Report No. 23351 -ET Ethiopia Focusing Public Expenditures on Poverty Reduction (In Three Volumes) Volume 1: Main Report December 20, 2001 World Bank Country Office in Ethiopia Country Department 6 Africa Region Document of the World Bank GOVERNMENT FISCAL YEAR July 8 - July 7 CURRENCY EQUIVALENTS Currency Unit: Ethiopian Birr (Br) Official Rate: US$1.00 - Br 8.56 (December 18, 2001) ABBREVIATIONS AND ACRONYMS ADB African Development Bank MEDAC Ministry of Economic Development and Cooperation ADLI Agricultural Development Led Industrialization MEFF Macroeconomic and Fiscal Framework ARM Annual Review Meeting MoF Ministry of Finance CAD Central Accounts Department MoE Ministry of Education CBE Commercial Bank of Ethiopia MOH Ministry of Health CBO Community Based Organization MOWR Ministry of Water Resources CDF Comprehensive Development Framework MTEF Medium-Term Expenditure Framework CFAA Country Financial Accountability Assessment MTR Mid Term Review CJSC Central Joint Steering Committee NBE National Bank of Ethiopia CPAR Country Procurement Assessment Report NGO Non-Govenmmental Organization CRDA Christian Relief and Development Agency O&M Operation and Maintenance CSA Central Statistical Authority OECD Organization for Economic Cooperation and Development CSAE Center for the Study of African Economies PAP Program Action Plan CSRP Civil Service Reform Program PIM Program Implantation Manual DAG Development Assistance Group (of Donors) PEM Public Expenditure Management DFID Department for International Development (UK) PEP Public Expenditure Program DHS Demographic and Health Survey PER Public Expenditure Review DPPC Disaster Preparedness and Prevention Commission PIP Public Investment Program DSA Decentralization Support Activity PMO Prime Minister's Office ERSC Economic Rehabilitation Support Credit PPA Participatory Poverty Assessment EC/EU European Commission/European Union PRA Participatory Rural Appraisal EFY Ethiopian Fiscal Year PRGF Poverty Reduction and Growth Facility EMCP Expenditure Management and Control Program PRSC Poverty Reduction Support Credit EPRDF Ethiopian People's Revolutionary Democratic Front PRSP Poverty Reduction Strategy Paper ERA Ethiopian Roads Authority RHB Regional Health Bureau ESDP Education Sector Development Program RPER Regional Public Expenditure Review ESRDF Ethiopian Social Rehabilitation and Development Fund RRA Regional Roads Authority FIS Financial Information System RRTP Rural Roads and Transport Program FY Fiscal Year RRTS Rural Roads and Transport Strategy GDP Gross Domestic Product RSC Regional Steering Committee GFS Government Financial Statistics RSDP Road Sector Development Program GNP Gross National Product RTTP Rural Travel and Transport Program GoE Govermnent of Ethiopia SDP Sector Development Program HICES Household Income, Consumption and Expenditure Survey SNNP/R Southern Nations Nationalities and Peoples/Region HIPC Heavily Indebted Poor Countries TA Technical Assistance HMIS Health Management Information System TOR Terms of Reference HSDP Health Sector Development Program TWG Technical Working Group IDA Intemational Development Association UNCITRAL United Nations Commission on Intemational Trade Law IDF Industrial Development Fund UNDP United Nations Development Program IFMS Integrated Financial Management System USAID United States Agency for International Development IMF International Monetary Fund VAT Value Added Tax IPF Indicative Planning Figure VLTTS Village Level Travel and Transport Study l-PRSP Interim Poverty Reduction Strategy Paper WDC Woreda Development Committee JRM Joint Review Mission WMS Welfare Monitoring Survey LDC Less Developed Country WMU Welfare Monitoring Unit Vice President: Callisto Madavo Acting Country Director: Prafiil C. Patel Sector Manager: Frederick Kilby Task Team Leader: Duvvuri Subbarao PREFACE This Public Expenditure Review for 2001 entitled Ethiopia - Focusing Public Expenditures on Poverty Reduction - is a collaborative effort between the Government and a multi-donor team. From the Government side, the task was led by Ato Hailemelekot T/Giorgis, Vice Minister of Finance until early October 2001, and by his successor Ato Getachew Gebre, State Minister, thereafter. Ato Demelash Alem of Ministry of Finance handled the logistics and coordination. From the donor side, the task was coordinated by the World Bank with Duvvuri Subbarao (AFTM2) as the Task Team Leader. The topics for this review were determined by the Government in consultation with a core group of donors at a workshop in Brussels organized by the European Commission in November 2000. Among other topics, the Government suggested reviewing public expenditures at the regional level, and for three regions to start with. Thus this PER marks the first explicit attempt at Regional PERs. It was also decided at the workshop to have each topic led and managed by a different donor agency. The institutions involved, apart from the World Bank, were 1MF, ADB, EC and DFID-UK. The Government also nominated counterpart officials from the Ministries of Finance and MEDaC (since merged into a single Ministry of Finance and Economic Development) to work with the donor teams. The staff/consultants who worked on each of the topics is listed at Annex A at the end of this volume. The mission work for the various sub-tasks was carried out during January-April 2001 and the draft reports were ready by end August 2001. The review of fiscal performance was done during a joint Bank-Fund mission in September 2001. A workshop was held in Addis Ababa during 17- 18 September to discuss the analysis and findings of the draft reports. This integrated report draws from the draft reports as well as the discussion at the workshop. It has been put together by the Task Team Leader. The very generous and competent help from Stephen Lister (DFID Consultant) and Eyerusalem Fasika (AFTM2) in doing so is gratefully acknowledged. This report is being issued in three volumes. This first volume is the main report. The second volume contains Appendices and data tables. The third volume is the Oromia PER which was task managed by the World Bank (by Giovanna Prennushi of PRMPR). A draft version of this set of reports was discussed with the Government in November 2001. Comments of the Public Expenditure Thematic Group of the Development Assistance Group (DAG) were also invited on the draft. This task was carried out under the general guidance of Oey Astra Meesook, Country Director (AFCO6) until early October 2001 and by her successor Praful Patel thereafter. Frederick Kilby was the Sector Manager and Surjit Singh, the Country Office Manager. Benno Ndulu (AFTM2) and Anand Rajaram (PRMPS) were the peer reviewers. The extremely competent assistance from Fekerte Getachew of the Country Office and Roboid Covington, Patrick Mamboleo, and Tanisha McGill at Washington Headquarters by way of document processing and production is gratefully acknowledged. CONTENTS PR E FA C E .............................................................. I 1. OVERVIEW AND SUMMARY .............................................................1 I. CONTEXT .I II. RATIONALE AND COMMON UNDERLYING THEME .3 III. PER PROCESS .4 IV. SUMMARY OF TOPICS COVERED .6 V. REFORMS IN PUBLIC EXPENDITURE MAMAGEMENT .12 VI. PRIORITIES FOR FUTURE WORK .14 2. REVIEW OF FISCAL PERFORMANCE ................................... 15 1. FISCAL TRENDS .15 II. MEDIUM TERM FISCAL FRAMEWORK .17 III. THREE YEAR FISCAL REvIEW 1999-2002 .20 IV. ECONOMIC AND FUNCTIONAL CLASSIFICATION .21 V. FEDERAL AND REGIONAL SHARES OF EXPENDITURES .26 VI. SUSTAINABILITY OF CURRENT FISCAL STANCE .30 3. TRACKING PUBLIC EXPENDITURES FOR POVERTY REDUCTION ............................... 31 I. INTRODUCTION .31 II. TRACKING OF PUBLIC EXPENDITURES .32 III. POVERTY AND WELFARE MONITORING IN ETHIOPIA .36 IV. SECTOR EXPENDITURES AND POVERTY .42 V. INCREASING THE POVERTY FOCUS OF PUBLIC EXPENDITURES THE WAY FORWARD . 51 4. DONORS SHIFTING TO SECTOR AND BUDGET SUPPORT ................................................ 57 I. INTRODUCTION .57 II. CONCEPrs AND ISSUES IN AID MANAGEMENT .58 III. AID MANAGEMENT EXPERIENCES IN ETHIOPIA .61 IV. HIPC INITIATIVE: REQUIREMENTS AND OPPORTUNIES. .66 V. THE WAY FORWARD AND IMPLICATIONS FOR PRSP .68 5. PUBLIC EXPENDITURE - REGIONAL PERSPECTIVE ................................................ 75 I. INTRODUCTION .75 Il. FISCAL FEDERALISM - MAIN FEATURES AND DIMENSIONS ........................................................... 76 111. COMMON THEMES AND ISSUES FROM THE RPERs ........................................................... 87 IV. ADDITIONAL ISSUES AND CONCLUSIONS .......................................................... 98 ANNEX A ..........................................................1 02 BOXES Box 2.1: Tax Policy Reforms .......................................................... 19 Box 2.2: Ethiopia: Defense and Poverty-Targeted Outlays, 1997/98 - 2001/02 1/ .................................... 24 Box 3.1: Explanatory Power of Welfare Monitoring Data ........................... ............................... 39 Box 3.2: Reconciling Evidence on Poverty ........................................................... 40 Box 3.3: Regional Planning and Local Poverty Profiles ..................... ..................................... 44 Box 3.4: Capacity Building - the Role of the Civil Service Reform Program ............................................ 53 Box 3.5: A Paradigm for HIPC Expenditure Planning .......................................................... 54 Box 3.6: Setting Targets for Expenditure Shares ......................................................... 55 Box 4.1: Aid Disbursement Channels ......................................................... 59 Box 4.2: Some Issues and Paradoxes in Aid Management .......................................................... 65 Box 4.3: Proposed HIPC Disbursement Mechanism for Ethiopia ......................................................... 69 Box 4.4: PER 2000 Recommendation on Government/Donor Strategic Dialogue .................. .................... 71 Box 5.1: Goals and Objectives of the Regional Public Expenditure Reviews .............................................. 88 TABLES Table 2.1: Ethiopia - Fiscal Trends, 1986/87-2001/02 ................................................................. 16 Table 2.2: General Govermnent Expenditures - Economic Classifications 1998/99 - 2001/02 ................... 22 Table 2.3: General Govermment Expenditures - Functional Classifications 1998/99 - 2001/02 .................. 23 Table 2.4: Recent Trends in Social Sector Expenditures 1998/99 - 2001/02 .............................................. 26 Table 2.5: Real Per Capita Social Sector Expenditures 1998/99 - 2001/02 .................................................. 26 Table 2.6a: Federal and Regional Shares in Recurrent Expenditure, 1997 - 2001 ....................................... 28 Table 2.6b: Federal and Regional Shares in Capital Expenditure, 1997 - 2001' ......................................... 28 Table 2.6c: Federal and Regional Shares in Total Expenditure, 1996 - 2001 1 ............................................ 29 Table 4.1: Debt and Aid 1990 and 1998 for Selected Developing Countries .............................................. 57 Table 4.2: Summary of Government Finance 1/ ................................................................. 62 Table 5.1: Overview of Ethiopia's Regions ................................................................. 77 Table 5.2: Horizontal Imbalance Across Regions ..................... ............................................ 80 Table 5.3 Shift in Vertical Imbalances Over Time, 1997 - 2001' ................................................................. 81 Table 5:4: Shares of Federal and Regional Governments in Revenues 1997 - 2001 .................................... 83 Table 5.5: Federal and Regional Shares in Recurrent Expenditure, 1997 - 2001 ....................................... 84 Table 5.6: Regional Per Capita Expenditure FYOO/01 ' ..................................... ............................. 86 1. OVERVIEW AND SUMMARY 1.1 The end of the border conflict in June 2000 saw a rapid fiscal reorientation in Ethiopia. Supported by an Economic Rehabilitation Support Credit (ERSC) from the World Bank and a three-year arrangement under the IMF's Poverty Reduction and Growth Facility (PRGF), the Government formulated, and started implementing, a new economic program. The medium term policy framework agreed under the ERSC and PRGF supported program rests on maintenance of macroeconomic stability, laying the foundation for a strong tax revenue performance, redirection of resources from defense-related expenditures toward poverty-reducing outlays, strengthening public expenditure management, and improving decentralized frontline service delivery. This Public Expenditure Review (PER), the eighth in a continuous annual series since 1994, begins with a review of the fiscal situation. 1.2 The standard format of the PERs has been to supplement the fiscal review by focusing on some issues of current relevance and of mutual interest to the govermment and its external development partners. The theme and the focus topics for this PER emerged out of consultations between GoE and a core donor group at a workshop in Brussels organized by the European Commission (EC) in November 2000. There was consensus at the workshop that the content and process of this PER must be informed by three premises: (i) the PER must thematically be set in the context of PRSP and HIPC; (ii) given that over two-thirds of the anti-poverty expenditure is administered through the regions, the PER must focus on getting a better understanding of the budgetary institutions, systems and processes at the regional level; and (iii) the PER analysis and recommendations must harmonize with the sequence and pace of Government of Ethiopia's (GoE's) Expenditure Management and Control Program (EMCP). This PER exercise has made a conscious attempt to pursue that consensus. 1.3 This overview chapter (i) outlines the context in which this PER exercise has been carried out; (ii) explains the rationale for the topics addressed and the common theme running through them; (iii) describes the PER process; (iv) summarizes the issues and findings under each of the topics; and (v) takes stock of the reforms in public expenditure management (PEM) in the context of recommendations of the past PERs. A final section summarizes the way forward. I. CONTEXT 1.4 Political context: The EPRDF Government that first assumed office in 1991 was re- elected to a five-year term in the parliamentary elections held in May 2000. Even as there was widespread support for the Government from across the political spectrum during the border conflict, the signing of the peace accord in December 2000 was followed by vigorous, and occasionally violent, political dissent that resulted in crack-downs, arrests and defections. Party hardliners were particularly critical of the Government's policies on liberalization and globalization and its liberal attitude towards Eritrea. In order to contain discontent and give a democratic voice to dissenting opinion, the EPRDF conducted organized, structured consultations with party cadres during June-September 2000 which culminated in the Organizational Congress of the Party. In October 2001, the Prime minister reorganized his government and reconstituted his cabinet. Changes/decisions important in the PER context are the merging of the Ministry of Finance (MoF) and Ministry of Economic Development and Cooperation (MEDaC) into a single Ministry of Finance and Economic Development, the creation of a new Ministry of Revenue, decision to roll over reforms under the EMCP to the regions on an accelerated time-table and capacity building, especially at the decentralized levels. 1.5 Economic context: The fiscal erosion on account of the preemptive demands of defense expenditure was the main thrust of the last PER. Post-conflict, the Government is refocusing attention on reforms and fiscal consolidation. Macroeconomic performance in FY01 was quite satisfactory. Real GDP is estimated to have grown at 7.9 percent, inflation turned negative, reflecting the combined effects of a bumper cereal crop and large inflows of food aid, and the external current account balance declined to 4.9 percent from 5.2 percent of GDP in FY00. Coffee exports (which accounted for half the merchandise exports) were lower than expected as both volumes and prices fell. With slower than anticipated pace of project implementation, imports were also lower than projected. Aided by structural adjustment support from the IMF, the World Bank, ADB and the European Commission, foreign exchange reserves have been built up to provide a three month import cover. Fiscal performance in FY01 was marked by implementation of several tax reform measures and roll back of defense expenditures to 6.3 percent of GDP. Ethiopia, like other African countries, will be affected by the post-September global crisis. Preliminary estimates indicate further deterioration in terms of trade (higher oil and lower coffee prices) and lower remittances from the diaspora. The resultant current account losses of the order of USD44 million will completely absorb the projected interim debt relief under the enhanced HIPC. 1.6 PRSP and HIPC: Soon after the signing of the peace agreement, the Government quickly formulated its I-PRSP (January 2001) which reaffirms its commitment to poverty reduction within a framework of macroeconomic stability. The I-PRSP agenda includes (i) agriculture led development strategy; (ii) judiciary and civil service reform; (iii) decentralization and empowerment; and (iv) capacity building in public and private sectors. GoE is now in the process of formulating its full PRSP which is expected to be ready by April/May 2002. Ethiopia's case for a Decision Point under the enhanced HIIPC was approved by the Boards of the World Bank and IMF in November 2001. After the Completion Point, actual average debt relief to Ethiopia will be of the order of USD90 million per year on current dollar terms, equivalent roughly to 20 percent of the annual net foreign assistance (loans and grants) disbursed during FY98. The PER is expected to play an important role in evaluating the Government's performance under the PRSP and utilization of IIPC resources. 1.7 Decentralization: Ethiopia's fiscal decentralization is a complex hybrid of radicalism and centralist control. Its institutions of federalism are still evolving. As is characteristic of most federations, in Ethiopia too there are vertical and horizontal imbalances. There are wide disparities in the fiscal bases of the regions (the horizontal imbalance), and the regions' expenditure responsibilities far outstrip their own revenues (vertical imbalance) making them heavily dependent on federal fiscal transfers. Ethiopia is still grappling with the issue of fiscal decentralization with a variety of initiatives to accelerate overall development while ensuring inter-regional equity. In his recent statement to the parliament (October 2001), the Prime Minister said that decentralization (along with governance and capacity building) will be one of the main pillars of their development strategy.' Proposed initiatives include constitutionally mandating certain responsibilities to the regions (primary education, primary health, and rural roads), untied block grants to the regions, rolling over public expenditure reforms to regions on an accelerated schedule and improving the own revenues of the regions. This PER marks the beginning of an explicit regional focus in analyzing public expenditure management. ' It is expected that this development strategy will be more fully articulated in the full PRSP, now under formulation. 2 1.8 Aid relations: Relations with aid partners, which were strained during the border conflict, are beginning to repair, albeit slowly. Many donors, particularly bilaterals have been circumspect in returning to normal aid levels until there is an improvement in the Government's human rights record. The PRSP consultation process, which is now under way, is expected to pave the way for inspiring donor trust and confidence in the Government's development strategy and its sincerity of propose, and create a platform for increasing the levels of aid. Over the last year, the Government negotiated a macroeconomic program under the IMF's PRGF. Under an Interim Support Strategy, the World Bank committed fresh credits of the order of USD670 million for repair of war torn infrastructure, rehabilitation of war affected people, demobilization and resumption of economic reforms interrupted by the war. In May 2001, the EC released Euro9O million from its SAS/Stabex fund to aid the post-war reconstruction effort. Over the last few years, the PER provided the main platform for dialogue between the Government and the donors on public expenditure issues which has complemented the sector-specific dialogue under the SDPs. As the PER deepens its focus on the links between public expenditure and poverty reduction, it may well become the main determinant of donor perceptions of the Government's performance under its poverty reduction strategy. II. RATIONALE AND COMMON UNDERLYING THEME 1.9 Following this Overview chapter, the four topics covered in this report are: Review of Fiscal Performance (chapter 2); Tracking Public Expenditures for Poverty Reduction (chapter 3); Donors Shifting to Sector and Budget Support (chapter 4); and Public Expenditures - Regional Perspective (chapter 5). The common theme underlying all these topics is an inquiry into the prospects and problems for reorientation of public expenditures for poverty reduction in the context of Ethiopia's transition from war to peace, formulation of the full PRSP, and utilization of savings under the enhanced HIPC Initiative. 1.10 Thus the chapter on fiscal performance reviews the shift in public expenditures from defense to poverty reducing outlays, looks forward to the medium fiscal framework that should inform the PRSP, analyzes the trends in social sector expenditures as well as the shifting balance in federal-regional expenditure shares, and comments on the sustainability of the current fiscal stance. 1.11 Both the PRSP and the enhanced HIPC initiatives emphasize the expansion of poverty- related public expenditures to reduce poverty. Together they seek to tilt the composition of public spending toward poverty-related allocations; improve the efficiency and effectiveness of public spending; and use debt relief to finance pro-poor activities. An important prerequisite for this is that systems be in place for tracking the flow and impact of public spending. Tracking of expenditures is particularly complex in Ethiopia because of the depth of decentralization, the absence of a uniform and integrated financial information system across the vertically decentralized system, predominance of aid in the total resource envelope and severe capacity constraints. The chapter on 'tracking' evaluates the existing systems for tracking expenditures, monitoring poverty and assessing the impact and outcomes of public expenditure, and makes recommendations for improvement. 1.12 The PRSP demands that the government take a medium term view of its public expenditure allocations in order to make the decisions more poverty sensitive. Since, donors are partners in the PRSP process, they should accept an obligation: (i) to make medium-term aid pledges in conformity with the poverty reduction strategy; and (ii) to shift from traditional project aid to program/budget support. For a variety of political and administrative reasons, this is a difficult ideal to achieve in practice; nevertheless, it is a worthwhile goal to pursue. What are the key actions required of both the government and the donors to travel in this direction? That is the question addressed by the chapter on 'Donors Shifting to Sector/Budget Support'. 3 1.13 Ethiopia's Constitution mandates a federal structure with considerable autonomy devolved to the regions in administrative and fiscal matters. Although the share of the regions in the combined federal-regional revenue is only about 15-20 percent, they account for nearly 40 percent of the total expenditure. When only the anti-poverty intensive sectors such as health and education are considered, the regions' share in the combined expenditure goes up to as much as 60-70 percent. In the PRSP context, the regions are, therefore, positioned at the cutting edge in the delivery and monitoring of anti-poverty expenditures. Yet there is relatively limited documented analysis of the fiscal situation or the budgetary institutions and processes at the regional level. PER 2000 was the first to make a systematic effort to include a regional focus. This PER advances the agenda by studying the budgetary systems and processes in three pilot regions. An attempt has also been made to identify to what extent poverty impact informs regional public expenditure management. II. PER PROCESS 1.14 Over the last few years, a conscious attempt has been made to make the PER a more collaborative and consultative process. Most notably: (i) the agenda and focus topics were determined by GoE after discussion with a core donor group in the EC sponsored workshop in Brussels in November 2000; (ii) the Concept Paper was shared and discussed with the Government and donors (Development Assistance Group-DAG) before it was put to internal review within the Bank; (iii) the Government nominated officials from MoF and MEDaC to act as counterpart team members; and (iv) both the DAG and the higher management in MoF were kept updated on the status of the report and findings throughout the process. 1.15 However, the Government's involvement tended to remain largely reactive rather than proactive. The Government's, and more particularly MoF's, interest in the PER is likely to increase significantly in the near future for a number of reasons. First, the previous PER provided the fiscal underpinnings for defining some of the policy measures in the I-PRSP. Second, the PER was one of the main instruments in defining the policy matrix under the ERSC, and is likely to play an incensing role in that direction in the context of the proposed Poverty Reduction Support Credits (PRSC). Third, the analysis in the PER informed the formulation of the HIPC Completion Pint triggers. And finally, Government have indicated that the annual PER, rather than a new product, should be the basis for evaluating their performance in the implementation of the PRSP and utilization of HIPC resources. 1.16 Although regional officials were involved in the previous PER process (by way of mission visits, dissemination workshop), this PER marks the first attempt at Regional Public Expenditure Reviews (RPERs). Since MoF identified the pilot regions for this purpose at the Brussels Workshop, presumably without prior consultation with the regions, the regions were unprepared for the exercise. In order to take them on board at an early stage, a brainstorming workshop was held in February 2001 involving federal and regional officials to begin what turned out to be a very constructive process of engagement. The objectives of the workshop were to: (i) to get inputs from regional representatives into the proposed content and process of the RPERs; and (ii) to harmonize the methodology, content and structure of the three separate studies. In response to a specific request, all three regions established Regional PER Working Groups to act as the main counterpart team for the exercise. 1.17 In a departure from the past practice of a combined PER team and a single mission, this PER cycle experimented with the a new model assigning each sub-task to be led and managed by a specific donor agency by fielding staff and/or consultants operating under the umbrella coordination of the World Bank (see Annex A for details). Also, the practice in the past was to hold a dissemination workshop after issuing the gray cover report. Another innovation this year 4 has been to hold a discussion workshop based on the draft reports to make the exercise more participatory. During the Decision Meeting of the PER (December 2001) at the end of the PER process, there was criticism, mainly from donor partners that this approach resulted in a lack of coherence in the process and a fragmented approach. It was noted by way of response that this model has itself evolved in deference to criticism that the previous model of a single large mission was found to be unwieldy and inefficient. It was agreed that efforts should continue to be made to find an optimal approach that is effective, practical and inclusive. 1.18 The Discussion Workshop, organized jointly by the MoF and the World Bank and held in September 2001, brought together a total of 127 participants representing a wide range of stakeholders and provided a conducive forum for brainstorming on the PER. Participants comprised government officials from the federal ministries of Finance, MEDAC, Education, Health, Agriculture, Transport and Communication, officials from all the regions drawn from the Finance, Planning and Sector Bureaus as well as a large cross-section of the donor community. The reach of the workshop, and thereby of the PER, beyond the traditional boundaries of government, was made possible by the presence of members of Parliament and the Regional Councils, the academia (Addis Ababa University, Economic Development Research Institute, Ethiopian Economic Policy Research Institute) and media (print and electronic). The workshop generated stimulating discussion on a number of issues central to the PER such as the current status and way forward on public expenditure reforms, tracking and evaluation of public expenditures, and ways and means of making the PER a more useful and relevant exercise for all stakeholders. This report thus draws from the draft reports as well as the workshop discussion. 1.19 Given that this PER marked the first explicit effort to focus on the regions, the workshop provided an excellent opportunity to brainstorm on the role of regions in the PRSP/HIPC context and hence in the PER process. The Task Manager from the World Bank stressed that in response to past criticism, a conscious attempt has been made to harmonize the PER content and process with the EMCP agenda. The thrust of the PER has shifted from making specific recommendations to providing the analytical underpinnings which could in turn shape the direction, content and sequencing of the EMCP. There was general endorsement to this view. There was difference of opinion, however, on whether the present model of decentralizing the responsibility to participating institutions has been more efficient and effective than the earlier model of a single team and combined mission. Many donors felt that the process should be more inclusive. A concern was expressed about the failure to implement the past recommendation that the PER be synchronized with the planning half (the front end) of the financial calendar. The general refrain among donors was that the government should take a more proactive role in the PER process. They welcomed the government's suggestion of operationalizing the donor-government platform for consultation. 1.20 There was an interesting discussion at the workshop on the evolving role of the PER in the PRSP/H[PC context. There was a view that the PER could substitute for the external evaluation of the government's performance vis-A-vis PRSP targets and of the utilization of HIPC resources. In such an event, the PER has to be entirely an external effort and cannot be done in collaboration with the Government. An alternative suggested was that the PER could be divided into two segments: the first being an external evaluation of fiscal performance against PRSP targets, and the second, a collaborative effort with the Government on issues of mutual concern. It was agreed that this is an issue to be further explored as the PRSP/HIPC process evolves. 1.21 All three RPERs were discussed with the respective Regional Working Groups. However, as of the date of this report, only the ADB went a step further to disseminate the Benishangul-Gumuz RPER at a regional dissemination workshop. Similar workshops in the other two regions are planned to be held before starting the next PER cycle. 5 IV. SUMMARY OF TOPICS COVERED Review of Fiscal Performance 1.22 The Government embarked on fiscal consolidation in earnest soon after the peace agreement was signed. The twin fiscal policy objectives of the Government for the next three years are to redirect resources from defence-related expenditures toward poverty-reducing outlays (while addressing the country's post-conflict recovery needs), and to lay the foundation for a strong tax revenue performance. Preliminary estimates of budget outturn for FY01 indicate that the fiscal deficit (including grants) narrowed to 5.6 percent of GDP, much lower than the 8.4 percent envisaged under the macroeconomic program, mainly as a result of cautious fiscal management, lower than programmed external assistance, and slower than anticipated execution of emergency programs and capital expenditures. In mid-2001, an additional Birr 500 million (0.9 percent of GDP) was reallocated from defence to mostly social outlays. Defence outlays were limited to 6.3 percent of GDP (lower than the programmed 7.3 percent of GDP), while poverty-targeted outlays (health, education, agriculture & natural resources, and roads) are estimated to amount to the equivalent of 10.9 percent of GDP. 1.23 The FY02 budget reflects a continuation of the effort towards fiscal consolidation and expenditure reorientation with much of the results flowing from expenditure compression rather than revenue expansion. Defence outlays are to be further cut from 6.3 percent of GDP in FY01 to 5.1 percent of GDP (ahead of the PRGF program projection of 5.2 percent). The expected debt relief under the enhanced HIPC Initiative will be important to increasing social - and more generally - poverty-targeted expenditures. Preliminary estimations indicate that the possible interim assistance under the enhanced HIPC Initiative provided after reaching the Decision Point could finance additional poverty-targeted expenditures equivalent to around 0.7 percent of GDP in FY02. This estimate is, however, subject to the effects of global recession aggravated by the events of September 11, 2001. 