Ethiopia Public Expenditure Review (In Two Volumes) Volume 1: Main Report

Transcription

1 Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Report No ET Ethiopia Public Expenditure Review (In Two Volumes) Volume 1: Main Report November 30, 1999 World Bank Country Office in Ethiopia Country Department 6 Africa Region Document of the World Bank

2 GOVERNMENT FISCAL YEAR July 8 - July 7 CURRENCY EQUIVALENTS Currency Unit: Ethiopian Birr (Br) Official Rate: US$ 1.00-Br 8.00 (September 7, 1999) ACRONYMS AND ABBREVIATIONS AfDB ARM CAD CSC CSR DIFD DSA EC EFY EMC ERA ESDP FAGO FDRE FIS GDP GoE HIID HIPC HSDP IMF IPF MEDaC MEFF MoE MoF MoH MTEF PAP PEP PER PIM PIP PMO RMI SDP SIDA SNNPS TA UNDP USAID African Development Bank Annual Review Meeting Central Accounts Department Civil Service College Civil Service Reform Department for Intemational Development Decentralization Support Activity European Commission Ethiopian Fiscal Year Expenditure Management and Control Ethiopian Road Authority Education Sector Development Program Federal Auditor General's Office Federal Democratic Republic of Ethiopia Financial Information System Gross Domestic Product Govemment of Ethiopia Harvard Institute for Intemational Development Healthy Indebted Poor Countries Health Sector Development Program Intemational Monetary Fund Indicative Planning Figures Ministry of Economic Development and Cooperation Macroeconomic and Fiscal Framework Ministry of Education Ministry of Finance Ministry of Health Medium-Term Expenditure Framework Program Action Plan Public Expenditure Program Public Expenditure Review Program Implementation Manual Public Investment Program Prime Minister's Office Regional Management Institutes Sector Development Program Swedish Intemational Development Association Southem Nations, Nationalities and People's State Technical Assistance United Nations Development Program United States Agency for Intemational Development Vice President Callisto E. Madavo Country Director Oey Astra Meesook Sector Manager Peter Miovic Task Team Leader Hart Schafer

3 PREFACE The 1999 Public Expenditure Review (PER) for Ethiopia was prepared by a team from the government and a donor team. The government team was led by H.E. Ato Hailemelekot T/Giorgis, Vice Minister of Finance, and included Ato Asrat K/Work, Ato Amare Gebrewold, Ato Tesfaye Kassa, Ato Mesfin Asrat, Ato Anemwa Ebay, Ato Fantahun Belew (MoF), Ato Melaku Kifle, Dr. Meshesha Getahun (MEDaC), Ato Tesfamichael Nahusenay (ERA), Dr. Demeke Yigletu, Ato Kassahun Habte-Mariam (MoT), Ato Gebre-Medhin Belay (MoA), Ato Girma Asfaw, Ato Teferi Hagos (MoE), Ato Ketemaw Yimam (MoTC), Dr. Gebremaskal Habte-Mariam, and Ato Abebe Gesit (MoH). The donor team was led by Hart Schafer (World Bank) and included David Cowen (EMF), Kurt Cornelis (EC), Roland Martin (Netherlands), Stephen Lister (DFID Consultant), Stephen Olanrewaju (AfDB), Filiberto Sebregondi (EC), and Doug Addison, Nicholas Bennett, Eyerusalem Fasika, and Fred Kilby (World Bank). During the field work, the team liaised closely with Nigel Roberts (World Bank Country Office Managers/Ethiopia). The main field work for the 1999 Ethiopia PER was carried out jointly from May 3 through May 18, The PER teams visited regional authorities in Addis Ababa, Amhara, Oromiya, Harari, and Dire Dawa. This report was prepared under the general guidance of Oey Astra Meesook (Country Director) and Nigel Roberts (World Bank Resident Representative in Ethiopia). Peter Miovic was the Sector Manager (AFTM2), Vinaya Swaroop (DECRG) the Lead Advisor, and Gene Tidrick (AFTMI) the Peer Reviewer. Special thanks goes to Tanisha McGill for editing and desk-top publishing support of the final report. The draft 1999 Ethiopia PER was discussed with the government in workshops held in Debre Zeit, Ethiopia (September 13-15, 1999) and Brussels, Belgium (October 27-28, 1999).

4

5 Contents PREFACE I EXECUTIVE SUMMARY... I CHAPTER 1: INTRODUCTION... 1 SETTING... 1 FOCUS OF THE 1999 PER... 1 STATUS OF PREVIOUS PER RECOMMENDATIONS DATA ISSUES-AN IMPORTANT CAVEAT... 5 CHAPTER 2: DEVELOPMENTS IN BUDGETARY AGGREGATES... 7 FISCAL TRENDS... 7 RECENT FISCAL PERFORMANCE: BUDGET TARGETS VS. OUTTTURNS... 8 ECONOMIC AND FUNCTIONAL EXPENDITURE CLASSIFICATION REAL AND PER CAPITA EXPENDITURE CONSTRAINTS To MOBILIZING EXTERNAL RESOURCES FOR SPS IMPACT ON SOCIAL SECTOR EXPENDITURE CHAPTER 3: FISCAL IMPLICATIONS OF THE WAR FUNDING OF INCREASES IN DEFENSE BUDGET SECOND ROUND EFFECTS OF THE WAR CHAPTER 4: FISCAL SUSTAINABILITY OF SECTOR PROGRAMS BROAD MACROECONOMIC OBJECTIVES MACROECONOMIC SCENARIOS IMPLICATIONS FOR IMPLEMENTATION OF SECTOR PROGRAMS CONCLUSIONS CHAPTER 5: CAPACITY BUILDING IN PUBLIC FINANCE MANAGEMENT INTRODUCTION ONGOING REFORMS IN PUBLIC FINANCE MANAGEMENT FISCAL DECENTRALIZATION REFORM OF EXPENDITURE MANAGEMENT SECTOR DEVELOPMENT PROGRAMS RECOMMENDATIONS ON CROSS-CUTTING ISSUES CHAPTER 6: PER PROCESS ANNEXES AND ANNEX TABLES ANNEX 1: DESCRIPTION OF THE CGE MODEL ANNEX 2: OUTCOME AND RECOMMENDATIONS OF DEBREZEIT WORKSHOP ANNEX 3: GOVERNMENT OF ETHIOPIA COMMENTS ON PER ANNEX 4: OUTCOME AND RECOMMENDATIONS OF BRUSSELS WORKSHOP... 81

6 Tables, Figures, and Boxes Table 1: Fiscal Trends... 8 Table 2: Budget Execution... 9 Table 3: Economic Classification of Public Expenditures Table 4: Functional Classification of General Government Expenditure Table 5: Real Expenditure Table 6: Real Per Capita Expenditure Table 7: Sectoral Expenditure-Actuals/Projections vs. Targets Table 8: Road Transportation Cost Addis Ababa to Assab and Djibouti Table 9: Macroeconomic Scenarios-Outcomes Table 10: SP Expenditures as Share in Total Spending Table 11: Base Case Scenario - Per Capita Spending in Education, Health, Roads Figure 1: Capital Budget Implementation Rate in Priority Sectors Figure 2: Trends in Defense Expenditure Figure 3: Per Capita Spending in Social Sectors for Selected SSA Countries Figure 4: Disbursement of Capital Expenditures by Source Figure 5: Modified Implementation Path for Sector Programs Figure 6: The Civil Service Reform Program Figure 7: Alternative Disbursement Channels for SPs Box 1: Economic Cost of Chronic Underspending on the Capital Budget Box 2: The Limited Effectiveness of Supplementary Budgets on Spending Levels Box 3: EC Structural Adjustment Support and Stabex Box 4: Important Deviations in FY99/00 Budget from Base Case Box 5: Budget Management Issues in Previous Studies Box 6: Fiscal Devolution Below the Regional Level Annex Tables Annex Table 1: Progress on Previous PER Recommendations Annex Table 2: Fiscal Trends FY98/87-FY98/ Annex Table 3: Functional Classification of General Government Expenditures Annex Table 4: Functional Classification of Recurrent Expenditures Annex Table 5: Functional Classification of Capital Expenditures Annex Table 6: Economic Classification of Public Expenditure Annex Table 7: Developments in Defense Expenditure Annex Table 8: Real Per Capita Expenditure Annex Table 9: Financial Calendar... 94

7 You cannot build a house for lastyear's summer (Ethiopian Proverb) Ethiopia 1999 Public Expenditure Review An Exercise by the Government ofethiopia anddonors EXECUTIVE SUMMARY Focus of the 1999 PER 1. Since May 1998, the war between Ethiopia and Eritrea has been the center of attention of the Ethiopian government and its donors. In addition to the human toll of the war, there have been concerns that the sharp rise in defense expenditure could jeopardize the substantive macroeconomic and structural progress made in recent years. Consequently, one major area of attention of the present report is to review the macroeconomic impact of the war and the financing of the associated defense expenditures, and to determine whether defense spending pressures have diverted resources from priority development programs. At the time of writing this report, efforts to reach a peaceful resolution of the war are ongoing. Although the available fiscal data for the past two years are still incomplete or preliminary, enough is available to allow some important conclusions to be drawn'. The 1999 Public Expenditure Review (PER) In the PER discussion workshops in Debre Zeit and Brussels, the Government made available revised projections for FY98/99. In comparison to the PER data base, the latest fiscal estimates indicate that defense expenditures may reach a share in GDP close to 7%, and 1.2 percentage points higher than the

8 ii Executive Summary concentrates on: (a) recent fiscal developments with a specific focus on the fiscal impact of the war; (b) fiscal sustainability of the expenditure program in general and the sector programs in education, health, and roads in particular; (c) public finance management; and (d) the PER process itself Recent Fiscal Developments 2. Ethiopia has followed a policy of not permitting the war to interfere with the consolidation of macroeconomic stability and continuation of structural reforms. Indeed, except for the floor on net foreign assets of the banking system, all quantitative targets under the 1998/99 program with the IMF were met. Despite the shortfalls in official loans and grants official reserves at the end of FY98/99 were slightly higher than one year ago (in terms of months of import coverage). The government has appropriately protected core spending in priority sectors, while financing the increased defense budget through extraordinary revenue efforts, including one-time financing measures such as utilization of privatization receipts. 3. But, the fiscal stress posed by the war is reflected in the projected fiscal outcome for FY98/99, which is a clear outlier to the trend and deviates substantially from the government's intentions and objectives as reflected in the original and revised budgets for the fiscal year. Unless timely corrective actions are taken, performance over the past two years could put the long-term positive trend in jeopardy. While capital spending will remain below budgeted figures, total recurrent expenditure is likely to exceed budget targets because of the excess in defense outlays. Although real per capita spending increases are still projected for education, health, and roads, the chronic underspending in these sectors (compared to budget targets) appears to be more pronounced in FY98/99 than in previous years. This reflects capacity constraints and teething problems with the sector programs in education, health, and roads, as well as lower-than-expected aid disbursements (overestimation of the availability of funds), rather than deliberate budget cuts. So, while pre-war gains have been preserved, the past two years expose missed opportunities to accelerate progress toward a more poverty-sensitive expenditure composition. Reasons for Underspending on the Capital Budget 4. The puzzling chronic under-execution on the capital budget is a recurring theme of this PER because of concerns with regard to the ambitious sector programs (SPs) in education, health, and roads. The issue is reviewed from various angles, including whether the binding constraints to disbursement are external resource availability, administrative bottlenecks, limited implementation capacity, or fiscal constraints. 5. Observed delays in the disbursement of external resources for SPs point to startup problems during FY97/98 and FY98/99 in concluding external assistance agreements. There were also significant delays in achieving effectiveness of aid agreements because PER estimate. Despite the increase in defense spending (compared to earlier projections) it appears that core outlays for education, health, and roads are in line with the PER projections. Therefore, one of the basic PER conclusions--that the government has so far protected per capita and real spending levels in priority areas-remains valid.

9 Executive Summary iii the specific financial reporting system-which is required by donors and has been agreed to by the government-was not operational (such as, IDA support to the education and health sectors). In addition, there are technical bottlenecks within the government to move funds to the beneficiary level, for example, the insufficient capacity to handle a substantive increase in civil works contracts and tenders for procurement of goods. 6. Delays in the release of external resources due to political considerations are certainly another reason for the slowdown in disbursement of external financing in FY98/99. In a number of cases, donors have adopted a wait-and-see approach in finalizing their support to sector programs because of concerns about the fungibility of resources and the possibility to indirectly fund the defense budget, combined with hesitation in donor capitals to support countries in conflict. A quick survey of donors showed that the most serious impact has been on counterpart funds from balance of payment support and on new aid commitments, while disbursements under existing commitments have only been affected marginally. However, future commitments could be seriously affected if peace remains elusive in FY99/00. Expenditures Related to the War 7. The long-term decline in Ethiopia's defense expenditure was reversed in FY97/98 with the outbreak of the war with Eritrea. Defense expenditure as a share of GDP is now approximately triple its historically low level of the mid-1990s, but still significantly below the war-time levels in the 1980s. The PER mission estimated the costs of secondround effects of the war between Birr 1,100 million and Birr 1,600 million annually, comprising immediate assistance to the displaced, replacement of assets, and higher transport costs due to the rerouting of freight from Eritrean ports through Djibouti. This is equivalent to 2.5 percent and 3.3 percent of GDP per annum in FY97/98 and FY98/99, respectively. Together with the additional defense spending over and above levels of the mid-1990s (about 3.8 percent of GDP in FY98/99), the fiscal burden of the war would be equivalent to between 6 percent and 7 percent of GDP per annum in the two years considered. In addition, because of donors' concerns about the war, external loans and grants of some Birr 1,500 million have been delayed in FY98/99. 2 Fiscal Sustainability of Sector Programs 8. Ethiopia faces the challenge of having to balance the need for tremendous increases in per capita spending in poverty-sensitive areas with the need to maintain fiscal stability. In the past, fiscal constraints have not been binding, because implementation capacity and external resource flows were the primary factors that determined expenditure levels in priority areas. If a peace agreement is reached, the binding constraints are likely to change. External resource flows will probably not remain a binding constraint if all aid pledges are turned into formal agreements. With time, implementation capacity should also improve and help overcome the teething problems of the first few years of SP implementation. If so, the overall fiscal sustainability will become the major determinant for the balance between recurrent and capital expenditure and between spending on SPs and in non-sp areas. 2 After the completion of the PER draft, the Govemnment shared revised projections for FY98/99 with the PER team. The revised data are reflected in Annex 2.

10 iv Executive Summary 9. A base case scenario-premised on Ethiopia's ability to reach lasting peace, rebuild investor confidence, and reduce the vulnerability to external shocks-assumes substantial GDP growth of between 5 percent and 6 percent per annum over the next five years. This outcome will require a gradual reduction in defense expenditures and a timely and realistic acceleration of development expenditures in infrastructure, education, health, and social services. Macroeconomic stability will also be underpinned by achieving overall fiscal objectives, including single digit budget deficits (including grants) supported by increased foreign exchange inflows and balance of payment support. Conversely, if the war continues, the government's long-term economic growth objectives and poverty reduction strategy will be jeopardized. The main risks result from the cumulative negative impact that the war would continue to have on government finances, public and private investment, and the external current account. These risks add to the challenges that Ethiopia already faces, including weakening of the terms-of-trade and recurrence of drought in The base case provides a fiscal envelope that has to accommodate funding requirements for SPs and non-sp activities. The appraisal documents for the three SPs lay out ambitious annual expenditure targets, which may be feasible individually, but are not consistent with a sustainable macroeconomic framework if taken together. Even with sufficient external funding, the government's entire resources would not be sufficient to meet the domestic financing share under the SPs. By FY01/02, the SPs would completely exhaust the notional capital expenditure envelope and there would be no room left for capital expenditure outside the three SPs. This is clearly not a feasible option. 11. Alternatively, the base case could provide room for healthy growth in non-sp spending and still allow for substantial increases in real per capita spending in education, health, and roads, if the implementation timeframes for the three SPs were modified. Originally, all three SPs were heavily front-loaded with the lion's share of disbursements expected in the second and third year of a five-year implementation period. These targets seem no longer attainable in view of the encountered obstacles to a timely mobilization of external resources. In addition, these assumptions are not consistent with overall fiscal constraints. It is therefore appropriate to envisage a more pronounced back-loading of capital expenditures, with gradually increasing disbursements for the education and health SPs-thereby maintaining the five-year implementation period-and the extension of the implementation period for the roads SP from five to seven years'. Capacity Building in Public Finance Management 12. To some degree, the underspending on the capital budget reflects limited implementation capacity which is closely linked to the quality of budget planning and the financial relationship between the center and the regions. Public finance management in Ethiopia is complex because three important reform exercises are proceeding 3Since the fiscal constellation in September 1999 was off the base case and worse than Scenario 3, which had assumed higher defense spending and a continued slow-down in aid disbursements, a 4h Scenario was computed to gauge the impact of higher defense spending with a resulting higher fiscal deficit (financed by domestic credit expansion). The results for that scenario are presented in the Annex 2, as part of the outcome of the Debre Zeit PER workshop.

