made available on the separate commitment and disbursement databases (CRS 1 and 5 respectively), linking programmes by transaction and programme number. This line of enquiry proved complicated to pursue due to gaps in the data, particularly when either commitment or disbursement data was unavailable for programmes. The lack of corresponding disbursement data to match commitments in some of these instances could have indicated that no disbursements had taken place. For the UK, Sweden, Norway and The Netherlands, there was stand-alone commitment data with no disbursement data for 4, 4, 1 and 6 programmes respectively. 5 Out of the programmes rejected due to lack of data some of these may have had a closing date before 2002 or after 2004. 25 Graph 8: Cross-Country Comparison of the Number of Budget Support Disbursements Made Within the Expected Programme Duration 0 2 4 6 8 10 12 14 16 2002 2003 2004 2002 2003 2004 2002 2003 2004 2002 2003 2004 Netherlands Norway Sweden UK NumberofDisbursements Equal to 0% Less than 25% but greater than 0% Greater than or equal to 25% but less than 50% Greater than or equal to 50% but less than 75% Greater than or equal to 75% but less than 100% Greater than or equal to 100% Table 3: Number and Percentage of Budget Support Programmes Disbursing Their Total Commitments on Time or Late The Netherlands Norway Sweden UK 2002 2003 2004 2002 2003 2004 2002 2003 2004 2002 2003 2004 Total Programmes 8 4 7 4 8 11 4 15 11 15 9 11 Total disbursing 100% of funds by completion data 5 3 6 1 6 9 2 8 6 6 1 8 Percentage 63% 75% 86% 25% 75% 82% 50% 53% 55% 40% 11% 73% Total disbursing some proportion of funds late or not at all 3 1 1 3 2 2 2 7 5 9 8 3 Percentage 37% 25% 14% 75% 25% 18% 50% 47% 45% 60% 89% 17% Graph 9 on the following page compares the values of disbursements taking place for programmes due to end in each year between 2002 and 2004, to paint a picture of the total volume of budget support funds that are being disbursed late. It is immediately apparent that the UK is disbursing a far greater quantity of funds as budget support compared to the other donors. The sheer size of the funds being disbursed shows the serious negative financial impact that a few instances of late disbursement can have. For example, for programmes due to complete in 2003 when the largest volume of funds were disbursed by the UK, the proportion of funds disbursed on time was the most variable with only one third of programmes receiving more than 50% of their committed funds on time corresponding to 57% of the total volume of budget support. The amount that remained undisbursed totalled more than the combined budget support programme commitments made by the Netherlands, Norway and Sweden together so would have had a significant negative impact on the recipient countries. The UK does show a considerable improvement for programmes due to complete in 2004, disbursing around 98% of funds on time, although only 8 out of 11 programmes fully disbursed on time. However, the volume of aid that remained undisbursed at the end of 2004 for the UK was twice that of the volume of aid that remained undisbursed at the end of 2003 for Norway even though this corresponded to 5.8% of Norway's programme commitment total and only 1.8% of the UK's programme commitment total. Without data for 2005 and 2006, it is not clear that the UK's good performance in 2004 is yet 26 a trend, and even if it is, it is important to note that a few percent of the total budget support commitment remaining undisbursed by the end of the programme can correspond to a significant volume of aid, thus having a negative impact on the recipient country's recurrent budget. By contrast, when comparing absolute values and number of programmes Norway shows a positive trend of improvement of the both the total volumes of aid disbursed and the total number of programmes disbursing funds in full. It is worth nothing that Norway's poor performance in 2002, is due to the largest of the four programmes due to end in 2002 only disbursing 13.3% of total funds committed. Sweden is consistently showing around half of its programmes committing all their funds on time corresponding to around 75% of the total volume of programme aid being disbursed on time but this percentage has been decreasing slightly each year thus leaving room for improvement. Finally, The Netherlands is performing more consistently than the other donors, but still exhibits some variability in performance ranging from 85% to 100% across the three years examined. Graph 9: Cross-Country Comparison of the Value of Disbursements Made on Budget Support Programmes, 2002-04 100%85.3% 92.6% 100%94.2% 38.5% 71.7% 74.8% 98.2% 57.4% 90.9% 78.4% 61.5% 5.8% 0.002% 42.6% 1.8% 9.1% 21.6% 25.2% 28.4%0.02%14.7% 7.4% 0 100000 200000 300000 400000 500000 600000 2002 2003 2004 2002 2003 2004 2002 2003 2004 2002 2003 2004 Netherlands Norway Sweden UK DeflatedUS$thousands(2004) Value of disbursements made late or not at all Value of all disbursements made prior to completion date NB The percentages in graph 9 are rounded to 1 decimal place (with the exception of Netherlands and Norway in 2004) thus they may not always add up to 100%6 . 6 When calculating the data for the value of disbursements made late or not at all, if any negative figures were produced where donors had over-disbursed on programmes, these figures were discounted and presumed to be zero, as the programme had received at least 100% of committed funds, and including this data would have underestimated the values still to be disbursed, or those that were disbursed late. Please see the appendix for further information on how exchange rates were calculated. 27 To summarise, Table 4 presents the average percentages across the three years of the number of programmes and the value of those funds that were disbursed late or not at all. Table 4: 2002-04 Average of Number of Programmes and Their Value that were Not Disbursed on Time Netherlands Norway Sweden UK Average percentage, 2002-04, of the number of programmes not disbursed on time 25.6% 39.4% 47.4% 55.3% Average percentage, 2002-04, of the value of disbursements not disbursed on time 7.37% 22.4% 25.06% 17.83% 6.3 Evidence on EC Disbursement Delays It is a significant constraint that the EC does not make available complete data for its programme disbursements. This means that it is not possible to track programmes to monitor disbursement predictability. Specific evidence for EC disbursement delays are referred to in Odedokun 2003, (p.144-5) where the NGO Population Concern experienced two significant delays in receiving EC funding which resulted in the partner NGOs requesting Population Concern not to seek future financing from the EC: (i) there was a 13-month delay in the disbursement of funds for four mini-projects funded by the EC in Bolivia and Peru; (ii) there were serious delays in the funding of a community-based distribution programme in Karachi leading to the director of a local NGO taking out a personal loan to pay staff. Odedokun points out that over the five years to 1999, the average delay in disbursing EC funds for an already committed programme has increased from 3 to 4.5 years with a few programmes having a backlog of undisbursed aid commitments totalling more than 8.5 years' payments (cited in Odedokun, 2003, p.145). Odedokun concludes that there are three main reasons for disbursement delays by the EC: (i) the normal time lag between commitments and disbursements; (ii) conditionalities not being met by recipient countries; and (iii) slowness in EC spending mainly related to administrative or bureaucratic delays, which he argues is the main factor. He concludes: "The fundamental mistake has been to allocate excessive funds in the first place for predominantly political reasons." (cited in Odedokun, 2003, p.144) A subsequent review of 34 programmes in 20 Africa, Caribbean and Pacific countries between 2001 and 2003 found that on average 71% of the variable tranche was disbursed (European Commission, 2005, p.37). Despite this, there were still delays in the disbursements. The average delay between the planned disbursement date and the government request for the variable tranche disbursement averaged 9 months, though this average was largely driven by two or three cases of very long delays (over a year) due to macroeconomic problems which led to the IMF programme going off-track. The main other reason for delays relates to issues concerning conditionalities. On a more positive note, the EC managed to reduce the time taken between the Government's request and the EC headquarters decision from 4 months in 2003 to 6 weeks in 2004, largely due to new procedures adopted at the end of 2003 (European Commission, 2005, p.44). An SPA survey showed that the EC was more likely to disburse on time than the other multilaterals, and less likely than the bilaterals. In 2003, 72% of EC commitments were disbursed on time, compared with 25% for the IMF and 28% for the World Bank. However, when examining overall eventual disbursement, 21% of EC funds committed for 2003 were 28 eventually disbursed in 2004, with only 7% totally lost, compared to 25% lost fund for the IMF and 28% lost funds for the World Bank (SPA 2004 cited in European Commission, 2005, p.44). The EC gave the reasons outlined in Table 5 for disbursement delays, which differed only slightly to recipient countries' perspectives, which did not recognise administrative problems in their own countries as a valid reason. Table 5: EC Given Reasons for Disbursement Delays Reasons for Disbursement Delays EC Administrative processes of the EU 40% Failure to satisfy conditions 35% Administrative problems in recipient countries 25% Source: European Commission, 2005, p.44 An Oxfam survey of donor practices across 11 developing countries in 2004, found that only in one in three cases does aid arrive on time, with the EC being the worst performer, only 14% of its aid arriving on time and 20% of its aid arriving over one year late. In 25% of cases, aid disbursements arrived between six and twelve months late (cited in Oxfam, 2005, p.9, p.56-58). 