1.24 GoE's financial strategy assumes that a large part of the fiscal deficits in the medium term will be financed with disbursements of highly concessional loans and debt relief. The projected reliance on strong financial donor support takes account of the major challenge Ethiopia is prepared to tackle: rebuilding core development spending from the compressed levels of the recent years and healing the economy from conflict-related damages while sharply rolling back military expenditure. Over the medium term, financial policies and structural reforms, as well as a gradual improvement in Ethiopia's terms of trade, are expected to strengthen the external position, so that project loans would remain as the only debt-creating financing flows. Tracking Public Expenditures for Poverty Reduction 1.25 An evaluation of Ethiopia's budgetary systems against the World Bank and IMF HIIPC checklist is at Appendix I (Vol II). Salient issues for an understanding of these systems are Ethiopia's large size, its decentralization programme, the development of several SDPs, and a comprehensive and home-grown civil service reform programme (CSRP) whose major component is the Expenditure Management and Control Programme (EMCP). The EMCP adopts an evolutionary approach to improving and expanding Ethiopia's unsophisticated but effective budgetary and accounting systems 1.26 Given Ethiopia's strong tradition of expenditure discipline, the main challenge in improving public expenditure management is in obtaining better management information for improved public service delivery. Under the EMCP, cost centres with a common expenditure classification across recurrent and capital budgets are being identified and introduced, at federal level initially. Continuing problems include: (i) reporting systems which serve the purposes of control rather than management; (ii) reporting delays, which reflect difficulties in aggregating 6 reports across an under-resourced decentralized system and (iii) difficulties in including donor funded expenditures. The introduction of a hybrid channel I reporting system has retarded progress in reform, without achieving the expected standards of accountability or speed of disbursement. Progress towards medium term expenditure planning has been less fruitful than budget reforn, with some controversy about the design of this element of EMCP [detailed in Appendix I (Vol II)]. The proposed reform actually contains the essential elements of a medium term expenditure framework (MEFF), its real problem having been its failure to attract the high level political support it needs to be put into operational practice. 1.27 Under the impetus of the PRSP process, the Government is now committed to observing the financial calendar required for medium termn planning. The Government has also initiated the formulation of regional PEPs. The Federal Government will provide advance forecasts of the federal subsidy, so that regions can make three-year expenditure plans within a defined resource envelope. Annual revisions will enable expenditure projections to be reviewed in the light of poverty strategy, and the annual roll-over should also allow medium term plans at both federal and regional levels to become more robust as experience is gained. 1.28 Recent GoE policy documents, particularly the I-PRSP, show increasing awareness of the multi-dimensional nature of poverty. A range of welfare monitoring data addressing the four principal dimensions of poverty [economic opportunities, capabilities, vulnerability and empowerment) and at different levels of disaggregation is available (see Appendix 3, Vol II)]. The Household Income, Consumption and Expenditure Survey (HICES) and the Welfare Monitoring Survey (WMS) are the main sources of data on poverty. They constitute an adequate foundation for welfare monitoring, provide scope for meaningful disaggregation, and allow links between consumption and welfare outcomes to be explored. 1.29 The Welfare Monitoring Unit (WMU) in MEDAC was intended to be the focal point of a coherent welfare monitoring system, but it lacks managerial and analytical capacity and has fulfilled only a small part of its intended role. This has contributed to the fact that to date poverty information has played a very limited role in policy formulation and resource allocation. With the increased demand for effective poverty monitoring generated by the PRSP, both government and donors are keen to revitalize the WMU and other welfare monitoring activities. 1.30 These stakeholders have different interests and expectations, however, which include: (i) the WMU's more quantitative approach versus donor preferences for more qualitative and multi- dimensional poverty analysis; (ii) the WMU's focus on overall welfare trends versus the demand by some donors and NGOs for a focus on the poorest; (iii) the gap between the need for immediate results to feed into the PRSP versus the need to build the capacity of the WMU and (iv) unrealistic expectations on all sides about the extent to which analysis of HICES and WMS data can definitively answer whether economic reforms have worked. 1.31 WMU capacity needs to be built in analysis and research management with a strategy that contains: (i) a plan for analyses aimed at more users, with more emphasis on disaggregation, poverty profiles, and links between different dimensions of poverty; (ii) technical assistance geared towards long-term capacity building; (iii) the development of clear relationships with users and potential users; (iv) a dissemination strategy which aims to stimulate evidence-based debates and (v) flexibility to commission further analysis, e.g. regional or zonal studies, studies that link WMS data to other sources from social sector ministries, studies that provide contextual information in zones with apparent problems, studies aiming to explain certain trends (e.g. changes in enrolments or dropouts), etc. Over the past year, steps have been taken to strengthen the WMU and enhance its capacity along these lines. Sector expenditures and poverty - Health and Education 1.32 Mid Term Reviews (MTR) of Health and Education Sector Development Programmes (HSDP & ESDP) do not identify major fiduciary problems or failure of funds to reach their 7 intended uses, but timely reporting and closing of regional accounts is a problem. Strategies focus on primary services and extending coverage so, prima facie, both sectors have a pro-poor orientation. Both SDPs have a range of indicators and government-donor annual monitoring and review systems. It is difficult to track the impact of expenditure, however, because the SDP component categories do not match the budget classification. This should be resolved by the development of cost centre budgets. 1.33 Both MTRs comment on the burdens imposed on limited capacity by having to follow donor-specific requirements, including Channel lb. Reporting mechanisms designed to facilitate within-year reimbursement of funds have hampered progress in bringing regions' regular accounts up to date. Regions are handicapped when donor funds are not allocated in advance of the fiscal year when expenditure is to take place; and have been similarly handicapped by late notification of the federal subsidy. Five-year expenditure programmes, developed jointly by federal and regional bodies, demonstrate that coherent sector expenditure programmes can be developed. In practice, regional strategies are quite tightly specified by federal policies and standards and the MTRs argue that regions should have more latitude to adapt policies and expenditure pattems to local circumstances. 1.34 The impact of SDPs on poverty depends on the distribution of expenditure within each sector, as well as on the volume of expenditure to the sector. Generally, sector planning has over- emphasised access (particularly the physical expansion of services), a trend that is supported by donor preferences for capital projects. This has sometimes compromised service quality, however, with knock-on effects on equity: lower quality reduces benefits to service users, and also reduces usage of services, especially by poorer people. Periodic user satisfaction surveys can be a useful-means of checking the relative importance of access and quality aspects of services. 1.35 The SDPs provide lessons on capacity building for better expenditure tracking and the more effective use of resources to address poverty: These include (i) more resources should be committed to capacity building; the ESDP MTR argues that 10 percent of donor resources should be used for this; (ii) it is as important to restrict demands on limited capacity as to increase that capacity; for example, there is currently considerable duplication in reporting and review activities in both SDPs; (iii) capacity building strategies must be long term; in particular, rolling out reforms to the regions where expenditure and implementation happen, will take several years; and (iv) capacity building includes changing decision-makers thinking, e.g. to encourage strategic choices across recurrent and capital budgets, and increasing their awareness of issues of quality, access, usage, and poverty impact. It is also important to encourage proactive measures to involve communities in managing health and education facilities. Sector expenditures and poverty - Roads 1.36 The potential impact of roads on poverty includes effects on economic growth, on equity, on reduced vulnerability, and on income opportunities from labor intensive road construction and maintenance. Neither the appropriate level of expenditure on roads, nor the best balance between different components of road expenditure is simple to determine: While there is evidence that it is rural feeder roads which create greater economic opportunities for the poor, these roads must connect to a main road network to have any real impact. 1.37 The bulk of expenditures under the first phase of the Road Sector Development Programme (RSDPI) have taken place at the federal level, but RSDP 11 (2002-2007) stresses the development of rural roads. Monitoring federal road expenditure has been relatively straightforward but it may be more difficult to track spending under rural roads programmes which have multiple sources of funds. Monitoring of RSDPI performance has been through 16 indicators, none of which is a reliable measure of the outcome of road expenditures on welfare. A more rigorous focus on how spending on rural roads might benefit the poor would require relatively sophisticated, and therefore expensive, surveys. 8 1.38 Improving roads is only one way of reducing the transport burden. Under the Rural Roads and Transport Programme (RRTP) funds will be made available at village level to support locally appropriate and locally determined solutions to transport problems. The RTTP has outcome indicators which assume an expansion of the low level road network related to the contribution of new facilities (water points etc.) and of the availability of non-traditional means of transport. It will not be easy to track expenditures under the RTTP and to assess their poverty impact because: (i) it is hard to develop an indicator for the transport burden, which also takes account of the effect of siting facilities closer to villages or of the cost of transport; (ii) the RTTP has multiple sources of funding; (iii) the large number of small projects within the RTTP; and (iv) the need for village level surveys to monitor the average hours walk from a road. It should be enough to review a small sample of projects to verify the programme's assumptions, however. 1.39 There is a direct trade-off between the requirements of federal capital expenditure programmes and the funds available for the regional subsidy, and hence for other sectors dominated by regional spending. The PRSP will need to propose a sustainable balance between sectoral (and hence between federal and regional) expenditures. The RSDP has a backlog of road rehabilitation; this implies that expenditure on roads will be above its long-run trend for a period. Once the backlog is made up, roads may attract a smaller share of public expenditures. Increasing the poverty focus of public expenditures - the way forward 1.40 Two key themes in poverty analysis and expenditure management are disaggregation and decentralization. Disaggregated information needs to be used at decentralized levels to be effective. It should be used to analyze the diversity of local livelihoods, and address issues of expenditure incidence, service usage and quality, not just the broad allocation of resources. Each region needs to develop strategies (and adapt national policies) to address their specific circumstances. The usefulness of cross sector collaboration and review increases the more decentralized it is. 1.41 It should also be noted that: (i) collecting information has costs which need to be weighed against the benefits of the tracking information gained; (ii) standards of evaluation will vary depending on what the information is required for - if it is required to trigger within-year disbursements (or reimbursements), the speed and specificity of information required will be much more demanding. 1.42 HIPC guidelines stress the importance of expenditure planning as a benchmark for expenditure tracking. The I-PRSP policy matrix includes an undertaking to introduce the Macroeconomic and Fiscal Framework in the coming fiscal year. (As previous PERs have reiterated) this needs to become the benchmark for annual discussion of expenditure allocations, including the use to which HIPC and other aid resources are put. 1.43 The SDPs have demonstrated that it is perfectly feasible for the federal government and the regions to collaborate in drawing up expenditure programmes. Regional representatives have indicated that they have no objection to the Federal Government prescribing some guidelines towards earmarking HIPC resources for poverty reducing sectors provided they have the flexibility on intra-sector allocations. The longer and therefore slower reporting chain in a decentralized system, reinforces the importance of developing a joint federal-regional system for medium term planning of expenditures. 1.44 GoE has an excellent foundation for expenditure tracking and strengthening public expenditure management in its existing poverty strategy, expenditure reform programme and welfare monitoring system. The principal challenges under the full PRSP will be: (i) to present a more comprehensive picture of the nature, causes and dimensions of poverty in Ethiopia; (ii) to spell out priorities and a programme for research and analysis to supplement existing data; (iii) to put forward a convincing timetable for the on-going reforrns in public expenditure management; (iv) to show how the capacities for welfare analysis and expenditure planning will be linked and 9 strengthened at each level from the centre down to woredas; and (v) to explain how the Government intends to reconcile the imperatives of decentralization and of medium term expenditure planning. Donors Shifting to Sector and Budget Support 1.45 External loans and grants account for as much as a third of the total Government expenditure, and their share is likely to increase in the years ahead as donor confidence is restored consequent to the Government reorienting public expenditure from defence to social sectors. This alone makes aid an important aspect of Ethiopia's public finance management. Aid management is also complex because of the depth of decentralization, the number of multilateral and bilateral donors involved, the diversity of systems and procedures, and the objective of maintaining equity across regions in the distribution of public resources. Also food related expenditure (including food aid) is not integrated into the Government's planning and budgeting cycle. Aid management has therefore been a focus topic in several past PERs. This review revisits the issue from the perspective of PRSP and HIPC. 1.46 The predominant form of aid in Ethiopia, like everywhere else, has been project aid. Yet there are good arguments for donors shifting increasingly towards programme/budget support; indeed the PRSP/IHPC context makes the case for such a shift quite compelling. There are several potential benefits that would arise as a consequence. Most notably, both the Government and the donors will be able to focus attention on broader issues of fiscal strategy, efficiency of expenditure management and strengthening of institutions for public expenditure management and public service delivery; there will be potential for simplification and harmonization of procedures; and aid disbursements will accelerate. In the Ethiopia context, decentralization reinforces the case for such a shift. The crucial point is that budget support, like the SDPs, requires an agreed institutional framework of procedures and forums linked by a binding calendar. It will not happen automatically, but needs to be designed and put in place as part of the PRSP framework. 1.47 All donors acknowledge the potential advantages of programme/budget support. Yet they are circumspect in effecting the shift for a variety of political, programmatic and operational reasons. There are several actions required of both the Government and the donors to making this shift possible. For the Government, the PRSP formulation process provides both an opportunity and a challenge to encourage the donors in this direction. The Government can use the opportunity of the PRSP to build partnership and dialogue with the donors. Other prior actions required of the Government are to move towards medium term programming of expenditures and improve the systems for tracking and evaluation of public expenditures. From the donor side, prior action required will include making medium term aid commitments, making such commitments conditional only on the quality of implementation of PRSP, routing all aid through the budget and assisting in capacity building. Public Expenditure - Regional Perspectives 1.48 Ethiopia's regions vary enormously in size, population and ethnic diversity. Although decentralization was mandated by the 1994 Constitution, the institutions of decentralization have yet to take root. The Constitution is explicit on the division of tax sources between the federal government and the regions, but not so on the expenditure side. For example, although economic and social development responsibilities vest largely with the regions, the federal government has the reserve right as well as obligation to formulate and implement strategies for overall development. Like in all other federations, there are horizontal fiscal balances across regions, and vertical imbalance between the federal and regional levels as regards own revenues and expenditure obligations. Indeed the imbalances are larger in Ethiopia than typical levels, making the regions' dependence on federal grants deeper than elsewhere. These grants are made annually 10 on the basis of a formula which is a function of population, level of development, poverty and revenue raising effort. 1.49 Since as much as 60-70 percent of poverty reducing outlays are administered by the regions, they are positioned in the frontline in the PRSP-HIPC context. Important issues in this regard include: - the interaction between federal and regional levels in setting expenditure priorities and monitoring expenditure performance (this is especially relevant for HIPC and for other forms of non-project aid); - ability to track expenditures at regional level (the detail and the timeliness of expenditure information available etc); - the quality of public expenditure management at regional level (including the fiduciary standards of expenditure control and reporting, but also the efficiency and effectiveness of resource use); - capacities and capacity building requirements at various levels. 1.50 With regard to budgetary systems and procedures, the main findings of the RPERs are as follows: - Regions suffered because of reduced federal grants and aid flows during the period of border conflict. The brunt of adjustment had to be borne by discretionary recurrent expenditures and capital spending. There has been a reversal of that trend starting FY01 when cuts in defense expenditure were directed to social sector spending through the regions. - Although there is an elaborate system of priority setting by the communities, it does not feed in any meaningful way into expenditure allocation in the regional budget as such priority setting is done with no reckoning of any resource constraint (the so called 'blue sky' budgeting). The regions receive intimation of their resource envelope very late in the budgeting cycle which leaves them with no alternative but to resort to incremental budgeting. The decision to adhere to the financial calendar starting with the FY03 budget cycle should help repair this situation. - Regions are also unable to plan expenditures in a medium term perspective because of annual setting of resource envelopes. Plans under the EMCP to roll over PEP to the regions by allowing them to leapfrog over the PIP stage and giving them three year resource ceilings should help more efficient resource planning. - Regions have an explicit preference for 'federal money' as opposed to aid money because of the uncertainty of aid flows and the complexities of implementation, procurement and reporting procedures which put additional demands on what is already overextended capacity. This bias is further reinforced by the 'offset system' (explained in Appendix 7 of Vol II). The decision last year to offset only 70 percent of external loans and 30 percent of external grants has alleviated this grievance and bias to some extent. In any case, the long term solution lies in building capacity at the regional level and donors streamlining their procedures to harmonize with those of the Government. - The Constitution permits the regions to design their own policies and systems for intra-regional decentralization. Consequently the systems and practices in this regard vary from region to region. Given the capacity constraints, smaller regions such as Benishangul find the cost-benefit calculus of intra- 11 regional decentralization working against them as opposed to larger regions such as Oromiya. - In general budget integrity and discipline at the regional level are good. Budget outturns conform to the outlays, and reappropriations from one budget head to another require legislative approval. There are expenditure reporting systems which are more or less complied with, but these are oriented more towards control rather than management. Regional audit bureaus are hampered by capacity constraints, audits are in arrears, and audits have not proven to be an effective means of enforcing accountability. The implementation of the new chart of accounts at the regional level and below as well as the admittedly gradual introduction of IFMS should improve the timing, accuracy and effectiveness of the tracking systems. 1.51 The mission work for the RPERs and the drafting of the reports were completed by end August 2001. Subsequently, in October 2001, the Prime Minister presented his (re-elected) government's development strategy to the parliament. Among other things, one of the central pillars of this strategy is decentralization. The new initiatives of the Government include the following: (i) devolution of untied block grants to woredas to effect a better match between priorities expressed by the communities and expenditure allocations; (ii) constitutional amendment giving exclusive responsibility to the regions for primary education, primary health and rural roads; (iii) rolling over reforms under the CSRP to the regions on an accelerated time table; (iv) tax reform one of whose objectives will be improvement of the regional revenue base; and (v) merging the bureaus of finance and planning into a single entity to effect coordination between capital and recurrent budgets. V. REFORMS IN PUBLIC EXPENDITURE MAMAGEMENT Progress and prospects 1.52 The past year witnessed a significant advance in the acceptance and implementation of several recommendations made in the past PERs. This has happened for a number of reasons. First, with the demanding fiscal management during the conflict behind them, the Government has been able to devote serious attention to reforms. Second, the dialogue on I-PRSP, HIPC Completion Point triggers and the policy matrix under the ERSC provided a conducive platform for reviewing the recommendations. Third, the PER itself has made conscious effort to harmonize its recommendations with the pace and sequencing of the ongoing EMCP rather than set a fresh agenda and time-frame. Public Expenditure Management Reform 1.53 One of the main refrains of previous PERs has been that GoE should develop a systematic medium term framework for expenditure planning. At a minimum, GoE should press ahead with the program it has already adopted for the institutionalization of the MEFF on which the PIP is to be based. As noted earlier, the MEFF has so far remained a shadow exercise without any political support. As part of the ERSC policy framework, GoE has undertaken to present the MEFF and PIP leading to the budget of FY03 to the Council of Ministers and then to the parliament. A corollary to this is GoE's simultaneous commitment to adhere to the financial calendar for the FY03 budget cycle so that regions get an indication of their resource envelope in time for them to effect a better match between priorities expressed by the communities and expenditure allocations. This is expected to enhance the poverty sensitivity of public expenditures. Moving forward, MoF has prepared a design document for making the next crucial shift from PIP to a PEP which would involve integrated medium term programming of both recurrent and capital expenditures. Moreover the merging of the MoF and MEDaC into a single 12 Ministry of Finance and Economic Development (MOFED) acknowledges a long standing recommendation of past PERs to effect integration of capital and recurrent budgets. Tracking of Public Expenditures 1.54 The new chart of accounts introduced at the federal level effective FY02 removes several egregious anomalies in the accounting system including misclassification of recurrent and capital expenditures. The new chart facilities better tracking of expenditures across functional and economic classifications, and also links to sources of funding. When this is rolled over to the regions starting FY03, the new system should enable better tracking of expenditures across the decentralized system. In its I-PRSP, GoE has undertaken to define a time-table for introducing cost-center budgeting, and the new chart of accounts has the built in capability to facilitate this shift to cost-center budgeting once the new accounting system is effectively stabilized. This reform on the accounting side will be complemented by the IFMS which will enable accurate, effective and timely flow of financial information thereby providing the basis for efficient financial control and management. GoE has also undertaken to reduce the backlog in accounts from the current arrears of two years at the federal level and up to five years at the regional level. Indeed two regions are already effectively current. Auditing systems, both at the federal and regional levels, continue to be weak and beset with capacity constraints. A recent EC sponsored study on financial control and audit systems should pave the way for reforms in this area. In another significant development, under the HIPC Completion Point triggers, GoE has undertaken, as part of the budget exercise each year, to consolidate federal and regional budgets for both the past year and the budget year - including all extra-budgetary funds and accounts - and make it available to donors. Besides enhancing transparency, this will help donors make an evaluation of the budget priorities and expenditure allocations. Aid Management 1.55 One of the main problems faced by GoE in moving over to medium term expenditure planning has been the uncertainty of aid flows. Given that external loans and grants account for as much as two thirds of federal capital budget plus subsidy appropriations to the regions, getting an accurate estimate of aid flows is critical to expenditure planning over a longer horizon. Undoubtedly, part of the problem has been the interruption to aid flows during the conflict period. Another major impediment has been the absence of a platform for Government-donor dialogue on PE issues, beyond what is contextually provided by the PER. This impasse has been resolved to some extent by GoE constituting a Committee to oversee public expenditure management issues and mandating that committee to meet with and brief a donor sub-group public expenditure at least every six months. It is expected that this dialogue through the Committee will graduate into a larger forum for the Government presenting its budget strategy and expenditure priorities to the donors as part of the budget cycle each year. Another major area of concern in aid management has been the slow progress on the SDPs. The Govermnent has undertaken to reformulate the SDPs in line with the PRSP priorities. A regular topic of previous PERs has been the offset system, which was considered to be a significant disincentive for the regions to uptake aid. The decision last year to move from 100 percent offset to only 70 percent for loans and 30 percent for grants was a significant step forward in remedying this malady. Reorienting Expenditures Towards Poverty Reducing Outlays 1.56 Previous PERs recommended (a) increasing the budget shares of education and health; (b) increasing the share of O&M expenditures; and (c) conducting a simultaneous review of the sustainability and mutual consistency of the SDPs. As chapter 2 on fiscal review shows, significant progress as been achieved under (a) and (b) with defense outlays yielding to social sector expenditures ahead of schedule. Moreover in its I-PRSP, the Government has agreed to 13 maintain minimum expenditure shares for important anti-poverty sectors, as well as consistency between current and capital expenditures. The PRSP Technical Committee has reportedly met with the budget committee of the parliament to discuss, among other things, a policy framework for effecting this recurrent-capital expenditure balance. Discussion on the medium term resource envelope as part of the PRSP formulation exercise should pave the way for restructuring the SDPs to fit within the resource envelope. Federal-Regional Fiscal Relations 1.57 Several reforms described above should make planning and budgeting at the regional level more effective and efficient. First, the regions will benefit from the adherence to the financial calendar and in-time notification of the resource envelope. Second, they should also begin the move towards medium term perspective with the federal decision to intimate their subsidy ceilings to the regions for two years to start with. Third, the EMCP agenda includes moving the regions over to PEP straight away bypassing the PIP stage. Fourth, the regions' bias towards external aid should be neutralized to a significant extent by the revision to the offset formula. Fifth, the introduction of the new chart of accounts and the IFMS should help the regions improve the efficiency and effectiveness of their financial management while enabling better tracking of expenditures. Sixth, the decision to accelerate the roll over of reforms under the CSRP should enable the regions to improve their financial management as well as service delivery. Finally, the decision to have the regional constitutions amended to make it mandatory for them to pass on at least some funds to the woredas as block grants should enhance grass-roots autonomy and effect a better match between community priorities and expenditure allocations. Revenue Mobilization 1.58 Strengthening both tax base and tax administration is one of the main pillars of GoE's PRGF Arrangement with the IMF. Even though the focus so far has remained restricted to the federal level, some of the measures under contemplation such as presumptive taxation, income tax codification, increasing the top sales tax rate will have scope extending to the regions. As appendix 6 (Vol. II) of this report on federal-regional revenue shares shows, the share of the regions in the combined revenue has remained static at around 18 percent. The are several changes in the offing such as introduction of VAT, privatization and private sector development which will alter the federal-regional distribution of taxes, although it is not possible yet to precisely estimate the direction and extent of shift. In the long term, however, regional autonomy will be sustainable only if regions are able to increase their own revenues and reduce their dependence on federal grants. It is proposed to review regional revenues either as an independent study or as part of the next PER to identify avenues for strengthening it to better position the regions to implement the PRSP. VI. PRIORITIES FOR FUTURE WORK 1.59 During the course of the missions, discussions at the workshop and based on the findings in this report, the following have been identified as potential focus areas for future PER work: (i) A more disaggregated analysis of social sector expenditures to examine the basis for intra-sector allocations, and identify avenues for improvement; (ii) Examination of the linkages between expenditures and physical results in key sectors, and on that basis, analysis of how that information can be used to improve the poverty sensitivity of public expenditures; (iii) Deeper study of priority setting by the communities as pat of the planning process and examine how that feeds into the budget. This is especially important in the context of deepening decentralization and proposed allocation of untied block grants to woredas; (iv) Analysis of own revenue base of the 'gions to identify ways of strengthening it; and (v) Costing of some of the important prc. ums underlying the PRSP to provide a resource framework for its implementation. 