11 Executive Summary v simultaneously: fiscal decentralization, civil service reform, and comprehensive sectoral budgeting and sector development programs. The PER reviews each area of reform with regard to its implications for effective expenditure execution. Decentralized Budgeting 13. In general, budgeting procedures in the regions broadly mirror the federal system and echo federal financial regulations. Besides exhibiting the considerable strength of a disciplined, well-documented, systematic process, there are also inherent weaknesses in regional budgeting. These weaknesses are reflected by the pervasive separation between recurrent and capital budgets. Budget preparation is often done without the knowledge of ceilings (so-called "blue-sky budgeting"). The one-year budgeting horizon creates inflexibility. There are capacity limits at zonal and especially woreda levels, and timely reporting on budget implementation is weak. There are also problems with incorporating aid into the regional planning and budget system. To incorporate donor funds into their budgets, regions and zones need to know the amount, the source, and any special conditions attached to such funds. This information needs to be available well before the fiscal year begins to include it in the budget and allow the timely implementation of donor-funded activities. This PER develops the observation that the "offset" system, as it is currently operated, leads local governments to prefer central government funding relative to aid-financed projects, because of simpler procedures associated with the utilization of resources. Reform of Expenditure Management 14. In parallel with the decentralization, Ethiopia embarked on a comprehensive Civil Service Reform (CSR). The PER focuses in particular on the Expenditure Management and Control component (EMC) which aims at strengthening financial management systems. * In 1996, the Financial Legal Framework was promulgated. At present, the legislative drafting and approval process for numerous required regulations, directives and guidelines is underway. Priority should be given to issue the remaining financial directives, in particular those on budget management, including the budget calendar, procedures of budget control, and procedures for submission of supplementary appropriation. * The 1997 and 1998 PERs recommended the preparation of a three-year Medium-Term Expenditure Framework (MTEF) to strengthen the weak link between policy formulation, planning, and budgeting. The government has accepted this recommendation and has launched a process that will eventually culminate in an MTEF by expanding the Public Investment Program (PIP) into a Public Expenditure Program (PEP), which would cover both capital and recurrent expenditures and would be based on a Macroeconomic and Fiscal Framework (MEFF) that accounts for total revenue and expenditure commitments. The government also plans to roll out the federal PIP to the regions. The

12 vi Executive Summary first round of the PIP (FY98/99-00/01; EFYs ) was considered a dry run. The preparation of the second round PIP has been obstructed by uncertainties about the war. Operationalization of the PIP/MEFF exercise is extremely important and should be the top priority for the FY0O/Ol budget exercise. The Ministry of Economic Development and Cooperation (MEDaC), in conjunction with the Ministry of Finance (MoF), should take the lead: intensive coordination with the donors will be indispensable. + The aim of the budget reform component is to produce annual budgets of a higher quality on a federal and regional level in a more timely manner by moving towards activity-center (also called costcenter) budgeting. The two Budget Reform Design Manuals that have already been completed lay out the revised procedures for capital and recurrent budgeting at federal and regional levels, and introduce changes in the structure of the budget with regard to classification, expenditure codes, revenue codes, financial calendar, and budget formats. It is critical that the new principles and guidelines for budgeting permeate through all government institutions, at a central and regional level. Specific follow-up is required on the implementation of the financial calendar, which covers both planning and budgeting cycles, and aims at the adequacy and the timeliness of the budgeting exercise. Some difficulties encountered during the preparation of the budget for FY99/00 were to a large extent due to the absence of such an authoritative calendar. + The components on accounting and audit are the least advanced within the EMC program because of a shortage of donor funding and delays in contracting consultants. It is important that EMCmanagement accelerates the implementation of these components. In particular, the government needs to operationalize the financial reporting system on the use of donor funds under the SPs for education and health. Sector Development Programs 15. The preparation of SPs in education and health has materially assisted the definition of policies and expenditure programs. However, less attention seems to have been paid to the integration of SP implementation with regular government budget procedures, and in particular to the programming of donor funds. The PER mission found large discrepancies between the government's estimates/projections of external resource flows and donors' planned aid disbursements. This raises a number of issues. Inconsistency between donor and government records is clearly a problem (reflecting the extent to which disbursement is still taking place outside the government budget) because incomplete recording of aid flows make budget planning and the calculation of regional offsets a matter of chance.

13 Executive Summary vii Cross-Cutting Issues 16. The three areas of reform-decentralization, strengthening of expenditure management and control, and sector development programs-are all making good progress. All, however, involve major institutional change which naturally takes years to consolidate. Moreover, donors have an important role to play in supporting the reforms that the government is undertaking and in helping to ensure a coordinated approach. The PER identifies steps to strengthen the government's strategic planning of public expenditure and makes recommendations for better management of aid within the government's strategic planning framework. * Important steps to improve the strategic planning of public expenditure have already been taken through the preparation of the SPs and the launching of the MEFF/PIP exercise. It is now important to move swiftly to a full MTEF/PEP to ensure that medium-term expenditure plans are consistent with resource availability, and that budget holders are able to plan and implement expenditures effectively and efficiently. To operationalize the medium-term planning of public expenditure, the government now needs to adopt and adhere to the proposed financial calendar. There needs to be broad acceptance of an MTEF or PEP as a new way of budget planning. The MoF and MEDaC need to collaborate systematically in preparing the MEFF and in proposing indicative planning figures. When the PIP is extended to the regions, sector planners should be required from the outset to prepare consistent medium-term plans for both recurrent and capital expenditure, and regional and zonal councils should be required to undertake joint reviews of recurrent and capital budgets for each sector. + The collaboration of aid agencies is essential to achieve more rational planning and utilization of public expenditure. Credible projections of aid flows are an essential input into the MEFF and are needed for three or four years ahead, not just for the imminent budget year. An essential foundation for MEDaC is to maintain, with donor assistance, a reliable and comprehensive database of ongoing and expected aid projects and programs, which can be clearly mapped into the SPs that have been agreed, as well as into the PIP and successive annual budgets. In addition, the government and donors need to collaborate in making projections of aid flows available to the regions with sufficient advance notice to take them into account for budget preparation (by end-december if the proposed financial calendar is to be observed). The government should continue to encourage aid agencies to move towards unearmarked Channel 1 support. * The budget "offset" system continues to be an obstacle to the rational utilization of aid. The Government has taken steps to address this

14 viii Executive Summary issue and has carried out an analysis with the aim to improve the system. Without departing from the principle of regional equity, the following improvements could be adopted. The planning bodies' projections of aid flows should be published when ceilings are announced, so that the offset is transparent. In subsequent years, these projections should be compared with actual aid flows, and, if there are significant divergences, adjustments should be made. The aim should be to achieve equity in the actual flow of resources to regions (and within regions to zones) not just in the expected resource flows. Ceilings for Treasury funds and for aid should be clearly separated, and it should be made clear that if regions are unable to utilize available aid in a given budget year, it will normally remain available in subsequent years. If implementing agencies can clearly see that aid is additional, and can be drawn down over a practical time frame, they should be more ready to invest the additional effort needed to utilize aid funds. PER Process and Focus 17. The Ethiopia PER is an annual exercise carried out by the government and donors. The 1999 PER expands on previous PERs by focusing more on a continuous process and dialogue on public expenditures than on the production of a final report. With this objective in mind, two follow-up workshops are planned with the government and with the donors to explore the implementation of the PER recommendations and to link the PER process closer with the budget planning cycle. The partnership aspect of the PER process is reflected by the close collaboration on analytical issues and the 4 commitment to a more systematic follow-up process. 18. There are implications both for the scope and for the timing of future PERs. As regards timing, it is proposed to undertake future PERs much earlier in the fiscal year in order to support the planning phase, rather than the budgeting phase, of the financial calendar. As regards their scope, future PERs can be important in reconciling government and donor estimates of aid flows, and could help extend the sector dialogue associated with the SPs into a broader dialogue on overall public expenditures and enhancing the poverty impact of government and donor assistance. This could support further progress towards provision of aid in the form of unearmarked budget support. While recent PERs have focused mainly on budget inputs, it will be necessary to shift the focus of the next PER to the effectiveness and efficiency of public spending. This aspect will become particularly relevant in the context of Ethiopia's envisaged participation under the HIPC debt initiative. 4Summaries of the main findings and agreements of those two workshops are attached in the Annexes.

15 CHAPTER 1: INTRODUCTION SETrING 1. Since May 1998, the war between Ethiopia and Eritrea has been the center of attention of the Ethiopian government and its donors. In addition to the human toll of the war, there have been concerns that the sharp rise in defense expenditure could jeopardize the substantive macroeconomic and structural progress made in recent years. At the time of writing this report, efforts to reach a peaceful resolution of the war are continuing. Although the fiscal data for the past two years is still incomplete or preliminary, it allows some important conclusions. 2. Ethiopia has followed a policy of not permitting the war to interfere with the consolidation of macroeconomic stability and continuation of structural reforms. Indeed, with the exception of the floor on net foreign assets of the banking system, all the quantitative targets under the 1998/99 program with the IMF were met. Despite the shortfalls in official loans and grants official reserves at the end of FY98/99 are expected to be slightly higher than one year ago (in months of import coverage). The government has appropriately protected core spending in priority sectors, while financing the increased defense budget through extraordinary revenue efforts, including one-time financing measures such as utilization of privatization receipts. 3. Yet the fiscal stress posed by the war is reflected in the projected fiscal outcome for FY98/99, which is a clear outlier to the trend and deviates substantially from the government's intentions and objectives as reflected in the original and revised budgets for the fiscal year. Prudent fiscal policies and the timely resolution of the conflict are two critical elements that will help Ethiopia limit the short and medium-term economic impact of the war. Conversely, if the war continues, the govemment's long-term economic growth objectives and poverty reduction strategy will be placed in jeopardy. The main risks result from the cumulative negative impact that the war would continue to have on government finances, public and private investment, and the external current account. These risks add to the challenges that Ethiopia faces including weakening of the terms-of-trade and recurrence of drought in Focus OF THE 1999 PER 4. The Ethiopia PER is an annual exercise of the government and donors. The joint effort is reflected by the strong collaboration during the field mission, sharing of information, a common approach to the analyses, and a strong buy-in to the follow-up process and sharing of recommendations of the report. For the next PER, it is envisaged

16 2 Chapter ] that the government will also share in the drafting of background notes for the PER mission. 5. The PER has become a key analytical tool for allocation of public expenditure in line with the overall resource envelope and the broad range of development priorities. In line with Ethiopia's partnership approach to external assistance, PERs have become increasingly important in the dialogue between the government and the donor community. Typically Ethiopia's PERs center on the resource framework, budget systems, and expenditure patterns. In addition, each PER includes a specific analytical focus on key issues or recent developments with a view to developing specific recommendations for improving the design and implementation of priority sector expenditure programs. 6. The 1999 PER-the sixth PER since builds upon and expands the partnership approach started in The 1999 PER expands on previous PERs in that it is no longer focused so much on the production of a final report, but aims at a more continuous process and dialogue on public spending. With this objective in mind, the government shared the Aide Memoire of the main PER mission with the donor community at large in May The donor community responded well to the Aide Memoire and followed up on a number of findings individually, as well as in the context of the Annual Review Meeting (ARM) of the Sector Programs 6 (SP) for education and health. Two follow-up workshops are planned with the government and with the donors to pursue the PER recommendations and to link the PER process closer with the budget planning cycle The 1999 PER has a more limited scope than originally envisaged. The main field work for the PER was originally scheduled for mid-february 1999, with the intention to build upon the outcome of the IMF mission that, among other things, was to assess the government's current macroeconomic stance. As the IMF mission left Addis Ababa early because of security concerns, the PER mission was postponed. During the continuation of policy discussions between the IMF and the Ethiopian authorities in April 1999, the government presented a revised set of fiscal projections for FY98/99, which provided the basis for the analyses during the May 1999 PER mission. 8. The main PER mission focused on the original issues presented in the Concept Paper (January 15, 1999), except for the fiscal assessment of debt relief under the HiPC 5The previous reviews of Ethiopia's public finances were: Ethiopia: Public Expenditure Policy in Transition, Report No ET, October 1994 (1994 PER); Ethiopia: Public Expenditure Review, World Bank "White Cover" Report, May 1995 (1995 PER); Paul Collier (ed.), Ethiopia: Social Sector Review, Center for the Study of African Economies, Oxford University, July 1996 (1996 PER); Ethiopia: Public Expenditure Review, Report No ET, November, 1997 (1997 PER); and Ethiopia: Public Finances, Report No ET, December, 1998 (1998 PER). 6 The term Sector Programs is used to refer to Sector Investment Programs (SIP) as well as Sector Development Programs (SDP). Outcome and main findings of these two workshops are attached in the Annex.

17 Introduction 3 Debt Initiative 8. The 1999 PER also deviates from the general trend of the "new generation" PERs, which focus more on effectiveness and efficiency of public spending rather than on budget composition and inputs. In view of the war, there is an overriding interest in the impact of the war on public spending. Consequently, one major area of attention of the present report is to review the macroeconomic impact of the war and the financing of the associated defense expenditures, and to determine whether defense spending pressures have diverted resources from priority development programs. The major areas of focus of this report follow: * Recent Fiscal Developments: Has the trend toward fiscal stabilization continued? How has the war with Eritrea affected fiscal aggregates? * Fiscal Sustainability: In light of lessons learned from the first two years of implementation of SPs in education, health, and roads, what are the implications for their fiscal sustainability? What are the options for modifications in the financing and timing framework of the SPs? + Public Finance Management: The PER reviews the progress in implementing measures to improve budget formulation and execution, and identifies areas for additional measures to enhance the effectiveness and efficiency of public spending by focusing on the relationship between the center and the regions. * PER Process: The follow-up process to the PER and a more continuous dialogue on public expenditures between the government and the donors are central components of the PER. The report highlights the implications for future PERs. STATUS OF PREVIOUS PER RECOMMENDATIONS 9. Considerable progress has been made with regard to recommendations made in previous PERs. A summary of previous PER recommendations and their status of implementation is presented in Annex Table 1. The main focus of the 1997 and 1998 PERs was-in addition to the review of fiscal trends-on Ethiopia's expenditure management system, tax and expenditure assignments in the context of fiscal decentralization, the sustainability of public expenditure, and revenue mobilization and aid management. 10. With a view to improve public expenditure management within a Medium- Term Expenditure Framework (MTEF)-a core recommendation of the 1997 PER- the 8The compilation of debt data for the HIPC Debt Initiative has been delayed with the IMF ESAF Mid- Term Review.

18 4 Chapter I government has embarked on a complex process that will expand the Public Investment Program (PIP) into a Public Expenditure Program (PEP). This will provide a mechanism for linking the sector programs with the overall macroeconomic framework and capital with recurrent expenditures. Several PER recommendations are being implemented in the context of the Civil Service Reform (CSR) program launched in Under the Expenditure Management and Control (EMC) component, good progress has been made with regard to enhancing transparency of the budget process. For instance, two Budget Design Manuals have been prepared, the chart of accounts has been updated, and a budget calendar has been designed. In the next phase, these new procedures will need to be implemented on the ground. 11. The fiscal relations between the center and the regions were addressed in both, the Regionalization Study and previous PERs. Only limited progress has been made on core recommendations, such as the need to clarify the borrowing powers of regional institutions, illuminate the responsibilities for setting local taxes, and improve the budget offset system. A study supported by UNDP focuses on the inter-governmental grant regime, and budget offset is ongoing. The other issues have not been resolved and probably will need to be addressed in the context of a center-region dialogue. 12. Good progress has been made with regard to the implementation of a number of revenue mobilization measures. The Federal Inland Revenue Authority and Customs Office were initiated and the computerized customs reporting and recording system (ASYCUDA) was introduced at the main customs station. It is now being made operational along with the introduction of taxpayer identification numbers. Penalties and interest charges are levied on late tax payments, and the tax fraud unit is being expanded. The expansion of cost recovery measures, user charges, and a greater role of the private sector are under consideration, however, without visible progress to date. 13. The improvement of aid management has taken a major step forward with three SPs in education, health, and roads. Policy discussions and aid flows are better coordinated and options for harmonization of implementation arrangements are being explored. However, there is still a long way to go toward providing budget support and adopting one set (preferably the government's) of implementation procedures. The agreed financial reporting system for the SPs is not yet operational. 14. The fiscal sustainability of the three SPs and the required inter-sectoral expenditure allocations have been investigated in the two previous PERs. Budget provisions for the three SPs have been increasing in recent years; yet these lag behind the ambitious targets set out in the appraisal documents. Actual spending has not reached budgeted levels. There are clearly costs plus lost credibility-associated with overoptimistic implementation projections. In addition, it is necessary to identify the binding constraints to implementation. It has not become clear whether these are deficient implementation capacity, insufficient external resources, or overall fiscal constraints within a sound macroeconomic framework. This aspect will be explored in the current PER.