6.4 Comments on World Bank Disbursement Data Availability It was very difficult to find detailed data on World Bank aid disbursements by type (budget support, project support, loan and grant). This information is not currently available in the DAC database, and the only information readily available is aggregate data on commitments and disbursements in the World Bank 2005 Annual Report summarised in Table 6 below. Table 6: IDA Commitments and Disbursements 2001-2005 IDA FY2001 FY2002 FY2003 FY2004 FY2005 Number of Projects 134 133 141 158 160 Of which Development Policy Loans (DPLs) 15 23 24 23 322 Amounts (millions US$) Commitments 6,764 8,068 7,282 9,035 8,696 Of which DPL 1,826 2,443 1,831 1,698 2,301 Gross Disbursements 5,492 6,612 7,019 6,936 8,950 Of which DPL 1,280 2,172 2,795 1,685 2,666 Net Disbursements 4,495 5,549 5,651 5,538 7,330 % Gross Disbursements 81% 82% 96% 77% 103% Of which DPL 70% 89% 153% 99% 116% % Net Disbursements 66% 69% 78% 61% 84% Source: World Bank, 2005, p.2 The World Bank has a long way to go in the reporting of its data, in that this commitment data is not the commitment made to recipient governments at the beginning of projects, but rather the commitments that are made by the Bank board at the time of disbursement. This is why commitment and disbursement data do not deviate from each other significantly. 29 7. Review of Global Funds 7.1 Health "The share of total bilateral ODA allocated to the health sector (including health and population) increased from 3.8 percent in 1990 ($2.2 billion) to 6.8 percent in 2002 ($2.9 billion)...Total DAH [Development Assistance to Health] from major selected sources increased...from $6.4 billion on average between 1997-1999 to $8.1 billion in 2002. Most of this increase was due to new funds committed by both public and private sources to the Global Fund to Fight AIDS, Tuberculosis and Malaria (GFATM)." (Michaud, 2003, p.1-2) There are three main large-scale funds for health: the Global Fund to Fight AIDS, Tuberculosis and Malaria (GFATM), the US Government President's Emergency Plan for AIDS Relief (PEPFAR) and the Bill and Melinda Gates Foundation. The GFATM gives money directly to governments, with the other two funds providing finances to a mixture of NGOs, CSOs, research bodies and US government departments. 1. GFATM The Global Fund to Fight AIDS, Tuberculosis and Malaria (GFATM) was established in January 2002 to draw in, manage and disburse additional funding to fight the three main health pandemics. It has had five rounds of funding approving a total of US$4.9 billion to over 350 grants in 131 countries. By 22 February 2006, the Fund had signed agreements with countries for 84% of approved grants (US$3.56 billion) and it had disbursed a total of US$1.99 billion to 127 countries (GFATM, 2006, p.1). The GFATM approves funding for five years, but commits funds initially only for two years to the recipient government, with any funding after this being conditional on satisfactory progress and performance during these first two years. In the event of insufficient finances in the future, phase two funding for existing grants takes priority over the signature of new grants. Disbursements usually take place quarterly, and will always lag behind commitments, as commitments are made at the beginning of a two-year period for the whole period, whereas disbursements happen quarterly in each quarter of the two-year period. Initial disbursements are often quite small as they are used to strengthen programme capacity and prepare procurement plans that will in turn trigger future disbursements for medicines. Despite, this the GFATM is broadly on-track in relation to its disbursements, with a mean disbursement percentage of 62% with 65% mean time elapsed for the total funding commitments (GFATM, 2006, p.4). 2. PEPFAR The President's Emergency Plan for AIDS Relief (PEPFAR) was created by the US government in May 2003 as a five-year, US$15 billion global initiative to fight HIV/AIDS. The US Congress has approved the overall funding of US$15 billion but the actual amount that will be provided each year is subject to annual agreement and approval by Congress. Of 30 this overall commitment, $1 billion has been earmarked to the GFATM, on the condition that the GFATM shows good progress and results. The main focus of PEPFAR (55%) is to provide antiretroviral treatment for individuals with HIV/AIDS, though there are also elements for palliative care (15%), the prevention of HIV/AIDS (20%), and assisting orphans and vulnerable children (10%). Fifteen countries are the focus of PEPFAR: Botswana, Cote d'Ivoire, Ethiopia, Guyana, Haiti, Kenya, Mozambique, Namibia, Nigeria, Rwanda, South Africa, Tanzania, Uganda, Vietnam and Zambia. However, there are other non-focus countries, such as India who are also recipients of PEPFAR money. PEPFAR distributes money through a number of US government agencies, including the US Agency for International Development (USAID), the US Department of Defense, the US Department of Health and Human Services (HHS) (who in turn pass money onto other agencies and research bodies), the Department of Labor, the Peace Corps and the Census Bureau (www.avert.org/pepfar.htm). Each of these agencies then passes funds down to "prime" partners who may then give grants on to "sub-partners" to implement PEPFAR's plans (www.avert.org/pepfar.htm, p.10). This makes it very difficult to track PEPFAR funds as a whole. Due to safety concerns, all pharmaceutical products paid for with PEPFAR fund have to be approved by the US Federal Drug Agency (FDA) or an equivalent regulatory agency in Canada, Japan or Western Europe. This meant that initially, most generic drugs were excluded as whilst they may have been approved by the World Health Organisation (WHO), they had not been approved by one of these other regulatory agencies. Action Aid and other have been very critical of this stating that it therefore puts contracts into the hands of US pharmaceuticals (Action Aid, 2005, p.19). This situation has slowly changed, so that by the end of 2005, 15 generic drugs had been approved by the FDA including two Fixed Dosed Combinations (www.avert.org./pepfar.htm, p.8). This decision has the potential to enable more people living with HIV/AIDS to access drugs as generic brands are usually cheaper. During 2004, a total budget of US$570.2 million was committed, and for 2005, there was a planned budget of US$1.032 billion (www.avert.org/pepfar.htm, p.11). However, there are no publicly available figures on actual disbursements, and then on whether funds that have been disbursed to partners have already been spent, so it is not possible to comment on the predictability of PEPFAR fund flows. 3. The Bill and Melinda Gates Foundation The Bill and Melinda Gates Foundation was established in 1994 with an endowment of US$29.1 billion, with total grant commitments since inception of US$10.2 billion. The Foundation supports four main areas: global health, education, public libraries and the Pacific West area housing, community service and early learning services. Whilst the latter three are focussed on marginalized and deprived communities in the US, the global health fund specifically targets developing countries and the global health issues of HIV/AIDS, Malaria, Tuberculosis (TB) and other infectious diseases. Table 7 on the following page outlines total commitments made by the Bill and Melinda Gates Foundation in the area of global health. 31 Table 7: Total Commitments Made by the BMGF in Global Health to June 2006 Global Health Programmes (Total) $6,509,318,891 HIV, TB and Reproductive Health $1,900,821,297 Infectious Diseases $1,592,621,889 Global Health Strategies $2,424,965,606 Global Health Technologies $443,286,269 Global Health Research, Advocacy and Policy $147,623,830 Source: www.gatesfoundation.org/GlobalHealth Most of the resources are channelled through NGOs and research institutes, with 767 different grants having been made to date. In addition, $150m has been contributed to the GFATM for onward disbursement. Many of these grants are of several years' duration. During 2005, grant payments totalling $1.36 billion were made. The Bill and Melinda Gates Foundation is of similar magnitude in terms of financing health initiatives as the GFATM. 7.2 Education In 2002, the Education for All Fast Track Initiative (EFA-FTI) was launched as a global partnership between developing countries and donors to ensure progress in achieving the MDG of Universal Primary Education (UPE) by 2015. For countries to receive FTI- endorsement, they must satisfy two criteria: (i) to have a poverty reduction strategy or equivalent in place; and (ii) to have a sound education sector plan in place that has been endorsed by in-country donors Once a country has received endorsement, there are two main channels for increased donor funds to flow to that country to assist financing the education sector: (i) for donor orphans (i.e. those countries with four or fewer bilateral donors each giving over US$1m to the education sector), the Catalytic Fund (CF) was established at the end of 2003 to provide transitional funding for two to three years until more donors come on board; or (ii) for countries with more donors, the onus is on the local donor group to provide increased levels of better-coordinated aid to the education sector. For those countries that do not have education plans, a further fund, the Education Program Development Fund (EPDF) was established at the end of 2003 to provide technical support and capacity building for countries to develop a sector plan. The EPDF is a much smaller fund than the CF, with total donor commitments of US$30m for the period 2005-2007 (FTI Secretariat, 2006c, p.3). A total of 20 countries7 had been FTI-endorsed by February 2006, with 54 countries receiving technical support through the EPDF including 14 of the FTI-endorsed countries (FTI Secretariat, 2006c, p.1). The FTI appears to have had a positive impact on external financing for education in developing countries. Commitments to basic education have risen from $1.8 billion in 2002 to $3.4 billion in 2004, almost the minimum estimated level of aid of $3.7 billion needed annually by the Zedillo report to support UPE. However, disbursements have been much lower than commitments, with only $1.3 billion being disbursed in 2004. Despite this low disbursement rate, 70% of committed funds were disbursed in LICs in 2004, an increase from 7 Burkina Faso, Djibouti, Ethiopia, Gambia, Ghana, Guinea, Guyana, Honduras, Kenya, Lesotho, Madagascar, Mauritania, Moldova, Mozambique, Nicaragua, Niger, Tajikistan, Timor Leste, Vietnam and Yemen. 