14 2. REVIEW OF FISCAL PERFORMANCE' 2.1 The end of the border conflict in June 2000 saw a rapid fiscal reorientation in Ethiopia. Supported by an Economic Rehabilitation Support Credit (ERSC) from the World Bank and a three-year arrangement under the IMF's Poverty Reduction and Growth Facility (PRGF), the Government formulated, and started implementing, a new economic program. The medium term fiscal policy framework agreed under the ERSC and PRGF supported program rests on the redirection of resources from defense-related expenditures toward poverty-reducing outlays, and laying the foundation for a strong tax revenue performance. Fiscal management was cautious in FY OO/Olbecause of uncertainties in the flow of project and program aid. This chapter reviews the fiscal perfornance in FYO1 and the medium term outlook in the context of this reorientation. 2.2 The layout of the chapter is as follows. The first section will analyze the fiscal trends since FY 86/87 and relate them to the Government's policy stance. Section II is an outline of the medium term fiscal framework. The following section is devoted to a review of fiscal performance in FY 99/00, program and preliminary estimates of budget outturn for FY 00/01, and the budget for FY01/02. Section TV presents an analysis of the economic and functional classification of public expenditure. Section V shows the trends in federal and regional shares of expenditures to help appreciate the trends during the border conflict and the reorientation thereafter. The final section concludes with a brief evaluation of the current fiscal stance of the Government. I. FISCAL TRENDS 2.3 The period FY 86/87-00/01 can be sub-divided into five distinct sub-periods for analyzing the fiscal trends: late civil war (86/87 - 90/91), stabilization (91/92 - 94/95), fiscal consolidation (95/96-97(98), border conflict with Eritrea (98/99-99/00), and post-conflict reconstruction (00/01-01/02) (Table 2.1). Starting with a high fiscal deficit in the first sub-period (86/87-90/91), there was a steady reduction in the fiscal deficit, both before and after grants, over the next two sub-periods evidencing the Government's strong commitment to fiscal discipline. This trend was reversed during the sub-period 98/99-99/00, largely as a result of the fiscal impact of the conflict. With the emergence of Ethiopia from the conflict in the second half of 2000, the Government has renewed its commitment to fiscal discipline, while redirecting defense expenditures towards poverty-targeted outlays, and embarking on a major tax reformn to strengthen revenue. 2 This section is based on the findings of the IMF- PRGF Supervision mission (May 2001) in which the World Bank participated and the joint Bank-Fund HIPC mission (September 2001). 'rable 2.1: Ethiopia - Fiscal rrends, 1986/87 - 2001/02 _ _ Avera,ge 1986/87 1991/92 1995/96 1998/99 (as a percentage ofGDP) -1990/91 -1994/95 -1997/98 -1999/00 1998/99 1999/00 2000/01 2001/02 Actual Actual Actual Pre.Act Pre. Act Pre.Act Prog. I/ Pre. Act. Budget 2/ Total revenue and grants 24 6 16 5 21.8 21.5 21.4 21 6 24.7 24 2 24.7 Total revenue 21.0 13.7 18.7 1I 0 17 8 18.2 18.S 19.3 19.5 Tax revenue 143 96 12.4 12.0 11.5 12.4 14.0 14 1 14.6 Non-Tax revenue 6 8 4.1 6 3 6 0 6.3 5.8 4.8 5.2 4 8 Extemal grants 3.6 2.8 3.1 3.5 36 33 6.0 5.0 53 Total expend. (including net lending, 31.7 22.4 25.4 31.8 306 33.0 31.2 29.1 29.5 excluding emergency programs) Of which. Recurrentsexpenditure 22.1 14.9 147 23.6 20.8 26.4 21.5 19.6 19.3 o/w Defense 96 2.7 30 10.9 8.7 13.1 7.3 6.3 5.1 Capital expenditure 9.6 7.5 9.7 8.2 9.8 6.6 9.6 9.5 10.2 Net lending 1.0 Fiscal balance (cash basis) -No e "gEriyprogra m Afler grants (excl. emergency program) -6.4 -4.8 -4.7 Before grnts (excl. emergency program) .12.4 -9.8 -10.0 Fiscal balance (cash basis) After gnts -7.1 -5 9 .3.6 -10.3 -9.2 -11.4 -8.4 -5.6 -7.6 Before grnts *10.7 -8.7 -67 -13.8 -12.9 -14.8 -14.3 -10.6 -12.9 Financing 7.1 5.9 3.6 10.3 9.2 11.4 8.4 5.6 7.6 Extemral(net) 3.0 3.2 2.4 2.6 3.5 1.7 7.1 3.9 6.9 Domestic 4.0 3.7 0.2 6.3 3.1 9.6 0.4 0.1 0.0 o/w Banking system 3.8 3.5 -0.3 6.2 1.8 10.6 -0.3 -0.4 -0.7 Privatization 0.5 1.4 1.6 1.2 0.9 0.8 0.7 Other and residual 01 -1.0 1.0 0.0 1.0 -1.0 0.0 0.9 0.0 Memoreasdm Items: Capitaleexpediturebysourcesoffinrncing:: 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 1000 Domestic treasury 50.5 60.8 71.6 53.8 51.9 55.7 47.7 51.8 46.8 Externalassistance 11.6 10.0 6.5 11.4 11.1 11.8 17.0 13.4 21.7 Extemalloan 379 29.2 22.0 34.8 37.0 32.5 35.4 34.8 31.5 Domesticborrowing/fiscaldeficit(aftergrants) 53.1 61.6 -8.4 -58.6 -337 -83.5 -5.1 -.8 0.0 GDP atcuanentmatketpfices(inmillionbinr) 17,139 27,419 41,414 50,381 48,688 52,074 57,746 52,872 59,109 Real GDP growth rate 6.3 5.4 7.8 7.9 7.0 Exchbnge rate (biff/USS, period average) 7526 8152 ... 8337 8.601 Reserves in months ofimports N/A 5 3 7.1 2 45 2.8 2.1 2.6 2 3.6 Source: Ministry of Finance, MEDAC, National Bank of Ethiopia, and WB & IMF Missions. I/ Program under the three-year IMF PRGF arrangement 2/ Consistent with program under PRGF arrangement, with lower defense expenditure, and excluding Enhanced HPC assistance. 2.4 The dominant factors responsible for fiscal improvement varied across the first three sub- periods. Between the first two sub-periods, both total revenue and total expenditure declined as a percentage of GDP, but expenditure declined more than revenue, owing to the transition from war to peace. Consequently, fiscal deficit before grants declined from an average of 10.7 percent of GDP during 86/87 - 90/91 to 8.7 percent during 91/92 - 94/95. Correspondingly, the deficit after grants declined from 7.1 percent of GDP to 5.9 percent of GDP. Between the second (91/92 - 94/95) and the third (95/96 - 97/98) sub-periods, both total revenue and total expenditure registered an increase as a percentage of GDP, but revenue increased faster than expenditure yielding a reduction in fiscal deficit from 8.7 percent of GDP to 6.7 percent before grants and from 5.9 percent to 3.6 percent after grants. During 95/96-97/98, the government's borrowing from the banking system was - 0.3 percent of GDP, evidencing its determined effort at fiscal consolidation. This fiscal discipline was, however, impaired during the period of border conflict (1998-2000) when defense expenditures increased from 3.0 percent of GDP during 95/96 - 97/98 to 13.1 percent of GDP in 99/00. As a result, fiscal deficit before grants increased to 14.8 percent of GDP in 99/00, and that after grants to 11.4 percent of GDP. Despite late disbursements of foreign assistance, and taking into account exceptional spending related to post-conflict emergency programs, fiscal deficit before grants declined to 10.6 percent of GDP in 00/01, and that after grants to 5.6 percent of GDP. II. MEDIUM TERM FISCAL FRAMEWORK I-PRSP 2.5 Immediately following the end of the border conflict with Eritrea, the Government redirected its attention towards fiscal consolidation and reforms in public expenditure management. In its Interim Poverty Reduction Strategy Paper (I-PRSP), endorsed by the Board of the World Bank in February 2001, the Government took on the following commitments: (i) give priority to education, health, agriculture and roads in programming and budgeting by raising their shares in total expenditure to 14.7, 7.0, 10.2 and 10.1 percent respectively; (ii) maintain consistency between current and capital expenditures; (iii) base planning and budgeting on the macroeconomic and fiscal framework; (iv) adhere to the budget calendar so as to effect a better link between spending priorities set by the people and the budget; and (v) define a time frame for introducing cost-center budgeting. The Government reconfirmed its commitment to macroeconomic stability in a program supported under a three-year IMF PRGF Arrangement encompassing the fiscal, monetary and external sectors of the economy. Macroeconomic Program 2.6 The twin fiscal policy objectives of the Government for the next three years are to redirect resources from defense-related expenditures toward poverty-reducing outlays (while addressing the country's post-conflict recovery needs), and to lay the foundation for a strong tax revenue performance. The Government recognizes that a major improvement in domestic revenue mobilization will be the key to accommodating the level of public services needed to support Ethiopia's economic development and avoiding an unsustainable accumulation of debt. Public savings are targeted to increase at a pace that will allow the authorities to substantially reduce dependence on external funding of the budget in the medium term. The budgeted overall fiscal deficits would be financed by concessional foreign assistance, including debt relief, and by privatization receipts, thus permitting the Government to phase out domestic borrowing. 2.7 As a priority, the Government intends to bring its investment spending back to pre- conflict levels, which would require resumption of donor support for the sectoral development programs in education, health, roads, and utility services. These programs are seen as crucial for 17 increasing the income-earning capacity of the poor and improving Ethiopia's growth potential. The Government is committed to maintaining a fiscal stance over the medium term consistent with a sustainable external current account deficit and with a monetary policy that targets low inflation, while allowing exchange rate flexibility and a gradual recovery in the level of international reserves. The Government emphasized that macroeconomic stability needed to be protected through efforts to raise domestic savings and to solicit considerable support from foreign sources-including grants, highly concessional loans, debt relief, and foreign direct investment. 2.8 Overall expenditure (excluding outlays under the emergency programs) is targeted to decline from 33.0 percent of GDP in FY 99/00 to 29.5 percent of GDP in FY 01/02. Within this envelope, a steep decline in defense expenditures (from 13.1 percent of GDP in FY 99/00 to 5.1 percent in FY 01/02) will help reduce the overall deficit and make room for an increase in poverty-targeted expenditures from 8.4 percent of GDP in FY 99/00 to 13.9 percent in FYO1/02 (see Box 2.2). These expenditures, which comprise outlays on social services, agriculture, and priority infrastructure, will be allocated through sectoral development programs formulated with technical and financial assistance from the World Bank and other development partners. The expected interim debt relief under the enhanced HIPC Initiative (expected to be of the order of USD40 million in FY02), which will be additional to the programmed resources, will also be important to put these expenditures on the envisaged higher trajectory.3 Other Government expenditures that were compressed during the conflict period have risen moderately as a ratio to GDP in FY 00/01 but would remain broadly unchanged in real terms thereafter. Tax Reform 2.9 To lay out the foundation for a strong tax revenue performance, the Government has embarked on a comprehensive tax policy reform, and an overhaul of tax administration. Tax policy reforms under the macroeconomic program are geared towards increasing efficiency and equity by streamlining the income tax, improving the efficiency of the incentive system, and shifting emphasis toward indirect domestic taxation so as to make room for a further reduction in external tariffs; these actions should culminate in the introduction of a value-added tax (VAT) in January 2003. 2.10 The revenue strategy calls for an increase in tax revenue of 21/2 percentage points during the three-year program period; it takes into account the likelihood that non-tax revenue will decline as a ratio of GDP, given the depletion of the petroleum fund and the Government's ongoing privatization program. 2.11 Specific revenue-yielding measures in the FY 00/01 budget include an increase in the sales tax rate from 12 percent to 15 percent and the introduction of a 5 percent advance tax on imports (to be credited against income tax) (see Box 2.1). These measures have partly compensated for the revenue loss from the elimination of the 10 percent surcharge on imports. 3Ethio.-ia's case for a Decision Point under the enhanced HIPC was approved by the Board on November 8, 200i. 18 Box 2.1: Tax Policy Reforms Policy Measures Effectiveness INCREASE EFFICIENCY OF INCOME TAX LEGISLATION Impose a five percent withholding tax on imports (credited against income) Jan. 1, 2001 Enact legislation for presumptive taxation, comprising: Feb. 1, 2001 -2 percent minimum tax on turnover of the largest enterprises; - transitional computation rules for smaller enterprises; - standard assessments: and - 5 percent withholding tax on payments to hard-to-tax payers. Introduce taxation of interest income on bank deposits. Feb. 1, 2001 Income tax codification. 2001/02 INCREASE SCOPE AND EFFICIENCY OF INDIREcT TAXATION Increase the top sale tax rate from 12 percent to 15 percent. Jan. 1, 2001 If necessary, expand the coverage of excise taxes and selectively raise rates. 2001/02 onward Introduce VAT in 2003. Jan. 2003 REDUCE EXTRNAL PROTECTION WHILE IMPROVING EFFICIENCY OF THE TAX SYSTEM Eliminate the 10 percent import duty surcharge. Jan. 2001 Review and streamline the tax incentive system. End 2001 Reduce the effective rate of external protection. 2001/02 onward 2.12 The ultimate success of tax policy reforms will be determined by a major strengthening of Ethiopia's capacity to administer tax laws and to ensure a high level of compliance by taxpayers. To this end, the Government has adopted an ambitious schedule for building an effective tax administration with IMF technical assistance and a multi-donor Trust Fund, which involves the establishment of a task force to oversee the implementation of tax reform in March 2001, the creation of a fully operational large-taxpayer unit in June 2001, and the introduction of taxpayer identification numbers in the second half of 2001. The recruitment of additional audit staff, and the development of new procedures, and training of staff for the newly established large-taxpayer unit is to be completed shortly. Meanwhile the World bank is planning a study of regional revenues to analyze the possibilities for increasing regional autonomy. Post-Conflict Emergency Program 2.13 With the emergence from the conflict with Eritrea in the second half of 2000, the Government stressed that for the immediate period ahead, its strategy would be to reconcile the country's security needs with the requirement to finance both the emergency programs for reconstruction and demobilization and the improvement of priority public services, all within a sustainable fiscal framework. 2.14 Following the cessation of hostilities and in response to a request from the Government, the World Bank prepared an emergency reconstruction package comprising four operations. The first, Emergency Recovery and Rehabilitation Project, for which the IDA committed credit of US$257 million, is to provide humanitarian assistance to war-affected people and to rebuild the key infrastructure damaged in Tigray and Afar, thus reducing the war toll on private sector development. The second, the Emergency Demobilization and Reintegration Project, for which the IDA committed US$171 million, is aimed at demobilizing an estimated 150,000 war veterans, 19 helping them reintegrate into civilian life, and resume income-earning activities, and thus foster a socially stable environment. Third, the Economic Rehabilitation Support Credit of US$150 million provides initial support for the Government's efforts in stabilizing the economy and moving back to a sustainable path of growth. Finally, the Fertilizer Support Credit of US$44 million is aimed at helping Government maintain uninterrupted import of fertilizer pending preparation of a new Agricultural Input Project. 2.15 The envisaged pace of execution is tied to the Government's implementation capacity and the objective of macroeconomic stability, with costs of 6Y2 percent of GDP to be spread over the FY 00/01 - 02/03 period. Disbursements under the emergency programs only amounted to 0.8 percent of GDP in FY 00/01, but are programmed to increase to 2.9 percent of GDP in FY 01/02, and 1.5 percent of GDP in FY 02/03, resulting in large, but transitory, overall fiscal deficits of 7.6 percent of GDP in 2001/02; the deficit would fall to 6.3 percent in FY 02/03. III. THREE YEAR FISCAL REVIEW 1999-2002 Fiscal Performance During 1999-20004 2.16 The deterioration of the fiscal situation accelerated during FY99/00. Tax revenue increased by 0.9 percentage points of GDP as a result of the 10 percent surcharge on imports introduced in December 1999. This was substantially offset by a decline in external grant flows as well as a drop in non-tax revenues occasioned by decline in transfers from (i) public enterprises; (ii) sugar auction sales; and (iii) fuel stabilization fund. The net result was a modest increase in total revenues and grants from 21.4 percent in FY 98/99 to 21.6 percent in FY 99/00. However, total expenditure rose from 30.6 to 33.0 percent of GDP as a result of the preemptive demands of defense expenditure. Consequently fiscal deficit after grants rose to an all time high of 11.4 percent of GDP forcing the Government to resort heavily to borrowing from the banking sector. Preliminary Fiscal Outturn for 2000/01 2.17 Preliminary estimates of budget outturn for FY 00/01 indicate that the fiscal deficit (including grants) narrowed to 5.6 percent of GDP, much lower than the 8.4 percent envisaged under the macroeconomic program, mainly as a result of cautious fiscal management, lower than programmed external assistance, and slower than anticipated execution of emergency programs and capital expenditures. Despite lower than anticipated inflows of external assistance, the net government position with respect to the domestic banking system declined by 0.4 percent of GDP. 2.18 Total revenues increased from 18.2 percent of GDP in FY 99/00 to 19.3 percent in FY 00/01. Even as non-tax revenues continued their declining trend mainly owing to lower realization from dividends and lower inflows from the fuel stabilization and sugar auction funds, this was more than offset by increase in tax revenue of as much as 1.7 percent of GDP as a result of several measures taken to increase tax rates, and improve tax administration. Also external grants improved substantially from 3.3 percent of GDP in FY 99/00 to 5.0 percent in FY 00/01. 2.19 On the spending side, the Government contained expenditure because of uncertainty regarding the disbursement of program aid, while redirecting resources from defense expenditures to poverty-targeted outlays. In mid-2001, an additional Br 500 million (0.9 percent of GDP) was reallocated from defense to mostly social outlays. Defense outlays were limited to 6.3 percent of 4The outturns of capital expenditures for FY 99/00 and FY 00/01 do not include the sectoral allocation of project grant and the actuals are therefore likely to be higher. 20 GDP (lower than the programmed 7.3 percent of GDP), while poverty-targeted outlays (health, education, agriculture & natural resources, and roads) are estimated to amount to t.he equivalent of 10.9 percent of GDP. However, spending under the emergency programs was only 0.8 percent of GDP against the programmed 1.9 percent. Regarding the envisaged rapid recovery in non- military core expenditures in FY 00/01, the Government thought that it was possible to accelerate investment spending by concentrating on project preparation early in the fiscal year, notwithstanding the severe constraints on committing expenditure that had been in effect until the FY 00/01 budget was approved by parliament in late November 2000. However, largely because of delays in the disbursement of external assistance,5 capital outlays were limited to 9.5 percent of GDP, slightly below the program target of 9.6 percent of GDP. Fiscal Projection for 2001/02 2.20 The FYO 1/02 budget reflects a continuation of the effort towards fiscal consolidation and expenditure reorientation. Total revenue and grants are projected to increase from 24.2 percent of GDP in FY 00/01 to 24.7 percent in FYO1/02. While a modest decline in non-tax revenue will be offset by a corresponding increase in grant flows, tax revenues performance will benefit from the full year impact of measures implemented since early 2001 and from improved tax administration. As a further effort in this direction, the Government has established a large tax payer unit and has plans to streamline the Federal Inland Revenue Authority. 2.21 On the expenditure side, the budget envisages containing total expenditures to 29.5 percent of GDP. Defense outlays are to be further cut from 6.3 percent of GDP in FY 00/01 to 5.1 percent of GDP (ahead of the PRGF program projection of 5.2 percent). Poverty targeted spending will increase by over 25 percent to 13.9 percent of GDP. Savings in external debt servicing under the enhanced HIPC (estimated at 0.7 percent of GDP) will further contribute to poverty reducing outlays.6 Capital expenditure is projected to grow to 10.2 percent of GDP, up from 9.5 percent in FY 00/01 while recurrent expenditure will decline to 19.3 percent yielding a fiscal deficit after grants of 4.7 percent of GDP. However, as a result of spending under the emergency programs of 2.9 percent, overall fiscal deficit, including emergency programs and grants, will go up to 7.6 percent of GDP. IV. ECONOMIC AND FUNCTIONAL CLASSIFICATION Economic Classification 2.22 The composition of expenditure according to the economic classification changed substantially during the last two years, largely as a result of the sharp increase and then decline in defense expenditures (Table 2.2). In FY 00/01, the recurrent-capital expenditure balance reverted to its FY 98/99 level after a significant decrease of capital expenditure in 99/00. The share of recurrent expenditure in total expenditure increased from 67.9 percent in FY 98/99 to 80.0 percent in FY 99/00, and is estimated to decline to 67.4 percent during FY 00/01. The share of wages and salaries declined from 19.7 percent in FY 98/99 to 19.1 percent in FY 99/00, but is estimated to have risen significantly to 23.4 percent in FY 00/01, owing, among other factors, to increases in allowance. The proportion of materials and supplies, driven by defense expenditures, 5 Some donors had delayed aid disbursements until the establishment of the Temporary Security Zone in April 2001. 6 Ethiopia, like other African countries, is expected to be adversely affected by the post-September global crisis. Preliminary estimates indicate that the combined effect of an increase in oil prices to about USD30 per barrel (until the second quarter of 2002) and a decline of 7 percent in commodity prices would be an increase of the current account deficit by 12 percent (USD50 million). In that case, it would absorb most of the interim debt relief of the order of USD40 million leaving no scope for expenditure expansion. 21 rose steeply from 31.7 percent in FY 98/99 to 42.5 percent in FY 99/00 and is estimated to decline to 25.0 percent in FY 00/01, as military expenditures decrease significantly. 2.23 Capital expenditures have inevitably yielded to recurrent expenditures in FY 99/00, their share dropping from 32.1 percent in FY 98/99 to 20 percent in FY 99/00, and are estimated to have recovered to 32.6 percent in FY 00/01, despite delays in disbursements of foreign assistance and lending. Excluding defense expenditure, the share of capital expenditures decreased from 44.8 percent in FY 98/99 to 33.3 percent in FY 99/00, and is estimated to have increased to 41.6 percent in FY 00/01. These trends lead to two conclusions: first, during the conflict with Eritrea, defense expenditures have crowded out public investment, and second, these trends have begun to be reversed in FY 00/01. Table 2.2: General Government Expenditures - Economic Classifications 1998/99 - 2001/02 Share of total expenditure 198/99 1999/00 2000/01 2001/02 (in percent) Pre. Act. Pre. Act. Prog. I/ Pre. Act. Budget 21 Total Expenditure 100.0 100.0 100.0 100.0 100.0 Total Recurrent 67.9 80.0 69.1 67.4 65.4 Wages and salaries 19.7 19.1 22.4 23.4 22.2 O/w Defense 5.5 7.8 6.2 7.1 6.1 Materials and supplies 31.7 42.5 26.5 25.0 25.1 Grants, contr. trans. and others cur. transfers 2.6 2.4 3.1 3.5 3.2 Pensions 2.1 1.8 1.8 2.1 2.2 Subsidies 0.0 0.0 0.0 0.0 0.0 Interest and Charges 6.4 6.5 7.9 7.0 6.8 External assistance 5.4 7.5 7.5 6.4 5.9 Miscellaneous 0.0 0.0 0.0 0.0 0.0 Total Capital Expenditure 32.1 20.0 30.9 32.6 34.6 Capital expenditure /total non-defense expenditure 44.8 33.3 40.3 41.6 41.8 Source: World Bank-IMF Mission estimates. 1/ Program under the three-year IMF PRGF arrangemenL 2/ Consistent with progran under PRGF arrangement. 22 Table 2.3: General Government Expenditures - Functional Classifications 1998/99 - 2001/02 In Percent 1998/99 1999/00 2000/01 2001/02 Pre. Act. Pre. Act. Prog. 1/ Pre. Act. Budget 2/ Share of total expenditure General Administration 37.6 48.4 32.6 32.7 27.9 O/w Defense 28.4 39.8 23.4 21.5 17.2 Economic Infrastructure 10.3 6.7 13.5 11.4 14.7 O/w Roads constructions 8.0 5.2 10.2 9.0 12.1 Social Services 19.5 16.4 21.9 22.5 25.3 O/wEducation 11.5 9.6 13.2 13.9 17.0 Health 4.5 3.1 6.9 5.2 6.2 Economic Services 14.4 10.0 12.2 12.9 14.1 O/w Agri & Nat. Res 10.6 7.5 10.4 9.6 11.8 Others 18.2 18.5 19.8 20.6 18.0 O/w Interest and charges 6.4 6.5 7.9 7.0 6.8 External Assistance 9.0 9.9 7.5 10.7 5.9 Share of recurrent expenditure General Administration 55.3 60.6 47.2 48.6 42.7 O/w Defense 41.8 49.8 33.8 31.9 26.3 Economic Infrastructure 1.4 1.2 1.6 1.8 1.7 O/w Roads constructions 0.7 0.7 0.8 1.0 1.1 Economic Services 6.2 4.7 6.2 7.3 9.0 O/w Agri & Nat. Res 5.2 3.9 4.8 6.1 6.9 Social Services 18.7 15.3 19.3 21.8 22.5 O/w Education 12.2 9.5 13.9 14.8 16.4 Health 4.5 2.9 4.1 4.6 4.7 Others 18.2 18.3 25.8 20.6 24.0 O/w Interest and charges 9.4 8.2 11.4 10.4 10.4 External Assistance 8.0 9.4 10.8 9.4 9.0 Share of capital expenditure Economic Services 31.6 31.1 25.8 24.6 23.6 O/w Agri & Nat. Res 21.8 21.9 22.8 17.0 21.2 Economic Infrastructure 29.1 28.8 40.4 31.1 39.3 O/w Roads constructions 23.3 23.2 31.1 25.5 32.9 Social Development 21.1 20.7 27.5 23.8 30.5 O/w Education 9.8 9.9 11.7 11.9 18.2 Health 4.6 4.1 13.2 6.4 9.1 Others 18.2 19.3 6.3 20.6 6.6 O/w External Assistance 3/ 11.1 11.8 13.4 Source: Ministry of Finance and WB and IMF Missions estimate. 1/ Program under the three-year IMF PRGF arrangement. 21 Consistent with under program under PRGF arrangement. 3/ Extemal assistance under capital expenditures is not distributed across sectors for FY99, FY00 and FY01 (preliminary actual). However, under program (FY01) and budget (FY02), the extemal assistance has been allocated across sectors. 23 Box 2.2: Ethiopia: Defense and Poverty-Targeted Outlays, 1997/98 - 2001/02 1/ 1997/98 1998/99 1999/00 2000/01 2001/02 2/ Before HIPC After HIPC (In millions of birr) Defense 2,190 4,233 6,842 3,307 3,000 3,000 Poverty- targeted expenditure 4,408 5,154 4,355 5,787 8,207 8,662 (In percent of expenditure) 3/ Defense 19.3 28.4 39.8 21.5 17.2 17.1 Poverty- targeted expenditue 38.9 34.6 25.3 37.6 47.1 49.5 (In percent of GDP) Defense 4.9 8.7 13.1 6.3 5.1 5.1 Poverty- targeted expenditure 9.8 10.6 8.4 10.9 13.9 14.7 Defense andPoverty-TargetedOutlays, 1997i98 - 2001102 so0. 60 50 .4 40 - Poverty targeted 40 empenditure (in percent of total 30 - 30 Defense fin percent oF total \s ex~~~~~~~~~~~~~~pendkture) 20 O DeFense (in percent of GDP) 20 Povertq-targeted outlap on percentt GDP) 0 0 1SS7)98 198891 1999100 20o0101 2001102 ource: Ethiopian authorities; and Fund-Bank staff estimates and projections. 1! Fiscal year ending July 7. Poverty-targeted outlays include on health, education, agriculture and roads. 2/ Projections for 2001/02 reflect lower defense spending and increased poverty expenditure in the budget compared to the program (EBS/01/108, 7/5/2001). 31 Excluding special programs (demobilization and reconstruction). Functional Classification 2.24 The shifts in defense expenditures have also altered quite dramatically the composition of public expenditures according to its functional classification during the two year conflict period 1998-2000, and in the aftermath of the conflict. From a proportion of 20 percent of total expenditures of the general government during FY 97/98, defense-related expenditures increased to over 28.4 percent during FY 98/99, and further to 39.8 percent in FY 99/00, and are estimated to have declined to 21.5 percent in FY 00/01 (see Table 2.3). The fluctuations in the share of defense-related expenditures are even more dramatic with respect to recurrent expenditures. In this case, defense-related expenditures increased from 31 percent of recurrent expenditures during 24 FY 97/98 to 41.8 percent during FY 98/99, and to 49.8 percent in FY 99/00, before declining to 31.9 percent in FY 00/01. They are projected to decrease further to 26.3 percent of recurrent expenditure under the budget for FY 01/02. The trends in defense-related expenditures have been mirrored by opposite trends in poverty-related expenditures (see Box 2.2). Trends in Social Expenditures 2.25 While the Government was unable to protect critical expenditures on social services (education and health), and particularly their capital portion in FY 99/00 (Table 2.4), the priority given to expenditures on social services in FY 00/01 was reflected in an estimated increase of 1.3 percent of GDP for the two sectors combined. In the education sector, the shares of recurrent and capital expenditures declined to 2.5 and 0.7 percent of GDP respectively in FY 99/00, but are estimated to have increased to 2.9 and 1.1 percent of GDP respectively in FY 00/01. In the health sector, while the share of recurrent expenditure declined marginally from 0.9 percent of GDP in FY 98/99 to 0.8 percent in FY 99/00, capital expenditure fell sharply from 0.5 percent of GDP in FY 98/99 to 0.3 percent in FY 99/00. Capital expenditure on health are estimated to have recovered to 0.6 percent of GDP in FY 00/01. During FY 01/02, total expenditures in both sectors are estimated to increase - from 4.0 percent of GDP in FY 00/01 to 5.0 percent in FY 01/02 for education, and from 1.5 percent of GDP in FY 00/01 to 1.8 percent of GDP in FY 01/02 for health. 2.26 As a proportion of total government expenditures, the shares of both education and health have fallen during FY 99/00, but are estimated to have increased during FY 00/01. The share of the education sector in total expenditures declined from 11.5 percent during FY 98/99 to 9.6 percent during FY 99/00, and is estimated to have increased to 13.9 percent in FY 00/01. The share of the health sector dropped from 4.5 percent in FY 98/99 to 3.1 percent in FY 99/00, and is estimated to have increased to 5.2 percent during FY 00/01. 2.27 The recent emphasis on the social sectors is also apparent in figures of real per capita expenditures (Table 2.5). In the education sector, real per capita expenditure declined from Birr 8.8 in FY 98/99 to Birr 8.1 in FY 99/00 before increasing to an estimated Birr 10.8 in FY 00/01. Trends in health sector real per capita expenditures are even more pronounced with a decline from Birr 3.5 in FY 98/99 to Birr 2.6 in FY 99/00, and an increase to Birr 4.0 in FY 00/01. Given Ethiopia's low expenditure levels in these sectors, reversing these past trends is particularly critical for the realization of the poverty reduction objectives of the Government. 2.28 The expected debt relief under the enhanced Heavily Indebted Poor Countries (MPC) Initiative will be important to increasing social - and more generally - poverty-targeted expenditures. Preliminary estimations7 indicate that the possible interim assistance under the enhanced HIPC Initiative provided after reaching the Decision Point could finance additional poverty-targeted expenditures equivalent to around 0.7 percent of GDP in FY 01/02. 7 Preliminary Enhanced HIPC Initiative Decision Point Document (EBS/0 1/15, 2/8/01; IDA/R2001-0022, 2/12/01). 25 Table 2.4: Recent Trends in Social Sector Expenditures 1998/99 - 2001/02 Social sector expenditure 1998/99 1999/00 2000/01 2001/02 Pre. Act. Pre. Act. Prog. 1/ Pre. Act. Budget 2/ Education As % of GDP 3.5 3.2 4.1 4.0 5.0 Of which: Current expenditures 2.5 2.5 3.0 2.9 3.2 Capital expenditures 1.0 0.7 1.1 1.1 1.9 As%ofTotalexpenditures 11.5 9.6 13.2 13.9 17.0 As % of current expenditures 12.2 9.5 13.9 14.8 16.4 As % of capital expenditures 9.8 9.9 11.7 11.9 18.2 Health As % of GDP 1.4 1.0 2.1 1.5 1.8 Of which: Current expenditures 0.9 0.8 0.9 0.9 0.9 Capital expenditures 0.5 0.3 1.3 0.6 0.9 As % of Total expenditures 4.5 3.1 6.9 5.2 6.2 As % of current expenditures 4.5 2.9 4.1 4.6 4.7 As % of capital expenditures 4.6 4.1 13.2 6.4 9.1 Source: Ministry of Finance and WB & IMF Mission estimates. 1/ Program under the three-year IMF PRGF arrangement. 2/ Consistent with program under PRGF arrangement. Table 2.5: Real Per Capita Social Sector Expenditures 1998/99 - 2001/02 1998/99 1999/00 2000/01 2001/02 Pre.Act. Pre.Act. Prog. 1/ Pre. Act. Budget 2/ Education expenditure (million birr) 1,709.0 1,646.0 2,384.0 2,130.0 2,964.0 Health expenditure (million birr) 678.3 535.0 1,239.0 794.0 1,080.0 GDP Deflator (1980/81 = 100) 3.15 3.19 3.32 3.01 3.14 Population (million) 61.7 63.5 65.3 65.3 67.2 Education real expenditure (million birr) 542.7 515.3 717.9 708.3 943.3 Health real expenditure (million birr) 215.4 167.5 373.1 264.0 343.7 Education real expenditure per capita (birr) 8.8 8.1 11.0 10.8 14.0 Health real expenditure per capita (birr) 3.5 2.6 5.7 4.0 5.1 Source: Ministry of Finance and WB & IMF Mission estimates. 1/ Program under the three-year IMF PRGF arrangement. 