19 Introduction 5 DATA ISSUEs-AN IMORTANT CAVEAT 15. The main mission for the 1999 PER took place in May At that time, detailed fiscal accounts for FY98/99 were available for seven months of the fiscal year. For the mission's Aide Memoire, the PER teams decided to use the government's projections for the fiscal year. In May 1999, the government had projected a fiscal deficit (on a cash basis and including grants) of 2.1 percent of GDP and defense expenditure of 4.6 percent of GDP. 16. As the fiscal year progressed and additional fiscal data were made available, it became clear that actual performance would probably not meet the government's projections. As a result, the IMF's 1999 Article IV consultation staff report (issued in mid-july 1999) projected a fiscal deficit of 4.3 percent of GDP and defense expenditures of 5.8 percent of GDP. From the most recent federal government data for 11 months of the fiscal year, it appears that defense expenditures might even reach a level of more than 6 percent of GDP. Consequently, the fiscal deficit might be slightly higher than projected by the LvIF. 17. The quantitative analyses of this report are based on federal budget data for the first 11 months of the fiscal year, and projections made by the PER teams. The projections for fiscal aggregates are broadly consistent with the data in the IMF's Article IV consultation staff report and will be revised in light of the planned discussions with the government in September. 18. In the PER discussion workshops in Debre Zeit and Brussels, the Government made available revised projections for FY98/99. In comparison to the PER data base, the latest fiscal estimates indicate that defense expenditures may reach a share in GDP close to 7 percent, and 1.2 percentage points higher than the PER estimate. Despite the increase in defense spending (compared to earlier projections) it appears that core outlays for education, health, and roads are in line with the PER projections. Therefore, one of the basic PER conclusions--that the government has so far protected per capita and real spending levels in priority areas--remains valid.

20

21 CHAPTER 2: DEVELOPMENTS IN BUDGETARY AGGREGATES FISCAL TRENDS Despite the war, Ethiopia has managed to avoid a significant reversal in overall fiscal trends. Nevertheless, fiscal performance over the past two years could put the long-term positive trend in jeopardy. 19. Public expenditure trends underscore Ethiopia's improved fiscal performance and progress towards fiscal stabilization over the past 12 years, although FY98/99 is a clear outlier compared to the trend. Therefore, a timely correction in expenditure levels and composition is necessary-preferably in FY99/00-to help preserve Ethiopia's medium to long-term progress towards fiscal sustainability. Details on fiscal aggregates are presented in Table The period FY86/87-98/99 can be divided into three distinct sub-periods: the late civil war period (FY86/87-90/91), the reconstruction period (FY91/92-94/95); and the stabilization period (FY95/96-98/99). A comparison of fiscal performance over the three sub-periods reveals a steady decline in the fiscal deficit on cash basis (before and after grants), thus underscoring Ethiopia's strong commitment to fiscal discipline. The average general government fiscal deficit before grants declined from 10.7 percent of GDP to 8.6 percent and 6.0 percent over the three sub-periods. Similarly, the corresponding fiscal deficit after grants also declined from 7.1 percent of GDP to 5.8 percent and 3.1 percent. 21. On the revenue side, the performance has been uneven over the three sub-periods. The total ratio of revenue and grants to GDP decreased from 24.6 percent during FY86/87-90/91 to 16.5 percent during FY91/92-94/95, but increased to 21.6 percent during FY95/96-98/99. Total expenditure, on average, declined from 31.7 percent of GDP during FY86/87-90/91 to 22.4 percent during FY91/92-94/95, but increased to 24.9 percent of GDP during FY95/96-98/99 due to a commendable rise in the share of capital expenditure from 7.5 percent of GDP to 9.5 percent, but more recently mainly on account of increased defense spending. Over the period FY86/87-98/99, the capital expenditure to GDP ratio rebounded to levels equivalent to those in the late 1980s. Recurrent expenditure however, has been reduced from 22.1 percent of GDP in the late 1980s to between 14.9 percent and 15.4 percent over the past two sub-periods.

22 8 Chapter 2 Table 1: Fiscal Trends Average- FY FY FY Share of GDP (%) 86/87-90/91 91/92-94/95 95/6-98/99 FY9S/96 FV96/97 FY97/98 Ff98/ Actual Actual Pre-Actual Pre-Actual PER Proj. Tot. Revenue & Grants Total Revenue External Grants Total Expenditure Recurrent Expenditure Capital Expenditure Fiscal Deficit(cash) Before grants After grants Source: Ministry of Finance, World Bank Database, and IMF staff estimates (July 1999); Note: Expenditure excludes net lending. RECENT FISCAL PERFORMANCE: BUDGET TARGETS VS. OUTTURNS The projected outcome for FY98/99 deviates significantly from the government's intentions and objectives as reflected in the original and revised budgets for the fiscal year. Moreover, underspending on the capital budget (estimated at 3.5 percent of GDP) is more pronounced than in previous years. 22. In May 1999 at the time of the PER mission, the government projected fiscal improvement and progress towards fiscal consolidation for FY98/99. The revised budget for FY98/99 envisaged a rebound in external grants and mobilization of additional domestic revenue that would allow capital expenditures to reach a historically high level of 12 percent of GDP. In parallel, a reduction in recurrent spending (assuming a decline in defense expenditures from FY97/98 levels) would have allowed the government to reduce the fiscal deficit before grants to below 5 percent of GDP and contain the fiscal deficit after grants at around 2 percent of GDP. 23. Additional data through May 1999 for the federal government budget (the latest available data) indicate that the actual outcome for FY98/99 will not meet the government's projections. Total expenditure and revenue-plus-grants for FY98/99 are now projected at Birr 10.4 billion and Birr 12.6 billion, respectively. Still based on partial data for FY98/99, the fiscal deficit before grants is projected at 6.5 percent of GDP, in line with the revised budget target and very close to the pre-actual for FY97/98 (Table 2). However, the projected fiscal deficit after grants (4.3 percent of GDP) is substantially above the target of 3.2 percent of GDP and also above the corresponding FY97/98 level.

23 Developments In Budgetary Aggregates 9 Table 2: Budget Execution Fiscal Deficit FY97/98 FY98199 Revised Pre-Actual Revised Pre-Actual Budget Budget Fiscal Deficit (cash basis) Excluding Grants Including Grants Total Revenue and Grants Total Revenue External Grants Total Expenditure Recurrent Expenditure Capital Expenditure Source: Ministry of Finance, World Bank Database, and PER estimates (July 1999). Note: Expenditure excludes net lending. 24. Total revenue and grants are projected at 21.2 percent of GDP (similar in magnitude to the 21.5 percent in FY97/98) and clearly below the FY96/97 pre-actual. The discrepancies between targets and projected outturns on the revenue side are comparable to FY97/98, with external grants projected to reach only two-thirds of the budget target and domestic revenues reaching 93 percent of the budget target. The shortfall on domestic revenues is mainly attributable to the lower-than-budgeted tax receipts. A closer examination of the past four fiscal years reveals the offsetting effects of a continuous and substantive decline in the ratio of external grants to GDP and an increase in the ratio of domestic revenue (tax revenue plus non-tax revenue) to GDP. The decline in external grants has been in the order of between percent of GDP between FY96/97 and FY98/ Total expenditure is projected at 25.7 percent of GDP, thus remaining around the same level as in FY97/98. Recurrent expenditure is projected to continue to increase from 15.9 percent of GDP in FY97/98 to 16.9 percent in FY98/99 while the share of capital expenditure is projected to further decline from 9.5 percent of GDP in FY97/98 to 8.8 percent of GDP in FY97/98. It is striking, however, that the offsetting effects of underspending on the capital budget (equivalent to 3.5 percent of GDP) and overspending on the recurrent budget (equivalent to 2 percent of GDP) are more pronounced in FY98/99 than in previous years. 26. The large discrepancy between capital expenditure budget targets and outcomes is of particular concern with regard to the ambitious sector programs in education, health, and roads. Disbursement of capital expenditures in the three priority sectors is expected to reach only two-thirds of the revised target in FY98/99, and will remain even behind the previous year's performance, thus exacerbating implementation delays (Figure 1).

24 10 Chapter /_ 80-/:0 60-0:;j4- /0-: Percent 50 o_- 0 _I1FY79 Z' i0i i //j Helt Education Reads /Z_ 0 FY98199 Health Education Reads Sectors Figure 1: Capital Budget Implementation Rate in Priority Sectors 27. In theory, the chronic underspending on the capital budget could be (a) the reflection of chronic over-optimism when setting budget targets and/or (b) the effect of deliberate budget cuts (during the fiscal year) to increase funding for other budget items 9. The former effect could be the result of either continuous overestimation of implementation capacity or recurrent overestimation of available resources. The latter effect (deliberate budget cuts), because of the war, is a particular concern to donors, who were alarmed by the rising financing needs for defense during the fiscal year. In any case, chronic overestimation of budget allocations carries significant economic costs which are summarized in Box 1. 9It should be noted that 100% implementation of capital budgets is unlikely to ever be achieved. Expenditure discipline means that all projects are prevented from spending more than the budgeted amounts. Inevitably some projects will experience delays. The overall implementation rate will therefore never reach 100%.

25 Developments In Budgetary Aggregates 11 Box 1: Economic Cost of Chronic Underspending on the Capital Budget Regardless of whethert.chronic.'undrspending' onthe capdil budget is the result of overly-optimistic' budget allocations:or deliberate budgetc..uts, there are serious long-term costs to economic growth and poverty objectives,.because of insufficient investment now fbr growth: in later years. In addition,. there are specific isecondary-but equally farreaching-costs. As -a minimum, chronic over-optim,ism has costs in: loss of credibility and a loss of faith in the planni.ng and budgeting proess, As long as underperformance on the revenue and expenditure sides. of the.budget are of similar magnitude, short-term distributional effects m,.ay be small, thou h the lon-term costs in frfeited growth are not negligible. ' If, however, persistent u Ierspending on :the. cpital budget results in the accumulation of "unplanned"' public savings, thenathe public sector is taxing the private sector more than::what.is. needed to sustain matcroeconomic stability (thus, exacerbating the dampening effect:on economic growth). Further, there.are distributional losses if an unrealisic(too optimisc) budg't iallcation in one sector (Sector A) implicitly reduces the funds available to another sector (Sectr B). In.thatfcase, resource availability instead of::abs,optiv,e capacity becomes the constraining factor fo:r expenditures in :Sector B. Sector B.culd spend.more than w'hat.has been. allocated, but the 'rigidity of the budget process does.not provide effective.options fobr in-year reallocation of unused funds. 28. The following analysis will show that the underspending on the capital budget is mainly the result of overestimating the availability of funds and implementation capacity, and not the result of deliberate budget cuts. In some cases aid disbursements were delayed because of administrative bottlenecks; in other cases the aid flows never materialized because of donors' concerns about the war. In addition, the firm budgeting system, which does not allow for provisional budget allocations unless aid agreements have been finalized, hinders budget implementation if critical provisions for social sector expenditures are only voted into the budget at mid-year through supplementary budgets, as was the case in FY98/99 (Box 2). Box 2: The Limited Effectiveness of Supplementary Budgets on Spending Levels In FY98/99:the government-introduced three :supplementary.budgets, which increased the original budget-allocations.'. Two of ithe sup lementary budgets represented additional capital appropriations. for education and health funded from.external. assistance,. the third was additional recurrent expeindit.re fr dedense funde from n on-tax revenue and budget contingencies. These supplementary budgets were implemented in:a transparent manner in accordance with the relevant legal provisions. 1lHowever, :the capital subsidy appropriations to, the regions appear to have come too late in. the fiscal year for them to reasotnably undertake associated -projects and: thus.will probably lead to shortfalls in expenditure agains,t the revised targets fbr the fiscal year.

26 12 Chapter The analysis on whether under-execution on the capital budget is the result of chronic over-optimism or deliberate budget cuts, however, is only indicative because of the dearth of data on the availability of aid and the lack of quantitative data on implementation capacity. Nevertheless, the following steps should help shed more light on the issue: + As a first step, the development of expenditure shares and real per capita spending could indicate if there has been a deliberate and major reversal in the expenditure composition compared to previous years. The analysis of year-to-year changes in the shares and in real per capita spending is presented in the following section. + Second, a review of external aid flows could provide information on whether external resource availability was overestimated. This part of the analysis will be presented at the end of this chapter. + Finally, only qualitative information is available with regard to implementation capacity constraints or over-rigid procedures which unduly delay implementation of development projects; this analysis is presented in the chapter on budget management.

27 Developments In Budgetary Aggregates 13 ECONoMIC AND FUNCTIONAL EXPENDITURE CLASSIFICATION Notwithstanding an overall shift toward recurrent expenditures, especially defense, over the past two fiscal years, the functional expenditure classification reveals that the spending shares for priority areas have been by-and-large protected. Over the past three years, the expenditure shares for education, health, and roads-for total as well as capital spendinghave continued to increase. 30. The sharp rise in defense expenditure has resulted in a substantial shift in the expenditure composition. The year-to-year changes in expenditure shares are much more remarkable than the development of nominal and real expenditure aggregates net of defense. Because of the defense-related increase in recurrent spending, the share of capital expenditures in total spending has dropped significantly. Higher defense expenditure is also reflected by the striking increase in the share of wages and salaries and of materials and supplies from a combined 38 percent of total expenditure in the prewar years to over 50 percent in the past two fiscal years (Table 3). Table 3: Economic Classification of Public Expenditures FY FY FY In Million of Birr 86/87-91/92-95/96- FY95/96 FY96/97 FY97/98 FY98/99 90/ Actual Actual Pre-Actual Pre-Actual PER Proj. Nominal Expenditure Total Expenditure 4,937 6,223 10,828 9,206 10,078 11,405 12,624 Recurrent Expenditure 3,461 4,076 6,713 5,644 5,778 7,140 8,291 o/w Wages & Salaries 1,259 1,577 2,548 2,101 2,173 2,653 3,266 Materials & Supplies 1, ,329 1,551 1, ,300 Capital Expenditure 1,476 2,147 4,115 3,563 4, ,333 Share in Total Expenditure Recurrent Expenditure o/w Wages & Salaries Materials & Suppies Capital Expenditure Source: Ministry offinance, World Bank Database, and PER projections for FY98/99. Note: Covering general government expenditure and excluding net lending. 31. Nevertheless sectoral expenditure reflects the government's commitment to increase expenditure shares for social services and economic infrastructure. At present, the three priority areas (education, health, and roads) account for almost one-third of total spending and half of capital expenditures. The dramatic increase in spending shares for these three sectors was accompanied by the long-term decline in the expenditure shares for economic services (excluding agriculture) and-until recently-defense. The analysis of sectoral expenditure shares, covering both federal and regional government expenditures, is summarized in Table 4.

28 14 Chapter 2 Table 4: Functional Classification of General Government Expenditure Average In Percent FY FY FY 86/87-90/91 91s92-94n S199 FY95/96 FY96197 FY97198 FY98/ Actual Actual Pre-Actual Pre-Actual PER Proj. Share in Total Expenditure General Administration Defense Economic Infrastructure o/w Road Construction Economic Services o/w Agric. & Nat. Res Social Services o/w Education Health Other Share in Recurrent Expenditure General Administration Defense Economic Infrastructure o/w Road Construction Economic Services o/w Agric. & Nat. Res Social Services o/w Education Health Other Share in Capital Expenditure Economic Infrastructure o/w Road Construction Economic Development o/wagric. &Nat.Res Social Development o/w Education Other Health Source: Ministry of Finance, World Bank Database, and PER projectionsfor FY98/99. Note: Covering general government expenditure and excluding net lending. Defense 32. The long-term decline in Ethiopia's defense expenditures was reversed in FY97/98 with the outbreak of the war with Eritrea. Defense expenditure as a share of GDP, is now approximately triple its historically low level of the mid-1990s, but still significantly below the war-time levels in the 1980s. Three distinct trends in defense expenditure can be identified over the past decade: * The civil war period (FY86/87 to FY90/91), characterized by large defense outlays, accounting on average for 30.8 percent of total public spending or 9.6 percent of GDP.