32 62-63% in 2002-2003 (FTI Secretariat, 2006c, p.4). If budget support to education and technical assistance are included, then the overall aid committed to basic education in LICs over 2003/04 was around US$2.6 billion, but only US$1.4 billion was actually disbursed (FTI Secretariat, 2006b, p.5). Catalytic Fund Progress on Commitment and Disbursements There are currently 14 countries receiving funds from the Catalytic Fund, and 9 donors who have made financial commitments to the Fund totalling $165 million over the last three years with future commitments of $279.7 waiting to be released to the CF. Table 8 below gives an overview of these commitments and disbursements made. Table 8: Commitments and Disbursements Made by Donors to the Catalytic Funds (US$m equivalents) Country 2003-04 2005 2006 2007 Total 2003-07 Cumulative Payments Balance to be paid* Belgium 1.2 2.6 1.2 1.2 6.2 3.7 2.5 EC 0 0 37.8 37.8 75.6 0 75.6 Ireland 0 0 1.5 0 1.5 1.5 0 Italy 2.4 2.4 0 0 4.8 4.8 0 Netherlands 39.5 56.1 60.0 72.0 227.6 93.8 133.8 Norway 5.9 8.1 25.5 3.0 42.5 39.5 3 Spain 0 6.0 0 0 6.0 6.0 0 Sweden 0 5.3 10.4 0 15.7 15.7 0 UK 0 0 32.4 32.4 64.8 0 64.8 Total 49.0 80.5 168.8 146.4 444.7 165.0 279.7 * This balance mostly relates to pledges made for 2006 and 2007. Source: Taken from Table 1 of FTI Secretariat, 2006a, p.3. Whilst it appears that some donors have made commitments that they have not yet released to the CF for onward commitment and disbursement at country level, this is due to the unwillingness of many bilateral donors to have idle funds that are not being disbursed sitting in an account managed by the World Bank. A total of $320.7m has been committed for EFA spending in 14 different country programmes as shown in Table 9 below. Table 9: Commitments and Disbursements Made to Countries from the Catalytic Funds (US$m) 2003-07 2003/04 2005 Planned Disbursements Remaining Sum to Disburse Country Commitments Allocated Disbursed Allocated Disbursed 2006 2007 Djibouti 8.0 3.0 3.0 2.0 Ghana 30.0 8.0 8.0 6.0 5.0 11.0 Guyana 12.0 4.0 4.0 4.0 4.0 4.0 Kenya 72.6 24.2 24.2 12.1 12.1 24.2 Lesotho 11.9 3.6 3.6 4.7 Madagascar 60.0 10.0 6.0 19.0 35.0 Mauritania 9.0 7.0 7.0 2.0 2.0 Moldova 8.8 2.2 2.2 4.4 Nicaragua 21.0 7.0 7.0 7.0 7.0 7.0 Niger 21.0 13.0 9.0 8.0 4.0 4.0 4.0 Tajikistan 18.4 4.6 4.6 9.2 The Gambia 12.0 4.0 4.0 4.0 4.0 4.0 Timor- Leste 6.0 1.5 1.5 3.0 33 Yemen 30.0 10.0 10.0 10.0 10.0 10.0 Total 320.7 45.0 41.0 77.2 42.2 83.0 36.0 118.5 Source: Taken from Tables 3 and 4 of FTI Secretariat, 2006c, p.5. From this table, we can see that 91% of all funds planned for 2003 and 2004 were disbursed in 2003/04 (separate figures are not available for the two years separately), but only 55% of all 2005 funds were disbursed during 2005. Whilst the time lag between commitments and disbursements has halved over the last 2 years, it is still taking 3-4 months on average between allocation decisions and funds disbursement taking place (FTI Secretariat, 2006a, p.3). The two main challenges to the CF are: (i) the low disbursement to commitment ratio in 2005; and (ii) the fact that whilst overall pledges to the CF for the next few years based on original disbursement schedules, currently total US$198.5m, the estimated financing gap in these 14 countries over the next three years is a huge US$1.1 billion, hence the CF can only meet around 17% of their needs. In addition to the 14 CF eligible countries, the other 6 FTI- endorsed countries that are not eligible for CF finances face a financing gap of nearly US$1 billion too (FTI Secretariat, 2006a, p.6). 7.3 The Millennium Challenge Account The Millennium Challenge Account (MCA), which is managed by the Millennium Challenge Commission (MCC) is a Compact which was set up by the US government in early 2004 to reduce poverty through sustainable economic growth in developing countries and funds flow from the this account to recipient country governments. Its focus is specifically on developing countries that have and maintain sound policy environments, and funds can be used for investments in education, infrastructure, private sector development and agriculture. There are currently 23 countries eligible for MCA assistance, twenty of which are low- income countries, and three of which are LMICs. Four are based in Latin America, six in Eurasia, and thirteen in Africa. During 2005, compacts of 4-5 years were signed with five countries: Cape Verde, Georgia, Honduras, Madagascar and Nicaragua together totalling US$905 million over the five-year duration. Three other countries have had funds approved by the MCC board (Armenia. Benin and Vanuatu) and should have formal compacts signed during the first half of 2006, along with a further three countries (MCC, 2006). However, much of 2004 and the early part of 2005 involved setting up the MCC and the MCA and ensuring relevant staff were in post and negotiations for funding began. This has meant that the MCA has been slow to disburse to date, with no official figures available, but a commitment that: "MCC disbursements will accelerate in FY 2006 and 2007 as the first group of signed compacts move beyond the initial stages of implementation." (MCC, 2006, p.4) To date only one country, Madagascar, has actually received any funding, meaning that the MCA does not yet seem to be a predictable source of long-term funding for developing countries. 34 8. Reasons for Aid Volatility There are a plethora of reasons for aid volatility, some of which should be blamed on donors, others of which are more related to recipient countries. The two main categories of reasons relate to (i) technical and administrative delays and (ii) conditionalities set by donors. In addition, recipient countries can face exogenous shocks through no fault of their own which cause them to fail to meet conditionalities and they can sometimes disagree with the content of an IMF programme whilst still managing their economies sufficiently. SPA 2005 Assessment asked donors why aid was unpredictable, found the following reasons as shown in Table 10 below. Table 10: Donor Assessed Reasons for Unpredictable Aid Reasons Percentage Failure of Policy Conditionality 40% Donor administrative problems 29% Recipient country administrative problems 25% Political problems 4% Other factors 2% Total 100% Source: Celasum and Walliser, 2005, p.3. This led Eifert and Gelb (2005) to conclude that around half of the volatility of programme aid might be performance related with the half being linked to administrative delays and other exogenous factors such as low rainfall leading to drought and poor performance in the agricultural sector negatively impacting economic growth targets. Exploring these two areas in more depth, we find the following reasons prevail as outlined in Table 11 below. Table 11: Main Reasons for Delays in Aid Disbursements Donor Side Recipient Country Side Technical and Administrative Delays 1. Transactions costs and cumbersome administrative procedures in donor countries 2. Different parts of the donor government having responsibility for aspects of decision-making 3. The tendency for donors to make short- term rather than long-term commitments 4. Fluctuating donor budget allocations to aid 5. Exaggerated optimism by donors on how much aid can be disbursed over a given period 1. Weak procurement systems 2. Lack of willingness to sign long-term aid agreements 3. Over-optimism by government planners about the levels of aid that can be disbursed in a given time period Conditionalities 1. Excessive conditionality policies attached to aid given by donors 2. No formal system to hold donors to account for slow disbursement 1. Political concerns such as human rights violations and anti-democracy issues 2. Corruption, weak governance and the lack of a transparent budget process 3. The in-country IMF programme going off-track 4. Low absorption capacity for existing aid 35 Other 1. Exogenous shocks (cannot be blamed on either donor or recipient country) 2. Recipient country disagreeing with content of IMF programme but still managing its economy sufficiently (cannot be blamed on either donor or recipient country) 8.1 Technical and Administrative Delays Donor Side 1. Transactions costs and cumbersome administrative procedures in donors' annual budgetary allocations can lead to a time lag between commitments and disbursements. Likewise, time spent working on donor coordination and harmonisation, albeit a good thing, can cause delays in the short-term. This seems to be one of the main causes of delays in within-year disbursements. 2. There are three different parts of donor government usually responsible for aid: parliaments approve it, development agencies manage it, but Ministries of Finance are sometimes in charge of disbursing it. This can lead to time lags between the donor development agency in-country approving the aid disbursement, and the headquarters office relaying the information to the Ministry of Finance in the donor country to authorise the release and transfer of funds to the recipient country. 3. Donors tend to make short-term commitments of one to three years and many are unable to commit guaranteed funds for more than one year in advance due to legislative constraints and the need for parliamentary approval for each annual allocation. In 70% of cases, donors commit aid for three years or less (cited in Oxfam, 2005, p.9, p.55). In addition, indications of long-term financial support are less binding than formal commitments. 4. Donor budget allocations sometimes fluctuate due to donors changing their priorities or facing budget pressures in their own countries. The war in Iraq has put pressure particularly on the US and UK governments spending. Likewise, the French pension crisis during 2003 led to a freeze on overseas aid in certain countries (personal communication with French aid officials in Rwanda during 2003). 