2/ Consistent with program under PRGF arrangement. V. FEDERAL AND REGIONAL SHARES OF EXPENDITURES 2.29 Tables 2.7 a-b-c show the federal and regional shares in the recurrent, capital and total expenditures respectively. The most striking trend is the decline in the expenditure share of the regions from 42.5 percent in FY 96/97 to 33.4 percent in FY 00/01 which is largely explained by the pre-emptive demands, over these years, of defense expenditure which is an exclusive federal responsibility. The declining share of the regions in health and education sectors is an accounting 26 anomaly as all capital expenditures out of external loans and assistance have been attributed to the federal government as disaggregated accounts have yet to be finalized. It is expected that regions will gain in both sectoral shares of health and education as well as overall capital expenditure as Ethiopia reorients to public expenditures to poverty reduction following the transition from war to peace. 2.30 Since international debt service is met entirely from the federal budget ('interest and charges' in Table 2.6a), debt relief, by reducing this item of expenditure, will also tend to increase vertical imbalance. 27 Table 2.6a: Federal and Regional Shares in Recurrent Expenditure, 1997 - 2001 1996/97' 1997/98 1998/99 1999/00 2000/01 2000/01 (Program) (Prel. Actual) Categoly Federal Regional Total Federal Regional Total Federal Regional Total Federal Regional Total Federal Regional Total Federal Regional Total Share Share (mrn. Birr) Share Share (min. Bir) Share Share (ruin. Birr) Share Share (min. Birr) Share Share (min. Binr) Share Share (nln. Bin) General administraton 66.2 33.8 1880.6 77.5 22.5 3323.9 83.8 16.2 5604.0 89.4 10.6 8322.2 84.4 15.6 5865.0 794 20.6 5025.8 olw Defense 100.0 0.0 834.8 100,0 0.0 2189.5 100.0 0.0 4232.9 100.0 0.0 6842.2 100.0 0.0 4204.4 100.0 0.0 3307,0 Economicinfrasticture 72.8 27.2 179.7 24.8 75.2 104.5 57.9 42.1 145.7 54.6 45.4 165.3 60.7 39.3 193.1 604 39.6 185.1 o/w Road construction 1/ 76.4 23.6 169.8 25.2 74.8 91.3 28.5 71.5 75.6 27.0 73.0 93.9 35.3 64.7 96.5 33.3 66.7 99.1 Economic services 19.9 80.1 483.2 22.4 77.6 553.5 23.5 76.5 632.1 28.4 71.6 643.2 32.5 67.5 767.9 25.9 74.1 7S2.7 o/w Agricult. & nat. res. 13.9 86.1 414.4 17.4 82.6 473.3 17.0 83.0 529.8 21.1 78.9 530.6 21.8 78.2 596.2 18.3 81.7 630.4 Social services 19.0 81.0 1501.6 22.2 77,8 1706.6 20.8 79.2 1897.3 29.6 70.4 2102.4 19.6 80.4 2403.6 22.3 77.7 2257.5 o/wEducation 14.3 85.7 1034.0 14.9 85.1 1115.0 14.2 85.11 1240.0 16.7 83.3 1305.0 18.9 81.1 1733.5 15.9 84.1 1533.4 Health 18.1 81.9 334.3 20.0 80.0 394.2 19.3 80.7 455.8 17.4 82.6 395.2 11.0 89.0 505.9 18.9 81.1 475.0 Others 87.5 12.5 1692.9 83.5 16.5 1392.1 96.6 3.4 1847.2 97.1 2.9 2508.7 97.5 2.5 3209.3 97.6 2.4 2130.9 o/winterest&charges 100.0 0.0 918.7 100.0 0.0 835.7 100.0 0.0 956.9 100.0 0.0 1121.9 100.0 0.0 1416.0 100.0 0.0 1079.8 extemral assistance 100.0 0.0 256.5 100.0 0.0 160.0 100.0 0.0 812.6 100.0 0.0 1289.5 100.0 0.0 1349.0 100.0 0.0 978.2 Total current expenditure 56.4 43.6 5737.9 60.3 39.7 70806 70.2 29.8 10126.3 78.4 21.6 13741.7 71.7 28.3 1243S.9 66.4 33.6 10352.0 [I] Includes urban development. [2] Figures for 1996/97 through 1999/00 are prelirninary actual expenditures. Table 2.6b: Federal and Regional Shares in Capital Expenditure, 1997 - 20011 1996/97 1997/98 1998/99 1999J00 2000/01 2000/01 (Program) (Prel. Actual) Category Federal Regional Total Federal Regional Total Federal Regional Total Federal Regional Total Federal Regional Total Federal Regional Total Share Share (mln. Birr) Share Share (mrn. Birr) Share Share (rlnt. Bier) Share Share (min. Birr) Share Share (min. Birr) Share Share (ndn. Bier) Economic infrastructure 71.8 28.2 1168.0 59,3 40.7 1093.0 73.1 26.9 1395.9 70.2 29.8 992.6 77.8 22.2 2242.7 68.4 31.6 15610 o/wRoadconstructiork 58.3 41.7 776.3 51.5 48.5 898.3 66.8 33.2 1117.3 63.4 36.6 796.9 71.7 28.3 1729.9 61.7 38.3 1280.0 Economic development 64.6 35.4 1990.4 66.7 33.3 1239.6 68.0 32.0 1513.9 71.2 28.8 1071.7 71.2 28.8 1432.7 63.8 36.2 1231.7 o/wAgricult. & nat. res. 24.5 75.5 865.8 43.4 56.6 722.6 54.0 46.0 1044.1 59.5 40.5 752.4 67.9 32.1 1269.3 48.1 51.9 852.4 Social development 26.6 73.4 694.4 41.1 58.9 1013.5 53.1 46.9 1009.3 60.1 39.9 713.7 74.3 25.7 1530,0 53.6 46.4 1192.7 o/wEducation 39.3 60.7 392.6 44.5 55.5 436.7 61.3 38.7 469,0 77.7 22.3 341.0 79.5 20.5 649.8 46.3 53.7 597.4 Health 8.7 91.3 221.2 209 79.1 276.9 25.4 74.6 222.5 21.3 78.7 140.2 81.8 18.2 732.6 65.8 34.2 319.2 Others 45.4 54,6 315.6 77.4 22.6 800.6 76.5 23.5 871.1 78.0 22.0 664.0 44.0 56.0 349.9 83.8 16.2 1032.7 o/w External assistance 100,0 0.0 495.0 100.0 0.0 531,0 100.0 0.0 404.7 100.0 0.0 673.0 Totalcapitalexpenditure 588 41.2 416B.4 60.6 39.4 41466 67.9 32.1 4790.1 699 30.1 3442.0 730 27.0 55553 66.9 33.1 5018.1 I11 Regions' capital expenditures do not include external loan and assistance for FY98, FY99, FY00 and FY01 disaggregated figures are not available. [2] Figures for 1996/97 through 1999/00 are prelintinmy actual expenditurets. Table 2.6c: Federal and Regional Shares in Total Expenditure, 1996 - 2001 1 1996/977 1997/98 1998/99 1999/00 2000/01 2000/01 (Program) (Prel. Actual) Category Federal Regional Total Federal Regional Total Federal Regional Total Federal Regional Total Federal Regional Total Federal Regional Total Share Share (min. Birr) Share Share (min. Birr) Share Share (mrin. Birr) Share Share (min. Bir) Share Share (min. Biff) Share Share (min. Birr) General Administration 66.2 33.8 1880.6 77.5 22.5 3323.9 83.8 16.2 5604.0 89.4 10.6 8322.2 84.4 15,6 5865.0 79.4 20.6 5025.8 o/wDefense 100.0 0.0 834.8 100.0 0.0 2189.5 100.0 0.0 4232.9 100.0 0.0 6842.2 100.0 0.0 4204.4 100.0 0.0 3307.0 Economic Infrastructure 71.9 28.1 1347.7 56.3 43.7 1197.5 71.6 28.4 1541.5 68.0 32.0 1157.8 76.4 23.6 2435.8 67.5 32.5 1746.1 o/wRoadConstruction3/ 61.5 38.5 946.2 49.1 50.9 989.6 64.4 35.6 1192.9 59.5 40.5 890.8 69.8 30.2 1826.4 59.7 40.3 1379.3 EconomicServices&Dev. 55.9 44.1 2473.6 53.0 47.0 1793.0 54.9 45.1 2146.1 55.2 44.8 1714.9 57.7 42.3 2200.6 49.4 50.6 19844 olwAgricult.&Nat. Res. 21.0 79.0 1280.3 33,1 66.9 1195.8 41.5 58.5 1573.9 43.6 56.4 1283.0 53.1 46.9 1865.5 35.4 64.6 1482.8 SocialServices&Dev. 21.4 78.6 2196.0 29.3 70.7 2720.1 32.0 68.0 2906.5 37.3 62.7 2816.0 40.9 59.1 3933.6 33.1 66.9 3450.2 o/wEducation4/ 21.2 78.8 1426.5 23.2 76.8 1551.7 27.1 72.9 1709.1 29.3 70.7 1646.0 35.4 64.6 2383.3 24.4 75.6 2130.8 Health 14.4 85.6 555.6 20.4 79.6 671.1 21.3 78.7 678.3 18.4 81.6 535.4 52,9 47.1 1238.5 37.7 62.3 794.2 Others 80.9 19.1 2008.4 81.3 18.7 2192.7 90.2 9.8 2718.3 93.1 6.9 3172.7 92.2 7.8 3559.2 93.1 6.9 3163.6 o/winterest &charges 100.0 0.0 918.7 100.0 0.0 835.7 100.0 0.0 956.9 100.0 0.0 1121.9 100.0 0.0 1416.0 100.0 0.0 1079.8 external assistance 100.0 0.0 256.5 100.0 0.0 655.0 100.0 0.0 1343.6 100.0 0.0 1694.2 100.0 0.0 1349.0 100.0 0.0 1651.2 TotalExpenditure 57.5 42.5 9906.3 60.4 39.6 11227.2 69.5 30.5 14916.4 76.7 23.3 17183.7 72.1 27.9 17994.2 66.6 33.4 15370.1 [ II Regions' capital expenditures do not include extemal loan and assistance for FY98, FY99, FY00 and FY01 as disaggregated figures are not available. (21 Figures for 1996/97 through 1999/00 are preliminary actual expenditures. (31 Road constructions under recurrent includes urban development. 14] The declining share of the regions in health and education sectors is an accounting anomaly as all capital expenditures out of external loans and assistance have been attributed to the federal govenment as disaggregated accounts have yet to be finalized. Note: Regions capital expenditures do not include extemal loan and assistance for FY98, FY99, FY00 and FY01 (expenditure). However, the total goverment expenditurea include extemal assistance and loan for the mentioned years above. VI. SUSTAINABILITY OF CURRENT FISCAL STANCE 2.31 The Government financial strategy assumes that a large part of the fiscal deficits will be financed with disbursements of highly concessional loans and debt relief on concessional terms. The projected reliance on strong financial donor support takes account of the major challenge Ethiopia is prepared to tackle: rebuilding core development spending from the compressed levels of the recent years and healing the economy from conflict-related damages while sharply rolling back military expenditure. Over the medium term, financial policies and structural reforms, as well as a gradual improvement in Ethiopia's terms of trade, are expected to strengthen the external position, so that project loans would remain as the only debt-creating financing flows. In the interim years, when savings capacity is still being built, in addition to the expected provision of debt relief under the enhanced HlPC initiative, substantial balance of payments support, at highly concessional - preferably grant - terms, will be needed. 2.32 The Government will need to sustain its efforts to implement the tax reform program to achieve the targeted increase in tax revenue. While in FY 01/02 tax revenues should benefit from the full-year impact of the measures implemented since early 2001 and from improved tax administration, should there be a revenue shortfall, the Government needs to be ready to expand the coverage or raise selectively the rates of other indirect taxes. 2.33 Non-tax revenues will continue to decline as the Govemment pursues the privatization program. Since 1995, the Government has embarked on a comprehensive privatization program, with 158 entities sold for a total value of Br 3.4 billion, at end FY 00/01. The Government intends to complete the privatization program by bringing an additional 114 entities to the point of sale (including units of whole enterprises, particularly in the agriculture, textile, and services sectors) by 2003/04. 2.34 There may be a need for changes in the civil service pay scale, including a decompression of wages, when the Government elaborates further its agenda of civil service reform. The budgetary implications of changes in the wage bill strategy will need, however, to be mapped out within a medium-term expenditure strategy that builds on a coherent macroeconomic framework. 2.35 The Government's preparedness to compress expenditures under special budgetary provisions should the overall fiscal situation so require is reassuring. To be able to take such decisions in a timely manner, the Government will need to proceed decisively with its plans for refining budget monitoring and reporting systems. Crucial steps in this direction will be the enhancement of systems - also at the regional level - that track expenditures, and assess impacts and outcomes. 30 3. TRACKING PUBLIC EXPENDITURES FOR POVERTY REDUCTION I. INTRODUCTION 3.1 This module8 of the PER 2001 had two main sub-components: - to evaluate the existing systems for tracking expenditures, and the availability and quality of data for assessing expenditure performance in the light of the government's key policy objectives, and make recommendations for a program of improvement; - to evaluate existing systems for monitoring poverty and for assessing the impact and outcomes of public expenditure, and make recommendations for improvement. 3.2 The work involved revisiting some of the themes from previous PERs, but from the perspective of utilization of HIPC resources, and by the requirement for GoE to prepare a Poverty Reduction Strategy Paper (PRSP).9 The most recent Public Expenditure Review (PER 2000)10 examined the results-orientation of public expenditures, but this report goes further in considering the links between public expenditures and poverty reduction. The 1MC guidelines" recognize that there are many complications in assessing the links from public expenditure to poverty reduction. At the same time, it is urgent to get basic mechanisms into place to allow debt relief to begin. The impetus of the HIPC initiative is towards (a) identifying a sub-set of public expenditures which are 'pro-poor', (b) using HlPC resources to increase the level and/or share of resources allocated to these 'pro-poor' expenditures, while (c) tracking expenditures so as to verify that incremental resources have in fact reached the sub-set of pro-poor expenditures (and, ultimately, have helped to reduce poverty). 3.3 In practice none of these elements is straightforward. This chapter tries to separate issues into three main sets: (a) expenditure tracking per se (which is independent of any particular objectives of expenditure) (Section II); (b) definition, measurement and monitoring of poverty in Ethiopia (Section III); and (c) links between public expenditures and poverty in the three sectors covered by Sector Development Programs (SDP) (education, health, and roads) (Section IV). Finally Section V draws together the principal conclusions and recommendations. Conclusions that are especially relevant to aid management are carried over to Chapter 4. 8 The block of work that is reported in this chapter was led by DFID, and is reported more fully in Mokoro Limited, Tracking of Public Expenditures for Poverty Reduction, Ethiopia PER 2001 Discussion Paper, August 2001. 9 Ethiopia's full PRSP is expected to be ready by AprilMay 2002, while a Decision Point under enhanced HIPC was approved by the World Bank Board on Nov 8, 2001. 1' World Bank, Ethiopia: Public Expenditure Review, report no. 2081 0-ET (two volumes), August 31, 2000. See in particular Volume 1, Chapter 5 and Volume II, Appendix 5. "1 See IMF/IDA, Tracking of Poverty-Reducing Spending in Heavily Indebted Poor Countries (HIPC), Prepared by the IMF's Fiscal Affairs Department and the World Bank's Poverty Reduction and Economic Management Network, March 27, 2001. II. TRACKING OF PUBLIC EXPENDITURES Issues and Approach 3.4 This section looks at 'expenditure tracking' in the narrow sense of establishing how funds have been used, not what they have achieved. Understandably a first concern for the HIPC initiative (as for all forms of aid) is to make sure that funds have been used as agreed.12 Without this assurance there is little basis for discussing their effectiveness in reducing poverty, or in meeting any other agreed objective. However, the desire to evaluate the impact of additional expenditures leads to two other demanding questions: * The first is a technical question: how well can funds be linked to particular programs and projects, so as to allow a specific analysis of the links between expenditure and outcomes? * The second is a broader question: how to make a valid comparison between the situation where additional funds are provided and the (counterfactual) situation as it would have been without those funds? 3.5 . The answer to the first question requires a review of the format of the budget and accounts, while the answer to the second requires a comprehensive view of public expenditure. It is not enough to know, for example, that particular funds were directed towards education. It is also important to examine whether educational expenditure has increased; and whether education received a bigger share of public expenditures. A complete picture of public expenditures should include: (a) sub-national as well as federal expenditures; and (b) aid-funded expenditures, whether or not they formally pass through the government budget and accounts. Both these dimensions are of great practical importance in Ethiopia. 3.6 Moreover, in order to judge the difference made by additional funds, tracking public expenditures as a whole is necessary, but not sufficient. There are both conceptual and practical problems in establishing any counterfactual. In reviewing the public expenditure management system, it is important to understand the planning and budget formulation process as a basis for making judgments about how additional funds have, in practice, been deployed. This point is further examined in the final section of this chapter. 3.7 In connection with the enhanced FHPC initiative, the World Bank and IMF have prepared a basic checklist to assess a country's capacities to track expenditures effectively, and to identifv ways in which its public expenditure management systems may need to be strengthened. The checklist reflects a broad approach to expenditure tracking, and deals systematically with budget formulation, execution, reporting and evaluation. An evaluation of Ethiopia's budgetary systems against this check list is at Appendix I (Vol II).'3 This chapter focuses on a number of principal coinclusions and issues derived from this assessment.'4 12 This is often referred to as afiduciary concern - i.e. it is a matter of trustworthiness. 13 The most recent version consulted is Guidance for HIPC Tracking Missions and Finalisation of Assessments andAction Plans (AAPs), April 25, 2001. Appendix I was first prepared as part of this PER exercise. It was subsequently updated during the joint Bank-Fund HIPC Expenditure Tracking mission in October 200 1. 14 In many cases, the findings reported here were reinforced by the Regional PERs described in Chapter 5. 32 Context - Decentralization and Expenditure Reforms Expenditure reform 3.8 It is important to evaluate Ethiopia's public expenditure management system in context. Ethiopia is a large country, which is also implementing a demanding decentralization program. In parallel, it has developed sector-wide development programs for a number of key sectors, while simultaneously pursuing a comprehensive civil service reform program (CSRP). A major component of the CSRP is the Expenditure Management and Control Program (EMCP), which is a government-owned strategy for the reform of all of the main elements of the public expenditure management system. The starting point for reform (i.e. the system inherited by the present government) is a budgeting and accounting system that is in many ways rudimentary and unsophisticated (e.g. line item budgeting, single-entry bookkeeping, and dependent on manual systems) but remarkably disciplined and effective. 3.9 The EMCP strategy takes a long view about ownership and sequencing of reforms.'5 It is based on the premise that improvements to the financial management system must be evolutionary, building on procedures with which practitioners are already familiar, and must follow a logical sequence in which, for example, "administrative" approaches precede "management" approaches, and effective external controls are established before a system of internal controls can be introduced.'6 Considerable effort has been devoted to the understanding of existing systems, as the basis for steps to improve and expand them. The Concept Note for this PER stressed the importance of taking the EMCP as the frame of reference for proposals to strengthen public expenditure management. This does not mean that the EMCP strategy should be treated as perfect or immutable, but (a) additional proposals to strengthen aspects of public expenditure management should be integrated with the EMCP, and (b) it should not be assumed that there are easy short-cuts for the strengthening of public expenditure management. Decentralization 3.10 Ethiopia's system of decentralization, which is extensively discussed in Chapter 5, has important implications for expenditure tracking. In the first place, the regions dominate the delivery of basic services, so that tracking of poverty related expenditures has to focus strongly on regional services. Secondly, the regions have considerable discretion in the details of budget formulation, so that targets for poverty-related expenditures are, in effect, set jointly by the federal and regional governments. Thirdly, decentralized delivery inevitably lengthens the chain for the reporting of expenditures, and hence affects the speed with which details of expenditure can be reported. Finally, both service delivery itself, and also the tracking of expenditures, is highly dependent on capacity at local levels. Fiduciary Issues and the Quality of Expenditure Reporting 3.11 Ethiopia has a strong tradition of expenditure discipline, reflected in the fact that expenditure outturns correspond closely to budgets and there is little evidence of improper diversion of funds or unauthorized expenditures. There are, however, a number of areas where standards of transparency and accountability could be improved. As detailed in Appendix I (Vol 15 See Financial Reform in a Devolved African Country: Lessons from Ethiopia, Stephen B Peterson, April 25, 2000. 16 As Peterson, ibid. p5, puts it: To reach the stage of "managing outputs", financial systems must be formalised, first through the imposition of effective and efficient external control and, second, through the evolution of internal control. There are no shortcuts to improving financial management in a developed or developing country. 33 II), improvements are needed in relation to a number of extra-budgetary funds, in cash management, in the presentation of accounts which show financial assets as well as transactions, and in the strengthening of internal and external audit functions. Procurement procedures also need some strengthening to reach acceptable international standards. However, the strengths of expenditure control are such that the main challenge in improving public expenditure management is not to prevent the diversion or misapplication of funds, but to obtain better management information on expenditures, and to use this information to help improve the efficiency and effectiveness of public service delivery.'7 A recent review of financial controls'8 has confirmed the view that expenditure control mechanisms are quite strong. The main weaknesses arise from problems in staff complement and training, especially at lower levels Detail and timeliness of expenditure reporting 3.12 There are a number of weaknesses in expenditure reporting. The budget classification has been unsatisfactory. Under the budget reform a new classification is being adopted, and will facilitate the identification of cost centres, with a common expenditure classification across the recurrent and capital budgets.'9 The new classification is now being introduced at federal level, but will take 2-3 years to roll out completely to the regions. Meanwhile, although systematic reports are collected vertically through the financial bodies, these are highly aggregated, and serve the purposes of overall financial control rather than management information. Moreover, there are substantial delays in obtaining both in-year and end-of-year reports. These reflect the difficulties of aggregating reports across a decentralized system, with limited computerization and severe capacity constraints, especially at lower levels. The inclusion of donor funded expenditures continues to be a major difficulty. There is substantial progress with EMCP efforts to bring accounts up to date, and to reforrn the budget classification, but this will not transform reporting capacities overnight. The effort to develop a hybrid Channel I reporting system (Channel 1 b, where World Bank funds are passed through the financial bodies, but continue to be earmarked, separately programmed and passed through separate accounts within the government budgetary system) retarded progress on the main accounts reform but did not achieve either the standards of accountability or the speed of disbursement that had been hoped for. Public Expenditure Planning Medium Term Planning and Budget Formulation 3.13 EMCP embraces reform of medium term expenditure planning as well as budget formulation and accounting, but this component of EMCP has, to date, been less fruitful. As noted in Appendix 1 (Vol. II), there has been some controversy about the design of this element, which was based on sequential movement from medium-tern programming of the capital budget (PIP) to programming of the entire budget (PEP), starting with the federal government and then incorporating the regions. However, the proposed reform contains all the essential elements of a medium term expenditure framework. An annual Macroeconomic and Fiscal Framework (MEFF) is supposed to provide the basis for the setting, and approval by the Council of Ministers, of ceilings - Indicative Planning Figures (IPFs) - for the capital expenditures of federal public bodies, which are consolidated into a three-year rolling Public Investment Program. According to regulations, only projects that are prioritized in the PIP are eligible for inclusion in the annual 17 As put by the Joint IDA/IMF Staff Assessment of the I-PRSP (1I 1): Although the efficiency and transparency of Ethiopia's public expenditure management is relatively high, continuedprogress n the Expenditure Management and Control Programme will be needed to maximize the benefits from the HIPC initiative. (in World Bank Report No. 21769-ET, January 30, 2001). 18 Diagnosis ofExisting Control Capacities in Ethiopia -Provisional Report, Group 2 AC, August 2001. Prepared for the European Commission. '9 The importance of the cost centre approach is recognised in the I-PRSP policy matrix. 34 capital budget. A detailed financial calendar, developed as part of the budget reform, spells out the responsibilities of different bodies as well as the annual timing of each stage of the exercise. Medium term planning by the regions cannot take place without timely medium term projections from the Federal Government of Ethiopia, because the regions need to know the projected level of the federal subsidy, and the availability of external aid (both of which are under federal control). 3.14 The fundamental problem has not been the design of this reform, but its failure to attract the high level political support that is essential for it to be effective. As successive PERs have documented,20 preparation of MEFFs and PIPs has so far been only a shadow exercise. The MEFF and PIP have never been submitted to the Council of Ministers for approval, and neither the procedures nor the timing mandated in the financial calendar have been adhered to. The Government's principal explanations for the failure to implement these reforms have been (a) the uncertainties associated with the war, and (b) difficulties in forecasting of external resources. Previous PERs however, have argued that medium term expenditure planning should be seen as a way of addressing such uncertainty.21 3.15 From the standpoint of public expenditure management to address poverty reduction, Ethiopia's deficiencies in medium term expenditure planning are a much more serious weakness than the shortcomings in expenditure reporting. Non-marginal changes in the pattern of expenditures take time, and effective discretion in the context of a single annual budget is very limited. It is only through a medium term planning process that the issues of appropriate expenditure composition (the balance between sectors and sub-sectors, between federal and regional levels, between recurrent and capital expenditures, etc) and the targeting of expenditures (their efficiency and distributional impact) can be properly addressed. Only through medium term scenarios can the government and potential donors effectively explore the difference between 'with-aid' and 'without aid' scenarios, as a benchmark for monitoring whether aid does, in the event, make the difference to expenditure composition that was intended. Two important corollaries are (a), for the government, the need to adhere to the financial calendar,22 and (b) for the donors, to facilitate the medium-term programming of their aid. The need to strengthen the ex ante planning of expenditures, and to avoid concentrating exclusively on ex post reporting, is reinforced by the decentralization framework, since, as discussed in Chapter 5, the broad allocation of public expenditures is jointly determined by federal and regional governments. 3.16 Under the impetus of the PRSP process there are now signs of real progress in the medium term planning of public expenditures. The Government has committed itself to the observance of the financial calendar, including the Cabinet endorsement of the MEFF and IPFs, and has initiated the formulation of PEPs at regional level.23 A revised strategy24 avoids the 20 See in particular PER 2000, Volume I, Chapter 3, and Volume II, Appendix 2. 21 As PER 2000 put it (¶3.1 1): GoE interprets the problems in terms of uncertainty and lack of information. The border conflict makes economic forecasting more difficult, makes pre-emptive demands on expenditure, and has disrupted aid flows. Even without the conflict, it was proving difficult to make reliable medium- term projections of aid flows. These problems are serious, but it is perverse to argue that such uncertainties make medium term planning impractical. A major reason for undertaking systematic medium term expenditure planning is to address such uncertainty (this year's MEFF, in fact, attempts this by exploring a number of possible scenarios), and the preparation of medium term forecasts could be an important focus for dialogue with the donor community about aid commitments and their potential impact. 22 A point that is now recognised in the policy matrix of the I-PRSP. 23 GoE has included this agenda in the policy matrix accompanying the Economic Rehabilitation Support Credit (Report No. P7441 -ET dated April 30, 2001) 35 preparation of regional PIPs as an intermediate step, recognizing that capital and recurrent expenditures need to be planned jointly from the outset. The Federal Government has undertaken to provide advance forecasts of the federal subsidy, so that regions will be able to make rolling three-year expenditure plans within a defined resource envelope. The three-year plans will be revised annually. This will provide an opportunity for discussion and review of expenditure projections in the light of poverty strategy, and the annual roll-over should also allow medium term plans at both federal and regional levels to become more robust as experience is gained. m. POVERTY AND WELFARE MONITORING IN ETHIOPIA Approach 3.17 This section reviews the definition, measurement and monitoring of poverty in Ethiopia. Given the complexity in analyzing the causes and dimensions of poverty, it is unrealistic to expect poverty monitoring to give simple, unambiguous and rapid feedback on policies. Nevertheless it is extremely important to understand and monitor poverty. This section covers: (a) the different dimensions of poverty and the extent to which they feature in public policy debate; (b) the systems for measuring and monitoring poverty in Ethiopia; (c) issues in the interpretation of poverty data, and potential tensions between different parties interested in poverty monitoring; and (d) directions for strengthening poverty monitoring and linking it more effectively to public policy. Concepts of Poverty 3.18 The traditional definition of poverty is a lack of income and assets, or of economic opportunities, through which people obtain minimum nutritional and other material requirements. The most common way of assessing this dimension of poverty is by setting a poverty line of income or consumption, with households falling below the line deemed to be poor. The first of the International Development Goals (to reduce by half the proportion of people living in poverty by 2015) relates to this concept of poverty. At the same time, the IDGs recognize other important dimensions of poverty. A second dimension, often labeled capabilities, is concerned with people's access to basic education and health. A third dimension, vulnerability concerns the degree to which individuals, households or communities are exposed to risks or 'shocks' that threaten incomes. 'Shocks' may include regular and predictable events, such as seasonal changes in food supply, or periodic less predictable events, like drought, illness, war or civil disorder. Poor people's ability to withstand shocks may be strongly linked to their possession of assets such as savings or livestock. Finally, empowerment, or the degree to which people are able to participate in the decisions that affect their lives, is increasingly being seen as an important determinant of welfare. Empowerment has a direct effect on people's sense of well- being, but also has indirect effects on welfare. For example, poor people's participation ('voice') may lead to more effective targeting of health and education services to meet their needs. 3.19 In order to understand the nature of poverty and its causes, it is important to keep all its dimensions in view. They are complementary, not substitutes. They interact in different ways, producing patterns of poverty that are varied and location-specific, and where individuals and households move into and out of poverty over time. Policy makers, planners, and development partners need to be aware of the different dimensions of poverty and to look out for relevant interactions between them. Although the income poverty concept has tended to dominate perceptions, Ethiopia's policy documents, and especially the I-PRSP show increasing awareness 24 See Perran Penrose, Medium Term Expenditure Framework (MTEF) and Public Expenditure Programme (PEP), DSA Project Report P43, April 2001. 36 and concern for all the principal dimensions of poverty [see Appendix 2 (Vol II)]. At the same time, considerable tension and misunderstanding arises in poverty discussions, much of which stems from different emphases on different dimensions of poverty. Welfare Measurement and Monitoring in Ethiopia Availability of data 3.20 Appendix 3 (Vol. II) provides an overview of the welfare monitoring data that are available in Ethiopia. Boxes 1 and 2 in that Appendix summarize the main surveys available, indicate the lowest level of disaggregation each can support without compromising statistical representativeness, and show how they can address the four principal dimensions of poverty discussed above. The Appendix also describes institutional responsibilities for collecting and analyzing welfare data, including the intended pivotal role of the Welfare Monitoring Unit (WMU) in MEDAC. 