29 Developments In Budgetary Aggregates 15 + The reconstruction and stabilization period (FY91/92-FY96/97), when defense expenditure decreased dramatically, firstly on account of demobilization, falling to percent of total public spending or between percent of GDP in FY91/92, and thereafter further declining to around 8 percent of total public spending or 2 percent of GDP in FY95/96. The historically low level of defense expenditure observed during this period was possible only on account of a highly commendable demobilization effort. * The Ethio-Eritrean War period (FY97/98-FY98/99), characterized by an increase in defense spending to around 23 percent of total spending or 5.8 percent of GDP in FY98/ ~~~~Defense Spending 30,0 sl as Share in Total S ~~~~Public Expenditure Defense Spending 20 as Share in GDP ,' ' ~~~~~~~~~~~~~~~~~~~~~~~ / / / / / / / / / / t99 Figure 2: Trends in Defense Expenditure 33. By comparison, Ethiopia's current defense spending (5.8 percent of GDP) is clearly above the level for Sub-Saharan Africa (SSA), where defense spending was on average 2.8 percent of GDP in Roads 34. The share of expenditure in the roads sector within total expenditure has been increasing over time (Table 4). Over the three sub-periods, expenditures on roads

30 16 Chapter 2 increased from 2.8 percent of total expenditure during FY86/87-90/91 to 6.3 percent during FY91/92-94/95 and 9.7 percent during FY95/ In recent years, this trend has continued with the share of roads expenditures rising from 8.7 percent in FY95/96 (actual) to 10.4 percent in FY97/98 (pre-actual), and is projected to further increase to 10.8 percent in FY98/99. Despite the increasing trend, actual spending continues to fall short of budget targets, and future budget allocations may need to become more realistic with regard to implementation capacity. Education 35. Total expenditure on education has followed an upward trend over the past decade. On average, sectoral expenditure shares increased from 9.5 percent of total expenditure during the civil war period to 13.6 percent during the reconstruction period and 14.5 percent during the stabilization period. Despite the increasing trend from FY86/87 to FY98/99, concerns were expressed in the 1998 PER report over the decline in the share of education expenditure in recent years. Table 4 shows that expenditure on education declined from 15.0 percent of total expenditure in FY95/96 to 14.0 percent in FY97/98. The government is addressing this concern and the share of expenditure on education is projected to rise to 14.6 percent of total expenditure in FY98/99, with most of the increase resulting from capital expenditures under the education SP. Health 36. The share of expenditure on health in total expenditure increased, on average, from 3.3 percent of total expenditure during the civil war period to 5.0 percent during the reconstruction period and 6.0 percent during the most recent five-year period. The trend has also continued in recent years-from 5.3 percent in FY95/96 (actual) to 6.0 percent in both FY97/98 (pre-actual) and FY98/99 (projection). Again, most of the increase is attributable to the capital budget where expenditures under the health SP are recorded. REAL AND PER CAP1TA EXPEND1TURE Real expenditure levels for capital expenditures and for non-defense spending are comparable with pre-war levels. The good news is that prewar gains have been preserved However the past two years expose missed opportunities to accelerate progress toward a more povertysensitive expenditure composition. 37. As indicated earlier, the recent nominal increases in total expenditures are mainly the result of the increases in defense spending. Between FY96/97 and FY97/98, overall public expenditure increased in nominal terms by about Birr 1.3 billion, which was broadly equivalent to the increase in defense spending. In FY98/99, an increase of total expenditure by about Birr 1.2 billion is projected, of which the lion's share is again on account of the increase in defense expenditure in the fiscal year. Therefore, it is more appropriate to look at real expenditure levels. Table 5 reveals that real non-defense expenditure has remained broadly constant at levels equivalent to pre-war years.

31 Developments In Budgetary Aggregates 17 FY98/99 real capital expenditure is comparable to FY97/98, yet about 10 percent lower than the historically high levels in FY96/97. Table 5: Real Expenditure Average I million of Birr FY FY FY FY95/96 86/ / /99 FY96197 FY97/98 FY98/ Actual Actual Pre-Actual Pre-Actual PER Proj. Nominal Total Expenditure o/w Non-Defense Recurrent Expenditure Capital Expenditure Real (GDP Deflator 1991/92 = 100) Total Expenditure o/w Non-Defense Recurrent Expenditure Capital Expenditure Source: Ministry of Finance, World Bank Database, and PER projections for FY98/99. Nlote: Cavering general government expenditure and excluding net lending. 38. Despite the low budget execution rate (63 percent of the capital budget) and the overall stagnation of real expenditure levels, FY98/99 projections indicate that real per capita spending in education, health, and roads has continued to improve over previous years. While real per capita capital expenditures have declined from Birr 55 (in 1991/92 Birr) to Birr 48 in FY98/99 (Table 6), the sectoral equivalent for the three priority sectors have more than doubled since the early 1990s and have continued to rise in recent years-reflecting the government's commitment to safeguard the three sectors from the impact of war-related expenditure shifts. Nevertheless, there is no reason for complacency since Ethiopia's levels of per capita spending are lagging significantly behind SSA averages (Figure 3). Per capita education and health expenditures in the majority of low-income Sub-Saharan African countries (for instance Kenya, Ghana, or Malawi) are two to three times higher than Ethiopia's current per capita spending on health and education of about $1.7 and $3.8, respectively.

32 18 Chapter 2 Table 6: Real Per Capita Expenditure Aver*ge--- - Blrr FY FY FY 861/7-90/91 91/92-94/95 95/96-98/99 GDP Deflator (1991/92= 100) FY95/96 FY96t97 FY97/98 FY98/ Actual Actual Pre-Actual Pre-Actual PER Proj. Total Expenditure Capital Expenditure Total Education Capital Education Total Health Capital Health Total Roads Capital Roads Source: Ministry oftfinance, World Bank Database, and PER projectionsjfor FY9/&99. Note: Covering general government expenditure and excluding net lending.

33 Developments In Budgetary Aggregates 19 Education Kenya (1997) Togo (1997) Ghana (1998) Zambia (1996) Uganda (1997) Rwanda (1997) Benin (1996) Malawi (1997) Burkina Faso (1996) Madagascar (1998) Burundi (1996) Tanzania (1996) Ethiopia (1998) USS per capita p.a. Health Ghana (1995) Cote d'lvoire (1997) Zambia (1996) Togo (1997) Madagascar (1998) Kenya (1995) Burkina Faso (1997) Benin (1995) _ Niger (1997) Uganda (1997) Ethiopia (1998) US$ per capita p.a. Figure 3: Per Capita Spending in Social Sectors for Selected SSA Countries

34 20 Chapter 2 CONSTRAINTS To MOBILIZING EXTERNAL RESOURCES FOR SPS The analysis of available-but still partial-fiscal data for FY98/99 does not support concerns that the government initiated deliberate budget cuts to fund the increased resource requirements for the defense budget. A substantial shorlfall is projected for external resources. Estimates of actual disbursement of external resources are expected to fall short by approximately Birr 1,300 million, which would account for almost three quarters of the underspending on the capital budget. The reasons for this shortfall are a combination of (1) the delays in turning aid pledges into formal agreements due to donor 's concerns about the war; (2) procedural delays in taking formal aid agreements to effectiveness; (3) complex financial and reporting requirements on donors' side, particularly in the case of SPs; and (4) technical capacity constraints within the government particularly related to an increased amount of civil works. These problems will need to be resolved within the government/donor partnership. IPACT ON SOCIAL SECrOR EXPENDITURE 39. A comparison of disbursement rates between external funds and the government's own funds-cumulative for the third quarter of the fiscal year-confirms that low disbursement of external resources is a major factor for underspending on the capital budget. At the time of the May 1999 PER mission, expenditure data for three quarters of the fiscal year indicated that the disbursement rate for external resources on the FY98/99 capital budget expenditures was significantly lower than the comparable figure for previous years. Conversely, the disbursement rate for the government's own funds was above that for external resources and higher than in FY97/98 (Figure 4). Estimates of actual disbursement of external resources are expected to fall short by approximately Birr 1,300 million, which would account for almost three quarters of the underspending on the capital budget. 40. The comparison of the planned average funding shares for the three SPs (as agreed between donors and the government during the program appraisal) with actual/projected funding shares is presented in Table 7. In the case of education, it was planned that the government would provide on average 73 percent of sectoral expenditures and external resources would cover 27 percent. The government's share of the financing burden is 55 percent for health and between 40 percent and 50 percent for roads. In the past two years, the government's actual/projected financing shares have exceeded targets across all sectors and external resources have lagged behind. For example, in education where external resources were expected to provide some 27 percent of funding, only 6 percent and 8 percent of external funding was achieved in FY97/98 and FY98/99, respectively.

35 Developments In Budgetary Aggregates ,40 * l la. 25 FY94/95 FY95/96 FY96197 FY97/98 FY9BW9 Fiscal Year INDomeic Resoure lextemal Resource. Figure 4: Disbursement of Capital Expenditures by Source Table 7: Sectoral Expenditure-Actuals/Projections vs. Targets Education Health Roads FY97F98 FY98/99 FY97/98 FY98/99 FY97/98 FY98/99 Pre-Act. PER Proj. Pre-Act. PER Proj. Pre-Act. PER!Proj Planned Average Funding Shares GoE Resources 73% 73% 55% 55% 50% 40% External Resources 27% 27% 45% 45% 50% 60% Pre-ActuallProjected Average Funding Shares GoE Resources 94% 88% 82% 68% 670% 58% Extem al Resources 6% 12% 18% 32% 33% 42% 41. The delays in mobilizing extenal resources for the SPs could be the result of capacity bottlenecks, extensive donor procedures, or political considerations associated with the war. The PER mission collected information from donors on expected and actual disbursement flows and on the reason for the slowdown in aid disbursements. There is quantitative and qualitative evidence to conclude that delays in closing exteeal assistance agreements (converting donor pledges into formal agreements) and taking

36 22 Chapter 2 those agreements to effectiveness were the most important factors behind underspending on the capital budget in the past two years Observed delays in the commitment of external resources for SPs point, first, to teething problems in the sense that FY97/98 and FY98/99 were special cases reflecting start-up difficulties in concluding external assistance agreements in support of the SPs. IDA support to health and education is an example of external assistance which was delayed because pre-requisites for effectiveness were not met before the start of FY98/99. Two supplementary appropriations in January 1999 were necessary to accommodate the support in the budget. However, the regions were not informed until a couple of months after such support was approved. Consequently, disbursements under the two IDA operations lag behind projections. 43. There are, in addition, systemic delays resulting from difficulties in meeting specific-and sometimes complex-donor procedures and extensive reporting requirements (to which the government has agreed). For instance, the introduction of a new financial reporting system-agreed with donors in the context of the education and health SPs-is still not fully operational, and consequently effectiveness and disbursements have been delayed. These issues will be addressed in a later section of the PER. 44. There are also technical bottlenecks within the government to moving funds to the beneficiary level. The government has indicated that in some sectors where SPs were launched during the past years, the capacity to handle a substantive increase in civil works contracts, tenders for procurement of goods etc., has become a constraint to absorbing external resources. 45. Delays in the approval of external resources due to political considerations are certainly a reason for the slow-down in external financing. In a number of cases, donors have adopted a wait-and-see approach to finalizing their support to sector programs, because of concerns about fungibility of resources and the possibility to indirectly fund the defense budget, combined with a negative public opinion in donor countries on supporting countries in conflict. The PER mission tried to assess the extent of a "warrelated" delay of external aid flows. Although the picture which emerged from the quick survey of donors is difficult to quantify, it is obvious that the most serious impact has been on counterpart funds from balance of payment support and on new aid commitments, while disbursements under existing commitments have been affected only marginally. The impact of this will be felt in the coming 1-3 years. The delay in disbursement of EC support (Adjustment Support and STABEX) is a prime example of 10 There are three distinct stages in the possible delay of disbursement of external assistance. The first stage is the conclusion of the assistance agreement between the GoE and the donor. The second stage is the effectiveness of the agreement, which often requires compliance with donor-specific procedures and accountability standards, as well as the detailed programming of aid (its allocation to specific projects in the capital budget). The third stage is at the implementation level where either physical implementation capacity or administrative capacity to ensure timely reimbursement of expenditures rnay be the limiting factor.

37 Developments In Budgetary Aggregates 23 how donors' concerns have affected fiscal aggregates in FY98/99 (Box 3). Moreover, future commitments could be seriously affected, if peace remains elusive in FY99/00. Box 3: EC Structural Adjustment Support and Stabex When the war between.. Ethiopia and Eritrea erupted in May 1998, the European Commission had prepared, in collaboration with the government,.financing proposals on structural adjustment support and STABEX, to be submitted to the EU Member States for endorsement in July The second structural adjustment support program (SAS II) comprises a total amount of Euro.75,8 million of which Euro 74,4 million should be disbursed in two tranches of' Euro 50.million and Euro 24.4 million, in the form of balance of payment support. Generated Counterpart Funds were; to be used as budgetary support. The remaining. balances and..'interests of STABEX transfers to. Ethiopia during accounted for Euro 65 million to which the.same modalities as SAS II would be applied It was foreseen. that a total amount of counterpart fu,nds of approximately Bir,r 1.2 billion (equivalent to 17 percent of recurrent expenditures.and 2.5 percent of' (DP) would become available to the budget in FY98/99. However,-the prevailing concerns within the European Commission and its Member States about the funibility of funds, but also about the ongoing ESAF negotiations with the IMF_ at that time, delayed the signing of the grant agreements until December 18., 1998, when a first.tranche of Euro 50 million was disbursed. In addition, the release scheme of countierpart. funds from the double signature account was modified to provide for a series of small installments: (equivalent to Euro 12.5.million for SAS,II and Euro 15 million for STABEX). Each installment would be conditioned on a positive assessment by the European Commission "on. the evolution of the conflict and its impact on public expenditures, in particular within the social sectors." To this end, only one. SAS IT and one STABEX installmhent of counterpart funds (equivalent, in total, to Euro 27.5 million) have been released. The consequence is that instead of the planned Birr 1.2 billion, Ethiopia has probably received only Birr :213 million during FY98/99. Source. Delegation of the European Commission in Ethiopia

38

39 CHAPTER 3: FISCAL IMPLICATIONS OF THE WAR The PER mission estimated the costs of second-round effects of the war between Birr 1,100 million and Birr 1,600 million annually, comprising immediate assistance to the displaced, replacement of assets, and higher transport cost due to rerouting road freight from Eritrean ports through Djibouti. This is equivalent to 2.5 percent and 3.3 percent of GDP per annum in FY97/98 and FY98/99, respectively. Together with the additional defense spending over and above levels of the mid-1990s (about 3.8 percent of GDP in FY98/99), the fiscal burden of the war would be equivalent to between 6 percent and 7 percent of GDP per annum in the two years considered. In addition, because of donors' concerns about the war, external loans and grants of some Birr 1,500 million have been delayed in FY98/99 (equivalent to approximately 3 percent of GDP). 46. This section of the PER focuses on the fiscal impact of the war beyond the immediate financing requirements that have led to a sharp increase in the defense budget. First, the funding sources for the increase in defense spending are reviewed, then some of the second round effects of the war are assessed. FuNDING OF INCREASES in DEFENSE BUDGET 47. During the past two fiscal years, increases to the defense budget have been funded through a combination of expenditure cuts in non-priority sectors, allocation of budget contingencies, additional non-tax revenues, such as transfers from extrabudgetary funds, and domestic financing. 48. In FY97/98, the originally budgeted level of military expenditure was Birr 847 million compared to a pre-actual outturn of Birr 2,090 million. The difference of Birr 1,247 million between budget and pre-actual was covered by higher-than-budgeted domestic financing of Birr 742 million (of which Birr 664 million was bank financing) and under-execution of the domestically financed portion of the capital budget (in nonpriority sectors) by Birr 500 million. 49. In FY98/99, the original defense budget was Birr 983 million (equivalent to 2 percent of GDP). At the time of the PER mission in May 1999, the government projected an increase in the defense budget to Birr 2,033 million as a result of the escalation of the conflict with Eritrea to a full-fledged war in early The authorities had identified Birr 1,050 million in extraordinary resources and spending allocations through supplementary budget appropriations to cover the higher than originally

40 26 Chapter 3 budgeted defense expenditures. The increase in defense spending was expected to be covered by Birr 800 million in additional non-tax revenue identified in a supplementary appropriation in January 1999 and Birr 250 million (of an original Birr 306 million) in the allocation of several budget contingencies. Non-tax revenues were as follows: Birr 362 million from Public Enterprises Privatization Fund; Birr 300 million released from additional dividends collected by the State-owned Enterprise Supervisory Authority; Birr 81 million from the Fuel Stabilization Fund; and Birr 57 million from the Sugar Auction Fund. 50. Since the May mission, however, it has become evident that defense expenditure is likely to exceed the revised target of Birr 2,033 million because of continued fighting. Through May 1999, total federal defense spending was Birr 2,871 million. The continued increase appears to have been covered by higher domestic financing and further non-defense spending cuts. 51. The use of extrabudgetary funds raises concerns about opportunity costs of funding. Revenue from privatization, as well as from other extrabudgetary funds, would have been used in principle for developmental or revenue stabilization purposes. The opportunity cost of the resources implies foregone spending on future investments-and possibly spending in poverty-sensitive areas-with probably higher rates of return. 52. It is also important to highlight an additional caveat that may lead to an underestimation of actual defense expenditure. A part of the war effort is being funded by "patriotic contributions." This voluntary and individual support to the country's war effort is not included in public accounts and may therefore mask the full magnitude of the war-related costs. Because of a lack of sufficient data, it is not possible to quantify this effect. SECOND ROUND EFFECTS OF THE WAR Slow-down in Domestic Investment 53. The war has had a dampening effect on economic activity, owing to the decline in investor confidence and possibly some crowding out of private sector credit. The pick up in the overall level of economic activity appears to have been less than originally targeted in FY98/99. At best, real GDP expanded by 6-7 percent (compared with an original target of percent), owing largely to a recovery in agricultural output from droughtaffected production in the previous year. Nonagricultural output, which covers mainly the formal sectors, grew at a considerably lower rate because of a loss in investor and consumer confidence associated with the protracted war. Evidence of an economic slowdown in this area can be found in the lack of growth in non-aid imports and decline in tourist arrivals. Furthermore, the change in money demand was subdued in FY98/99. Broad money is estimated to have grown by 6 percent, around one-half or the originally targeted rate of growth. Moreover, the financial health of several commercial banks was impaired by the seizure of imports and exports at Eritrean ports, which adversely affected some local borrowers.