5. There is an exaggerated optimism by donors about the level and speed of likely disbursement of aid commitments as evidenced by empirical research mentioned in the literature review. Recipient Country Side 1. Weak procurement systems can delay financial disbursements and project schedules which Action Aid (2005) cites as being a problem in Ethiopia and Zambia (p.27). 2. There is sometimes a lack of willingness for recipient countries to sign up to longer-term agreements with donors. An example is the Government of India not permitting donors to make multi-year budget support commitments in Andhra Pradesh for unknown reasons (Lawson et al., 2003a, p.72). During the 2001/02 fiscal year, budget support was released at the end of the fiscal year, which meant that the Government of Andhra Pradesh had to borrow US$250m from the Government of India repaying it when the budget support was finally 36 disbursed. This delay seems to have been due to the length of time taken from the donor organisations to go through the approval process (Lawson et al., 2003, p.63). 3. Government planners, partly led by donors, can be overoptimistic about the level of likely disbursement of aid commitments as shown by empirical research already referred to in the literature review. 8.2 Conditionalities Donor Side 1. Excessive conditionality policies attached to aid given by multilateral donors undermines its predictability. World Bank conditions were seen to be excessive in 60% of the cases recorded in the Oxfam survey of donors' practices: in 2004-05, the Ethiopian government had 85 policy actions to fulfil, in 2005-07 there are 84 policy action for the Vietnamese government to fulfil, and the Tanzanian government had 78 policy actions to complete during 2004-05 with some additional ones added by individual bilateral donors (Oxfam, 2005, p.60). 2. There is no formal system to hold donors to account on aid disbursements if they are disbursing late due to administrative delays. Recipient Country Side 1. Political concerns within the recipient country such as human rights violations and anti- democracy issues lead donors to suspend or delay aid. In June 2005, the UK suspended budget support to Ethiopia after 36 people were killed in protests over the election results. In January 2006, Hilary Benn, UK's international development minister, announced that the UK could no longer provide budget support to Ethiopia given the on-going concerns about human rights, but that the UK would explore other options for supporting poverty reduction in Ethiopia including a possible basic services grant covering health, education and water (DFID, 2006). Ireland, the UK and The Netherlands cut budget support to Uganda during 2002/03 due to disagreements with government about military spending. Ireland reallocated its budget support to the Poverty Action Fund (PAF) a ringfenced account for priority poverty reducing expenditure. The UK suspended aid again for several months in 2004 related to dissatisfaction on the defence review. Various donors suspended budget support again when President Museveni modified the Constitution so that he could run for a third term. When he then jailed his main opponent on charges of rape and treason, the UK suspended 20m of budget support channelling 15m through as humanitarian aid to the north of the country through the UN agencies and reserving a decision on disbursing the last 5m until after the election in February 2006 (DFID, 2005b). 2. Corruption, weak governance and the lack of a transparent budget process in recipient countries lead to concerns by donors about public expenditure management issues. In Chad, the World Bank suspended all operations after Parliament approved amendments to the Petroleum Revenue Management Law which permitted the redirection of oil revenues away from priority poverty reduction expenditure as had originally been intended (De Renzio, 37 2006, p. 1). The undisbursed aid totalled US$124m and included both projects and programme aid across all sectors including education, transport and decentralisation (World Bank, 2006). Aid flows have since resumed. In 2001, Mozambique faced a serious corruption scandal and crisis in the banking sector which led the donor community delaying disbursements for several months until the government showed a commitment to four follow- up actions to address the banking crisis after which aid was disbursed as planned (Norad, 2005, p.7). Since then, donors have wanted to retain the ability to withhold or delay aid disbursements over concerns about national governance issues (Lawson et al., 2003a, p.61). Several budget support donors including Norway and the UK decided to postpone disbursements to Tanzania during 2002 until the details of the government's decision to buy a US$ 40m Air Traffic Control System were made open and the donors felt that an acceptable solution had been found. After this in 2004, it was felt that very slow progress had been made in establishing the Public Financial Management Reform Programme (PFMRP) which led Norway to decide to make its tranche release during 2004/05 dependent on progress on this reform (Norad, 2005, p.7). The World Bank delayed disbursements to Uganda during 2003 and 2004 over administrative problems, the lack of clarify on budget execution and questions about the implementation of the Leadership Code (Norad, 2005, p.18). In early 2005, the Global Fund to Fight AIDS, TB and Malaria (GFATM) suspended aid to Uganda for two months due to concerns about financial mismanagement (AFP, 2005). Since then, aid has been resumed, but under the careful watch of a well-known international auditing firm. However, it is worth noting that disbursements are not always delayed in similar cases in that aid disbursement is a very political process that is influenced by other economic and geo- political interests. 3. The in-country IMF programme goes off-track and has a knock-on effect on other donors disbursing their aid allocations. The 2001-2003 IMF programme in Malawi went off-track during 2002 leading the IMF, the World Bank, the EU, the UK, Sweden and Norway to suspend aid due to the government's lack of fiscal control. Aid flows were resumed in October 2003. The IMF programme went off-track again in 2004, but most donors continued to disburse budget support (Norad, 2005, p.7). Within year disbursement in Rwanda has fluctuated since 2001. In 2002, disbursements were delayed for six months due to the IMF programme agreement being delayed. As a consequence District governments received no funding at all for the first three months of the financial year and all Ministries were on a cash budgeting system for the year (personal correspondence with Ministry of Finance officials). Then in 2004, disbursements were delayed twice for three months due firstly to the IMF programme going off track, and secondly to the threat of military incursion by Rwanda into the DRC (Kanyarukiga et al., 2006, p.23). 4. There can be low absorption capacity of existing aid leading to slower than predicted disbursements. Often the issue of absorption capacity is raised by donors, yet it may be due to reasons such as weak procurement systems in the recipient country or a badly-designed donor project that is cumbersome to manage, rather than the blame lying directly with the recipient country. In addition to the reasons on the donor side and recipient country side mentioned above, there are two other reasons where blame cannot clearly be laid on one party or the other. 1. Countries hit by external shocks such as drought leading to a bad harvest having a knock- on effect on macroeconomic targets, often have their aid temporarily suspended as they have 38 delayed the necessary economic adjustment to meet IMF conditionalities. This ideally should not be punished by donors withholding or suspending aid, but is not always the case. 2. Disagreement between the recipient country and the IMF on the content of the IMF programme which leads to delays in aid disbursements, as mentioned above in the case of Rwanda in 2002. The recipient country should not be blamed for this provided that it is still managing its finances well. All these possible reasons are based on evidence from in-country reports and research analysis rather than being the results of a specific empirical study to identify cross-country reasons for aid volatility. Odedokun (2003, p.162) points out that no single attempt at a formal empirical study of this nature has been undertaken. Despite this, these examples show how volatile aid can be, even in countries that have been hailed as great development success stories macroeconomically, such as Uganda and Ethiopia. 9. Recommendations and Conclusion In light of the evidence showing that aid predictability is a genuine problem, this section recommends a series of different measures that could be taken to try to improve the predictability of aid from the donor side to enable donors to disburse on time, within-year, across years and over the long-term. Table 12 below gives an overview of these recommendations, with more details in the subsequent text. Table 12: Key Recommendations Primary Recommendations Details 1. Long-Term Aid Commitments and Graduated Responses ˇ Donors should ideally move towards medium to long term commitments of 5-10 years ˇ Donors should commit funds early enough in the year to coincide with the budget cycle and support countries operating a cash budgeting system ˇ Donors should make more accurate projections of future aid allocations 2. Public Reporting of Donor Performance and an International Aid Agreement ˇ The UN should improve its coordination role to ensure that donors report disbursement and commitment data fully and accurately 3. Applying Conditionality to Future Aid Commitments ˇ Donors should change the practice of applying conditionality to present aid commitments often leading to within-year delays in disbursements, but instead apply conditionality to the following years' aid commitments which would enable the recipient government to plan ahead more effectively 4. Strengthening Absorption Capacity and Using Alternative Funding Channels Especially in Fragile States ˇ Donors should provide a capacity building fund alongside budget support to ensure that