3.21 The centrepieces of the welfare monitoring system are (a) the Household Income, Consumption and Expenditure Survey (HICES) and the Welfare Monitoring Survey (WMS), as data sources, and (b) the Welfare Monitoring Unit, as the central analytical unit. However, there are also various auxiliary sources of information, including the food security monitoring system, more localized quantitative sources (e.g. the Oxford University/Addis Ababa University panel studies) and various qualitative studies. Health and education dimensions are also tracked in the mainly administrative monitoring systems associated with the ESDP and HSDP (see section III below for further discussion of these). Assessment 3.22 Welfare monitoring may be judged by three sets of criteria: - the relevance and adequacy of data that are collected; - the timeliness and quality of processing, analysis and publication; and - utilization of welfare information in policy and resource allocation. 3.23 Given the complexity of poverty analysis, available data will never satisfy all potential users. Nevertheless, Ethiopia has better information on poverty and a more systematic poverty monitoring program than many other countries. The HICES supported by the other data sources listed in Appendix 3 (Vol. II) constitutes a relevant and reasonably adequate foundation for welfare monitoring. The system has improved in certain respects in recent years. For example, sample sizes for HICES have been increasing. This enhances the scope for meaningful disaggregation but also increases the logistical problems in delivering the data to the responsible analysts in a reasonable period of time. Broadly speaking, although CSA experiences capacity constraints,25 the quality of the surveys it conducts is commendable. Available data are sufficient for a broad multi-dimensional analysis of poverty and welfare. For example, the nationally representative Welfare Monitoring Surveys interviewed in 1995 and in 2000 the same households as those covered by the HICES. Consequently, monetary and non-monetary welfare indicators can be compared. In other words, links between consumption (monetary indicators) and health, education or nutrition outcomes can be explored. Also, links between assets (e.g. livestock) and poverty can be explored in the most recent data. 25 The CSA's draft prograimne for the next 5 years envisages a near trebling of its professional staff and similar increases in other resources. (See MEDAC, January 2001, A Medium Term National Statistical Program for Ethiopia (2000/01 - 2004/05), Executive Summary, p23, Table 2.) 37 3.24 The timeliness of processing, analysis and publication is a much greater concern than the relevance and adequacy of the data itself. The CSA published data reports on the HICES/WMS of 1995/96 and the WMS of 1998. However, since the CSA's main role is to gather data and make it available to users, these reports do not go beyond a set of descriptive statistics and cross tabulations by reporting levels. As described in Appendix 3 (Vol II) (especially Box 3.3), the WMU was intended to be the focal point of a coherent welfare monitoring system. Up to now, the WMU has fulfilled only a small part of its intended role. Initially, the WMU identified the analysis of the HICES to derive consumption poverty information as its main priority. Analysis of the WMS was also on the agenda, but after the first task was completed. The WMU produced a report (Poverty Situation in Ethiopia), in March 1999, which analyzed the HICES 1995/96, but only with respect to consumption poverty and some basic descriptive statistics on the non- monetary data. For example, there was no attempt to link the non-monetary and monetary characteristics. The report also drew on the 1996 WMS and described national and geographical poverty levels based on an absolute poverty line and some limited analysis of the non-monetary welfare data. This is the WMU's only substantial report to have been completed, and the complementary activities that were envisaged have not yet been undertaken. 3.25 Thus, although the WMS, as designed, is conceptually sound, it has not, in practice, lived up to expectations. There were initially problems with timely data delivery, but in the end much of the data that has been produced has remained under-researched. A key problem has been the capacity (both managerial and analytical) of the WMU to conduct this analysis. Since the processing and dissemination of welfare data has been unsatisfactory, so too has been the use of poverty information in policy analysis and resource allocation. However, this should not be seen simply as a failure to supply decision-makers with adequate information. There has also been insufficient appreciation by decision-makers of the potential practical value of welfare data for policy formulation and resource allocation, and therefore insufficient demand for the information that was potentially available. Issues in the Use of Welfare Data 3.26 The emergence of the PRSP has now created a more urgent demand for information about poverty and poverty changes in Ethiopia. As a consequence, and as a matter of priority, both government and donors are keen to revitalize the WMU and other welfare monitoring activities. However, different stakeholders have a wide variety of interests and expectations which are not necessarily easy to reconcile. To list a few: (a) Methodology: There is a gap between the emphasis on a quantitative approach used by the WMU and donor preferences for more multidimensional approaches, including qualitative studies such as Participatory Poverty Assessments (PPAs). The problem is that the WMU's analysis is based only on CSA data which is derived from large data sets on nationally representative random samples. (b) Povertyfocus: There is a gap between the typical analysis plan by the WMU, with a focus on overall welfare and poverty trends in the country, and a concern by some donors and other groups that the focus should be on the group of poorest households and their experiences. Some donors and groups, such as NGOs, would seem to advocate a strict focus on some of the most marginal groups in rural and urban areas, for example groups that appear to be chronically dependent on food aid. The data used by the WMU may include such groups, but is more suitable to analyze overall patterns and less appropriate for investigating these groups in depth. (c) Unrealistic expectations: There is a gap between what the available data can legitimately be used for and the expectations of both donors and government. 38 Both sides would naturally like data to reveal whether the reform program has worked or not. This expectation is inflated. For this to be possible, one needs data comparable over time and the possibility to control for other possible reasons for specific evolutions in the data (see Box 3.1). Comparing two 'cross-section' data sets over time (such as the different HCES and WMS) is always a challenge, and, for reasons that cannot easily rectified, the comparability between some of the rounds is far from perfect. Furthermore, in general, analyzing the impact of policies on poverty convincingly requires more information than what is available, or at least what can be delivered in a credible way in a short period of time. (d) Urgency and capacity building: there is a gap between the desire for urgent results to feed into the PRSP versus the need the build long run capacity. Arguably, compared to many other countries, the data available for poverty analysis from within the CSA data sources is substantial, while there is much additional data available elsewhere. In principle, with sufficient resources and outside assistance, it would be possible to analyze much of this in a relatively short time. However, the only feasible model for such quick analysis would be outside the WMU. The result would be, again, to undermine the opportunity to build the capacity of the WMU to take charge of the welfare analysis agenda. 3.27 The implication of all this is that no solution can satisfy all interests at the same time. But an acknowledgement of these differing expectations and agendas may provide a first step, and help to make the dialogue among stakeholders more constructive. It is also important to recognize the scope for differing perspectives on the same data, and the danger that excessive generalization will lead to oversimplification of the debate about poverty (see Box 3.2). Box 3.1: Explanatory Power of Welfare Monitoring Data The welfare monitoring system in Ethiopia, just as in other countries, delivers a lot of useful information on welfare outcomes and some of the constraints faced by households to attain a decent standard of living. However, welfare monitoring systems are usually under pressure to answer the key question: have reform policies worked or not? A standard approach, usually found in policy documents, is first to state the changes in welfare outcomes - for example a decrease in income poverty over time, and then to provide a narrative of economic policies and how they have helped to obtain the desired outcome. This can only be a first step. Measuring the impact of economic policies on welfare outcomes is fraught with problems. First, economic reform policies usually consist of different measures, some of which may pull in opposite directions. Secondly, many other factors may also have influenced the observed outcomes: for example, weather conditions or specific local events. It is important to be able to separate the effects of reform policies from any other factors that may have caused welfare changes. Policies may have had adverse effects even if welfare outcomes improve, or been beneficial even if welfare outcomes deteriorate. A correct assessment of policies requires a comparison with a 'counterfactual': what would have happened if the policies had not taken place? Different approaches exist to do this. For example, some use large general equilibrium models to simulate counterfactual outcomes, but the structure and data needs for these models make them far more suitable for macroeconomic analysis than for welfare analysis. Altematively, if one has data from different periods in time, preferably in the form of panel data, statistical analysis may be able to control for factors not related to policy. In that case, one would need to quantify these different factors and indeed, the policies. The data requirements are high and not met by standard welfare monitoring surveys, including in Ethiopia. The lesson is that welfare outcomes by themselves are not conclusive evidence as to whether policies are working or not. Expectations to the contrary are not helpful. This does not lessen the importance of collecting welfare outcome data, and of trying to use it to inform policies. But policy analysis should include a careful review of the other possible factors that may help to explain the observed changes. 39 Box 3.2: Reconciling Evidence on Poverty The Welfare Monitoring Unit, but also numerous other institutions inside and outside the govemment services, produce data and evidence on poverty and poverty trends in Ethiopia Much of this evidence appears to be conflicting, giving rise to much debate. In itself this is not surprising. Poverty is multi-dimensional and there is no reason a priori why trends on one dimension should be the same as on another. However, discussions can be less heated and more constructive if care is taken to assess the quality and limitations of different data sources. This will help to avoid undue generalization from specific studies. It is also important to recognize that the same conclusions can seem different when expressed from different perspectives. This box illustrates the last two points. Generalization Data based on representative samples do not need to be consistent on average with data based on non- representative samples. Consequently, the pattems emerging from the two types of study can be different without being contradictory - they just set out to describe and analyze a different group of people. For example, national data may show primary school enrolments increasing in the 1990s - an indication of at least one dimension of the standard of living improving - but case studies may show deterioration of enrolments for specific communities. National data may show increased agricultural output per household, but case studies may well show that in some areas output per household is stagnant or declining. This problem can be illustrated by the important but difficult discussions about the overall trends in welfare and poverty. This is of course a difficult subject, sensitive and emotive in any country in the world. Evidence is rarely non-conflicting. However, a careful look at the samples involved is a first step. For example, the I-PRSP quotes evidence from a panel data survey from AAU/Oxford University, showing declining poverty in a sample of villages between 1989-1995. Furthermore, a larger panel covering 15 villages in 1994-1997 also found evidence of declining poverty. While this is indeed what can be found in these samples, the work has raised a lot of eyebrows. For example, the qualitative Consultations with the Poor study suggested stagnating or even declining living standards. Thinking about samples involved helps. The latter study clearly states that vulnerable villages were chosen. The panel data survey also does not claim statistical representativeness even though in terms of agro-ecological zones and livelihood systems, there was a careful attempt to reflect rural diversity. In the larger sample, a similar diverse experience showed up. But even the panel data showed diversity in experience: in four out of six villages poverty went down between 1989-1995; in two villages poverty went up. Both of these have serious access problems, land pressure and limited potential - communities and households that found it harder to respond to the new incentives generated by the reforms. Finally, the evidence is not simply that the rich gained and the poor stayed behind - far from it. Arguably, inequality increased but only marginally. The main point was that a sizeable group of the poor could grow out of poverty, even though others could not. Viewed in that way, the inconsistencies between different studies become less pronounced. So, who is right, on average for the country? Did rural poverty go up or down? The (statistically unrepresentative) panel survey, combined with the evidence on increasing enrolment suggests some decline, even though not everybody may have been able to benefit to the same extent. (Indeed, there is no development strategy that has ever succeeded in lifting everybody equally, anywhere in the world.) But the question on whether poverty increased or not on average may not be the best question to ask anyway. In any case, the answer is always more complicated than yes or no. Although there is a natural tendency to draw general conclusions from particular cases, there is no reason to suppose that experiences are the same in different areas or for people in different social circumstances. Perspectives Leaving aside issues about the definition of poverty and the quality of data, there is plenty of scope to come to different conclusions if the same data are viewed from different perspectives. For example, it is perfectly possible that simultaneously (a) per capita GDP could increase, (b) the percentage of people below the poverty line could fall, and (c) the absolute numbers below the poverty line could increase. With a high population growth (3%) and starting from about 45% poor, it would require a yearly decline in the poverty rate of about 1.3 percentage points a year (or an almost 3% annual decline in the rate) just to keep the number of poor people constant. Clearly, the same data could lead to quite different assessments about progress in reducing poverty. Directions for Improvement of Welfare Monitoring Analytical priorities 3.28 The output currently available from the welfare monitoring system does not do justice to the richness of the data and information available. There has been too much focus on the headline figure, in the form of the measure of national consumption poverty, and not enough effort to develop the additional analyses that the WMS can support. Although the summary figures for national consumption poverty are the ones that most quoted in policy documents, they have very limited value for the design, planning and evaluation of policies. The data that are available have the potential to provide much richer sets of information that are more directly relevant to policy and planning. For example, they could provide detailed information on regional differences in poverty, socio-economic profiles of the poor in different regions, matched 40 consumption and non- monetary welfare indicators (such as links between consumption, education, health), links between income, access to facilities and the use of facilities, and so forth. 3.29 Currently, expectations have been raised that the most recent HICES data can be used for assessing whether the reforms have worked or not. The basis for this is that the 2000 HICES data can also provide a national poverty estimate, which then can be compared with the previous estimate (based on HICES 1995/96). Even if the data were fully comparable, the data set does not contain the necessary data to relate the economic reforms or any other measures to the observed outcomes (see Box 3.1). The problem is compounded because lack of institutional memory related to the 1995/96 HICES is going to make it virtually impossible to make the data comparable. With extensive effort, it may be possible to do a welfare comparison over time, using the two HICES surveys, even though it will remain very unclear how much confidence can be attached to this. However, this should not be allowed to distract from more practically useful work, maybe attracting less headlines, which could contribute to a better understanding of poverty and be relevant to more users. In future, there should be much more emphasis on using the full range of available data and on geographical disaggregation of the analysis as described in Box 3.3. Ownership and capacity building 3.30 The lack of capacity in the WMU and its accumulated backlog mean that the available data are largely under-researched, especially the government sources. This is a familiar problem in many countries. In Ethiopia, the problem is, among others, a serious one of analytical capacity within government. Although Ethiopia has a growing number of researchers familiar with the analysis of household and other surveys who are willing to participate in this type of work, most of the relevant representative data are collected within the government system and there is a general reluctance to share data with researchers outside the government system. This is not uncommon. Problems include legal issues of confidentiality, the need for government agencies to first complete their mandatory reporting stage (e.g. data abstract), incentives to secure one's own funding situation, concerns about protecting one's capacity building opportunities and political sensitivities about the ownership of programs. Regarding the WMS and HICES, only CSA and the WMU in MEDAC have access to the data, but they simply do not have the capacity to analyze it fully. There is an urgent need to broaden the base of researchers with access to welfare data while recognizing the genuine importance of building up a sufficient level of government analytical capacity and ownership of the analysis. The Welfare Monitoring System: possible ways forward 3.31 New data is currently becoming available, in the form of a new consumption and expenditure survey (HICES 2) and WMS 4. The PRSP process needs also more poverty analysis to enter into the policy debate. There is therefore an urgent need to improve the functioning of the welfare monitoring system. The objective should be to support more evidence-based policy making, including a willingness to strengthen or even change policies on the basis of new data and information. However, no strategy will succeed without sufficient sensitivity to institutional issues, including ownership issues. and political sensitivities. The strategy for strengthening welfare monitoring should start with trying to strengthen the WMU, by building up its analytical capacity, but also strengthening other elements, including building up its capacity as a research management unit, developing a better interaction with users and a clear dissemination strategy. 41 3.32 This strategy should contain the following elements: - A clear analysis plan, that is tailored to more users. This should involve less emphasis on the headline figures (such as the national level of poverty) but more focus on disaggregation (e.g. zonal information on poverty dimensions), poverty profiles (characteristics and determinants of poverty), links between dimensions of poverty (health, education, income). - More technical assistance, geared towards long-term capacity building. - The development of clear relationships with users and potential users. Data analysis within government only makes sense if it is demand-driven. - A clear dissemination strategy. Information and analytical findings on welfare monitoring must be made accessible to stimulate an evidence-based debate. - More flexibility to commission further analytical studies after the national summary figures have been produced and published. This must include the WMU being allowed to contract specific studies to clarify general issues raised in the national report. Examples could be regional or zonal studies, studies that link WMS data to other sources from the social sector ministries, studies that provide more contextual information in specific zones with apparent problems, analytical studies aiming to explain certain trends (e.g. changes in enrolments or dropouts), etc. 3.33 Over the past year, a number of steps have been taken to strengthen the WMU and enhance its capacity along the lines suggested above. A practical training course has been provided for selected MEDAC, CSA and regional officials in multidimensional poverty measurement, the interpretation of welfare data and its use in policy design and planning. The aim is to establish a wider pool of staff on which WMU could draw if necessary, and also to create a more active user group. More specific training has been provided for MEDAC staff who will be involved in the next round of data analysis. Technical assistance is being provided to assist the WMU with analysis of the WMS and HICES, and a local external consultant (from AAU) is also assisting the WMU. A work plan has been adopted which involves working on both monetary and non-monetary welfare indicators from the outset. It is expected that a first basic draft (comparable to the earlier Poverty Situation report) will be completed before end 2001. IV. SECTOR EXPENDITURES AND POVERTY Introduction 3.34 Four sectors - education, health, roads and agriculture - have been tentatively labeled as ?pro-poor' sectors. This report reviews the tracking of expenditures and their implications for poverth in three of these sectors - education, health and roads.26 The methodology is extensively discussed in the background paper, and Appendices 4 and 5 provide basic background information about education, health and roads sectors. This section highlights the main conclusions arising from the sector reviews, concerning (a) the poverty relevance of expenditures; (b) issues in the tracking of expenditure; and (c) issues related to monitoring of performance and the impact of expenditures. 26 The background report prepared for this chapter by DFID also included a review of the water sector. 42 Education and Health Issues in perspective 3.35 As described on Appendix 4 (Vol II), both these sectors have been organized around Sector Development Programs (ESDP and HSDP) which cover both the recurrent and the capital expenditures of the federal and regional governments. Fundamental strategies focus on primary services and extending their coverage across the country. Prima facie, then, there is a clear pro- poor orientation, and a corollary is that expenditures are dominated by the regional govermments.27 Each SDP has an agreed set of indicators, and a monitoring system which includes annual Joint Review Missions (JRMs) conducted on behalf of both government and donors, with and Annual Review Meeting (ARM) to discuss JRM findings and future directions for the programs. This report has been able to draw on the comprehensive Mid Term Reviews (MTR) undertaken in the first quarter of the year.28 3.36 ESDP and HSDP need to be judged in perspective. Both sectors start from very low levels of service provision, and in a context of limited capacity and an ambitious decentralization program. Both have been handicapped by the effects of the border conflict and the withholding of donor support that resulted. The recent MTRs (like previous JRMs) can point to significant weaknesses, both in the tracking of expenditures and more generally in the management of resources for effective service delivery. At the same time, the fundamental strategies being followed are clearly motivated by a pro-poor agenda, and the reviews do not identify any major fiduciary problems or failure of funds to reach their intended uses. The ESDP MTR (pl 1), for example, comments: that schools are constructed, books distributed, and teachers paid on time, in areas up to a thousand kms from the regional HQs is a considerable achievement. That all this is done in areas where there are no telephones, no banks, and no roads, is outstanding. 27 See Chapter 5 for details of the balance between federal and regional expenditure responsibilities. 28 See Report of the Mid Term Review Mission, 7 February to 8 March 2001, Ethiopia Health Sector Development Programme 1997/98-2001/02 (EC 1990-1994), 8th March 2001, and Mid Term Review Mission February 13'h - March 3'd 2001, (2 volumes), Ethiopia Education Sector Development Programme 1997/8 - 2001/2. 43 Box 3.3: Regional Planning and Local Poverty Profiles Avaabke data There is a tendency for welfare monitoring systems to becomne too preoccupied with a 'headline' measure of the national extent of (consunption) poverty. This is understandable, but it is not the most useful contribution to policy making. The most recent Welfare Monitoring Surveys (starting in 1998) are based on a sampling frmne that allows statistically representative results to be reported down to the zonal level. Even though for some zones the samples will still be relatively small (resulting in significant sampling errors), one could calculate welfare indicators for each zone. The information from the welfare monitoring system becomes most useful when informnation on available infrastructure, health, water and education provision are presented for a direct comparison with the indicators. Furthermore, much other administrative data, such as on food aid needs, could also be matched. Thus, it is now possible to provide for each region a comparison between all the zones on crucial information such as welfare outcomes, access to social services, water and infrastructure, health care use, education, enrolment and drop-outs, and more. A systematic presentation of these indicators allows for a welfare and deprivion ptofile of each zone, in comparison to other zones. Such information can support evidence-based policy making and resource allocation at both regional and zonal levels. Local perty prles Data from the welfare monitoring system could provide a useful basis for informing policy makers about local opportunities for a pro- poor focus of-public expenditure. The starting point is a local poverty profile. This is a set of characteristics of the poor, relative to the non-poor, based on the data from the welfare monitoring system. For example, the data available from the Wdlfare Monitoring Surveys give information on assets owned by households, on access (in terms of distance in kin) to roads, market, health facilities, safe water, schools, on use of facilities and on reasons for not using health facilities, education facilities, etc. Furthermore, data available on actual spending outputs (e.g. the number of schools, clinics, teachers, water points, etc.) could be obtained from the relevant line ministries and bureaus and added to the poverty profile. By comparing the indicators for the poor felative to the non-poor, one can get a sense of which types of expenditure could prioritized if one wants to increase the poverty focus of the expenditure. For example, if the poor typically live about double the distance in kilometers from health clinics, but the distance to all- weather roads is only a fifth more for the poor, then building health clinics may be preferred to building roads, at least if one want to reduce the gaps in access to public infrastructure between the rich and the poor. Note that the welfare monitoring surveys cannot help with the exact location. Nor is there any guarantee that building clinics will in any case be helping the poor. For example, the same average distance to a heahh clinics can be obtained in a zone with a very high density of clinics in a few woredas and none in others, as well as in a zone in which clinics are equally spread. Finally, the decision criterion is to remove the largest bias in expenditure against the poor. It is not necessanly the spending with the highest poverty-impact For example, an additional road could reduce poverty by more than the additional clinic. Even so, local poverty profiles could provide the first step for local-level evidence-based poverty policy planning. Geogr4phical equity The welfare monitoring data could assist further in the geographical allocation of spending. The poverty profiles of differnt regions could be compared to assess which zones appear to have more deprivation in general or specifically by the poor in terms of variables such access to roads, schools or clinics. Relative deprivation of a zone in one of these characteristics may help to prioritize spending in a sector towards this zone, while for other characteristics, spending could be prioritized to other zones. Zonal level poverty profiles are only the starting point for evidence-based policy making. Within zones, important differences will exist. It is unfeasible to expect the national welfare monitoring system to provide more detailed data than what is currently being collected. At the local level, this information will have to be auginented with local level studies and knowledge. However, as a starting point, the representative data can form a useful basis. 3.37 Both MTRs note that there has been learning from experience (e.g. in the more timely preparation. of donor sub-programs), and concerns about financial and other performance information relate as much to the use of available information as to the absence of additional desired information. The HSDP MTR (pl9) noted: that except for the emerging regions, reporting systems on major health indicators and health sector activities are in place and are backed up by supervision missions. 3.38 The most salient issues fall into a number of interlocking categories: - Issues concerning expenditure tracking and management. - Issues concerning decentralization. - Issues concerning quality, access and poverty. - Implications for capacity building. 44 Expenditure tracking and management 3.39 Deficiencies in expenditure tracking relate both to the capacity to provide timely reporting through existing systems and, more fundamentally, to weaknesses in those systems and in the initial design of the SDPs. Timely reporting and closing of regional accounts has continued to be a problem, but making use of budgetary information to track the SDPs is more difficult because the component categories used in preparing the SDPs did not match the budget classification. In time, the reformed chart of accounts and the development of cost center budgets (see Section II above, and Appendix 1 [Vol II]) should enable SDP projections and actual budgets/expenditure reports to converge at an appropriate level of detail. Meanwhile, comprehensive tracking of sector expenditures is hampered by the continued difficulty in capturing donor spending as well as the weaknesses in reporting on budgeted expenditures. Additional lessons concern the earmarking and the programming of available funds. 3.40 Earmarking of donor funds has costs, whichever channel is followed. Both MTRs comment on the burdens imposed on limited capacity by having to follow donor-specific requirements. Channel lb as employed by the World Bank, was intended to be consistent with Government's own systems, and does pass funds through the financial bodies rather than the sector bodies; however, it amounts to a separate system within a system, and setting up reporting mechanisms designed to facilitate within-year reimbursement of funds had a high cost in terms of setting back progress in bringing the regions' regular accounts up to date. In ensuring that resources are used effectively, the year by year programming of available funds is at least as important as the initial broad component allocations. It is extremely difficult for regions to make effective use of available donor funds if these are not allocated and programmed well in advance of the fiscal year when expenditure is supposed to take place;29 Regions have been similarly handicapped by the late notification of the federal subsidy. Moreover, the disconnections between different components of expenditure make it difficult to optimize the composition of expenditures when funds are constrained. Decentralization 3.41 The SDPs also offer some important insights into the workings of Ethiopia's system of decentralization. The five-year expenditure programs of each SDP, worked out collaboratively by federal and regional bodies, demonstrate that, even though the federal government does not annually dictate regional budgets in detail, coherent sector expenditure programs can be developed. The SDP expenditure projections provide a crucial benchmark for the year to year review of performance. At the same time, regional strategies are in practice quite tightly specified by federal policies and standards. In fact, both MTRs argue that the regions should have more latitude for the adaptation of policies and expenditure patterns to local circumstances. Thus the ESDP MTR (pl9) recommends that regions should be clearly informed that they have the flexibility to choose the most appropriate school designs for their needs, while the HSDP MTR (p6) welcomed the flexibility allowed to the regions to adapt the federally prescribed norms to suit the local situation. 