41 Fiscal Implications Of The War Private sector credit growth was also slowed, from around 15 percent in FY97/98 to 10 percent in FY98/99, possibly reflecting the higher than planned recourse to domestic bank financing by the government. At the time of the PER mission, the government was still seeking to finance larger than originally budgeted military expenditure through supplementary appropriations (namely additional transfers from extrabudgetary funds) and budget contingencies. The use of extraordinary resources or non-priority expenditure cuts to cover additional defense spending was in line with understandings reached with the IMF in the context of the second annual ESAF arrangement. The principal aim was to avoid recourse to domestic bank financing and potential crowding out of credit to the private sector. Data that have become available since the PER mission indicate that the govemment had a fairly large domestic bank financing requirement in FY98/99, equivalent to around 2.2 percent of GDP. In light of considerably higher defense spending than envisaged at the time of the PER mission, it appears that at least some additional military expenditure was covered through recourse to domestic financing in FY98/99. Assistance to the Displaced and Returnees 55. About 315,000 people are displaced and there are 15,000 returnees from Eritrea. In May 1999, an additional 268,000 people were at risk of being displaced by the continued fighting. The annual cost of assistance to the displaced is estimated between Birr million (US$ million), depending on the severity and duration of the war". In any case, the displaced will require the replacement of lost assets (such as oxen, ploughs, implements). Such requirements are estimated at between Birr 400 million and Birr 550 million, based on Birr 5000 to 7000 per family. When the displaced will be able to return to their villages, additional funding will be needed for the repair and rebuilding of destroyed homes. To this end, donor response to the indicated needs has been positive, including US$3 million from USAID, US$2 million from the UN, and substantial funding from the EC and bilateral assistance. 12 Re-routing of Road Freight via Djibouti 56. The re-routing of freight from the Massawa and Assab Ports' 3 in Eritrea to the Djibouti Port has resulted in additional costs of at least Birr 130 million and Birr 180 million in FY97/98 and FY98/99, respectively. The incremental expenses on transport are equivalent to 0.3 percent of GDP in FY97/98 and 0.5 percent of GDP in FY98/99. The incremental transport cost could result in higher consumer prices, lower profit margins, and a lower trade volume. World Bank Poverty Study - Displaced, Disempowered and Distressed: a Focus on Poverty in Crisis, by P. Middlebrook and F. Conzato (first draft), Addis Ababa, February Much of the information reported in this section is drawn from this study, as well as from the United Nations Country Team Rapid Assessment Mission of 7-8 April Since the population affected by war originates from drought-affected areas (actually 200,000 of the total displaced people were already declared in need of food assistance), the risk exists that estimates do not properly distinguish between war-related assistance and other assistance needs. 3 About USD 100 million is the estimate of goods and merchandise seized in the port of Assab at the start of the conflict.

42 28 Chapter The unit cost of road transportation increased from Birr cents/ton/km (variation due to differing rates in peak and low seasons) for Assab/Massava to Birr cents/ton/km for Djibouti. Higher unit costs are the result of the higher truck maintenance costs because of the poorer road conditions along the corridor to Djibouti and a higher share of one-way freight because of the high lay-over cost in Djibouti. In other words, transporters prefer to have one empty trip to/from Djibouti rather than paying high costs for storage or parking at the Djibouti port (Table 8). Considering that the distance between Addis-Assab and Addis-Djibouti is almost the same and that almost total substitution has taken place between the two traffic paths, the overall transportation 14 cost followed the same trend as unit costs Table 8: Road Transportation Cost Addis Ababa to Assab and Djibouti Assab Djibouti _FY96/97 FY97/98 FY98/99 Distance in Km Volume of Dry Cargo in '000 tons (peak season) Volume of Dry Cargo in '000 tons (low season) Unit cost Birr/ton/km (peak season) Unit cost Birr/ton/km (low season) Total cost of Dry Cargo (in million of Birr) (NB: 98/99 estimates are based on 3 rp Quarter data; distance from Addis to Djibouti is via Galafi) 58. In addition to redirecting roughly 55 percent of total dry cargo transport, Ethiopia has to redirect road transportation of roughly 1.2 million tons of petrol per year. Under the assumption that the cost structure and original distribution for petrol is similar to dry cargo, the additional cost of re-routing all road transport via Djibouti is estimated at Birr 152 million in FY97/98 and Birr 218 million in FY97/98. The rehabilitation of 40 km of roads into the Djibouti territory has been a further cost factor equivalent to Birr 10 million (paid for by the Ethiopian Road Authority). 59. The incremental sea freight and unloading charges are two cost effects that are offsetting each other. On the cost side, in 1999, a user fee of US$1 per ton was introduced at Djibouti Port. 15 The increased costs are outweighed by lower charges for unloading/loading of dry cargo in Djibouti because bigger vessels can serve its deep sea port. Sea freight for dry cargo to Djibouti is estimated at US$1,000 on average per container which is significantly less than the corresponding estimate of US$1,200 for Assab. The annual net savings effect of these offsetting charges and savings is estimated between US$3 million to US$4 million, equivalent to between Birr 24 million and Birr 32 million. 14 Information on traffic, volume and cost of transportation to/from Assab and Djibouti was collected at the Ethiopian Road Transport Authority from Monthly Reports and specific studies. Another source of information on the same topic is: USAID, Comnparative Transportation Cost Analysis Study, Addis Ababa, December A similar fee was in levied at Assab Port; it is a new cost element in Djibouti in 1999.

43 CHAPTER 4: FISCAL SUSTAINABILITY OF SECTOR PROGRAMS Ethiopia faces the challenge of having to balance the needfor tremendous increases in per capita spending in poverty-sensitive areas with the need to maintain fiscal stability. In the past, fiscal constraints have not been binding, because implementation capacity and external resource flows were the primary factors that determined expenditure levels in priority areas. If a peace agreement is reached, the binding constraints are likely to change. External resource flows will probably not remain a binding constraint if all aidpledges are turned into formal agreements. With time, implementation capacity should also improve and help overcome the teething problems of the first few years of SP implementation. Therefore, the overall fiscal sustainability will become the major determinant for the balance between recurrent and capital expenditure and between spending on SPs and in non-sp areas. Over the medium term, fiscal sustainability will hinge on (1) realistic projections of grants, (2) a substantial additional domestic revenue mobilization effort, and (3) an eventual compression of defense outlays towards levels observed in the mid-199os. This framework should help achieve the envisaged improvement in the execution of the capital budget and higher associated recurrent spending. 60. Over the past two years, the wide gap between external financing requirements and actual mobilization of aid resources had far-reaching consequences for the implementation of the SPs. The immediate conclusion is that, at the end of the second year of implementation, the three SPs are already significantly behind schedule. It is highly unlikely that implementation can be accelerated over the next three years to catch up and meet the five-year implementation timetable for the three programs. + The health SP is based on the assumption that Birr 2,024 million of foreign loans and grants will be used during the five-year period. Up to now, only Birr 251 million have been disbursed with about Birr 200 million projected during FY98/99. The most that could be expected in the near term would be a doubling of annual donor disbursements in this sector. * Similarly, with education, Birr 3,003 million of foreign loan and grant disbursements have been planned, of which Birr 2,741 million still remain to be disbursed. Expected FY98/99 disbursements are

44 30 Chapter 4 projected at Birr 216 million. Even a doubling of annual disbursements will only result in total disbursements of Birr 2,324 million over the five-year period. This would leave a shortfall of Birr 417 million, which will thus have to be met from government resources or be deferred. * Concerning roads, there are doubts about the feasibility of the government's own expenditure targets. The first PIP exercise has brought into question the feasibility of projected road SP expenditures, which exceed the total of Treasury resources expected to be available for all capital investment in all sectors during FY98/ Out of the three broad categories of constraints that determine the path of implementation of the SPs-availability of resources, implementation capacity, overall macroeconomic/fiscal constraints-the first two have been the binding constraints to implementation in the past. Over the medium term, teething problems and external resources could become less binding while macroeconomic prerogatives could become the determining constraints to implementation. * Teething problems are likely to become less of a constraining factor to implementation as the federal and regional authorities make progress toward putting in place the administrative infrastructure for the SPs, enhancing implementation capacity in the context of the CSR, and therefore being able to implement a larger volume of projects and absorb additional funds. * Similarly, premised on an end of the war with Eritrea, external resource mobilization is likely to accelerate if all pledges are turned into financing agreements. A quick survey of donors' support to the education and health SPs indicates that substantial amounts will be available and external funding constraints should not be binding. For education, financing of around Birr 1,100 million has been agreed but has not been disbursed, while another Birr 1,455 million have been pledged and could be turned into agreements. Similarly for health, support of approximately Birr 1,000 million has been formally agreed but not yet disbursed, and another Birr 500 million have been pledged but not yet confirmed in formal agreements. 62. With external funding and implementation capacity becoming less of a binding factor, fiscal constraints will require that the three SPs fit into the overall macroeconomic framework and still leave room for other expenditures. Looking ahead, the government is challenged by the need to balance solid fiscal policy with the need to protect and increase core outlays in priority sectors. This challenge requires prompt and sustained actions, starting with the FY99/00 budget. To keep the fiscal deficit within the parameters envisaged, mobilization of resources-foreign and domestic-is a prerequisite within Ethiopia's broad-based and ambitious development program. In this

45 Fiscal Sustainability Of Sector Programs 31 context, the government needs to strengthen the major domestic tax effort and maintain close cooperation with the donor community to secure an adequate flow of concessional external resources. This chapter illustrates some of the critical choices the government faces over the next years to sustain its ambitious expenditure program. BROAD MACROECONOMIC OBJECTIVES 63. The following broad macroeconomic objectives will provide the framework that will help secure sufficient funding for implementation of the three SPs while leaving room for healthy growth in non-sp expenditures such as agriculture: * sustained high rate of economic growth per capita; * low and stable inflation; * increased revenue performance, especially from domestic tax revenue; 1 6 * substantial increase in external support through grants and concessional loans; and v maintenance of a low, single digit fiscal deficit (after grants) over the short and medium term. MACROECONOMIC SCENARIOS 64. The strong outcome outlined here is within the reach of the government, but such an optimistic scenario is also fragile because of recurrent drought and volatile terms-oftrade. This section provides a comparison between the base case scenario and three less optimistic-but plausible-scenarios. The following scenarios are computed using a simple Computable General Equilibrium (CGE) model." The equations of the model are presented in Annex 1. The three alternative scenarios illustrate the impact of changes and shocks in key assumptions and explanatory variables. The prime objective is to show how public expenditure would need to be compressed in comparison to the base case and what the affect would be on overall growth and consumption. Base Case 65. The base case would allow Ethiopia to meet all debt service obligations and still provide room for real growth in non-sp expenditures. An ambitious real growth target of 16 The govermnent is encouraged to explore other revenue options as well. In particular, there may be scope for increased "cost-sharing" through cost-recovery for selected government services. 17 This is a slightly expanded version of the same model used in the 1998 PER.

46 32 Chapter 4 between 6 percent and 7 percent p.a. is sustained, monetary expansion is low, and the fiscal deficit (after grants) is contained below 5 percent of GDP. Therefore, macroeconomic stability under the base case assumption is underpinned by low inflation, not to exceed 3 percent yearly, and moderate net repayment by the government on domestic debt, thus allowing a sensible expansion of domestic credit to the private sector. The base case is built upon the following key assumptions: * Terms-of-Trade: The base case assumes further deterioration in Ethiopia's terms of trade by 17 percent in 1998/99 and another 7 percent in 1999/00. The terms of trade are assumed to slowly improve thereafter. * Additional Domestic Revenue Mobilization: The base case assumes additional revenue mobilization equivalent to 1.5 percent of GDP above the level of FY97/98. This increased revenue effort is ambitious since it is the net increase resulting from the offsetting effects of improved revenue collection efficiency and broadening of the tax base and expected revenue losses on account of the ongoing tariff reform and privatization program. The government's commitment to a gradual reduction in the effective tariff rate and to privatization, will improve the prospects for strong future growth, though these reforms will also temporarily reduce the revenue base until growth accelerates 18. The assumed annual net gain of 1.5 percent of GDP (sustained over the next five years) yields an average revenue effort equivalent to 20.4 percent of GDP-just above the average of 19 percent for Sub-Saharan Africa for * Grants and Loans: The base case scenario also assumes a sizeable increase in external grants equivalent to 0.5 percent of GDP above the FY92/93-FY97/98 average of 2.8 percent of GDP, and an increase in external loan drawings by another 1 percent of GDP over the FY92/93-FY97/98 average of 3.8 percent of GDP. By FY03/04, grants should reach a level of 3.1 percent of GDP and loan drawings should reach 5.5 percent of GDP. * Defense Expenditure: It is assumed that, with the reaching of a sustainable peace agreement, defense expenditure will be reduced from over 5 percent of GDP in FY98/99 to below 3 percent of GDP by FY03/ Under the base case assumption, the average real per capita GDP growth rate would reach 3.4 percent p.a. between FY99/00 and FY03/04. Details on the outcome of 18 As was illustrated in the 1998 PER, the reformn programs will need to be backed by increased revenues from domestic indirect and directaxation.

47 Fiscal Sustainability Of Sector Programs 33 the various scenarios are presented in Table 9. Growth in real per capita consumption by the private sector, where the private sector is a reference to the general population, would average 2.5 percent yearly. These parameters would augur well for helping Ethiopia reduce poverty levels in line with International Development Targets. Table 9: Macroeconomic Scenarios-Outcomes EFY FY 98/99 99/0 00/01 01/02 02/03 03/ Public Spendhg/GDP (%) a/ Base Case Scenario 1 -Weak Revenue Scenario 2 - ToT Shock Scenario3 -Hlgh Defense/Low Aid Growth, Real per Capita Spending (%/ ) a/ Base Case Scenario 1 -Weak Revenue Scenario 2 - ToT Shock Scenario3-Eligh Defense/Low Aid Growth, Real GDP per Capita (0/.) Base Case Scenario 1 - Weak Revenue Scenario 2 - ToT Shock Scenario3 -High DefenselLow Aid Growth, Real Cp per Capita (%/ ) Base Case Scenario 1 -Weak Revenue Scenario 2 - ToT Shock Scenario 3 - Hlgh DefenselLow Aid a/ Excludes domestic and extemal interest payments due. Population growth rate is assumed to be 2.5 percent per annmunl 67. The average level of public expenditure, excluding interest due, would be 25.7 percent of GDP. The average real per capita growth rate in public expenditure would be 5.9 percent annually. This would be made possible by a fiscal deficit financing (excl. grants) that increases from 2.8 percent of GDP in FY99/00 to 5.0 percent of GDP in FY03/04 and a trade deficit that increases from 8.2 percent of GDP to 9.5 percent of GDP over the same period. To this end, the FY99/00 general government budget does not meet the assumptions of the base case and shows undesirable deviations from the FY98/99 budget. The major deviations (Box 4) will be addressed during the September mission to Addis Ababa Subsequent to the Debre Zeit workshop a 4 h scenario was computed, assuming a more pronounced deviation for the base case on account of high defense spending. This scenario is presented in Annex 2.