state capacity is strengthened over the longer-term ˇ In fragile states, alternative funding channels such as through NGOs, Non-State Actors or Trust Funds may need to be considered to improve the predictability of aid, but ensuring that they do not create parallel systems but instead build state capacity 5. More Transparent Reporting on Aid Disbursements by All Donors, but Particularly the World Bank and the EC ˇ All donors should more fully report on details of programme data, such as end dates, which the EC does not report at all ˇ The World Bank should report disbursements and all donors should report more complete and accurate aid commitment and disbursement data, i.e. original commitments agreed with countries rather than what is agreed at the board/governing body just before disbursement is made Secondary Recommendations Details 1. Discounting of Aid ˇ Where donors disburse less than what they commit, recipient country 39 Disbursements government planners should be able to discount the projected aid disbursements for the following year by an appropriate amount to enable more accurate budget planning 2. Using Macroeconomic Policy as a Buffer Stock Tool ˇ The recipient country can use foreign exchange reserves as a buffer when aid disbursement is lower than expected ˇ NB the main challenge with this approach is for each country to distinguish between a temporary shortfall in aid receipts and more fundamental errors in forecasting, and to have a flexible fiscal framework, so it is not a strong recommendation overall 9.1 Primary Recommendations 1. Long-Term Aid Commitments and Graduated Responses Multilateral aid is generally less volatile than bilateral aid, as funding for multilateral agencies is agreed and committed to over a longer time frame and is subject to lower transactions costs, and is partly based on reinvestments from debt repayments and capital market borrowing providing more predictable inflows of cash to reinvest in developing countries. By contrast, bilateral aid is subject to short-term commitments generally of 1-3 years and is now nearly all provided as grant aid, with each year often subject to Parliamentary approval before funds can be firmly committed. If bilateral donors could move towards a 3-5 year financing commitment for individual countries, this would help make aid more continuous and less likely to be unpredictable due to administrative delays in signing new agreements every few years. In addition, global taxes (e.g. a carbon tax or international tax on flights) or funds raised through the International Finance Facility (IFF) or the existing Global Funds or Compacts could provide more stable and predictable longer-term financing provided they do not have a bureaucratic decision-making process that continues to delay disbursements. However, these new financing facilities are likely to be disbursed through existing funding channels, which may not take away bureaucratic delays where these exist. For aid given via budget support, funds need to be committed early enough to inform budget preparation, and should ideally be disbursed in one tranche early in the year to help countries operating a cash budgeting system. More accurate medium-term projections of future budget support should be given at the same time, even if they are only fully firmed up as each year passes. Even though countries are moving towards three-year commitments, these commitments are often only intentions and are currently not respected even within-year, leading to budget support being less predictable in general project support. A more ideal scenario would be one in which donors committed aid in a 5-10 year plan with the first few years being firm commitments and the outer years' commitments becoming firmed up in a rolling plan. The main challenge to donors of this idea is that they may be unwilling to commit longer-term financing as this will limit their ability to respond to unforeseen urgent priorities or the threat of domestic budget cuts (HLFH, 2005, p.4). However, DFID's recent announcement of its intention to provide 10-year funding for the education sector in developing countries might be a step forward for other donors to follow. Donors set conditionalities for budget support as they want to ensure that aid money supports good governance and a progressing economy. However, recipient countries want to see budget support become more predictable rather than it being so easy to turn on and off at a moment's notice. One compromise to achieve both these approaches together is to have a graduated response to budget support flows. This is an approach that has been spearheaded 40 by the European Commission and is based on releasing two components of budget support ­ a fixed component and a variable component. The fixed component is released in full or not at all based on whether or not the recipient country has met broad macroeconomic conditions, normally interpreted as the IMF programme being on-track, but sometimes connected to additional conditions around fiduciary risk. The variable component consists of additional resources that are released in a graduated form based on the level of progress and performance in selected sectors (usually health, education and public financial management). If only 30% of targets are met, then only 30% of the variable tranche funds will be released. The EC treats the two components separately so that in the case of sector budget support, if the IMF macroeconomic programme goes off track, the variable component can still be disbursed in full providing the sector in question has met the conditionalities thus not penalising the sector for macroeconomic issues beyond its control and making sector support more volatile than it needs to be. DFID is seeking to follow a similar approach. Usually the fixed component is around two-thirds of the total budget support in EC programmes, with the variable component being around one-third. These splits can vary from one country to another dependent on the recent track-record of the IMF programme, how dependent the recipient country is on budget-support for financing basic services, the level of fiduciary risk in the recipient country, the size of the budget support programme, and the size and scale of earmarked aid programmes to specific sectors. If there is a large budget support programme to a highly aid-dependent country, then predictability is a key issue and the fixed component should include a significant proportion of the total funds available (Norad, 2005, p.12). Other donors are in the process of considering the EC approach in some of their programmes. The World Bank was considering using a graduated approach for the 5th PRSC in Uganda. The UK is considering a longer-term approach to providing budget support where the early years would be a fixed component and the later years a variable component which would be firmed up based on the previous year's performance. The Netherlands follows a similar approach to multi-annual budget support in Burkina Faso and Mozambique. Sweden is currently considering a graduated approach in Tanzania with Norway having already entered into dialogue with other donors at country level in Uganda, Zambia, Malawi and Tanzania. Switzerland has adopted a similar approach in Mozambique, and in Uganda several bilateral donors are discussing the possibility of linking variable tranches to political governance indicators to address the deterioration in governance in the country (Norad, 2005, p.8-9). What is important for a graduated response to be successful at improving predictability, is for there to be a good level of dialogue and transparency between donors and recipient countries when there is the possibility of reducing aid due to a deterioration in the political, economic or public financial management areas. A DFID-funded study proposes the establishment of an Aid Guarantee Facility that poor aid- dependent countries could draw on if donors do not disburse what they have committed. It would be limited to budget and programme support and would not guarantee 100% of donor commitments, but a certain agreed in advance minimum level (HLFH, 2005, p.vii). A similar idea is proposed by Bulir and Hamann (2005, p.17) who propose an aid reserves buffer stock managed and provided by the IMF. As donors commit to providing longer-term, more predictable aid, it will be important that they respect the agreement made at Monterrey to delink debt relief from ODA. A recent 41 study by Kovach and Wilks (2006) calculates that nearly one third of reported ODA (equivalent to 13.5 billion) in 2005 has been spent on debt cancellation, refugee spending in donor countries, and foreign scholarships for developing country students to study in donor countries. Around 11.8 billion of this was spent on debt relief, most of this going to Nigeria and Iraq (p.11). Specific Recommendations: ˇ Donors should ideally move towards medium to long term commitments of 5-10 years ˇ Donors should commit funds early enough in the year to coincide with the budget cycle and support countries operating a cash budgeting system ˇ Donors should make more accurate projections of future aid allocations 2. Public Reporting of Donor Performance and an International Aid Agreement Whilst donors are responsible for submitting commitment and disbursement data to the DAC for insertion into the DAC or CRS databases, there is no formal public reporting of donor performance with regards to aid predictability. If the DAC, the UN or the World Bank could also take on the role of publicly reporting both commitments and disbursements from donors, this would help put information in the public domain which lobbyists could use to "name and shame" the worst performing donors and put pressure on them to be more consistent with the aid they disburse. However, this assumes that the main blame for aid volatility is on the donor side which may not be the case. As a more holistic approach, Action Aid argues for an International Aid Agreement (IAA) through which donors, recipients and civil society organisations can all be held accountable. The four key elements of such an agreement would be: 1. Clear policies from developing countries on the criteria for accepting aid 