3.42 The measures that will be required to redress equity imbalances and to ensure quality are likely to vary both within and across regions according to the specific character of local problems and needs. The particular choice of measures to be adopted will have implications for budgetary allocations within sectors and between capital and recurrent budgets, and flexibility will be needed to allow different choices to be made in different places. Structures for community 29 HSDP MTR (ibid.): The experience has been that realising donor funds (whatever channel is used) requires more than the twelve-month period of the budget cycle. The recent review of SDP implementation problems now acknowledges the importance of shifting to multi-year programming of donor funds (including the World Bank sub-programmes). 45 management have a potentially significant role in determining how those choices are to be made. However, both MTRs noted that little progress has been made in involving communities in the management of services or the mobilization of resources, with relatively few operative parents' or health councils. Moving ahead on this objective is a necessary step both to enhancing quality and to ensuring accountability at the facility level. Quality, access and poverty 3.43 The anti-poverty objectives of both SDPs, though not always explicit, are quite evident. However, their impact on poverty will depend not only on the volume of sector expenditure but also on the distribution of expenditure across components within each sector. Generally, the evidence from both Reviews is that there has been an over-emphasis on achieving increased access and that this has compromised the quality of the resulting services. Nor is there a simple trade-off between quality and equity: lower quality reduces the benefits to those (including the poor) who access services, and also reduces usage of services, especially by poorer people. This is reflected, for example, in fewer visits to health facilities and higher dropout rates from school. 3.44 The bias against quality has a number of components. Planners have a bias towards physical expansion of services - as the ESDP review noted, "up until now regional planners have tended to see education in mainly quantitative terms, which is partly the reason why insufficient attention has been paid to qualitative aspects, which are always harder to measure and report on". New facilities are visible, and geographical extension of services is an obvious demonstration of the concern to reach communities not previously served (which are most often relatively poor communities). Donor preferences for capital projects tend to reinforce this bias towards new facilities. The ESDP review noted that the generally low levels of planned activities of expansion and rehabilitation of classrooms and schools are insufficient to counter the widely noted trends in facility deterioration. It is likely that expansion of facilities would have outpaced the training of staff even without the effects of the war. 3.45 Compromising quality appears to be neither an efficient nor a cost-effective means of increasing utilization of services, which is a necessary step towards improving welfare indicators and reducing poverty. To the contrary, there is evidence to show that how much a service is used is as much a function of the quality of the service provided as of how far users have to travel to reach the facility. Indeed, the same evidence suggests that it may be much less expensive to increase usage of services by increasing the quality of existing facilities than by building new ones.30 Periodic user satisfaction surveys can be a useful means of checking the relative importance of access and quality aspects of services. 3.46 Both MTRs highlight other aspects of equity that need further attention, however. Specifically, they draw attention to a lack of success in promoting female education and women's health, with the result that welfare indicators continue to be worse for women than for men. There is a well-established positive relationship between women's educational and health status and broader indicators of development, and all regions need to improve their service delivery to girls and women. The specific factors affecting low take-up are likely to be culturally-embedded and therefore location-specific. For the same reasons, the nature of the solutions is also likely to vary by region and within regions. Other equity questions relate to differences between regions. While both Reviews identified lack of capacity as a national problem at all levels, capacity was far weaker in the Emerging Regions31 than in the four strongest regions. The Reviews recommended that the former should receive additional support to enable them to catch up. Since the Emerging 30 See Collier, P; Dercon, S; and Mackinnon, J., Density versus Quality in Health Care Provision: The Use of Household Data for Budgetary Choices in Ethiopia, CSAE Working Paper, Oxford. 31 Afar, Somali, Benishangul-Gumuz and Gambella are categorized as Emerging Regions. 46 Regions start from a more disadvantaged position as regards social service indicators, without such affirmative action, they will lag further behind as the programs advance with resulting widening rather than narrowing differences between them and other Regional states. Capacity building 3.47 ESDP and HSDP experience holds lessons about the capacity building that is necessary for better tracking of expenditures, and, more fundamentally, for the more effective use of resources in addressing poverty: * First, capacity building needs more emphasis and more resources than it has received up to now. The ESDP MTR (pl9) argues that at least 10 percent of donor resources should be used for capacity building activities involving: - engaging of contract hire national or foreign technical assistance; - mass training and retraining of staff in key implementation areas of procurement, accounting and reporting; - procurement of computers and mass training of staff in their use; and - selective procurement of motorcycles and vehicles. * Second, it is as important to limit demands on capacity as to increase the supply. As the ESDP review (pl2) points out: - Scarce planning and management capacity in the regions is often wasted through the duplication of efforts expected of them. Reports have to be sent to the regional councils, and different reports to the ESDP secretariat; government accounts have to be prepared as well as statements of expenditure for donor funds; and last year there was a review of the government's five year plan and this year a mid-term review of ESDP. Efforts need to be made to reduce these and other wasteful duplication. - These objectives are far from being achieved, cut-backs in the recurrent budget having compromised the quality of management as well as the quality of service provision. Consequently, though reasonably adequate data do exist for policy formulation, planning and budgeting, and for monitoring and evaluation, the data are seldom analyzed or used in this way. This is reflected in the number of recommendations from the regular annual reviews of both sectors that have not been followed up. * Third, capacity building strategies must be long term. Reforming the budget structure and classification is an essential reform. It will not only provide better, more timely information on resource allocation and use; it will also improve accountability of cost centres, enabling information to be used for more efficient, and more decentralized management of resources. However, rolling out this reform to all regions will take several years. There are few 'quick fixes' and, as the experience of Channel lb shows, interim solutions may have the effect of delaying the permanent reforms that are under way. 47 * Fourth, systems and attitudes are important. Capacity building is not only about training and resources within the existing systems. It also needs to change the way that people think about resources, which includes changing the framework within which budgetary decisions are made (allowing strategic choices across recurrent and capital budgets, for example) and increasing awareness of issues of quality, access, usage, and poverty impact. Another important facet is proactive measures to involve the communities in the management of health and education facilities. Roads Context 3.48 This sector too is dominated by a Sector Development Program, but the RSDP is rather different in character from its social sector counterparts. The bulk of the expenditures take place at the federal level, with the Ethiopian Roads Authority (ERA) as the main implementing agency, and aid to RSDP provided in conventional project form. As described in Appendix 5 (Vol. II), strategies for rural roads and local level transport in general have been evolving and will receive more emphasis in the next phase of RSDP. Poverty relevance of expenditures 3.49 The potential influences of roads on poverty are rather complex, and include both direct and indirect effects. As well as the effects of roads on economic growth (and hence the generation of resources to finance future public expenditures), one could consider the equity effects of extending communications to different parts of the country, the reduction in vulnerability that may result from better access to markets, and the income opportunities arising from labour intensive approaches to road construction and maintenance. It is tempting to assume that rural feeder roads are especially important in creating opportunities for the poor (some panel data identified good communications as a factor in enabling some villages to reduce poverty more than others), but the road network is inter-linked - feeder roads need a main road network to connect with.32 Moreover, expenditure on roads has an opportunity cost in reducing resources available for other sectors. Accordingly, neither the appropriate level of public expenditure on roads, nor the best balance between different components of road expenditure is simple to determine. Gathering more evidence about the effects of roads could be helpful, but would not remove the complexity of assessing any causal links to poverty. Expenditure tracking 3.50 Monitoring of actual expenditures in the sector is relatively straightforward. Construction expenditure is dominated by the ERA, whose systems of accounting and expenditure control are relatively strong.33 The bulk of maintenance expenditures derive from the Road Fund, which is described in detail in Appendix 5 (Vol. II). There should be no major problem tracking most public spending on roads, in terms of identifying budget appropriations and verifying that budgeted resources were spent as intended. Budget preparation and implementation have been much strengthened in the ERA and the Rural Roads Authorities (RRAs) of the more developed regions. The accountability of the ERA and Road Fund, in particular is considered satisfactory. 32 VLTTS: The rural development process in Ethiopia appears to be one of hierarchical diffusion in which change and opportunities spread to the rural areas through the woreda towns, which in turn depend on the trunk road network. (Village Level Travel and Transport Study in Ethiopia - Final Report, IT Transport for Ethiopian Roads Authority, September 1999, 117). 33 ERA was examined in detail by the recent Diagnosis of Existing Control Capacities in Ethiopia- Provisional Report, Group 2 AC, August 2001 (Prepared for the European Commission), and received a predominantly favourable review. 48 The characteristics of the roads sector are such that the bulk of funding on roads can be tracked and audited more confidently than in other sectors (including health, education and agriculture) where expenditures are more diffuse and decentralized. 3.51 Getting a comprehensive picture of aggregate spending on village level rural roads in the context of the Rural Roads and Transport Program (RRTP) may in future be complicated by the lack of information on other sources of funding, such as from NGOs and from communities. This may also apply to future spending on village level rural roads under the RTTP in view of the likely multiple sources of funding and the large number of Community Funds to be set up. The issue of consistency and coordination among various woreda-level initiatives is discussed further in Chapter 5. Issues in performance and impact of expenditure 3.52 Performance indicators. The ERA annually monitors the impact of the RSDP by tracking changes in 16 indicators (see Table 2 in Appendix 5, Vol. II). Baseline data for most of these indicators were collected in 1997. Five indicators have been selected for possible use in monitoring the impact of public expenditure on roads: road density by type of road, vehicle kilometers traveled, roughness and road condition, vehicle operating costs and freight and passenger tariffs. None of these can serve as a reliable measure of the outcome of road expenditures. For example, how much of the rise in vehicle operating costs is due to the increase in the cost and in taxes on fuel and how much to the deterioration in the road conditions? Nevertheless, the indicators can be helpful in identifying problems and the need for corrective measures. Two indicators of the transport burden on rural households are discussed below. 3.53 Expenditure on roads andpovert reduction. As noted earlier, the poverty impact of investment in, and maintenance of, the main road network depends on how important lower transport costs and better connections between markets are for economic growth and, more specifically, for the implementation of the Government's ADLI strategy and, in turn, on the success of that strategy in reducing poverty. The 16 indicators monitored annually by the ERA mostly reflect outputs as well as a few that measure outcomes (e.g. vehicle operating costs). However there is no outcome indicator for spending on roads that reflects welfare levels, even as a proxy. It might be particularly useful to focus more on how outlays on rural roads and, more especially, village level roads, may benefit the poor, but to do this in a rigorous way would require sophisticated - and therefore expensive - surveys, 3.54 Rural roads. Of all public expenditure on roads, the rural road program may be of most direct benefit to the poor. While the focus of RSDP I was on the federal road network, RSDP 11 (2002-2007) is intended to stress development of rural roads, including village level rural roads. The broad framework for rural transport development is set out in the Rural Road and Transport Strategy (RRTS). The findings of the Village Level Travel and Transport Study (VLTTS) helped to gain a better understanding of transport problems at the village level and have been used to develop an approach towards reducing the village level transport burden as part of the broader Rural Roads and Transport Program (RRTP) - see Appendix 5 (Vol II). 3.55 The RRTP has set targets in terms of reaching a certain improvement in the outcome indicators. To achieve these targets it has planned an expansion of the low level road network based on assumptions about the contribution of the development of new facilities (water points etc.) and the availability and use of non-traditional means of transport. For example, it is assumed that 20 percent of the reduction in travel time will come from the establishment of new facilities. These assumptions will have to be tested in practice. The choice between a road and the construction of a new facility is supposed to be made at the village level. It will not be easy to track expenditures under the RRTP and to assess their poverty impact for a number of reasons: 49 * The choice of suitable outcome indicators. The most useful indicator, a measure of the transport burden, is hard to develop. The proposed proxy indicators are 'percentage of population living within a day's walk of an all-weather road' and 'average walking distance in hours from any type of road'. The drawback of these indicators is that (a) they do not reflect the favourable impact on the transport burden of the siting of facilities near a village (i.e. drilling of a borehole would significantly reduce the transport burden without changing the outcome indicator); and (b) they do not take into account availability of rural transport services on the new roads or their affordability; * The multiple sources of funding of the RRTP (Government, donors, Community Development Fund, NGOs); * The large number of small projects; this will make it hard to get an aggregate view of the program; and * The need for village level surveys to monitor the average hours walk from a road. 3.56 On the other hand, the method of project selection and management, with local people choosing priorities and taking responsibility for managing improvements, is designed to bring local knowledge and preferences to bear from the outset. It should not be necessary for regional and federal agencies to aggregate and evaluate every scheme. It should be enough to review a small sample to verify the programme's assumptions. Conclusions 3.57 Assessing results of expenditures in terms of developing and monitoring output indicators has been good in the case of roads and is expected to be adequate as regards future expansion and maintenance of the lower level rural road network, given the planned close involvement of Rural Road Authorities in planning, monitoring and the provision of technical support. This does assume a strengthening of the capacity of RRAs in the poorer regions. 3.58 Outcome indicators for low level rural roads have been developed and can be monitored, though only through village level surveys with regard to the 'average hours walk from a road'. They are of limited use in measuring changes in the transport burden, however, because of the effect of external factors. While a reduction in the transport burden would be a useful indicator of the poverty impact of expenditures, it is hard to measure. With regard to assessing the results of future expenditures on village level roads, surveys will be needed to monitor changes in the transport burden of rural households. 3.59 As regards intra-sector allocation, given the rapid expansion of the regional road network and the sharp deterioration in its condition (as evidenced by the relevant indicators), the Road Fund should consider giving regions a bigger share in the overall allocation for maintenance. At the same time, strengthening the capacity of Rural Road Authorities in the poorer regions to plan and implement road projects and road maintenance should be a high priority. Village Development Associations and Woreda Development Committees will need long term support in implementing the RTTP. This is going to put an additional burden on the RRAs, which need to have adequate resources to assist them. 3.60 Finally, care is needed in setting targets for public expenditure on roads. It is not a simple case of "the more the better". There are at least three complications: the possible crowding out of other sectors, the time profile of capital and recurrent expenditures, and - particularly at local level - the need to consider alternative ways of reducing the transport burden. 50 3.61 The first Public Investment Program (PIP) exercise revealed clearly that targets for federal capital expenditure on roads are inconsistent with the funding projections for other sector programs.34 There is a direct trade-off between the requirements of federal capital expenditure programs and the funds available for the regional subsidy, and hence for other sectors which are dominated by regional spending. Because the Government's MEFF and PIP have never been made operational, this issue has not yet been properly addressed. The PRSP will need to confront it directly, and propose a sustainable balance between sectoral (and hence between federal and regional) expenditures. 3.62 The RSDP pays careful attention to the maintenance requirements of new and rehabilitated roads, and the Road Fund is a now-familiar approach to ensure the availability of maintenance funds by hypothecating certain revenues. However, there is an additional issue concerning the time-profile of capital expenditures on roads. The RSDP is making up a backlog of road rehabilitation; this implies that public expenditure on roads (and especially the share of public expenditure on roads) may need to be above its long-run trend for a period. The corollary is that, once the backlog is made up, roads may attract a smaller share of public expenditures, so as not to crowd out priorities in other sectors or generate unsustainable maintenance requirements. The Government's medium term expenditure planning needs to model road expenditures rather carefully, and avoid simple 'rules of thumb' (such as "maintain the current share of total public expenditure") which could be seriously misleading. 3.63 Finally, the VLTTS rightly highlights the fact that improving roads is only one way of reducing the transport burden at village level. It proposes to make funds available at village, level to support whichever solution to the transport problem is deemed locally appropriate. This has implications for target setting, since it implies that funds will be not be exclusively earmarked for roads, and hence the balance between roads and other expenditures will become apparent only ex post. It illustrates the more general point, that you cannot simultaneously decentralize genuine responsibilities for expenditure allocation and also specify in advance the sectoral allocation of decentralized funds. V. INCREASING THE POVERTY FOCUS OF PUBLIC EXPENDITURES THE WAY FORWARD Strengthening the Poverty Focus of Expenditure Planning and Allocation 3.64 While it is important to consider the poverty relevance of expenditures - both their distributional impact and their effect on opportunities and capabilities of the poor - there are few if any cases where a simple, direct and unambiguous connection can be established between a line of expenditure and a poverty outcome. It is worth raising the question, even as there are no simple answers. Accordingly, the problem should be approached at three levels: (a) how to strengthen the availability and use of poverty information in general; (b) how to strengthen public expenditure management and the quality of information on public expenditure in general; and (c) how to make a more effective link between poverty information and expenditure management. 3.65 On welfare monitoring and poverty analysis the key theme should be to explore the complexity and diversity of poverty in Ethiopia. Debates about poverty can be more fruitfuil if there is wider recognition of: (a) the different dimensions of poverty, (b) the technical and 34 See PER 1999, (World Bank report no. 20283-ET, November 30, 1999) Volume II, 134, which cites the first (draft) PIP for EFY 1991-93: The Indicative Planning Figures for the ERA were Birr 1.9 billion for Treasury financing. By comparison, the ERA submission, which is in line with the Roads Sector Development Program, includes Birr 3.6 billion of Treasuryfinancing. This would exceed the total of Treasury resources expected to be available for _I capital investments in all sectors during the PIP period 51 conceptual difficulties inherent in measuring and monitoring poverty and (c) the different perspectives and priorities of various parties to the debates about poverty. It is especially important to recognise the diversity of experience across geographic, economic and social groups. A corollary is that welfare information needs to be disaggregated and used at local levels of decision-making. Broad, multi-dimensional information on poverty and welfare is needed not only to see whether poverty is declining, but also to try to understand its nature, its variety (social, geographical, etc) and what determines movements in and out of poverty. Answers are not easy, but there is great value in posing the questions. Section III of this chapter has noted that the principal challenges are to ensure the processing of data that is already collected, to deepen analysis beyond headline figures of the poverty count, to involve more users at all levels of government and in the wider community, to produce disaggregated analyses and profiles at regional and zonal level and to encourage the decentralized use of welfare data - especially by the regional level. This is a long term agenda, but it can start soon. 3.66 Recommendations concerning processing, analysis and dissemination include: reinforcing the Welfare Monitoring Unit; adopting an analysis program which uses the full range of available data; commissioning additional, more detailed, studies of particular issues; providing a wider range of researchers with access to welfare data; and producing more geographicallv disaggregated reports. In turn, the provision of regional and zonal profiles should be used to strengthen planning and resource allocation at local levels. An explicit dissemination plan for welfare data should be prepared. 3.67 On the strengthening of public expenditure management, the Expenditure Management and Control Program should be the primary focus of support. The aim must be to ensure that budgets and plans are transparent, well-constructed and properly reported (so that the links between budgets and strategies are visible). As regards fiduciary concerns, continuation of accounting reforms and support to the audit function should be priorities (see Box 3.4). At the same time, Ethiopia's record is such that it deserves the "benefit of the doubt" on fiduciary management: acknowledged deficiencies that are being addressed should not be used as an excuse to defer budget support to some indefinite future date. Ethiopia should be encouraged to proceed as fast as realistically possible with refonns to the chart of accounts and reporting systems that will allow a cost-center approach to analysis and management of expenditures. These reforms will, inevitably, take time, but they are making demonstrable progress. 3.68 As stressed in Section II, from the standpoint of tackling poverty through public expenditure, Ethiopia's deficiencies in medium term expenditure planning are a much more serious weakness than the shortcomings in expenditure reporting. Meaningful discretion within the time-frame of a single annual budget is strictly limited. Therefore, it is only through a medium term planning process that the issues of appropriate expenditure composition (the balance between sectors and sub-sectors, between federal and regional levels, between recurrent and capital expenditures, etc) and the targeting of expenditures (their efficiency and distributional impact) can be addressed. Only through medium terrn scenarios can the government and potential donors explore the difference between 'with-aid' and 'without aid' scenarios, as a basis for monitoring whether aid does, in the event, make the difference that was intended. Government needs to give this reform the political support and priority that it has so far lacked, not only because of its inherent importance for the better implementation of government strategies, but also because of its role in facilitating dialogue with aid partners and unlocking more valuable forms of aid. Accordingly, current efforts to operationalise medium term expenditure planning at both federal and regional levels should be strongly supported. 3.69 In bringing together information about poverty and the processes of expenditure management, two key themes should be decentralization and disaggregation. Disaggregated information (e.g. geographically by region, zone and even woreda) needs to be used at 52 decentralized levels if it is to be effective. It should be used to analyze the diversity of local livelihoods, and address issues of expenditure incidence, service usage and quality, not just the broad allocation of resources. Each region needs to develop strategies (and adapt national policies) to address their own specific circumstances. The usefulness of cross sector collaboration and review increases the more decentralized it is. Box 3.4: Capacity Building - the Role of the Civil Service Reform Program Ethiopia's Civil Service Reform has suffered from a lack of acknowledgement and respect from many international development agencies which have not taken the time to understand this homegrown reform much less support it. Instead, some development agencies have proceeded with their own agendas even though the success of their agendas often depends on critical inputs from the Civil Service Reform. While external donors and creditors want to "build government capacity" so that assistance and loans can increasingly be provided through non-project support after four years no interrnational development agency has stepped forward to assist the government in developing the internal or external audit sub-programs of the Civil Service Reform. Audit capacity is essential to ensure that non- project assistance is accounted for. Audit capacity is also essential for building external and internal controls, which are the preconditions for the management reforms (e.g. output/outcome budgeting) development agencies are pressing for. Building capacity begins by recognizing existing capacity, a commitment to not duplicating capacity, and an appreciation of the importance of ownership by government in the capacity building process. Stephen B Peterson, Financial Reform in a Devolved African Country: Lessons from Ethiooia April 25, 2000, p23. 3.70 Some further points should be borne in mind when assessing the ability of a public expenditure management system to track expenditures: - Collecting information has costs which need to be weighed against the benefits of the tracking information gained. Too often benefits are taken for granted and the costs of collecting information are discounted. Where there are capacity constraints, the management and opportunity cost of providing information may be very high. - Standards of evaluation will vary depending on what the information is required for. In particular, if tracking information is required to act as a trigger for within-year disbursements (or reimbursements), the speed and specificity of information required will be much more demanding. Links Between Expenditure Planning and Expenditure Tracking 3.71 The HIPC guidelines rightly stress the importance of the expenditure planning that is necessary to provide a benchmark for expenditure tracking (see Box 3.5). As noted in section II, the EMCP has been working towards the development of the kind of medium term economic framework that is required. The I-PRSP policy matrix includes an undertaking to introduce the Macroeconomic and Fiscal Framework in the coming fiscal year. (As previous PERs have reiterated) this needs to become the benchmark for annual discussion of expenditure allocations, including the use to which HIPC and other aid resources are put. The MEFF can provide a suitable context for review of proposed expenditure targets. As Box 3.6 indicates, such targets need to be approached with caution, and the MEFF provides a suitable framework for keeping them under review. 53 Box 3.5: A Paradigm for HIPC Expenditure Planning Key guidelines from the World Bank and IMF35 include: * Tracking of poverty-reducing public spending under the HIPC Initiative should include a clear identification of the amount of resources released and an ability to assess the changes in the overall composition of spending. (1j0) * A first step in tracking the impact of the Enhanced HIPC Initiative is to measure resources released by HIPC assistance. Governments should also be encouraged to include data on HIPC assistance in their budget documents. (¶11 ) * Once the overall size of HIPC assistance is identified, a PEM system should be able to determine how the total amount and the detailed allocation of overall poverty-reducing spending has changed as a result. This requires a baseline against which to assess changes in overall spending. Ideally a medium termn expenditure framework (MTEF) should be in place prior to HIPC assistance to provide such a baseline. a full-fledged MTEF, a long-tern goal of many countries, could include sub-national governments, with expenditures classified comprehensively by function and by program. There would also be a clear understanding of the relationship between the MTEF, on the one hand, and intermediate targets and final objectives for poverty reduction, on the other hand. Tracking of the impact of the HIPC Initiative on poverty-related spending would then involve comparing a baseline MTEF, prepared before the receipt of HIPC assistance, with one that includes HIPC assistance. (116) Consistency with Decentralization 3.72 Ethiopia's federal system is unique and still evolving. Some aspects are not always well understood. In particular, the degree of effective autonomy of the regions is sometimes exaggerated and this could lead to misconceptions about the scope for setting and monitoring expenditure targets that affect the regions as well as the federal government. In practice, the Constitution gives the federal government a leading role in setting national policies and strategies, and it also sets the broad financial parameters for the regions through its determination of the size of the block grant (augmented by its control over aid funding). There are strong formal and informal mechanisms, including the political party structure, that ensure consistency in policy between federal and regional levels, so that in practice, regional discretion relates much more to implementation than to the underlying policies themselves. As the I-PRSP puts it: With the devolution of power to the regional governments, implementation of economic policies and development programs have, to a large measure, been shiftedfrom the center to the regions.36 As regards the setting of targets, the SDPs have demonstrated that it is perfectly feasible for the federal government and the regions to collaborate in drawing up expenditure programs that cover all tiers of the administration.3' During the PER Discussion Workshop (September 2001), the regional representatives said that they would have no objection to the Federal Government prescribing some guidelines towards earmarking HIPC resources for poverty reducing sectors provided they have the flexibility on intra-sector allocations. 