48 34 Chapter 4 Box 4: Important Deviations in FY99/00 Budget from Base Case The base i case is built : on optimistic prospects for resource mobilization and the concurrent expenditure increases ln prioritywsectors. To this end, the FY99/00 budget is neither consistentwvithathenvisaged improvement in tax revenuesa nor with the required incorease iln Cfederal capital expenditure. Federal $spendig ong social development is considerably smaller than planned.) This mayreflect the uncetaitnties regarding;0 the com mniitment of extemal resources, which-as in previous yearsw ould be appbropriate through budgett supplements. However, going by experience,.jitcould]againwbe that the supplementary budgetvapproprations come too. late to bet effctively applied, Of greater concerni4s the drop in regional transfersomparedato FY98/99which constrains& a mnorer accelerated implemenitation: of the isp Scenario 1 - Weak Revenue Efforts 68. Scenario 1 assumes that the revenue effort cannot be significantly expanded and the privatization program does not advance. Consequently, public spending would need to be adjusted downward if deficit financing is ruled out as a credible option. Scenario 1 applies the same tax revenue-to-gdp rates that were assumed for FY98/99 to all subsequent years. Non-tax revenues, however, are reduced to the FY97/98 level and then held constant because non-tax collection in FY98/99 was extraordinary due to the defense effort. All other scenario assumptions remain unchanged. These assumptions result in a revenue effort that declines from 19.5 percent of GDP in FY99/00 to 18.2 percent of GDP by FY03/04. Total revenues and grants over the period FY99/00-03/04 would average only 18.7 percent of GDP compared to 20.4 percent of GDP under the base case scenario. This is equivalent to a loss of 1.7 percent of GDP in revenue. 69. The loss in revenue is compensated by a reduction in expenditure to maintain the same fiscal deficit and domestic financing requirements of the base case scenario. Total spending, excluding interest, is constrained to 24.1 percent of GDP, compared to the 25-7 percent of GDP in the base case scenario. This adjustment in public spending is mainly affecting capital expenditure, thus slowing real growth and providing fewer resources for future spending. The average real per capita GDP growth rate between FY99/00 and FY03/04 is therefore 0.3 percent annually lower than under the base case scenario. However, real per capita private consumption grows 0. 1 percent yearly faster than Scenario 1, because the lower govemment taxation increases private income and consumption. Scenario 2 - Terms-of-Trade Shock 70. Ethiopia is vulnerable to terms-of-trade shocks because of its heavy reliance on coffee as the main export good. Therefore, this scenario shows how public spending would need to be reduced in response to sustained deterioration in the terms of trade. Import prices are assumed to rise by 5 percent yearly for FY99/00-FY02/04 rather than the base case forecast of an increase by 1.7 percent yearly. It is also assumed that aggregate export prices will rise by only 1.5 percent yearly rather than the base case

49 Fiscal Sustainability Of Sector Programs 35 assumption of an increase by 2.3 percent yearly. All other scenario assumptions remain unchanged compared to the base case. 71. The terms of trade decline will force the government to choose between preserving public sector purchasing power and softening the blow to the private sector. If the government acts to preserve public rather than private sector consumption, real per capita consumption would drop initially (by 3.1 percent) and would not recover to current levels before FY03/04. Clearly, in reducing poverty, this would be detrimental. Scenario 3 - Higher Defense Expenditure and Lower Aid Disbursements 72. This third scenario illustrates how public spending would need to be compressed in the event that defense expenditures cannot be quickly reduced and current delays and shortfalls in disbursing official grants and loans continue. In this scenario, defense spending falls much more gradually than in the base case scenario, reaching only 3.5 percent of GDP by FY03/04 rather than 2.8 percent of GDP. 20 The level of external grants in FY99/00 is assumed to remain at the current low level of 2.4 percent of GDP (though in the case of continued war, it is likely to drop even further because of donors concerns). 73. This scenario produces the lowest growth rate among the three scenarios for real per capita government expenditure and the second lowest rate for real per capita private consumption. The former reaches an average annual rate of 2.6 percent compared to 5.9 percent in the base case, and the latter averages at 1.6 percent p.a. compared to 2.5 percent p.a. under base case assumptions. TMPLICATIONS FOR IMPLEMENTATION OF SECTOR PROGRAMS 74. The base case scenario provides the overall fiscal envelope within a sustainable macroeconomic framework. Assuming that start-up problems and external resource constraints for the SPs are overcome, the fiscal envelope becomes the binding constraint to implementation of the SPs. Expenditure for the three SPs would compete with capital expenditure in other sectors. 75. The appraisal documents for the three SPs lay out ambitious annual expenditure targets. While these targets may each be feasible and realistic they are not consistent with a sustainable macroeconomic framework if taken together. Even with sufficient external funding, the government's own required contribution would exceed the available resources. In other words, the government would not be able to meet the anticipated funding contribution. Table 10 presents the shares of the three SPs in recurrent and capital expenditures that would be available under the base case scenario. Assuming the planned five-year implementation period is maintained, the SPs would absorb 58 percent of the FY99/00 capital budget (up from 48 percent in FY98/99). Even within the most optimistic expenditure envelope of the base case scenario, by FY01/02, the SPs would completely exhaust the notional capital expenditure envelope and there would be no room 20 It is assumed that all defense spending is recorded in the recurrent budget.

50 36 Chapter 4 left for capital expenditure outside the three SPs. This is clearly not a feasible option. The same analyses applied to the other scenarios would result in even more serious crowding out of non-sp expenditures by the three SPs. 76. Alternatively, the base case could provide room for healthy growth in non-sp spending if the implementation timeframe for the three SPs is modified. Originally, all three sector programs were heavily front-loaded with the lion's share of disbursements expected in the second and third year of a five-year implementation period. These targets seem no longer attainable in view of the obstacles to a timely mobilization of external resources. In addition, these assumptions are not consistent with overall fiscal constraints. It is therefore appropriate to envisage a more pronounced back-loading of capital expenditures, with gradually increasing disbursements for the education and health SP-and maintaining the five-year implementation period-and the extension of the implementation period for the roads SP from five to seven years. This easing of disbursement pressure would also leave sufficient scope for non-sp expenditures, which otherwise would need to be compressed to non-sustainable levels in order to preserve macroeconomic stability. The modified implementation path for capital expenditure under the three SPs is illustrated in Figure 5. Stretching out the implementation of the road SP to seven years provides for manageable increasing SP shares in the recurrent and capital budgets. The share of SP expenditure in capital expenditure would rise from 46 percent in FY99/00 to 56 percent in FY03/04 (Table 10). Table 10: SP Expenditures as Share in Total Spending Base Case- EFY9I EFY92 EFY93 EFF94 EFY9S EFY96 EFY SP as Share of... FY97198 FY98199 FY99/00 FY00/01 FY01102 FY02103 FY Five-Year Implementation Timeframe RecurTent Spending Capital Spending Modified Implementation Timeframe (515/7) Recurrent Spending Capital Spending a/ Total spending excludes interest due and defense. 77. The modified implementation path for the three SPs would allow a substantial increase in the real per capita spending on education, health, and roads (Table 11) thus, bringing Ethiopia's per capita spending figures closer to those of other SSA countries and providing the basis for improved service delivery and poverty reduction. Table 11: Base Case Scenario - Per Capita Spending in Education, Health, Roads Birr FY96197 FY97/98 FY98/99 FY99100 FYOO/O FY01/02 FY02/03 FY03/04 Education Health Roads Note: Base on GDP Deflator 1991/92=100. Source: PER Projections

51 Fiscal Sustainability Of Sector Programs 37 CONCLUSIONS 78. Each of the four scenarios illustrates the need for the government to make strategic choices about the sustainability of its expenditure programs. Such choices will be necessary under almost any conceivable circumstance, even with the most optimistic outlook for growth in revenues and external support of the base case scenario. In particular, it will be necessary for the government to balance an ambitious SP implementation schedule with non-sp expenditures. 79. The four scenarios presented in this chapter clearly illustrate the importance of increased resource mobilization if Ethiopia's ambitious SPs are to proceed in a timely fashion. External resources for implementation of the three SPs appear sufficient if all aid pledges are turned into formal agreements. Further, the ongoing donor/government dialogue may be used as a platform to revisit the implementation path of the SPs with a view to achieve more back-loading of the education and health SP and the expansion of the implementation timeframe for the roads SP to seven years, consistent with the projected fiscal envelope. 80. The key assumptions on revenue and grants both contribute to higher GDP growth and higher public expenditure. If the revenue assumptions do not materialize, government expenditure will have to be reduced by roughly 1.7 percent of GDP between FY99/00 and FY03/04. This would be detrimental for the envisaged increase in per capita spending in social sectors and would be counterproductive for helping Ethiopia reach the International Development Goals by If the grant assumptions do not materialize, real government expenditure will have to be reduced by roughly 0.8 percentage points of GDP between FY99/00 and FY03/04-with a larger reduction required if the government wishes to avoid a negative impact on private sector activity due to exchange rate pressures and lower net imports.

52 38 Chapter 4 Figure 5: Modified Implementation Path for Sector Programs Capital Eop.nditure. for Edij.on 120O a08 199i D 2rI1 2001lD2 I_1SP DOCLAiENT CAPITAL PROJ EDLCATION _ R PROJ.: CAPITAL EDUCATION I CapIta Expendltwes for Heath S as 1999UnD 20DOM1 2001n FE SPDJOCUWENTCAPITAL PROJ. HEALTH_-*PER PROJ:.C-A-P1TA-LH-EA-LTHIj Capital Expenditures for Roads d00 19ss9 s 19900O Dn112 0SP DOCLMENT CAPITAL PROJ. RSDE P -- PER PROJ.: CAPITALROEADSI

53 CHAPTER 5: CAPACITY BUILDING IN PUBLIC FINANCE MANAGEMENT INTRODUCTION 81. To a certain degree the underspending on the capital budget is related to the quality of budget planning and the financial relationship between the center and the regions. The fact that aid flows are far below expected levels is not only the result of donors' concerns about the war, but also the consequence of capacity constraints and insufficient communication between the institutions involved in budget management. 82. The aim of this chapter is to complement corresponding chapters in previous PERs and the Ethiopia Regionalization Study 2 ' (Box 5) by reviewing decentralized planning and budgeting in more detail. This chapter will first review progress in the areas where the govemment has taken action, then highlight priorities for immediate attention, and identify any new and cross-cutting issues that need to be addressed. Box 5: Budget Management Issues in Previous Studies Previous PERs have focused on aspects of public expenditure management, including links between planning and budgeting (with recommendations for.the development of, a MTEF) and fiscal decentralization. (with recommendations for refining the-mechanisms for sharing revenues within the Federation). The Regionalization Study focused on the evolving federal system, mainly from the perspective of the.division of tax and expenditure assignments between the- federal and regional governments and the related issues of budget transfers. The Regionalization Study noted that. the federal system is still evolving. The system builds upon strong macroeconomic coritrol inherent in the blocki grant system and the equitable nature of the grant formula. Among other points the study draws attention to the counterproductive incentive structure of the revenue formula (block grants to the regions), which essentially penalizes the regions for additional revenue efforts; the disincentive effects of the budget offset system, and.the inherent risks posed by contingent liabilities of regional guarantees for agricultural loans. ONGOING REFORMS IN PUBLIC FINANCE MANAGEMENT 83. Public finance management in Ethiopia is complex because three important reform exercises are proceeding simultaneously: 2 Ethiopia: Regionalization Study, World Bank, Report No , February 3, 1999.

54 40 Chapter 5 + A radical program of decentralization to regions and lower tiers of the government; * Civil Service Reform, including a variety of measures to strengthen expenditure management and control, and * Comprehensive sectoral budgeting and Sector Development Programs22. FISCAL DECENTRALIZATION Decentralized Budgeting 84. The principal features of Ethiopia's budget system 23 comprise of: the dual budgeting with separate government bodies in charge of the recurrent and capital budgets; an annual budgeting horizon which lacks a strategic framework; and a budget preparation process that is bottom-up and a resource allocation process that is top-down. Generally, budget discipline is enforced as budgets are prioritized to fit within expenditure ceilings, and expenditures conform to authorized appropriations. However, the planning process is characterized by a bias in favor of capital expenditure and recurrent expenditures are incremental. Line item budgeting is geared more toward expenditure control and not so much to public finance management. 85. Under the decentralized system, the center gives regions considerable autonomy over their budgets. Moreover, within the regions, there are moves to decentralize budgeting further, to zonal and even woreda level. In principle, the regions are autonomous in their budget preparation and execution since the federal 'subsidy' takes the form of an untied block grant. However, regional discretion is constrained since the federal government controls total expenditure and the constitutional framework obliges regions to follow federal policies 24 (for example, regional salary scales are tied to federal ones; school and health facilities follow national standards). Regional Budget Preparation 86. Regional budget planning in Ethiopia is a complex bottom-up process which involves multiple channels of coordination. 2 5 Traditional budgeting procedures in the 22 Three sector development programs are under way - for roads, education and health. The roads sector program is implemented primarily by the (federal) Ethiopian Roads Authority (ERA). The Education and Health Sector Development Programs are implemented at all levels of government, and this chapter therefore focuses more on them. 23 For more detail on present and proposed budget systems see, the Budget Reform Design Manuals. Version 1.0 (January 15, 1998) deals with Budget Processes and Preparation; Version 2.0 (March 25, 1999) covers Chart of Accounts. Budget Preparation and Presentation. 24 See Lister, December 1998, Chapter 2 (especially 2.3). 25 Regionalization Study.

55 Capacity Building In Public Finance Management 41 regions mirror the federal system and echo federal financial regulations (although caution is indicated in generalizing). 2 6 In addition, there are important dynamics in some regions with a shift from sector-wise to zone-wise budget planning (Box 6). Box 6: Fiscal Devolution Below the Regional Level For Oromia, the largest region, it is still ttue that regional sector bureaus are dominant in preparing sector budgets. But the two next-largest regions-amhara and SNNPR-have already decentralized budget preparation to the zonal level, and.a similar move is under consideration for Oromia. Thus, both Amhara and SNNPR. apply a formula to distribute available funds among zones, and the distribution of resources between capital. and recurrent budgets and. between sectors is decided at the zonal level, Both regions apply a formula very similar to the formula used for sharing federal subsidy among the regio,ns. Amhara plans to carry the process a stage further and allocate lump sum budgets also to woredas. Decentralized budgeting places greater responsibility on the fina'ce and planning bodies at the zonal level since they take the lead in supervising budget preparation, screening requests, and presenting consolidated budgets to- the zonal administration for approval. It is sometimes argued: that it is inappropriate.-to focus resource allocation responsibilities at the zonal level, because the zones do not have,the same constitutional status as regions and woredas. ' However, devolution of the capita+l. budget down to woreda level could result in,impracticably small allocations. Administrative capacity at woreda level is limited. There is no planning.body at woreda level, and the PER team visited two woredas:that are sharing the same administratin.. Zones, with average populations of around 850,000 do. offer at least the benefits of decentralization and a pragmatic compromise between top-down and bottom-up approaches, 87. The regional budgeting system has considerable strengths. + It is disciplined and activities are not included in the budget unless funding is assured 27 (this creates a degree of rigidity in dealing with aid funds, which is discussed later). + It is well documented and linked to implementation procedures. Budget documents themselves follow standard formats. Detailed annual work plans, which include specific financial and physical information, are prepared project by project at regional and zonal levels. 26 PER mission members visited a small sample of regional, zonal and woreda bodies in Amhara, Oromnia, Southern Peoples, and Harari Regions, and the Dire Dawa city administration. 27 The PER team asked one zonal sector head whether projects could be included in the budget in the hope that donor funding might be arranged in the near future; his response was that "planning without confirmed money is a crime".

56 42 Chapter S * It follows systematic channels, whereby sector bodies receive guidance from finance bodies in preparation of the recurrent budget, and from the planning bodies in preparation of the capital budget. * It takes account of local preferences, although lower levels are not autonomous, they are systematically consulted. 88. There are also inherent weaknesses in regional budgeting. * These weaknesses are reflected by the pervasive separation between recurrent and capital budgets. * Generally budget preparation is done without ceilings (so-called "blue-sky budgeting"). Guidelines for priorities are provided by the regions' five-year plans. However, these plans do not include detailed expenditure programs. This means either that the exercise has to be repeated once ceilings are known, or, more frequently, that the decisive round of budget cuts occurs at upper levels of the system. * The one-year budgeting horizon creates inflexibility. While there are systematic procedures for carrying out the projects that are included in each annual budget, it is very difficult to incorporate and implement projects for which funding is secured after the financial year has begun. * There are capacity limits at zonal and, especially, woreda level. There is no planning body at the woreda level. Even zones are often handicapped by poor communications, reliance on manual systems, and limited staff training. * Timely reporting on budget implementation is weak. Budgets and action plans provide a good framework for supervision and monitoring of implementation at woreda and zonal levels, and the budget coding used is clearly aligned with the federal coding system. Expenditures are reported monthly, and the finance bodies maintain consolidated records. However, the monthly reporting is very aggregated and does not generate the kind of detailed management information and analysis required, for example, for the SPs. Indeed, it is likely that the weakness of information flow upwards has led some observers to underestimate the quality of the budgeting and planning processes that take place at regional and zonal levels.

57 Capacity Building In Public Finance Management 43 Issues for Follow-up 89. The PER focuses on three systemic weaknesses of the system between the central and regional planning and budget system 28. Each area is important with regard to effective and efficient expenditure planning and execution: (a) incorporation of aid in regional budgets; (b) the budget off-set; and (c) the relevance of the financial calendar. * Incorporation of aid into regional budgets requires advanced notification of the amount, source, and specific implementation requirements of such funds. In order to fit in with the budget calendar, and allow timely implementation of donor-funded activities, this information needs to be available well before the fiscal year begins. Timely notification can be facilitated by a systematic recording and updating of all aid agreements and expected flows by MEDaC as well as routine reporting (to MEDaC) of all ongoing and envisaged aid agreements by donors. 29 * The "offset" system, as it is currently operated, leads local governments to prefer central government funding relative to aid financed projects, because of the simpler procedures associated with utilization of the resources. i Promulgation of the financial calendar (Annex Table ) will help ensure a timely and effective budget planning process. A practical link between federal and regional budgets, is that regional budget calendars are dependent on the federal one. The federal subsidy is pivotal for regional budgets, since the regions cannot issue firm ceilings for budget preparation until they know what the coming year's federal subsidy will be. Indeed, there is a cascade involved, since the volumes of resources available at woreda and zonal level also depend on the federal subsidy. 28 Aid principally affects the capital budget, since most donor finds are directed towards projects. However, this will not necessarily remain the case. Some donors are considering undifferentiated budget support, or channeling funds to the recurrent budget; and two traditional areas of donor support - text books and drugs - belong more in the recurrenthan the capital budget. 29 Aid has to be considered in the context of federal policies. Dealings with international aid agencies are a federal prerogative, with MEDAC as the lead agency. Financial regulations now require that all aid be accounted for as part of the consolidated fund. The federal subsidy is adjusted to take account of expected aid flows to the regions. Aid agencies are encouraged as far as possible to follow the govermnent's financial channels and procurement procedures.