2. Mutual commitments in place of one-sided conditionality, monitored transparently at the country level 3. National and international forums where donors and recipients can review progress on an equal footing, overseen by a UN Commissioner on Aid 4. New mechanisms to increase the volume and predictability of aid If the donor is providing lower aid than promised, or delaying disbursements for no good reason, the donor would be reported to the UN Commissioner on Aid with the result being that the donor in question would have a reduction in decision-making powers in the UN system or loss of voting rights in the IMF/World Bank. Whilst this is a good idea in theory, it may be very difficult to enforce actually altering decision-makings powers or voting rights within these international bodies. Specific Recommendation: ˇ The UN should improve its coordination role to ensure that donors report disbursement and commitment data fully and accurately 42 3. Applying Conditionality to Future Aid Commitments Currently, many donors apply conditionalities to on-going aid agreements, meaning there can easily be interruptions in aid flows due to recipient countries failing to achieve the stated conditions. If conditionalities were applied to future commitments rather than commitments already made, this would prevent within-year interruptions in budget support flows. An example of where this is being put into practice is Mozambique where commitments are made nine months in advanced by the G10 donor group and these commitments are then confirmed when the government budget is finalised providing the government reform programme remains on-track (Foster and Keith, 2004, p.99). This prevents within-year slumps in aid flows. Ideally, conditions should be agreed jointly between the donor and the recipient country and the recipient country should not be punished if external shocks take place which prevent the conditionalities being met. This will require transparency and mutual accountability between donors and recipient countries. Political conditionality has been left very vague in most aid agreements, making it difficult to make an objective decision if and when to suspend aid on political grounds. Given that aid given via budget support is directly supporting the government, this is the type of aid that is most likely to be suspended if there are concerns about human rights and democratisation issues. To make it more transparent, donors should either be clearer on what grounds they might consider suspending budget support aid for political reasons, or they should delink political conditionality from aid disbursements and leave such discussions to political dialogue rather than incorporating them as triggers for budget support (Norad, 2005, p.11). Specific Recommendation: ˇ Donors should change the practice of applying conditionality to present aid commitments often leading to within-year delays in disbursements, but instead apply conditionality to the following years' aid commitments which would enable the recipient government to plan ahead more effectively 4. Strengthening Absorption Capacity and Using Alternative Funding Channels Especially in Fragile States Most of the previous recommendations refer to donor related reasons for aid volatility. If on the other hand, the main reason for unpredictable aid is due to lower absorption capacity and weak public expenditure management systems, then a capacity building fund should be provided alongside budget or project support aid to assist recipient countries' in building greater government capacity in all aspects of financial management from planning to procurement to budget execution. There is a growing concern in fragile states about channelling aid through governments due to fiduciary risk concerns as well as human rights concerns, so an alternative is for NGOs, civil society groups and the private sector to receive funds either instead of providing funds through the government or in addition to channelling funds through government depending on the country government context. There has been a large increase globally in the number of NGOs since the early 1990s and many NGOs now play a significant role in funding humanitarian and development activities using donor funding. However, given that in the 43 long-term a stable government should be the focal point for managing aid flows, it is important that parallel systems are not created if funds are temporarily channelled through NGOs and other Non-State Actors, but rather state capacity is strengthened at the same time as Non-State Actors being a temporary channel for funds. "The private-sector component of NGO grants to all developing countries increased from $5 billion in 1990 to $10 billion in 2003 ­ about 15 percent of the value of total ODA." (World Bank, 2005, p.94). McGillivray (2006, p.15) recommends that where there is an absorption capacity constraint of increasing aid via government channels in fragile states, alternative channels should be explored including NGOs, civil society, and the private sector. There are also other alternative mechanisms which have been used in Afghanistan and East Timor where trust funds are created with countersigning or oversight by the World Bank or the UN, thus playing the very important role of building state capacity and systems that will be viable in the longer-term. These enable funds to be available to pay recurrent costs such as teachers' salaries. Specific Recommendations: ˇ Donors should provide a capacity building fund alongside budget support to ensure that state capacity is strengthened over the longer-term ˇ In fragile states, alternative funding channels such as through NGOs, Non-State Actors or Trust Funds may need to be considered to improve the predictability of aid, but ensuring that they do not create parallel systems but instead build state capacity 5. More Transparent Reporting on Aid Disbursements by All Donors The DAC databases containing commitment and disbursement data for all main donors is incomplete. The EC data on programme completion dates for budget support programmes is all missing making it impossible to comment on whether or not the EC is disbursing is commitments in full by the end of its programmes; the World Bank does not report on disbursements at all; and for all donors, there are individual data points that are missing or inconsistent between one database and another, meaning that some data needs to be rejected when trying to undertake an analysis comparing commitments with actual disbursements made. Specific Recommendations: ˇ All donors should more fully report on details of programme data, such as end dates, which the EC does not report at all ˇ The World Bank should report disbursements and all donors should report more complete disbursement data and more accurate aid commitment data, i.e. the original commitments that were agreed with countries rather than what is agreed at the board/governing body just before disbursement is made 44 9.2 Secondary Recommendations If donors cannot be held to account on improving aid predictability through the four recommendations outlined above, two other options for recipient countries to follow to mitigate the effects of unpredictable aid are outlined below. 1. Discounting of Aid Disbursements If budgets were based on a conservative estimate of likely aid disbursements and domestic revenue based on recent trends rather than ambitious statements from donors or recipient governments, then there would be greater consistency and coherence between actual disbursements and budget projections. The High Level Forum for Health recommends that the DAC or the World Bank could play this role of producing independent global, regional and country level forecasts of future aid disbursements based on past practice (HLFH, 2005, p.6). This practice already takes place in an informal manner in several countries. In Ethiopia, the Ministry of Finance discounts aid commitments on the basis of past track records, so the African Development Bank has its loan disbursements discounted by 80%, and EC aid is discounted by 75% at the beginning of the financial year. The Ministry of Finance in Uganda does a similar process discounting aid by up to 50% using donor specific coefficients (Action Aid, 2005, p.27). The Ministry of Finance in Rwanda follows a similar approach. Specific Recommendation: ˇ Where donors disburse less than what they commit, recipient country government planners should be able to discount the projected aid disbursements for the following year by an appropriate amount to enable more accurate budget planning 2. Using Macroeconomic Policy as a Buffer Stock Tool If aid is likely to continue to remain volatile, one way to mitigate the negative effects of this might be for the recipient country to use foreign exchange reserves as a buffer (Foster and Keith, 2004, p.96, Eifert and Gelb, 2005, p.10). The main challenge with this approach is for each country to distinguish between a temporary shortfall in aid receipts and more fundamental errors in forecasting and the accompanying implications this may have for running down reserves which may be needed for other purposes (Foster and Keith, 2004, p.100). It can also mean that non-priority expenditures become even more volatile if priority ones are protected. As well as building up reserves, Bulir and Lane (2002, p.25) also highlight the need to ensure that a country's fiscal framework is flexible enough so that taxation and expenditure can be adjusted easily when necessary. Given these potential downsides, this is not a strongly recommended option. Specific Recommendation: ˇ The recipient country can use foreign exchange reserves as a buffer when aid disbursement is lower than expected 45 9.3 Aid Effectiveness Whilst aid predictability is clearly a problem and limits aid effectiveness, other issues contributing to aid ineffectiveness are equally important to address if the MDGs are to be reached in developing countries. "Aid effectiveness is not solely a technical matter, therefore it can not be reduced to issues of procedures or "harmonisation" of donors. It must be admitted that dissipation of aid, its unforeseeable nature, the multiplicity of procedures, protagonists