3.73 It is worth noting that, although the unconditional federal subsidy is the centrepiece of transfers from federal to regional level, there are also important transfers which are tailored to the needs of particular regions and/or whose end-use is specified. Earmarked aid is passed on to M38 regions as such (although the offset system tries to ensure that the balance of total allocations 35 From IMF/IDA, Tracking of Poverty-Reducing Spending in Heavily Indebted Poor Countries (HIPC), Prepared by the IMF's Fiscal Affairs Department and the World Bank's Poverty Reduction and Economic Management Network, March 27, 2001. 36 I-PRSP p 1 3 (emphasis added; page numbering is from the version published in World Bank report no. 21769-ET, January 30, 2001). 37 Significantly, the Mid Term Reviews for the Education and Health Sector Development Programmes both argue that regions should have more effective autonomy in regard to the programmes, not less. 3 For an up-to-date description of the offset system see the Appendix 7 (Vol II) of this report. 54 among regions is not distorted). Food aid is the largest specific transfer and is not counted towards the offset. The way that the Road. Fund allocates resources to Regional Roads Authorities (see Appendix 5, Vol. II) is tantamnount to a specific purpose grant. Thus the federal system is neither as simple nor as rigid as sometimes portrayed. Box 3.6: Setting Targets for Expenditure Shares Overview Clearly, the allocation of public expenditures can make an important difference to poverty. The poor are likely to benefit if, for example, public expenditures are shifted away from defence and debt service and towards primary education and health. The I-PRSP commitment to increase expenditure shares for education, health, agriculture and roads should be broadly welcomed. At the same time, there are many reasons to be careful not to put too much weight on such targets - especially if they are to be precisely tracked and used as a main indicator of government performance and/or a condition for the release or continuation of HIPC funds and other similar support. Reasons for caution relate to (a) conceptual difficulties in defining an optimal composition of expenditures; (b) practical difficulties in measuring shares of public expenditures; and (c) institutional issues in the setting of such targets. Conceptual issues Identification of these four broad sectors omits some other key public expenditures that feature strongly in welfare (e.g. access to safe water). Intra-sector issues are glossed over: a target for a broad sector implicitly assumes that the whole sector is equally poverty relevant .In reality, as discussed in Section IV above, there are many issues concerning the intra-sector allocation of expenditures: for example, a current debate concerns the shift in balance between primary and tertiary expenditures in education.3' In every sector the balance between recurrent and capital expenditures is crucial, but it requires analysis at sub-sector level, not an arbitrary target for the capital/recurrent ratio of the whole sector. There is bound to be a degree of arbitrariness in targets. There is no "optimal" share of public expenditures for a particular sector that can be based on intemational experience.40 It does not follow from the identification of a sector as pro-poor that increasing the level or share of expenditures for that sector will always be appropriate, particularly in view of possible interactions and trade-offs between the "pro-poor" sectors. (For example, see Section IV above for reasons to be cautious about targeting ever-increasing levels of expenditure for the roads sector in Ethiopia). In focusing on 'headline' expenditure figures, such targets abstract from all the issues of quality, incidence, effectiveness and usage (discussed in Section 111) and ignore the scope for diversity in local priorities. Measurement problems Expressing a sector's target as a percentage of total public expenditure makes its achievement dependent on what may happen in unrelated sectors (under-expenditure in another sector could cause the target to be achieved quite spuriously). Comparing sector expenditures from one year to the next is tricky, particularly in a context when there are substantial resource flows that are not captured by the budget. Progress in capturing aid flows more fully into the budget may also create the illusion of an increase in public spending. The costs of measurement (gathering the information to show whether the target has been met) could be high, particularly if it requires administrative resources to be allocated to a special task. Institutional problems In the Ethiopian context, the respective roles of federal and regional governments in setting targets that incorporate both federal and regional budgets are not clear. Even if it is formally legitimate for the federal government to set such expenditure targets for the regions, it might be seen as undermining the principles and purposes of decentralisation. Furthermore, if targets are set in this way, and aid flows are made to depend on their achievement, a region could find itself penalised on account of the failure of other regions to meet their targets. 39 It should be stressed that, even when a high weight is put on poverty reduction, the 'correct' balance between primary and tertiary expenditures is not self-evident. The ESDP MTR noted a shift towards tertiary education spending, compared with original ESDP projections. However, the I-PRSP argues, also plausibly, that tertiary education is crucial to the capacity building which is one of the-pillars of the entire poverty reduction strategy. 55 3.74 Nonetheless, the federal system does have important implications for expenditure tracking. It is not straightforward for the federal government to make precise expenditure commitments as to how the aggregate of (federal plus regional) public expenditure will turn out. Inevitably, reporting chains are longer in a decentralized system, and this affects the speed with which detailed expenditure reports can be assembled. This reinforces the importance of developing a joint federal-regional system for the medium term planning of expenditures. Implications for the PRSP Process 3.75 The guidelines for PRSPs and for HIPC stress the importance of government ownership, including the responsibility of governments for expenditure tracking and for the strengthening of public expenditure management. GoE has an excellent foundation in its existing poverty strategy, its expenditure reform program and in its basic welfare monitoring system. The principal challenges as the full PRSP is prepared will be: * To use available data as far as possible to present a more comprehensive picture of the nature, causes and dimensions of poverty in Ethiopia.41 * In doing so, to recognize the limits of what can be learned from existing data and to spell out priorities and a program for additional research and analysis. * To put forward a convincing timetable for the reforms in public expenditure management that are under way (as a basis for a common understanding between government and donors as to what are the priorities, and what is a realistic pace of institutional reformn).42 * To show how the capacities for welfare analysis and expenditure planning will be linked and strengthened at each level from the center down to woredas.43 * To explain how the Government intends to reconcile the imperatives of decentralization and of medium term expenditure planning. How, within the framework of the MEFF and the eventual development of Public Expenditure Programs at central and regional levels, will mutually consistent expenditure targets be established? 40 A point that is made forcefully in Sanjay Pradhan, Evaluating Public Spending - A Frameworkfor I'ublic Expenditure Reviews, World Bank Discussion Paper 323, May 1996. 41 cf. Joint IDA/IIMF StaffAssessment of I-PRSP (¶5): .. a more comprehensive data analysis would be appropriate for the full PRSP; one that indicates the main causes of poverty and shows how these findings have influenced government policy for poverty reduction. A more detailed analysis is also required to identify particularly vulnerable groups, with disaggregation by region, gender and other relevant social categories. 42 A Country Financial Accountability Assessment (CFAA) is planned shortly. It should focus primarily on the adequacy, pace and sequencing of the reform programme. 43 In this context it should also indicate how the multiplicity of woreda-level initiatives that are in the pipeline (including funds linked to roads, HIV/AIDS and food security) will be reconciled with each other and with decentralisation strategy as a whole. 56 4. DONORS SHIFTING TO SECTOR AND BUDGET SUPPORT I. INTRODUCTION 4.1 Ethiopia is arguably the poorest country in the world, and has more people living in poverty than any other country in Africa. Yet per capita aid flows declined during the 1990s (from $20 in 1990 to $11 in 1998) and this level of per capita aid is the lowest for any country in Sub-Saharan Africa except Nigeria (see Table 4.1). The logic of the International Development Goals is that Ethiopia should receive substantially more aid than it does. The apparent underperformance in the provision and utilization of aid should be of concern equally to the government and to its aid partners. Table 4.1: Debt and Aid 1990 and 1998 for Selected Developing Countries External debt Official development assistance Total $millions Value '98 Dollars per capita % of GNP 1990 1998 %ofGNP 1990 1998 1990 1998 Ethiopia 8,634 10,352 135 20 11 15 10 Kenya 7,058 7,010 45 50 16 14.7 4.2 Mali 2,467 3,202 84 57 33 20 13.5 Nigeria 33,440 30,315 74 3 2 1.0 0.5 Namibia - - - 90 108 5.1 5.8 Ghana 3,881 6,884 55 38 38 9.7 9.6 Tanzania 6,438 7,603 71 46 31 29.3 12.5 Uganda 2,583 3,935 35 41 23 15.8 7.0 Source: World Development Report 2000101 (as presented in Philippa Bevan, Poverty in Ethiopia, paper prepared for DFID, 6 September 2000) 4.2 Aid management in Ethiopia is complex because of the number of multilateral and bilateral donors and the diversity of systems and procedures that results. Further complications stem from the depth of decentralization, and the objective of maintaining equity across regions in the distribution of public resources. Food aid has been a major resource flow, and there has been controversy about whether and how food-related expenditures (including food aid) should be integrated into the government's planning and budgeting cycle. Aid management has therefore been a focus topic in several past PERs. 4.3 The volume and the effectiveness of aid are closely linked to the aid modalities used. There is international recognition that traditional project-oriented modalities have serious deficiencies, and a consequent trend towards sector wide approaches and more general budget support. This trend is gaining additional momentum from the need to program the resources released by HIPC debt relief. For PER 2001, therefore, it was agreed to revisit the topic of aid management from the perspective of PRSP and HIPC, and with the specific objective of identifying the measures necessary to encourage donors to shift away from project-focused forms of aid towards budget and sector support. 4.4 Ethiopia has made substantial progress towards better ways of managing aid, both through the approaches associated with the Expenditure Management and Control Program, and more especially through the Sector Development Programs (SDPs) which GoE initiated. Unfortunately, full realization of the SDPs' potential has been hindered by the hiatus of the war, but they have established an institutional framework which can be further developed and which provides lessons for the design of more general budget support mechanisms. 4.5 This chapter reviews (a) relevant concepts and issues in aid management, including the case for shifting to sector and budget support; (b) lessons that can be drawn from Ethiopia's experience, and current donor and government attitudes the aid relationship and (c) the specific requirements and opportunities introduced by HIPC and the PRSP process. The final section considers the way forward, identifying specific requirements from both donors and the government. 11. CONCEPTS AND ISSUES IN AID MANAGEMENT Terminology 4.6 The following terms need to be used with care if confusion is to be avoided. - Budget support is used here to mean aid to the government whose use is not formally restricted to any particular project or sector. - Sector support is used here to mean aid to the government that is restricted to a particular sector (e.g. health or education) but is not more specifically earmarked to a particular project, region or budget line. - Project support is aid that is required to be used for a specific pre- agreed project. - Earmarking is the requirement that funds from a particular (donor) source should be applied to a specific project or budget line. Earmarking implies that the donor's funds will need to be tracked so as to demonstrate that they have been applied to the specific expenditures stipulated. (Note that, with the definition of budget support used here, "earmarked budget support" is a contradiction in terms.) - Fungibility: funds provided for a specific purpose, and applied as agreed to that specific purpose, may nevertheless release resources for other, possibly quite unrelated uses. If the government would itself have financed an aid funded project or activity, the provision of aid frees those resources for other uses.44 - Disbursement channels in Ethiopia have been classified in a very specific way (see Box 4.1). Note that Channels 1, 2 and 3 describe how funds are disbursed, and are quite independent of the manner in which aid has been negotiated. 4.7 In practice there is an entire spectrum from pure budget support to earmarked project funding; and there are several possible variants within each main type [e.g. sector support could be provided at the level of a sub-sector (e.g. primary education), but support to a very narrowly defined sub-sector could be tantamount to project support]. It is nevertheless important not to lose sight of the underlying distinctions. 44 PER 2000 (see Chapter 4, I4.30-4.33 and Box 4.5) includes a mnore thorough explanation of fungibility 58 Box 4.1: Aid Disbursement Channels Aid disbursements are classified into three main channels, as shown below. Funds disbursed through the federal Ministry of Finance, which are invariably captured in the budget, are categorized as Channel 1. Channel 2 represents donor aid that is disbursed through sector bodies. Funds disbursed directly by donors without being handled by any government agency comprise Channel 3. (See Chapter 5 for explanation of the federal structure.) The government has an express preference for Channel I on the ground that aid flows through Channels 2 and 3 undermine the integrity of the planning, budgeting and accounting systems. CHANNEL I CHANNEL 2 CHANNEL 3 (via Finance bodies) (via Sector bodies) (direct) Nrinistry of Finance |t ni Regional Finance ...... egional Sector Bureau JBureau j V I | Zorial Finance | r Zonal Sector l Deparbnent 1 Deparbiint Woreda Finanlce |a Sco l ~~~~Offlice | >Of¢ Note: World Bank support to the Health and Education Sector Development Programs has used a variant of Channel 1, known as Channel lb. Funds are passed through the finance bodies but are kept in separate accounts and earmarked to specific projects. 4.8 Much confusion can arise between two distinct aspects of aid management: (a) the extent to which aid is incorporated in government budgets; and (b) the degree to which aid is earmarked to specific uses. Whether aid funds pass through, or are in some way reflected in, government's budget and accounts is a quite separate issue from whether aid is provided in unearmarked form to the budget as a whole (although aid of the latter type is, by definition, fully captured and incorporated in the budget). Good resource management practice requires a comprehensive, and multi-year, view of the resources available, so that relevant trade-offs can be taken into account, and so as to provide the predictability which is necessary for efficient use of resources. Good aid management practice therefore requires that aid should be incorporated in this overall view of resources. Ideally aid would be fully incorporated in the budget, as part of the overall appropriation, but, where this is not possible, aid should at least be included in planners' overall assessment of resource availability, and its use should be reported at least as memorandum informationi in the government's accounts. Ethiopia's financial regulations set a high standard in defining aid as part of the broad definition of public resources all of which ought to be reflected in the budget. In practice, as discussed in Appendix 1 (Vol II), much aid remains uncaptured 59 either at the planning or the reporting stage, and there is a continuing need for donors to improve the notification of reporting of aid flows. Arguments for Moving from Project aid to Sector and Budget Support 4.9 The arguments for shifting to sector and budget support start with an acknowledgement of the two principal weaknesses of project aid. First, the earmarking of donor funds to specific projects has high transaction costs, especially in a context where there are many different donors with different preferences and administrative requirements. This PER (in its review of sector and regional expenditures - see Chapters 3 and 5) is only the latest in a whole series of studies which have shown that absorption capacity is much higher for treasury-funded than for aid-funded activities; the extra administrative burden associated with project earmarking results in much lower implementation rates for aid-funded than for treasury-funded projects. A major justification for project earmarking is the "comfort" that it provides to donors that "their" funds are being put to proper use and oftentimes that there is 'visible physical evidence' to their aid effort. But therein lies the second weakness of project aid. Because of the potential for fungibility, this comfort is illusory. Rational donors should instead seek comfort from a review of public expenditures as a whole, not from the narrow targeting of their own funds. In practice, a further cost of project approaches has been the very limited predictability of aid flows, and the greater likelihood of project aid to follow disbursement channels outside the mainstream government budget. This in turn undermines government efforts to move towards contemporary standards for comprehensive medium term planning of public resources. 4.10 The potential benefits of a shift to sector and budget support thus include: - Increased focus on the broad strategies and policies that ultimately determine the effectiveness of public expenditures. This involves explicit attention to aggregate fiscal discipline, to the optimal allocation of resources across sectors and sub-sectors, and to technical efficiency in the utilization of resources. - A similar focus on the strengthening of institutions for public expenditure management and public service delivery. - The potential for simplification and harmonization of aid procedures. - Arising from all of the above, the potential to enhance and accelerate aid disbursements, in a context where their effectiveness in realizing the objectives of the poverty reduction strategy can be kept under regular joint review by government and its aid partners. 4.11 In the Ethiopian context, decentralization reinforces the case for sector and budget support. Few donors doubt the virtues of decentralization - indeed many are strong advocates of greater local empowerment. Yet one effect of decentralization is to lengthen the administrative chains for disbursement and accounting. This in turn increases the transaction costs associated with earmarked aid. Moreover, the federal system does not allow the central government to dictate to regions the details of their public expenditure allocations (as opposed to the broader policies and strategies that should be followed across the nation). Sector and budget support thus has the potential to be both more efficient and more in tune with the decentralized political and administrative framework. A Continuing Role for Project Aid 4.12 However, there remains a valid role for project aid. For government itself, projects often remain a sensible management approach to the planning and implementation of capital expenditures. Many donors (including for example most UN agencies) are obliged by their 60 mandates to provide assistance in project form and do not have the option of shifting towards unearmarked forms of disbursement. A number of donors value the close contact with implementing agencies and with beneficiaries that can result from involvement at the project level. (Thus some donors who plan to shift towards sector and budget support, nevertheless intend to provide some assistance as project aid, as a means of "keeping in touch" with development in Ethiopia.) The relationship between project aid and sector or budget support can be complementary. The policy dialogue, the review of overall resource allocations, and the definition of clear sector strategies that are associated with the broader forns of support should help to ensure that project aid is focused on priorities, that the policies are in place to maximize the prospects of project effectiveness, and that project resources are not, through fungibility, supporting inappropriate patterns of expenditure. 4.13 The influential World Bank study, Assessing Aidi, which heightened awareness of the significance of fungibility, and hence the futility of much narrow conditionality, also drew attention to the non-financial dimensions of aid. Aid is characterized not merely by financial transfers but also by accompanying technical assistance, technology, ideas and approaches to development. The way aid is managed also has importance in terms of feedback to the constituencies which support aid in the first place. It is important to design aid management mechanisms which allow the non-financial benefits of the aid relationship to be realized, and which also take into account the political and psychological factors which can affect support for the continuance of aid. Studies in Ethiopia and elsewhere have repeatedly shown the high cost of earmarking of funds. There is growing support for moves towards broader forms of support. As this happens, the challenge will be to maintain the added value of non-financial dimensions of aid, and also to maintain adequate links to the donor-country constituencies for aid. m. AID MANAGEMENT EXPERIENCES IN ETHIOPIA The Importance of Aid 4.14 External loans and grants account for nearly a third to half of the total government expenditure, and their share is likely to increase in the years ahead consequent to the government reorienting public expenditure from defense to social sectors (Table 4.2). This alone makes aid an important aspect of Ethiopia's public finance management. 4.15 The introduction section already drew attention to the disappointing levels of aid to Ethiopia Reasons for more recent stagnation lie in donor reactions to the war with Eritrea, before which Ethiopia was one of the early leaders in the adoption of sector wide approaches and appeared poised for a substantial increase in levels of support. Lessons of the Sector Development Programs 4.16 The SDPs provide important lessons concerning the management and effectiveness of aid in Ethiopia. In Ethiopia, SDPs for health, education and roads sectors were initially developed by the government and then refined in consultation with the relevant aid agencies. They cover the first five years of a twenty-year strategy starting in 1999. The SDPs are accompanied by a Program Action Plan (PAP) that summarizes the strategy and its associated expenditure targets, more detailed education and health SDP documents for each region, and a Program Implementation Manual (PIM) describing implementation procedures for each sector. Besides the general characteristics of SDPs, a unique feature of the Ethiopian SDPs is that they were prepared 45 Assessing Aid: What Works, What Doesn't and Why, World Bank Policy Research Report, 1998 (Published by Oxford University Press) 61 by the federal government, in a somewhat pioneering manner, in collaboration with the regional administrations. 4.17 There is an agreed framework for annual consultations, in which an Annual Review Meeting is preceded by a Joint Review Mission to report on the implementation of each SDP. The roads SDP is based on an overall sector strategy, but is financed through conventional project modalities and mostly implemented by the federal government. The ESDP and HSDP were intended as a more radical move towards a sector wide approach. It was expected they would pave the way for high quality policy dialogue between the government and the donors, substantially simplify procurement and disbursement procedures and support a major boost to spending on these sectors. The reality has fallen far short of these expectations. On the positive side, their preparation has helped to define sectoral policies and expenditure programs. There has also been some progress in designing common financing channels and procurement procedures. However, ESDP and HSDP essentially stagnated as a result of the interruption to donor support on account of the border conflict. The non-appearance of promised donor funding seriously undermined implementation and at the same time made the Government disinclined to engage the same donors in policy dialogue. Table 4.2: Summary of Government Finance 1/ 1997/98 1998/99 1999/00 2000/01 2000/01 2001/02 PER Program 2/ Estimate Budget 3/ Percentage of GDP General Govemment Recurrent 15.8 20.8 26.4 21.5 19.6 19.3 Capital 9.2 9.8 6.6 9.6 9.5 10.2 Total 25.0 30.6 33.0 31.2 29.1 29.5 Federal Govemment Recurrent 9.5 14.6 20.7 15.4 13.0 12.5 Capital 5.6 6.7 4.6 7.0 6.4 7.5 Total 15.1 21.3 25.3 22.5 19.4 20.0 Regional Govemment 4/ Recurrent 6.3 6.2 5.7 6.1 6.6 N/A Capital 3.6 3.2 2.0 2.6 3.1 N/A Total 9.9 9.4 7.7 8.7 9.7 N/A Extemal Aid Grant 2.8 3.6 3.3 6.0 5.0 5.3 Loan 2.5 4.4 2.6 8.4 5.0 8.0 Total 5.3 8.0 5.9 14.4 10.0 13.3 Percentage of govemment expenditure External Aid. Grant 11.3 11.8 10.0 19.2 17.1 17.9 Loan 10.0 14.4 7.9 27.0 17.4 27.3 Total 21.3 26.2 18.0 46.1 34.4 45.2 1/ Figures for 1997/98 through 1999/00 are preliminary actual expenditures. 2/ Program under the three-year IMF PRGF arrangement. 3/ Consistent with program under PRGF arrangement, with lower defense expenditure, and excluding Enhanced HIPC assistance. 4/ As disaggregated figures are not available, regions' capital expenditures do not include external Joan and assistance for FY98, FY99. FYOO and FYO1 expenditure. 4.18 Nevertheless there are some very positive aspects to the experience gained from the SDPs; they represent a platform that can be built on and provide important lessons for the design of mechanisms to manage budget support. Chapter 3 (and Appendix 4, Vol. II) describe the ESDP and HSDP from the perspective of poverty reduction and summarize some of the key findings of 62 the recent Mid Term Reviews (MTR). The SDP experience offers the following lessons for the design of appropriate aid modalities: * It has to be recognized that project-focused earnarked modes of aid will continue to co-exist with unearmarked flows, and therefore systems need to accommodate both. * The formulation of the SDPs demonstrated how federal and regional levels can work together (and with donors) to develop a coherent expenditure program that has a multi-year perspective and addresses both recurrent and capital expenditures. They may be seen as a precursor of the PEP preparation process that is now being initiated. * A weakness of the SDPs, however, is that they did not from the outset sufficiently involve the finance and planning bodies, either at the federal or regional level. A result was to strengthen the federal sector ministries' role in aid coordination and program administration (instead of confining them to the policy and support roles envisaged by the federal system) and to bypass the intended role of regional planning and finance bodies. In the process this has resulted in a somewhat centralizing tendency. * In the absence of an MTEF or equivalent, there was no mechanism for linking sector-level discussions to the assessment of aggregate expenditure targets, and known inconsistencies between the expenditure targets of the various sector programs were, as a result, allowed to persist. The need to link the SDPs to wider discussion of overall resource flows represents an opportunity in the context of HIPC/PRSP to complete the necessary institutional architecture. * The SDPs show the importance of a fully articulated governance structure, comprising not only the comprehensive initial planning documents, but also procedures andforums for joint planning and review. A number of commentators have referred to the SDPs as if they are a failure. As a system of collaboration this is certainly not the case. The fact that joint MTRs have come up with a very incisive critique of SDP performance, is highly commendable, and is a direct outcome of the collective focus on the sectors as a whole, rather than separate projects. The challenge now is to see that the issues that have been raised are seriously addressed. * The SDP experience also demonstrates the importance of attention to the ongoing programming of funding programs: even where funds are not earmarked ex ante, there has to be an efficient mechanism for applying available funds to particular projects and activities. Moreover, it is impractical to carry out this programming exercise within the confines of an annual budgeting process.46 * The SDPs have also demonstrated the high costs of earmarking and of setting up special tracking systems that are 'bolted' on to the basic government implementation systems. 46 This point is graphically illustrated by the World Bank's experience with the preparation of annual sub- programmes to utilize the its funding for ESDP and HSDP. At the design stage the Bank ignored the advice that annual (as opposed to multi-year) programming of funds was certain to be a serious and unnecessary bottleneck. The topic of programming was omitted from the Programme Implementation Manual. Most recently the report of the SDP Special Task Force has again emphasised the importance of moving away from an annual to a three-year rolling budgeting system. 63 Current Donor Attitudes 4.19 Budget support recommends itself on account of the several advantages discussed earlier. Like other governments, GoE too has a decided preference for budget support on the main ground of its potential to enhance the development impact of external aid. Donors, while being appreciative of this argument at a broad theoretical level, have been shy of conceding the request mainly on the ground that they feel only marginally involved in the development agenda of the government. The PRSP context has the potential to change that as donors will be among the stakeholders actively consulted in the formulation and implementation of the PPSP. 4.20 A satisfactory PRSP process is a necessary, but not sufficient, condition for changing donor perceptions and encouraging them to shift to budget support. There are other concerns troubling the donors, and they can be conveniently grouped into three broad areas: political, programmatic and operational. * Political: In the aftermath of the tragic and costly war with Eritrea, donors are very sensitive to the progress of the peace process. The reduction in defense spending as well as demobilization of soldiers has gone a long way to inspire their trust and confidence in the sustainability of peace. Several donors, who suspended aid during the war years, have resumed assistance while a few are waiting further reassurance. Whether, and how soon, donors will shift to budget support will depend on the robustness of the peace process. (A number have assumed that a resumption of project support is logical stepping stone towards budget support. The validity of this assumption is challenged in Box 4.2 below.) * Programmatic: Many donors are wary of the government's reluctance to engage in a transparent and comprehensive dialogue on the policy framework. That they have not been taken into confidence on the I-PRSP has also been a cause for dissatisfaction among some donors. There have, however, been some encouraging developments since. Both the apex Steering Committee and the operational level Technical Committee for the formulation of the PRSP have held several meetings with the Development Assistance Group (DAG) of donors. In the most recent meeting in early November, 2001, the Government promised more regular and structured meetings. Being part of the PRSP consultative process is important to the donors, but equally important, from the point of view of budget support, will be that the larger civil society is involved in an open and broad based manner, and that their views are reflected in the final PRSP. * Operational: A necessary pre-condition for donors to shift to budget support will that they get timely and reliable reporting on the expenditures as well as outputs. The concern here is both fiduciary and management accountability. Although the relevant procedures and mechanisms are already in place, such as in the case of SDPs, the main deficiency has been that the government has not been able to produce the reports. Improvement in this area will be important. These issues have been reviewed earlier in Chapter 3. 