58 44 Chapter 5 REFORM OF EXPENDITURE MANAGEMENT Civil Service Reform-Context, Structure, and Organization 90. In parallel with the decentralization, Ethiopia embarked on efforts to strengthen the efficiency and performance of the civil service. The CSR comprises five reform areas: Top Management Systems; Ethics; Human Resources Management; Service Delivery; and Expenditure Management and Control (Figure 6). The PER focuses in particular on the Expenditure Management and Control component (EMC). The progress made on CSR in general and EMC in particular, does not seem to be well known within the donor community, in spite of its comprehensive and important nature. It could be beneficial-also in view of obtaining complementary external assistance-to organize a briefing on this subject and to keep the donors informed regularly. Expenditure Management and Control Program 91. The overall objective of the EMC is to bring financial management systems in line with current policy and redress structural weaknesses. Therefore, the EMC consists of the following components: * Development and adoption of a new financial legal framework; * Implementation of a Public Investment Program to reform the budget planning systems; * Reform of the budgeting system; * Capacity building in accounting and audit, including the development of a Financial Information System at federal and regional level and upgrade of computer facilities, clarification of the roles and responsibilities of all audit institutions (internal and external) within the civil service; and development and improvement of the auditing and accounting profession.

59 I EXPENDITlRE HUMAN RESOURCES TOP MANAGEMENT SERVICE ETHICS MANAGEMENT & MANAGEMENT SYSTEMS DELIVERY CONTROL * Legal Framework * Job class & grading * Strategic & annual * National Policy * Codes of practice & * Public Investnent * Staff performance Planning * Complaints & legislative Programmes appraisal * Strengthening of Redress Framework * Budgeting * Recruitment, delegation & * Best practice * Establish central * Accounting selection & transfer accountability of development bodies on ethics in * Financial Information * Remuneration & line institutions government Systems conditions of service * PMO reform * Strengtien capacity * Internal Audit * Human Resources * Senior mggmt. of Police & * External Audit planning Service Judiciary * Development of * HRM IS * Policy analysis * Strengthening accounting & audit * Time management logical framework capacity of mass profession * Legal framework & media grievances * Education on ethics * Development of in goverrnent HRM profession service Figure 6: The Civil Service Reform Program

60 46 Chapter 5 Financial Legal Framework 92. In 1996 the Financial Legal Framework was drafted and adopted by the Council of Peoples' Representatives 30. The framework calls for the preparation and drafting of regulations, directives and guidelines. At present, the drafting and approval process for these various pieces of legislative follow-up is in full swing and considerable progress has been made. The Financial Regulation were passed by the Council of Ministers in 19973'. Of the 21 financial directives that need to be issued, nine-including some of central importance 3 2 -have already been completed, and most of the others are in an advanced stage of preparation. A considerable effort is currently being undertaken to train and acquaint federal and regional staff with the new provisions and legislation and their consequences. 93. Next Steps. Priority should be given to issue the remaining financial directives, in particular those on budget management, including the budget calendar, procedures of budget control, procedures for submission of supplementary appropriation, and calendar for supplementary appropriation. Public InvestmentProgram andaidmanagement 94. The 1997 and 1998 PERs recommended the preparation of a three-year MTEF to address the weak link between policy formulation, planning, and budgeting. The aim of an MTEF is to ensure a strategic approach to public expenditures, which takes account of inter-temporal and inter-sectoral tradeoffs, as well as ensuring consistency between recurrent and capital budgets. The MTEF would help ensure consistency between SPs and the overall macroeconomic framework. 95. The government has accepted this recommendation and has started launching a process that will eventually culminate in an MTEF. The government decided to begin by introducing a Public Investment Program (PIP). The PIP will be followed by a Public Expenditure Program (PEP), the difference being that a PIP programs only capital expenditures, while a PEP programs both capital and recurrent expenditures. However, since the expenditure guidelines for the PIP have to be based on a Macroeconomic and Fiscal Framework (MEFF) that considers total revenue and expenditure commitments, the PIP is regarded as an MTEF 'under construction'. The government also plans to roll out the federal PIP to the regions over the medium term. 96. Although, PIPs are rather outdated and inappropriate, it is important to see Ethiopia's PIP in perspective. Modern planning and budgeting systems emphasize the 30 Federal government Financial Administration Proclamation No 57/1996, 19 December Council of Ministers Financial Regulations No 17/1997, 1 July Directives issued until May 1999 covered: financial responsibilities, debt management, public property, procurement, fees &charges, retention of records, short term salary advances to civil servants, writeoffs and government safes.

61 Capacity Building In Public Finance Management 47 importance of an integrated approach to recurrent and capital expenditure, whereas PIPs may perpetuate a bias towards capital expenditure. If the only intention was to prepare three-year capital plans and budgets for the federal public bodies, it would be of very limited value. However, the Ethiopia PIP is intended as a staging post towards a more comprehensive Public Expenditure Program. The MEFF provides an appropriate framework for strategic planning of public expenditures. It introduces a multi-year perspective, it requires analysis of overall resource availability and provides the opportunity to address tradeoffs between federal/regional, recurrent/capital, and alternative sectoral expenditures. 97. The PIP process is embedded in the budgetary system through the financial regulations, and would encourage both donors and government agencies to follow agreed national priorities. The problems the PIP approach addresses are almost equally relevant at regional level: here too a strictly annual perspective leads to inferior expenditure choices, inefficient project implementation, and sub-optimal use of aid. The question of how to adapt the PIP to regional requirements is important. 98. At the time of the 1999 PER mission, one round of PIP preparation had been completed and the second was in process. * The first version of the federal PIP covers the FY98/99-00/01 (EFYs ).33 It was decided to treat the exercise as a dry run. The output is in two volumes. The first explains the process and describes the MEFF. The second includes project profiles for federal public bodies, which are maintained on a database developed for the purpose. Instead of preparing the first PIP ahead of the budget, call circulars for the first PIP were issued simultaneously with the capital budget call. The output is dated October 1998, and was thus completed well into the first fiscal year it covers. This PIP has not been submitted to Cabinet or to Parliament, nor has it been circulated beyond MEDaC. Moreover, submissions from public bodies were well in excess of Indicative Planning Figures (IPF), and the exercise of prioritization was not followed through. * Uncertainties about the impact of the war have frustrated preparation of the FY99/00 PIP. It was planned to get the second round PIP (EFY ) started earlier, and this time ensure at least Cabinet if not Parliamentary approval. At the time of the PER mission Indicative Planning Figures (IPFs) had still not been issued. * The PIP thus far only covers federal projects. However, calculating the funds available for federal capital expenditure requires projections of federal recurrent spending and of the federal transfer to the regions. 33 FDRE: Public Investment Program FY's , draft, October 1998 (2 Volumes).

62 48 Chapter 5 Preparation of PIPs at regional level would require the federal government to give the regions forward projections of their expenditure ceilings and anticipated federal subsidy. The initial inclination was to defer the regional PIPs until the federal PIP was bedded in. There is now more support for involving the regions sooner, partly because they too have made it a legal requirement, but also because the current regional Five Year Plans will shortly expire, and it would be sensible to harmonize the five-year plan and PIP work. * The PIP work has been primarily located in MEDaC. Revenue forecasts were sought from MoF, but MoF is not used to forecasting more than one year ahead, and MNEDaC therefore made its own assumptions about the medium term. In order to develop the revenue/expenditure forecasts, MEDaC also prepared (apparently for the first time) a comprehensive set of projections of external assistance. These are crucial to the exercise: without reliable and complete donor forecasts it is difficult to produce an accurate MEFF since external funding accounts for percent of capital expenditure. 99. Next Steps. Operationalization of the PIP/MEFF exercise is extremely important. It is important to recognize the MTEF as a new way of doing business that shifts the focus of budget planning from being investment driven to being comprehensive, with adequate recurrent funding for existing infrastructure taking priority over new capital expenditures. The next round of the PIP should be started early--with a view to moving quickly to a PEP and should involve all players concerned. Lessons learned from the first "dry run" could be disseminated and applied. A review of the used MEFF methodology will be useful. MEDaC could develop a process through which it will obtain from the donors reliable and timely forecasts of foreign assistance, to be used as an important input into the MEFF. An intensive coordination with the donors will be indispensable. A strategy for expansion of the PIP/PEP exercise to the regions will need to be developed. Reform of the Budgeting System 100. The overall aim of the budget reform component is to produce annual budgets of a higher quality on a federal and regional level in a more timely manner. The reforms aim at budget preparation compatible with resource ceilings and development priorities, improved budget management that will lead to a more appropriate expenditure composition, and more effective budget control through a strengthened budget process and revised budget codes. The underlying theme of the approach is to move towards activity-center (also called cost-center) budgeting. Activity centers will allow public bodies to know the total cost of an activity and make budgeting decisions accordingly. The approach combines the virtues of a line item budget system that Ethiopia is used to, but aims to improve it as a tool of management and not just as a tool for expenditure control. Priority actions for improvement include:

63 Capacity Building In Public Finance Management 49 * Development of a more appropriate budget calendar, that allows strategic planning and detailed budgeting to be properly sequenced during the year. * Development and introduction of a better chart of accounts, with budget codes more suited to both management and reporting, and facilitating self-accounting by agencies. * Improved budgeting techniques, including the use of norms and standards, and the timely preparation of budget ceilings Up to now, two Budget Reform Design Manuals have been completed The first manual on Budget Process and Preparation 3 5 lays out the procedures for capital and recurrent budgeting at federal and regional levels and makes recommendations for improvement. The manual identifies a number of weaknesses that need be addressed, including: (1) considerable off-budget financing of capital expenditure; (2) chronic under-funding of the recurrent budget, in particular insufficient non-wage recurrent expenditures to sustain capital expenditures; (3) uncertainty and inappropriate budget practices because of the absence of an authoritative budget calendar; and (4) the uncertainty of externally financed capital expenditures. * The second manual on Chart of Accounts and Budget Presentation 36 was presented in March It introduces changes in the structure of the budget in five areas: classification, expenditure codes, revenue codes, financial calendar, and budget formats Next Steps. Implementation of the Budget Manuals is the top priority now. It is critical that the new principles and guidelines for budgeting should permeate through all government institutions, at a central and regional level alike. Specific follow-up is required on the implementation of the financial calendar, which covers both planning and budgeting cycles and aims at the adequacy and the timeliness of the budgeting exercise. Broadly speaking, the planning cycle would cover the first half, while the preparation and adoption of the budget would take place in the second half of the fiscal year. The difficulties encountered during the preparation of the budget for FY99/00 are to a large extent due to the absence of such an authoritative calendar. Targeted dates have not been 34 The next volumes of the budget reform design manual will address organization and staffing of budget institutions (volume 3.0) and aid management (volume 4.0). 35 Budget Reform Design Manual, Version 1.0, Budget Processes and Preparation, 15 January Budget Reforn Design Manual, Version 2.0, Chart of Accounts, Budget Preparation and Presentation, 25 March 1999.

64 50 Chapter 5 respected both during the planning and in the budgeting cycle. In the month of May for instance, the regions still had not received their budget ceilings, which complicated their timetable for presenting realistic budgets. What is now needed is to ensure adoption of the calendar and discipline in following it through the fiscal year. Accounting and Audit 103. The components on accounting and audit are the least advanced within the EMC program because of a shortage of donor funding and delays in contracting consultants. It is important that the EMC-management accelerates the implementation of these components. In particular, it is important to operationalize the financial reporting system on the use of donor funds under the SPs for education and health. The financial reporting system on the use of donor funds broadly follows the current accounting practice of the government. The major difference is the revised chart of accounts, especially developed for donor funding 7. Up to now, the interim reporting system is not yet operational. SECTOR DEVELOPMENT PROGRAMS 104. The present PER and previous ones have questioned the realism and sustainability of the expenditure targets in SPs, but the striking innovative element is the level of detail at which it is now possible to conduct this analytical debate. Few countries have developed comprehensive, resource-constrained sector programs that enable such issues to be systematically considered. While expenditure patterns under the SPs are reviewed in other chapters of this PER; the following section is about underlying implementation issues. This section first provides overviews of the SPs and of the financing channels for aid disbursement, then turns to a more detailed review of the programming of aid funds. Overview of Sector Programs 105. Sector strategies for roads, health, and education were first prepared by the government and then further developed jointly with aid agencies involved in the social sectors. 3 9 The SPs are the first 5-year slice of 20-year strategies, and cover EFYs The SPs are characterized by a Program Action Plan (PAP), which summarizes the strategy and its associated expenditure targets, more detailed Education and Health Sector Development Program documents for each region and the center, and a Program Implementation Manual (PIM) which describes implementation procedures for each sector. Basic procedures are virtually identical for the two sectors, but each version also includes some sector-specific material. Important features of both SPs For ESDP, see pages of the ESDP Program Implementation Manual, December The primary focus is on the health and education SDPs. 39 For a comprehensive account of the preparation of the ESDP, see Preparation of the Education Sector Development Program in Ethiopia: Reflections by Participants, by John Martin, Riitta Oksanen and Tuomas Takala, Final Report 7 June Preparation of the HSDP followed the same pattern.

65 Capacity Building In Public Finance Management 51 include the joint preparation by regions and federal governments in the spirit of "cooperative federalism", the coverage of the entire sector, the prescribed discipline in planning and prioritizing within expenditure ceilings, and ongoing efforts to simplify and harmonize donor procedures. Financial Channels for Implementation 106. An important motive (especially for the government) was to improve the quality of aid and of the aid relationship. Instead of fragmented, low-trust micro-management of separate projects by donors, the SPs offer a platform for high-trust dialogue about policies and implementation. A further objective was to achieve full pooling of aid funds. However, it became clear that a step-by-step approach was inevitable. At a minimum, the aim is to harmonize donor and government procedures, and to focus donor efforts on the priorities embodied in the government program, which will encourage regional uptake of available donor funding Three main channels for disbursement of aid have been identified: * Channel 1 is the standard route for the flow of government funds. Because of the seamless integration of unearmarked donor funding with the government's own revenues, Channel 1 is the preferred modality. Donors who are prepared to use Channel 1 fall into two groups: those who nevertheless insist on being able to trace their funds to particular end-uses, and those who are prepared to consider unearmarked budget support. 40 * Channel 2 involves transmission of funds via the Sector Ministries. Since Channel 2 undermines the integrity of planning, budgeting, and accounting systems, it is to be avoided. * Channel 3 depicts direct disbursement of funds by donors, in which no government body is directly responsible for handling the funds. In line with the government's policy that all aid must be brought to account in the consolidated fund, Channel 3 funds directly disbursed by donors should be appropriately recorded. 40 A number of donors envisage dividing their support between different disbursement channels, at least for a transitional period. IDA Bank has been foremost among donors prepared to use the earmarked variant of Channel 1 - so-called "Channelb". To do this effectively depends on a strengthening of financial reporting down to regional and zonal levels, so that expenditures are coded according to source of funds, and prompt reporting on expenditures allows timely replenishment of initial advances. The Excellence Report was commissioned in order to accelerate this aspect of budget reform and install a functioning interim reporting system as rapidly as possible. Installation of the system is under way, but has taken longer than hoped and at the time of the PER mission was not yet delivering consolidated reports.

66 52 Chapter S CHANNELI CHANNEL2 CHANNEL-3 (via Finance bodies) (via Sector bodies) (direct) Minsty ffinance -SetorMinistry Regional Finance Bureau Regional Sector Bureau V Zonal Finance Departmnt I ZonalSector Departrnt Woreda Finance Otrice Woreda Sector Office Figure 7: Alternative Disbursement Channels for SPs To this end all three channels are being employed. It is recognized that some donors operate under constraints that require them to continue to use Channel 3, and that in any case it will take time to restructure established disbursement arrangements. Programming of Aid Funds for the SPs 109. The preparation of SPs in education and health has helped to define policies and expenditure programs. However, less attention seems to have been paid to the integration of SP implementation with regular government budget procedures, and in particular to the programming 4 2 of donor funds. The PIMs describe how donor funds are to be 41 See Lister, 1998, Chapter 4, for examples. 42 Programming refers to the matching of donors to projects/activities and fitting of projects/activities into budgets. During the.preparation of the SDPs this was sometimes referred to as "donor mapping": however, it is important that the term is understood in the active sense of matching donors to activities and not simply as passive discovery of what donors are doing. There has been a tendency to add up aid agency pledges, on the one hand, and the SDP expenditure targets, on the other; if pledges (including GOE's own expenditure commitinents) equal or exceed SDP expenditure targets, there is said to be no funding gap. (Tbe recent paper on Donor Support to the SDPs also adopts this approach.) This is naive: pledges may appear to be sufficient but still leave a gap in practice if either the pledges are not consummated or the fumds are spent on activities not incorporated in the sector programs.