and macroeconomic conditionalities are elements hampering aid effectiveness. But this is not the decisive point. These problems are only the consequence of the way aid is conceived: most donors subjugate it to their own interests and view of development." (Coordination SUD, 2005, p.2) Donors give aid to developing countries predominantly according to historical, colonial or other strategic reasons rather than according to need alone. This means that some countries have a plethora of donors in a given sector or overall, whilst others remain relatively underfunded. This is very true of aid to fragile states, where we noted above that there are two types ­ donor darlings and donor orphans, but it is also just as true of non-fragile states. In order to make aid more effective and rationalised, donors could make use of silent partnership agreements and be more committed to limit their role in countries that are already receiving large amounts of aid, and instead invest in those countries which are relatively underfunded. Aid could be more focussed on the poorest countries rather than due to donor political and strategic interests. In addition, Action Aid (2005) argue that aid flows include a significant proportion of what they define as "phantom aid", which is aid that falls into one or more of the following categories: ˇ Aid that is not targeted for poverty reduction ˇ Aid double counted as debt relief ˇ Aid given in the form of Technical Assistance (one quarter of total American aid was spent on Technical Assistance in 2003 (p.22). ˇ Aid tied to goods and services from the donor country ˇ Poorly coordinated aid with high transactions costs ˇ Unpredictable aid ˇ Aid spent on immigration-related costs in the donor country ˇ Aid spent on excess administration costs They estimate that 89% of French aid is phantom aid, and 86% of American aid is phantom aid (p.17). In addition they state that only 40% of aid goes to LICs, and only one-third of aid goes to Sub-Saharan Africa, even though the needs are greatest in these two groupings of countries. By contrast, 75% of all EC aid goes to MICs. If aid flows could be made both more predictable and more effective, then they would be much more likely to produce real improvements in the livelihoods of the poor in developing countries, contributing to the achievement of tangible and positive results in providing basic services. 46 9.4 Conclusion This research, both from the literature review and analysis of DAC data on commitments and disbursements, has shown that aid remains unpredictable and volatile with disbursement rates varying immensely from donor to donor and country to country. This is the case for both project aid and programme aid, loans and grants, though as this research has shown, the consequences of unpredictable programme aid are much more severe and have a greater impact on many recipient countries than unpredictable project aid. This is supported by research evidence from several developing countries. The research has found that aid is twice as volatile in fragile states as in other LICs, yet fragile states include some of the countries furthest away from achieving the MDGs. This is a critical group of countries where aid predictability is of utmost importance. With respect to individual donors, the Scandinavians are the leaders in terms of volumes of aid as a percentage of GNI, but have plenty of room for improvement in terms of timely disbursements of aid, with an average percentage of programmes not being disbursed on time over the period 2002-04 of 39.4% for Norway and 47.4% for Sweden. Norway is the only bilateral donor showing improvements in both the percentage of programmes that are disbursing funds in full each year, and in the total percentage of volume of committed aid actually being disbursed on time. By contrast, the Netherlands has seen improvements in the percentage of programmes that are disbursing funds in full each year, but has seen some variation in the total percentage of volume of committed aid actually being disbursed on time, though this has consistently been in excess of 85%, although a quarter of their programmes over the period did not disburse in full. Sweden continues to experience some fluctuations in both the percentage of programmes that are disbursing funds in full each year with on average, only 50% of all programmes disbursing all their funds on time, and in the total percentage of volume of committed aid actually being disbursed on time, with an average over 2002-04 of 25% not being disbursed on time. The UK exhibits the greatest variability in its performance across the three years. It is disbursing a far greater quantity of aid as budget support compared to the other donors, meaning that when it does disburse funds late, the negative monetary impact this has on recipient countries is significant. In 2003, only 57% of the total amount of the programme commitment was actually disbursed, with the undisbursed component totalling more than the combined budget support programme commitments due to end in 2003 of The Netherlands, Norway and Sweden. The UK does show significant improvement in 2004, disbursing over 98% of funds on time, although this only corresponded to 8 out of 11 programmes fully disbursing on time. Without data for the most recent years, it is not yet clear if this improvement is a trend or a one off, and even if it is a trend, it is important to point out that even a small percentage of the total budget support commitment remaining undisbursed by the end of a programme can correspond to a significant volume of aid given the sheer volume of the UK's budget support programmes, thus having a negative impact on the recipient country's recurrent budget. On average over the period, 17.8% of the volume of UK funds was not disbursed, and 55.3% of the number of programmes was not disbursed in full. Thus there is significant room for improvement in performance for the UK if it is going to continue to provide similar levels of aid as present in the form of budget support. This research uncovered the fact that there is no data available in the OECD DAC database on World Bank disbursements, and this information could not be found on the World Bank website either. This shows that the World Bank is the least publicly transparent donor in 47 relation to aid volatility. For the EC, data in the OECD DAC database was incomplete making it impossible to undertake specific analysis based on DAC data, although evidence from other studies clearly shows the EC's poor track record in providing aid in a timely manner. Hence, improvements in the way in which data is submitted and reported within the DAC database are certainly needed by all donors are to hold donors to account on timely disbursements of aid. In relation to the global funds, the GFATM health fund seems to have a good track record for disbursement with the EFA-FTI Catalytic Fund falling further behind and only disbursing around 55% of committed funds during 2005. Information was impossible to find on the US- initiated MCA and the PEPFAR, both of which seem to have been slow to disburse funds. By contrast, the Bill and Melinda Gates Foundation has disbursed approaching US$6 billion for global health initiatives in developing countries since its inception, though this money is not included in ODA, as it is a private foundation distributing largely through NGOs and research institutes. The authors primary recommendations are for donors to provide longer-term more predictable aid; for the establishment of a formal public reporting system on donor performance regarding aid disbursements; for conditionalities to be transparent and applied to future commitments in a sustained or graduated manner; and for donors where appropriate, to provide technical assistance and funding through a variety of channels if this is likely to increase the effectiveness of aid and countries' absorption capacities, a particularly important issue in fragile states. To make better use of existing aid and to mitigate the damaging effects of unpredictable aid whilst donors adapt their practices, recipient countries have two main options: to discount aid disbursements based on past performance, and to use foreign reserves to act as a buffer stock when aid flows remain unpredictable. These are very much second best solutions, and are not able to make aid any more predictable. Thus the onus is very much upon donors to take on board the initial set of recommendations with the warning that if they do not, this is likely to severely impact the ability of developing countries to reach the MDGs by 2015. 48 References Action Aid (2005). Real Aid: An Agenda for Making Aid Work. Action Aid, London. AFP (2005). Global Fund Against AIDS, TB and Malaria Resumes Aid to Uganda. Public Policy International News Update, 11th October 2005. http://www.hivdent.org/publicp/inter/ppinGFAA1105.htm Arellano, C., Bulir, A., Lane, T., and Lipschitz, L. (2005). The Dynamic Implications of Foreign Aid and Its Variability. IMF Working Paper WP/05/119. IMF, Washington DC. Booth, D. (2004). Budgets not Projects: A New Way of Doing Business for Aid Donors. ODI Opinions 9, February 2004. Overseas Development Institute, London. Bulir, A. and Hamann, A. J. (2001). How Volatile and Unpredictable are Aid Flows, and What are the Policy Implications? IMF Working Paper 01/167. IMF, Washington DC. Bulir, A. and Hamann, A. J. (2003). Aid Volatility: An Empirical Assessment. IMF Staff Papers Vol. 40 No. 1. IMF, Washington DC. Bulir, A. and Hamann, A. J. (2005). Volatility of Development Aid: From the Frying Pan into the Fire? Paper prepared for the Seminar on Foreign Aid and Macroeconomic Management, organised by the IMF Institute and African Department, March 14-15, Maputo. Bulir, A. and Lane, T. (2002). Aid and Fiscal Management. IMF Working Paper WP/02/112. IMF, Washington DC. Celasum, O. and Walliser, J. (2005). Predictability of Budget Aid: Experiences in Eight African Countries. Paper prepared for World Bank Practitioners' Forum on budget support, 5-6 May 2005, Cape Town. Chauvet, L. and Collier, P. (2004). Development Effectiveness in Fragile States: Spillovers and Turnarounds. CSAE, Oxford. Coordination SUD (2005). Reforming Aid to Enhance its Effectiveness. Position paper on the 2nd High-Level Forum on Aid Effectiveness, Paris 28 February ­ 2 March 2005. Accessed on 4 March 2006 at: http://www.realityofaid.org/themeshow.php?id=12 DFID (2004). Poverty Reduction Budget Support: A DFID Policy Paper. DFID, London. DFID (2005b). UK Cuts Direct Budget Aid to Uganda by 15 million, Withholds Further 5 million. Press Release, 20 December 2005. DFID, London. Available from: http://www.dfid.gov.uk/news/files/pressreleases/uganda-reduction.asp [Accessed 24th February 2006]. DFID (2006). Hilary Benn Visits Ethiopia. News Item, 18th January 2006. DFID, London. Available from: http://www.dfid.gov.uk/news/files/benn-visit-ethiopia06.asp [Accessed 24th February 2006]. 49 Eifert, B. and Gelb (2005). Improving the Dynamics of Aid: Towards More Predictable Budget Support. World Bank Policy Research Working Paper 3732. World Bank, Washington DC. European Commission (2005). EC Budget Support: An Innovative Approach to Conditionality. EC, Brussels. Foster, M. (2004). Linking Budget Support to Progress Towards Education MDGs and EFA Goals. DFID, London. Foster, M. with Keith, A. (2004). The Case for Increased Aid Volume I: Main Report. DFID, London. FTI Secretariat (2006a). Catalytic Fund Status Report Update, March 2006. FTI Secretariat, Washington DC. FTI Secretariat (2006b). Education For All Fast Track Initiative Analysis of Official Development Assistance: Draft. FTI Secretariat, Washington DC. FTI Secretariat (2006c). Fast Track Initiative Technical Meeting Overview, March 13-15, 2006, Moscow. FTI Secretariat, Washington DC. GCE (2002). Broken Promises? Why Donors Must Deliver on the EFA Action Plan? Global Campaign for Education Briefing Paper. GCE, Brussels. GFATM (2006). The Global Fund to Fight AIDS, Tuberculosis and Malaria: Monthly Progress Update, 22 February 2006. GFATM, Geneva. Grimm, S. (2004). European Development Cooperation to 2010: Aid Disbursement and Effectiveness. EDC Briefing July 2004. ODI, London. Harford, T., Klein, M. and Tilma, K. (2004). Donor Performance: What Do We Know and What Should We Know? Public Policy for the Private Sector Note No. 278. World Bank, Washington DC. High Level Forum (2005). Paris Declaration on Aid Effectiveness: Ownership, Harmonisation, Alignment, Results and Mutual Accountability. High Level Forum, February 28 ­ March 2, 2005, Paris. HLFH (2005). Fiscal Space and Sustainability: Towards a Solution for the Health Sector. High Level Forum on the Health MDGs, Paris, 14-15 November 2005. HM Treasury (2006). Government Announces $15 billion to Deliver Education for All. Press Release, 10 April 2006. HM Treasury, London. Ireland Aid (2004). Evaluation of the Uganda Country Strategy 2000-2003: Executive Summary. Ireland Aid, Dublin. 50 Kanyarukiga, S., van der Meer, E., Paalman, M., Poate, D., and Schrader, T. (2006). Evaluation of DFID Country Programme: Country Study Rwanda 2000-2005. DFID Evaluation Report EV660. DFID, London. Kovach, H. and Wilks, H. (2006). European Aid: Genuine Leadership or Misleading Figures? An Independent Analysis of Comparative Aid Figures. Eurodad, Brussels. Lawson, A., Booth, D., Harding, A., Hoole, D., and Naschold, F. (2003). General Budget Support Evaluability Study Phase 1 Synthesis Report Vol I. DFID Evaluation Report EV643. DFID, London. Lawson, A., Booth, D., Harding, A., Hoole, D., and Naschold, F. (2003b). General Budget Support Evaluability Study Country Case Studies: Mozambique & Andhra Pradesh, Vol II. DFID Evaluation Report EV643. DFID, London. Levin, V. and Dollar, D. (2005). The Forgotten States: Aid Volumes and Volatility in Difficult Partnership Countries (1992-2002). Paper for DAC Learning and Advisory Process on Difficult Partnerships. Lister, S. (2005a). Joint Evaluation of General Budget Support 1994-2004 Inception Report: Burkina Faso, Malawi, Mozambique, Nicaragua, Rwanda, Uganda, Vietnam. IDD, Birmingham. Lister, S. (2005b). Joint Evaluation of General Budget Support to Tanzania, 1995-2004. Revised Final Report. IDD, Birmingham. McGillavray, M. (2006). Aid Allocation and Fragile States. UNU-WIDER Discussion Paper NO. 2006/01. UNU-WIDER, Helsinki. Michaud, C. (2003). Development Assistance for Health (DAH): Recent Trends and Resource Allocation. Paper prepared for the WHO Second Consultation Commission on Macroeconomics and Health. WHO, Geneva. Millennium Challenge Corporation (2006). Budget Justification 2007. MCC, Washington DC. Minbuza (2004). Mutual Interests, Mutual Responsibilities ­ Dutch Development Cooperation En Route to 2015. Minbuza, The Hague. Mosley, P. and Abrar, S. (2005). Budget Support, Conditionality and Poverty. Sheffield Economic Research Paper Series, SERP Number: 2005012. Sheffield University, Sheffield. Narea, C. A. and Christensen, M. (2004). Development of Swedish General Budget Support 1990-2003. UTV Working Paper 2004:3. Department for Evaluation and Internal Audit, SIDA, Sweden. Naschold, F. (2002). Aid and the Millennium Development Goals. ODI Opinions. ODI, London. Norad (2005). Direct Budget Support, Disbursements Mechanisms and Predictability. Norad, Oslo. 51 Odedokun, M. (2003). Analysis of Deviations and Delays in Aid Disbursement. Journal of Economic Development Vol. 28 No. 1, June 2003. OECD (2004). Financing ICTs for Development ­ Efforts of DAC Members: Review of Recent Trends of ODA and its Contributions. Report to the UN Task Force on Financial Mechanisms for ICT for Development (TFFM). OECD, Paris. OECD (2006). 2005 Development Cooperation Report. Volume 1, No.7. OECD, Paris. http://oberon.sourceoecd.org/pdf/dac/statann.pdf [Accessed 11th August 2006]. Oxfam (2005). Paying the Price ­ Why Rich Countries Must Invest Now in a War on Poverty. Oxfam, Oxford. Quartey, P. (2005). Innovative Ways of Making Aid Effective in Ghana: Tied Aid Versus Direct Budgetary Support. UNU-WIDER Research Paper NO. 2005/38. UNU-WIDER, Helsinki. SPA Secretariat (2005). Towards Sector Support Alignment. SPA 2004 Sector Program Tracking Report. UN (2001). Technical Report of the High Level Panel on Financing for Development. University of Birmingham IDD (2005). Joint Evaluation of General Budget Support 1994- 2004: Inception Report. DFID, London. World Bank/IMF (2005). Review of the PRS Approach: Balancing Accountabilities and Scaling Up Results. World Bank, Washington DC. World Bank (2005). Global Development Finance: Mobilizing Finance and Managing Vulnerability. World Bank, Washington DC. World Bank (2006). The Chad-Cameroon Petroleum and Pipeline Development Project: Questions and Answers. World Bank, Washington DC. Available from: http://web.worldbank.org/WBSITE/EXTERNAL/COUNTRIES/AFRICAEXT/EXTREGINI/ EXTCHADCAMPIPELINE/0,,contentMDK:20531903~menuPK:1104029~pagePK:6416844 5~piPK:64168309~theSitePK:843238,00.html [Accessed 23rd February 2006]. 52 Appendix: Data Issues and Sources The data contained in this research is sourced from the International Development Statistics (IDS) online databases on aid and other resource flows, which can be accessed at www.oecd.org/dac/stats/idsonline. All data from the DAC database is based on the calendar year, and unless explicitly stated all financial figures are adjusted for inflation and expressed in US dollars at 2003 exchange rate levels. Definitions: ˇ Official Development Assistance (ODA) refers to all activities undertaken by official agencies include state and local government actors, or by their executing agencies at favourable financial terms, with the aim of promoting economic development and welfare of the recipient country. ˇ A commitment is a written obligation by a government or official agency to provide resources of a specified amount for a specific purpose for the benefit of the recipient country. ˇ Net commitments per year comprise new undertakings entered into in the year in question (regardless of when disbursements are expected) and additions to agreements made in earlier years. Cancellations and reductions of earlier years' agreements are not taken into account. ˇ A disbursement is the placement of resources at the disposal of a recipient country or agency, or in the case of internal development-related expenditures, the outlay of funds by the official sector. ˇ Bilateral aid includes activities undertaken directly with an aid recipient, or those carried out by a national or international non-governmental organisation on behalf of a DAC Member country (since it is the donor country that effectively controls the use of the funds). Bilateral aid also includes development related spending in the donor country such as technical assistance. ˇ Multilateral activities refers to aid activities financed from the multilateral institutions' regular budgets, such as that of the World Bank, the regional development banks and some UN agencies. ˇ Grants are transfers in cash or in kind for which the recipient incurs no legal debt. For DAC/CRS reporting purposes, it also includes debt forgiveness, which does not entail new transfers; support to non-governmental organisations; and certain costs incurred in the implementation of aid programmes. ˇ Loans are transfers for which the recipient incurs a legal debt and repayment is required in convertible currencies or in kind. This includes any loans repayable in the borrower's currency where the lender intends to repatriate the repayments or to use them in the borrowing country for the lender's benefit. For graph 10, in order to convert the value of disbursements made late or not at all, the total figure calculated in the donor country's currency was converted to deflated US$ (2004 prices). The average 2004 annual exchange rate was calculated by comparing the available disbursement data in US$ and the donor currency for each programme where there were disbursements in 2004, then averaging the results. The exchange rates calculated were as follows: 53 US$:UK 1.839 US$:Nkr 0.149 US$:Skr 0.136 US$: 1.254 For the Netherlands the Euro exchange rate was calculated from EC data as the Dutch data was available until 2002 in Euros, but from 2003 in US$. In order to calculate the values of disbursements made before the completion date there was data available in US$ which was converted into deflated 2004 prices using the DAC deflators. 54