67 Capacity Building In Public Finance Management 53 accounted for, but do not explain how donor funds are matched to particular components of the SPs in the first place In an ideal arrangement (from the government perspective), programming of donor funds would not be an issue. If all donor funds were provided as unearmarked budget support, the only task would be to forecast the total of government-plus-donor funds available and use this in determining budgetary ceilings. Budget holders would not need to know how much of the ceiling originated from aid, since implementation, procurement and accounting procedures for all components of the program would be exactly the same. In practice, budget holders do need to know the source of particular funds and the conditions attached to their use indeed, the quality of aid programming may have a profound effect on the speed and effectiveness of implementation of the ESDP and HSDP. However, the move to more flexible aid modalities is a two-way street and requires that donors' concerns about accountability of resource utilization are met The PER mission found large discrepancies between the government's estimates/projections of external resource flows and donors' planned aid disbursements.44 Inconsistency between donor and govemment records is a problem because the correct calculation of regional offsets becomes a matter of chance. Even if the discrepancies reflect the extent to which disbursement is still taking place outside the government budget, it is important to get as accurate as possible ex-ante disbursement information. It is, however, not clear whether all reported donor expenditures in the education and health sectors should be counted as part of the SPs. In principle, they clearly should, because the SPs are intended as comprehensive programs that embrace all (federal and regional) government activities from primary to tertiary level in each sector. It may well be that some donor-supported activities (especially for projects and activities conceived before the SPs) do not fit well with the stated SP priorities. During SP preparation it was recognized that some donor activities may need to be realigned to fit with the redefined priorities of the programs. For example, technical assistance is particularly difficult to capture within government expenditure records, and this may account for some of the difference between donor and government figures. While donors have a responsibility to provide timely and accurate information on their programs (especially when disbursement does not follow regular government channels), the government itself needs to maintain continuous and consolidated records across donors and sectors. Within the government, MEDaC is the appropriate agency to maintain such records, and should be able to relate all ongoing donor programs in health and education to the SPs. Options for IDA Support to the Education and Health Sector Programs 112. The experience with IDA funding is given special attention because it is pivotal in several ways. IDA is the largest single external funding agency for both health and education sectors and it has adopted the role of 'financier of last resort'. IDA coordinated 43 At the ARM2 meeting, regional participants stressed the importance of receiving such information as early as possible in the budget cycle. 44 See Donor Supportfor SDPs: Some Supplementary Information for the 1999 PER, (Second Draft) Wim Olthof, EC Delegation.

68 54 Chapter 5 the donors' preparation of the health and education SP. In principle, IDA funds are available to support any of the identified components of the two SPs and are disbursed via Channel lb (following the finance body channel, and being disbursed at regional and even zonal levels). IDA funds are earmarked to specific uses and IDA procurement and accounting standards have to be satisfied by those projects. Separate accounts have been opened by each region to facilitate the tracking and reimbursement of earmarked expenditures. To facilitate rapid and decentralized implementation, the World Bank took a number of steps, including the posting of the Task Manager for the IDA education project in Addis Ababa (so that necessary approvals could be processed more rapidly) and holding in-depth seminars on IDA procurement procedures Agreements on the IDA support were concluded only after the commencement of FY98/99 45 because conditions for effectiveness were not met as planned. At negotiation, projected annual expenditure was one contentious issue. The government sought high target rates of expenditure in the first year, and IDA acceded to an upper limit of US$35 million for the education SP with a projected disbursement target of US$15 million in FY98/99 (thought IDA expressed doubt that such a high figure was achievable in practice). The present projection is that only a fraction will have been utilized by the end of FY98/99. For IDA's contribution to the health SP, it is expected that FY98/99 disbursements will be negligible There are a number of contributing reasons for the slow start to disbursement. Impediments to meeting the pre-requisites for effectiveness delayed the implementation of the operation well into the fiscal year. In line with its proposed role as financier of last resort, IDA sought information not only on the projects to be included in its own subprogram, but also on funding from other sources, so as to be assured that complementary inputs were available. The Ministry of Education (MoE) was unable to provide rapid and comprehensive information on the overall funding situation of the education SP-partly because it was not tracking donor funds in the necessary detail, but also because information on regional budgets had to be gleaned from their budget documents For its part, the government had to obtain Parliament's approval for a supplementary budget to allow expenditures to commence. The federal government decided to allocate funds to the regions in proportion to their formula shares of federal subsidy, which meant that allocations took no account of different regions' pipelines of prepared projects or capacity to cope with IDA's administrative and accounting requirements. The Ministry of Education (MoE) decided on the types of projects each region was to include because there was no time for a bottom-up identification process of IDA subprograms. There was considerable confusion amongst regional Education Bureaus, who felt that they were given insufficient discretion, that they were not given sufficiently clear instructions, and that they were subject to repetitive and overlapping demands for project information. 45 The IDA credit for health is running several months behind the education credit, and the latter therefore provides more relevant experience.

69 Capacity Building In Public Finance Management The respective roles of the MoE and MEDaC were not clear; MEDaC and the planning bodies in the regions are the usual channel for information and instructions on the capital budget but MoE has been the channel for instructions on the use of the IDA credit. Moreover, the perceived benefits of IDA funding were also unclear, since regions expected that IDA funds would substitute for Treasury funds which are much more straightforward to deal with Now regions are concerned as to whether the IDA funds allocated for FY98199, which have not been spent, will be carried forward to the next fiscal year, and what precisely will be the mechanism for doing this. The normal procedure for government funds 46 is that an unutilized provision lapses and has to be re-voted in the following year's budget. At the time of the PER mission, the MoE had not decided the level of IDA funding it would request for FY99/00 and, as already noted, no ceilings or indications of overall federal subsidy had been issued Some lessons have already been drawn from this first year's experience. It has been decided that in future the regions will be responsible for proposing subprograms, and the need to ensure that IDA funds are reflected in government budgets is now understood (although the delay in considering requirements for FY99/00 does not augur well). Once the expenditure tracking system is operational, the process of claiming reimbursement for expenditures should proceed more smoothly The procedures for programming the available IDA funds should be reviewed to: * Ensure that the concept of annual programming of IDA funds provides timely information to central and regional government levels on aid flows in line with IDA's role as financier of last resort. The implementation cycle (taking account of procurement lead times and so forth) is considerably longer than one year for most projects. Efficient implementation requires that the implementing body knows the source and availability of funds as early in the project cycle as possible. For all practical purposes this information (even if only as a notional allocation) needs to be available some months before the commencement of the budget year. In doing so, the government could allocate IDA funds to components less likely to be sought by other earmarking donors, taking into account IDA's willingness to finance any component of the SP. In addition, IDA funds might be reallocated as the program is rolled forward. * Finalize the standard format for project information for the government and the donors, and implement the financial information system. 46 Apart from a month's grace before the annual books are closed, to allow the completion of payments already under way at the end of the fiscal year.

70 56 Chapter S * Frequently review the appropriate amount in the Special Account (disbursement buffer) of advanced (but not yet accounted for) funds. 47 Decentralization does lengthen the chain of fund disbursement downwards and expenditure reporting upwards. Because of bottlenecks at earlier stages in the project cycle, this constraint has not yet been felt by the IDA credit. However, it is implausible that an advance of only US$6 million can leverage annual expenditure of US$ 30 million or more. RECOMMENDATIONS ON CROSS-CUTTING ISSUES Overview 120. The three inter-linked areas of reform reviewed in this chapter-decentralization, strengthening of expenditure management and control, and sector development programs-are all making good progress. All, however, involve major institutional change which naturally takes years to consolidate. This final section first identifies crosscutting issues to strengthen the government's strategic planning of public expenditure and then makes recommendations for the better management of aid within the government's strategic planning framework. Strategic Planning of Expenditures 121. Important steps to improve the strategic planning of public expenditure have already been taken, through the preparation of the SPs and the beginnings of the MEFF/PIP exercise. The next step is to move ahead with a comprehensive MTEF *because strategic planning of expenditure cannot take place within an annual budget cycle. Annual budgets need to be drawn up within a medium-term perspective. Moreover, the multi-year planning and programming of expenditures needs to take place at each level of Ethiopia's decentralized system, where not only federal public bodies but regions and zones need indicative figures for the forward planning of expenditures. These requirements have already been recognized within the government's proposal for a fiscal calendar, which provides for multi-year planning and programming in advance of the drawing up of an annual fiscal plan and the detailed preparation of annual budgets. To operationalize the medium-term planning of public expenditure, the proposed financial calendar needs to be adopted now, with the following corollary initiatives: + Broad acceptance of the pivotal importance of the MEFF and its timely adoption (including discussion at cabinet level). * Systematic collaboration between MoF and MEDaC in preparing the MEFF and proposing indicative planning figures. The PIP design team has made proposals for the establishment of an appropriate inter- 47 Limited computerization makes it even harder to achieve the required velocity of circulation of funds.

71 Capacity Building In Public Finance Management 57 ministerial working group, which would also involve the Central Bank. * Strengthened macroeconomic modeling capacity to be able to run consistent scenarios on revenues and expenditures, and to be able to adjust these scenarios as events unfold. In preparing the MEFF, the federal government should aim to provide reliable projections of the federal subsidy to regions, so that regions, too, can prioritize their activities taking account of resource availability Expansion of the PIP into a fully-fledged multi-year PEP by providing recurrent as well as capital expenditure guidelines in the MEFF. When the PIP is extended to the regions, sector planners should be required from the outset to prepare consistent medium-term plans for both recurrent and capital expenditure, and regional and zonal councils should be required to undertake joint reviews of recurrent and capital budgets for each sector. Aid Management to Support Strategic Expenditure Planning AidAgency Collaboration 122. The collaboration of aid agencies is essential to achieve more rational planning and utilization of public expenditures: * Aid Projections. Credible projections of aid flows are an essential input into the MiEFF, and are needed for three or four years ahead, not just for the imminent budget year. An essential foundation is for MEDaC to maintain, with donor assistance, a reliable and comprehensive database of ongoing aid projects and programs, which can be clearly mapped into the SPs that have been agreed, as well as into the PIP and successive annual budgets. In addition, the government and donors need to collaborate in making projections of aid flows that extend beyond existing firm commitments so as to provide a reasonable basis for expenditure planning. * Notification of Aid Commitments. Against this background of better medium-term aid projections, implementing agencies also need confirmation of specific aid commitments for the coming budget year with sufficient notice to take them into account for budget preparation 48 The Budget Reform Design Team has also pointed out that, to enable regions to proclaim their budgets on time, it may be necessary to appropriate the federal subsidy separately and in advance of the rest of the federal budget. (See Budget Reform Design Manual, Version 2.0, Annex 3.)

72 58 Chapter S (by end-december if the proposed Financial Calendar is to be observed). + Aid Focus. Aid agencies and the government together should ensure that aid, through whichever channel of disbursement, is allocated only to the priorities defined in the government's strategic planning documents (SPs, PIP, etc). * Aid Modalities. The government should continue to encourage aid agencies to move towards unearmarked Channel I support. This is not simply a matter of exhortation. The government needs to address aid agencies' genuine concerns about accountability, reporting and dialogue. 49 At the same time, aid agencies should recognize that the earmarking of funds has high costs in additional administration and delay. Offset and Regional Expenditure Ceilings 123. The offset system continues to be an obstacle to the rational utilization of aid. Without departing at all from the principle of regional equity, the following improvements could be adopted: * The planning bodies' projections of aid flows should be published when ceilings are announced, so that the offset is transparent. 50 In subsequent years, these projections should be compared with actual aid flows, and, if there are significant divergences, adjustment should be made. The aim should be to achieve equity in the actual flow of resources to regions (and within regions to zones) not just in the planned resource flows. * Ceilings for Treasury funds and for aid should be clearly separated, and it should be made clear that if regions are unable to utilize available aid in a given budget year, it will normally remain available in subsequent years. 5 ' If implementing agencies can clearly see that 49 These issues will have additional relevance when the modalities for HIPC debt relief are negotiated. 50 The budget refonn design team has already made a recommendation along these lines. The Budget Design Manual, Version 1.0: 7.22 includes a recommendation to: require that the regional affairs department of MEDAC document the projects it has considered in adjusting a region's subsidy. Documentation would include the project agreement and proposed disbursements for the upcoming financial year. 51The standard formats for budget call circulars in Version 2.0 of the Budget Reform Design Manual already stipulate that: The ceiling for your Capital Budget Request..has four types of finance: (i) Treasury; (ii) Retained revenue; (iii) Assistance; (iv) Loan. Your request should strictly conform

73 Capacity Building In Public Finance Management 59 aid is additional, and can be drawn down over a practical time frame, they should be ready to invest the additional effort needed to utilize aid funds. 52 * The sharing of experience from other countries, who are operating some type of off-set, may be a useful illustration of options. Implications for Future PER Work 124. There are implications both for the scope and for the timing of future PERs. As regards timing, it is proposed to undertake future PERs much earlier in the fiscal year (so as to support the planning phase, rather than the budgeting phase, of the financial calendar). As regards their scope, future PERs be important in reconciling government and donor estimates of aid flows, and could help extend the sector dialogue associated with the SPs into a broader dialogue on overall public expenditures and enhancing the poverty impact of government and donor assistance. This could support further progress towards provision of aid in the form of unearmarked budget support. to these ceilings, and ceilings earmarkedfor one type offinance cannot be transferred to other types. A similar provision is in the call circular for the recurrent budget. 52 A similar point is made in the regionalization Study, 12.22:

74

75 CHAPTER 6: PER PROCESS The 1999 PER aims at adopting additional recommendations from the Guidance on Public Finance Management agreed by the Special Program of Assistance for Africa. Since 1997, the PER has already been carried out in partnership with the government and donors. The next move is to refocus the PER from a product-driven exercise (focused on the final report) to a process-driven exercise which is fully owned and led by the government. Therefore, the 1999 PER aims at initiating a continuous PER follow-up process, better linked with the budget cycle, that ensures the follow-up to PER recommendations, and improves the aid dialogue. It is proposed to undertake future PERs much earlier in the fiscal year to support the planning phase, rather than the budgeting phase, of the financial calendar. As regards scope, it will be necessary to shift the focus of the next PER from budget inputs to the effectiveness and efficiency of public spending. This aspect will become particularly relevant in the context of Ethiopia's envisaged participation under the HIPC debt initiative The government and donor teams have agreed that the annual PER should evolve into a coordination exercise running in parallel with the different stages of the annual planning and budgeting cycle. The government started in this new direction by sharing the May mission's Aide Memoire with the donor community and discussing some of the preliminary findings in a meeting chaired by the Vice Minister of Finance. This sharing of information was welcomed by the donor community. Subsequently donors prepared a note with issues for follow-up, comments on the findings of the PER, and suggestions for the final report. This very constructive dialogue provides a good basis for a continuous PER follow-up process With a view to ensure a systematic dissemination of the PER findings and action on the recommendations made, the government and the PER team agreed in principle-at the end of the main mission in May 1999-on subsequent activities, consultations, and events. This systematic follow-up should help strengthen the dialogue between the government and its external partners on expenditure priorities, aid allocation and management. The idea to organize one or more follow-up workshops a few months after the completion of the PER mission was welcomed by all parties involved More recently, the government's and the donors' PER teams explored the options for holding two workshops in early fall to focus on the recommendations and findings of the 1999 PERReport.

76 62 Chapter 6 * A two-day workshop to be held in Ethiopia, in September, 1999, with participation by officials from MoF, MEDaC, sectoral ministries (including Health, Education, Transport), and the regions A smaller workshop in Brussels, with participation of a high level Ethiopian Delegation (mainly MoF and MEDaC) at the end of October The September workshop in Ethiopia-besides presenting and discussing the conclusions and recommendations of the draft PER-report-will focus on the linkage between policy formulation, planning, and budgeting. The timing of this workshop is opportune since it coincides with the preparation of the MEFF and the IPF. Both, MEFF and IPF are inputs for the PIP, which will be prepared around December 1999 for presentation to the Council of Ministers. The workshop participants would discuss the current planning process and identify options for improvement in the following areas: + linkage between macro and sector expenditure targets; * linkage between the central and line ministries budget planning process; - information flows between the center and the regions and vice versa; and * information flow between the government and the donors and vice versa The proposed topics address a broad-based audience. In addition to the govemment's PER team, led by the Vice-Minister of Finance, it will be useful to secure participation of other main players from MoF, MEDaC, and line ministries, as well as regional authorities. It would be ideal to have a good representation from the PFP team as well as the PIP team. From the donors' side, one would expect the PER team (about 10 participants), basically operating as a resource team, to be available for inputs in the discussions. As to a wider involvement of the donor community represented in Addis Ababa, it may be preferable to limit the donor participation in the two-day workshop to those who collaborated on the PER team. This will allow the government to be the main beneficiary of the workshop and will facilitate frank and open discussions among government players. For a broader donor information, it could be useful to schedule a half-day briefing on the outcome the day after the workshop. Like the discussion of the Aide Memoire, the Vice-Minister of Finance may wish to chair this workshop. 53 The outcome and agreements of the Debre Zeit workshop are attached in the Annexes. 54 The outcome and agreements of the Brussels workshop are attached in the Annexes. The Brussels workshop served as the launch meeting for the 2000 PER.