64 Box 4.2: Some Issues and Paradoxes in Aid Management Aid management involves a complex interplay of economics, politics, rationality and psychology. There is much scope for misconceptions and paradoxes. Approaches that are intuitively attractive can have perverse effects and an influence on poverty contrary to what the donor intended. Some examples and issues: Capital bias: the donor preference for earmarking funds towards capital projects reinforces a bias that can work against the effective delivery of services to the poor. Chapter 3 of this PER draws attention to cases in education and health, where rehabilitation of existing facilities has received too little attention, and the preoccupation with capital expansion has been accompanied by neglect of quality and efficiency issues (which depend on adequate recurrent funding and effective management of facilities). Poor quality of services deters poor people from accessing facilities, and undermines their benefits when they do. High costs of earmarking: A whole succession of PERs, SDP reviews and other studies have highlighted the administrative costs of earmarking funds. Implementing agencies have shown that 'capacity constraints' on utilizing Treasury funds, are far less than those that apply when the extra bureaucratic procedures of donor funding are applied. No evidence has appeared of the large scale corruption, inefficiency or diversion of funds that would justify donors in steering clear of govemment channels. And the experience of earmarking funds through government channels (the World Bank funding for ESDP and HSDP) shows a direct trade-off whereby the focus on tracking of earmarked donor funds is likely to undermine and delay the strengthening of the government expenditure management system. Some donors resuming aid to Ethiopia are assuming that they should begin with traditional earmarked project aid, and only later consider non-project forns: the logic of this assumption is seriously questionable. Perverse effects of targeting aid: Donor preferences for tying their aid to sectors they believe to be 'pro-poor' may have an unexpected effect. Consider ESDP: donor preferences for primary level facilities left Government with a larger share of planned funding for tertiary. When these donors' funding was withheld on account of the war with Eritrea, tertiary expenditures were unaffected (and actually increased), and primary services bore all the effects of stringency. Misconceptions about fungibility: The idea that earmarking of donor funds prevents the recipient from indirectly increasing funding for other purposes is a persistent fallacy (see PER 2000 Box 4.5). As for the effects of withholding funds during the conflict, it will always be possible to argue that military expenditures might have been even higher if aid had continued to flow, but there can hardly be any doubt that expenditures on primary health and education were sharply lower because aid was withheld (even though Government broadly adhered to the sector expenditure levels it had projected before the conflict). Short term bias: The focus on poverty - an enduring problem which does not lend itself to short-term solutions - should reinforce the pressure on donors to adopt a longer horizon in the programming of their aid. NB. Many of these issues are relevant to the federal-regional relationship as well as to the donor-government one. For example, fungibility is a relevant issue when considering the targeting of federal funding to regions, and the bias towards capital expenditure is not limited to donors - it pervades the approach to planning and budgeting throughout government. And the government's own resource allocation system suffers from being too short-term and excessively geared to a strictly annual budget cycle. 4.21 The appropriate design of mechanisms for the disbursement and monitoring of budget support and for continuing dialogue between government and its aid partners are discussed in the remaining sections of this chapter. However, some of the pervasive donor attitudes to aid are based on misconceptions, the two most prominent of which are: (a) that earnarking prevents fungibility (which is true only if donors insist on giving Ethiopia things which the government doesn't want); and (b) that it is logical or sensible to recommence aid flows via traditional project approaches before thinking of budget support. Box 4.2 highlights the confusions and paradoxes that can affect aid management. 65 IV. HIPC INITIATIVE: REQUEREMENTs AND OPPORTUNITIES Donor and Government Obligations in the Context of IiPC and PRSP 4.22 The HIPC initiative is intended to reduce debt service obligations, allowing governments to redirect the resources saved into poverty reduction. PRSPs are a mechanism to ensure that appropriate strategies are in place that can guide the use of HlPC initiative savings in the context of a comprehensive approach to poverty reduction. Together, PRSPs and HIPC relief give renewed impetus to budget support for several reasons: (a) HIPC resources themselves need to be programmed into public expenditure plans through non-project modalities; (b) the PRSP creates a framework of medium-term strategy, including a medium term expenditure strategy, that should facilitate sector and budget support; and (c) a necessary, though not sufficient, condition, for the success of the PRS will undoubtedly be enhancing the quantum of aid flows, and the volume of aid that can be effectively absorbed is clearly a function of the disbursement modalities used. 4.23 HIPC and the PRSP approach thus have the potential to be a turning point, leading to higher levels of aid more effectively used for poverty reduction. Realizing that potential requires reciprocal behavior by donors and GoE. The PRSP has to have a medium term perspective, and to map out the public expenditure priorities that are consistent with the PRS. In turn this reinforces the pressure on donors to provide medium term aid commitments as a basis for the government's forward planning. Moreover, the consultative approach to PRSP preparation ought to facilitate a shift in the aid dialogue away from the nitty gritty of project selection and implementation towards policy and strategic issues. The monitoring of the PRSP and of the use of HIPC resources, as discussed below, will keep the entirety of public expenditures under review, enabling donors to judge whether fungibility is, in practice, undermining the value of their aid. 4.24 The opportunity afforded by PRSP and H[PC is especially important in a country like Ethiopia which as yet lacks a platform for comprehensive government-donor dialogue; the systems that currently exist are both fragmented and ineffective. Many donors feel that the government missed an important opportunity to establish true partnership with the donors in the formulation of I-PRSP.47 Expectations are high regarding the full PRSP. Moreover, as discussed in the final section, the PRSP must not be seen as a one-off but as instituting continuing mechanisms for practical dialogue. Modalities for Allocation of HIPC Resources 4.25 Overall guidelines have been drawn up for the HIPC initiative48 which recognize the need to account satisfactorily for HlPC resources while maintaining the integrity of overall public expenditure planning and avoiding the high transaction costs involved in earmarking funds to specific project level. A fundamental principle - recognizing the relevance of fumgibility - is that expenditure tracking under HIPC should take place in the context of the totality of public expenditures, since this is the only credible way to form a judgment as to whether HIPC resources have led to an increase in pro-poor spending. "4 Donors recognize that consultation is mandatory for the full PRSP, and not for the I-PRSP. Even so they felt the government could have used the opportunity of I-PRSP to re-establish constructive relationships with the donors following the peace agreement. 4S See IMF/IDA, Tracking of Poverty-Reducing Spending in Heavily Indebted Poor Countries (HIPC), Prepared by the IMF's Fiscal Affairs Department and the World Bank's Poverty Reduction and Economic Management Network, March 27, 2001. 66 4.26 Perhaps the most fundamental principle of HIPC tracking is that it is neither feasible nor desirable to try to track only HIPC funding. Instead all public spending on poverty reduction needs to be tracked, both to understand the impact of HIPC assistance and to encourage a shift toward more poverty-reducing public spending in the overall budget.49,50 4.27 As noted in Chapter 3, Ethiopia has already gone a long way towards setting in place the components of the medium term expenditure planning framework that is required. The development of Public Expenditure Programs at both regional and federal levels has been accelerated in recognition of the need for a credible mechanism to generate broad expenditure targets that are an aggregation of federal and regional expenditure plans. In the Ethiopian context, therefore, the requirement is to work out and agree in detail how fHPC planning will be integrated into the ongoing (but hitherto somewhat neglected) development of the Macroeconomic and Fiscal Framework and associated strategic resource allocation processes. 4.28 A related issue is whether HIPC resources should be disbursed via a real or virtual fund mechanism. The HlPC guidelines make a useful distinction between institutional poverty funds and virtual poverty funds. Institutional poverty funds are described as: "revenues set aside in a separate account, with expenditures occurring outside a country's normal budget execution and reporting system, subject to different reporting and accountability standards, and frequently with dedicated local staff hired and paid outside normal civil service systems.""1 In Ethiopia the ESRDF is a classic example of such a fund. The guidelines rightly state that such funds are not an appropriate mechanism for the disbursement of H1PC funds. Since such institutional funds operate outside the regular budget framework, using this channel would undermine the integrity of public expenditure planning and accountability and would be inconsistent with the principle of tracking all poverty related expenditures. 4.29 At the same time the guidelines also stress the importance of identifying and quantifying HIPC resources so as to be able to demonstrate their impact on the budget as a whole (see Box 4.3). A number of countries (e.g. Uganda and Tanzania) have tagged such resources through a 'virtual fund' mechanism, so as to highlight what additional expenditures they have made possible. A virtual fund (of which many variants are possible) involves keeping track of certain resources (e.g. HIPC relief or other budget and sector support) and making explicit which elements of the budget have been financed by these resources. Expenditure tracking for a virtual fund takes place through the normal accounting mechanisms of the government. From the point of view of the implementing agency, the modalities for using the funds and reporting on expenditures can be exactly the same as those that apply to treasury funding. 4.30 The best-known example of a virtual fund is Uganda's Poverty Action Fund (PAF). Its key function is to act as a vehicle for relating incremental resources provided by debt relief and donor support to expenditure increases for specific programs.52 The PAF does not eliminate the possibility of fungibility, but it provides an effective focus for discussion of how aid funds are to be used in the context of a continuing view of the entire budget, and thus enables donors to make an informed assessment as to whether their aid is resulting in increased aggregate support to pro- poor expenditures. It has been highly successful in encouraging donors to shift towards budget support and to increase their aid flows. A variant of the virtual fund approach is being used in Tanzania. 49 IMF/IDA, ibid., ¶6. 50 See Box 3.5 for a summary of the resulting paradigm for HIPC expenditure planning. SI IMF/IDA, ibid., ¶8. 52 IMfIDA, ibid., 17. 67 4.31 A number of bilateral donors in Ethiopia have indicated their interest in a similar mechanism, whose potential advantages were described at the PER workshop as including: a helpful way (for both government and donors) of visualizing issues; useful also in tracking of expenditures (by making more explicit where incremental resources are being directed); providing a focus for dialogue, and thereby facilitating donor movement from project to budget support; and enhancing coordination amongst donors. It should be stressed (a) that the potential benefits of a 'virtual fund' approach mainly arise not from the existence of such a 'fund' per se but from the structure of continuing dialogue and consultation that is built around it, and (b) that principles of coordination and efficiency require that different donors providing budget support should share the same structure of dialogue and the same basic disbursement mechanisms. Therefore the mechanisms to be adopted for H[PC disbursement in Ethiopia should from the outset be deliberately designed to be applicable to other sources of budget support. HIIPC Modalities for Ethiopia 4.32 Dialogue between GoE and staff of the World Bank and IiMF has continued in parallel with the PER 2001 process. Box 4.3 encapsulates the proposed mechanisms for HIPC disbursement and tracking that have emerged. Key features are (a) the channeling of the funds to the regional level; (b) no restriction on the application of funds as between sectors or between recurrent and capital expenditure - but a requirement for recipients to indicate how the funds will be applied and to demonstrate the poverty-relevance of their use; (c) emphasis on the use of the MEFF/PER as a framework for discussing overall resource allocation and hence the effect of incremental resources on the profile of expenditures as a whole, in parallel with (d) measures to strengthen welfare monitoring and the capacity to review the efficiency and effectiveness of service delivery. What is envisaged is thus a pragmatic variant of the virtual fund approach, with an emphasis on support of expenditures at regional level.53 There appears to be scope for broadening this mechanism so as to allow donors providing budget support to use the same modality. V. THE WAY FORWARD AND IMPLICATIONS FOR PRSP Overview 4.33 A fundamental principle of PRSP formulation is government ownership. Similarly the guidelines for the tracking of HIPC expenditures place the responsibility for tracking on the government. Nevertheless progress requires common efforts and reciprocal commitments. This final section highlights the directions in which both donors and government need to move in order to facilitate progress towards more effective use of higher levels of aid. " Of course, channelling of HIPC or other budget support funds to the regional level does not necessarily increase regional resources by the full amount, since there may be consequential adjustments in the federal subsidy to the regions. However, an additional benefit can arise from increasing the predictability of transfers to the regions. 68 Box 4.3: Proposed HIPC Disbursement Mechanism for Ethiopia The government has provided IMF and IDA staffs with an indicative medium-term framework for poverty- targeted expenditures, before and after interim HIPC Initiative assistance. Table 454 shows that debt relief under the enhanced HIPC Initiative would supplement the federal and regional governments' own efforts to boost social spending, by increasing budget allocations to education, health, roads, and agriculture and natural resources-sectors with a strong positive impact on poverty reduction and growth. Particular efforts will be made to increase the efficiency and effectiveness of public spending related to HIPC Initiative relief through accompanying capacity-building efforts in public administration over the coming years. The use of interim debt relief will be reported and discussed annually during the PRSP consultative process. IMF and IDA staff are planning to monitor the use of resources freed by the enhanced HIPC Initiative in the following ways: Analysis of changing composition of public expenditures through annual public expenditure reviews, and annual sectoral reviews in the areas of health, education, and roads, carried out by the government and donors. Data on budget formulation and expenditure execution at both the federal and regional levels should also improve as a result of actions to be taken in relation with completion point conditions on governance and public expenditure management. Improved incidence analysis of public expenditure on the poor through the strengthening of the welfare monitoring system (WMS) by carrying out household income and expenditure surveys, consultations with the civil society on the implementation of the PRSP strategy, and compilation of other relevant information. The government is currently reviewing options for monitoring the use of principal and interest rate savings associated with debt relief provided by enhanced HIPC Initiative interim assistance, taking into account the institutional constraints of Ethiopia's federal system. For 2001/02, since interim HIPC assistance will only be available in the last two quarters, the authorities are still considering whether they would adopt a supplementary budget or put the funds in a blocked account at the Central Bank, and incorporate them in the 2002/03 budget. The government intends to keep funds from HIPC interim assistance in a holding account with deposits taking place as and when agreements are concluded with creditors. The federal government would then transfer funds from the holding account to the regional governments by way of annual budget allocations or supplementary allocations. Each recipient regional government would have its own holding account from which it would make transfers to zonal and woreda administrations and to implementing agencies, in support of the government's sector development programs in health, education, and roads. Regional governments, zonal and woreda administrations would confirm the receipt (within one month of the transfer) of the transfer and the purposes for which it would be used. Regional governments would require their Auditor Generals to carry out random audits during the third quarter of each financial year on the use of the funds. These means to track public expenditures should be strengthened in the medium-term with the planned implementation of a new framework for monitoring poverty-related spending, based on (a) the expansion of the application of the three-year MEFF; (b) the introduction of cost-centered budgeting at the center and in the regions; (c) the maintenance of a uniform accounting systems and a computerized management information system for tracking financial flows; and (d) the improvement of auditing and financial control capacities. Obligations on the Donors in the HIPC-PRSP Context 4.34 The decline in per capita aid flows and the disturbingly low level of per capita aid to Ethiopia were noted at the beginning of this chapter. A necessary, although not sufficient, condition for the success of the PRS will undoubtedly be enhancing the quantum of aid flows. This requires that donors appreciate two obligations on themselves. First, a PRSP has a medium term horizon and it requires the government to factor in realistic estimates of aid in their medium term resource envelope. This in turn obliges the donors to shift from typically annual to medium term aid pledges." Second, since donors are among the stakeholders consulted on the PRSP, their 54 This refers to table 4 of the HIPC Decision Point document (October 2001) 55 A few donors already make medium term donor pledges on a rolling basis but this is not a common practice. 69 endorsement of the PRSP should greatly facilitate a shift from traditional project to program and budget support.56 4.35 There may be political and administrative difficulties to overcome, but the objective is very important. Moreover, realizing the benefits of sector and budget approaches does not depend on every donor making the shift. Increasing the proportion of aid that is provided as budget or sector support will enhance the effectiveness of the entire spectrum of public expenditures, and the early movers towards budget support thus have the opportunity to leverage disproportionate benefits. At the same time, all donors should be encouraged to enhance the quality of their aid, whether or not they are ready to contemplate budget support. 4.36 Changes and actions required on donor side include: - Indicate multi-annual commitments. As indicated earlier, the government will not be able to plan and implement a medium term PRS in the absence of reliable assurance regarding aid flows. In the PRSP context therefore, it will becomes even more important for donors to give medium term projections of their aid pledges. As per current experience, some donors, notably the multilateral institutions, are able to do so, while others are handicapped by institutional systems and constraints in their home governments. A way forward will be for all donors to give medium term projections on an indicative basis and firm commitments on an annual basis. The projections can be updated each year on a rolling basis. - Observe the Government's financial calendar and maintain disbursements along a pre-announced time-frame in the annual cycle. For the government to plan resource management and project implementation, it needs to be informed of aid availability at the time when it its formulating its annual three-year rolling plans/expenditure framework; is also necessary that the donors indicate the time-frame for release of funds during the year along with the specific terms and any preconditions to be met. - Let adherence to PRS be the only criterion. Donors should refrain from dictating an independent agenda or imposing specific conditionalities as that would undermine the efficacy of budget support. Their only concern should be to ensure that the overall resource allocation conforms to the PRSP and that sufficient systems are in place to ensure fiduciary control and technical efficiency in the use of resources. (This in turn requires a clear structure and calendar for collective dialogue between government and donors, as discussed in the final section below.) - Route all aid through the federal budget; refrain from extra-budgetary support: If donors persist with the present practice of channels 2 and 3 support, it will undermine both planning and resource allocation on an aggregate basis. The only exception to this will be donor support through NGOs. Even in that case, it should be incumbent on the donors to report the amount and purpose of grant so that it can be factored into overall spatial and resource planning. - Actively assist in capacity building. Donors should not make the existence of ideal systems and practices a pre-condition for their budget support. The 56 Endorsement by the donors of the PRSP does not necessarily indicate that they agree with all of the analysis, targets, or public actions set forth in the PRSP, or consider that the PRSP represents the best possible strategy for the country. Rather it indicates that the staff consider that the strategy provides a credible framework within which they can design their programs of assistance. 70 ideal itself is a moving target and it will take several years for developing countries to bring their systems of planning and budget management to present rich country levels. Rather, the donors should assist in building both institutional and human capacity to move in the desired direction. Improve coordination amongst donors. The donors' record in following up PER recommendations, for example, is not impressive. A case in point is PER 2000's recommendation on government/donor strategic dialogue (reproduced as Box 4.4), on which there appears to have been no effective action, in spite of the fact that the PRSP/HIPC context makes the issue more relevant than ever. Box 4.4: PER 2000 Recommendation on Government/Donor Strategic Dialogue AREA FINDINGS RECOMMENDATIONS REspoNSIBILITY AND TImING Govermnent/ donor Different elements of Develop streamlined As a first step, strategic dialogue dialogue (SDPs, PRSP, proposals for a major donors to PER and CDF etc) not coordinated set of coordinate in sufficiently coordinated; documents on which developing danger of overloading dialogue could focus, and proposals for senior government propose a consolidated discussion with decision makers. calendar for dialogue, government. linked to GoE's own fiscal (short tenn) calendar. Source: PER 2000: Box 3.7: Summary of conclusions and recomnnendations Actions Required on the Government Side 4.37 Changes and actions required on government side include: - Sustain the peace process. The durability of the peace agreement will be the dominant criterion that is going to determine donor sympathy towards Ethiopia. The faster than programmed reduction in defense expenditure as well as demobilization of soldiers has gone a long way towards allaying donor concerns. Trust and confidence will be further reinforced if the government stands by its commitment of reducing defense expenditures to 5 percent of GDP by FY03 and not exceeding that limit thereafter. - Formulate the PRSP following the guidelines in letter and in spirit. Country ownership and civil society involvement are at the heart of the PRSP process. The donors have indicated to the government on several occasions their willingness to assist both in the content and process of PRSP. The donors will be willing to pledge budget support if they are convinced that the PRSP reflects a credible strategy that fully reflects the aspirations and priorities of the civil society. The challenge and opportunity for the government lies in conducting the consultations in an open and transparent manner and reflecting them in the PRSP. GoE's invitation to the donors to witness the consultations at the woreda and regional levels has been a significant step forward in this regard. - Build partnership with the donors. The PRSP-HIPC context provides a valuable opportunity for the government to build renewed partnership with the donors. While donors have been assured that they, as an important stakeholder group, will be consulted in the formulation of the PRSP, the 71 precise nature of this consultation is not yet clear. What will, however, pay rich dividends to the government would be to keep the donors on board through the process of PRSP formulation rather than limiting their involvement to a one off consultation.5" Donor advice should explicitly be sought on the monitoring indicators as they become the benchmarks for performance evaluation. Involve donors along with civil society in monitoring PRSP implementation. The PRSP guidelines require that the civil society should be involved not just in the formulation, but also in monitoring the implementation of the PRS. The government can increase the credibility of the monitoring process by inviting and encouraging the donors to participate along with civil society agencies in the process. Maintain regular policy dialogue with the donors. An important lacuna, and one that most donors feel concerned about, is the absence of an institutional forum for government - donor dialogue on policy issues. The mechanisms that are in place, are fragmented and ineffective. The platforms for reviewing SDPs, for example, have often been bogged down by details of implementation rather than strategy and policy issues. Oftentimes, donors have felt that they are invited to these forums only to seek their aid pledges and not for involving them in the program design. In recent years, the PER has evolved into an important forum for reviewing, and helping to resolve, issues that are more systematic or macroeconomic than those which the SDP mechanisms focus upon. However, the PER forum has failed to maintain year round momentum. In May 2001, the government constituted an inter- ministerial committee comprising officials of MoF and MEDaC58 to coordinate on public expenditure issues and brief the donors periodically. Since no donor briefing has yet taken place, it is not possible to evaluate the value added of this forum. Improve systems for tracking and evaluating public expenditures. The importance of improving the systems and the way forward has been detailed in Chapter 3. Government should recognize that donors will be unwilling to commit budget support unless they are assured that (i) mechanisms are in place to monitor and evaluate public expenditures; (ii) that the existing fiduciary controls are effective and sufficient. An important area for reform will be to improve the efficacy of both internal and external audit systems, as well as the timeliness and reliability of reports on inputs and outputs. Develop the MTEF (PEP) along a pre-announced timeline. An MTEF is in may ways an important complement of the PRSP. It can provide a basis for medium term resource planning and for balancing recurrent and capital expenditures. It will also challenge donors to make long term commitments and improve the predictability of their financing. As noted earlier (Box 4.3), it is an important template against which to monitor the use of HIPC resources and other budget support. An MTEF, known as PEP in the Ethiopian context, is very much on the agenda of the ongoing Civil Service Reform Program (CSRP). The government can provide reassurance to the donors by indicating the preparatory steps and time-table for the transition to an MTEF. Participate more proactively in the PER. Although government involvement 57 The Government has indicated that they will hold regular meetings with the DAG on the PRSP process. S since merged into a single Ministry of Finance and Economic Development. 72 in the PER has increased in recent years, the PER remains largely a donor driven process. The government is thereby forfeiting an important avenue for partnership with the donors. Should the government participate more actively, the PER will provide an opportunity for both parties to jointly assess the fiscal situation and review important developments in public finance management. This collaboration will be even more important in the future as the government has decided that the PER will be the instrument for evaluating their track record under the HIPC Initiative. Coherent Ongoing Mechanisms for Dialogue and Review 4.38 As the above analysis shows, the changes in perception and actions, required of both government and the donors, in the transition towards budget support are demanding but perfectly feasible. There is already appreciation on both sides of the desirable way forward. However, the shift away from fragmented approaches towards a coordinated partnership based on collective dialogue will not occur simply as the result of a series of piecemeal actions by the various parties. There needs to be a well-understood mechanism for annual consultations, linked to the fiscal calendar, which would, inter alia, review past disbursements of HIPC relief and other budget support and agree on future tranches. The role of the PER is one important element to be defined within this context and that of the annual PEP preparation cycle. Indeed, the PER 2000 recommendations highlighted in Box 4.4 remain as relevant as ever. The discussions surrounding the PRSP provide a unique opportunity for government and donors to agree on the integrated framework for dialogue that is required. As one aspect of this, there should be urgent discussions with donors who are interested in budget support to determine how best to integrate such support with the emerging HIPC modalities. 73 5. PUBLIC EXPENDITURE - REGIONAL PERSPECTIVE I. INTRODUCTION 5.1 Ethiopia's 1994 Constitution mandates a federal structure with considerable autonomy devolved to the regions in administrative and fiscal matters. Although the share of the regions in the combined federal-regional revenue is only about 15-20 percent, they account for nearly 35-40 percent of the total expenditure. When only the anti-poverty intensive sectors such as health and education are considered, the regions' share in the combined expenditure may go up to as much as 60-75 percent. In the PRSP context, therefore, the regions are at the cutting edge in the administration and monitoring of anti-poverty expenditures 5.2 Yet there is relatively limited documented analysis of the fiscal situation or the budgetary institutions and processes at the regional level. The World Bank's Regionalization Study (June 2000) was the first comprehensive effort to provide an overview of Ethiopia's fiscal decentralization and indicate some policy areas which will require attention as the system evolves.59 PER 2000 made a systematic effort to shift focus from a federal to a federal-regional perspective.fi Members of that PER mission visited most of the regions, albeit briefly, in order to get a better understanding of public expenditure management issues at the regional level. The PER 2000 report paid attention to the regional dimensions of information flows and aid management, and regional representatives were invited to the PER Workshop in Debre Zeit. However, the analysis of budgets and expenditures remained essentially at national level. 5.3 Both the Debre Zeit Workshop (September 2000) and the subsequent Brussels Workshop (November 2000) of the 2000 PER cycle strongly underscored the need for an increased regional focus in subsequent PERs, especially as the regions will be in the forefront of implementing the PRSP and in utilizing the HIPC resources. Accordingly, it was decided that PER 2001 would carry out pilot regional-level PERs in three regions (the institution leading the sub-task indicated in parenthesis): Benishangul-Gumuz (African Development Bank), Oromiya (World Bank) and Tigray (European Commission). The studies were conducted through field missions during January-March 2001 and the individual reports of the three regional PERs (RPERs) are being issued separately as independent reports by the lead institutions.6' 5.4 The rest of this chapter is divided into three sections. Section II presents an overview of the dimensions and main features of Ethiopia's fiscal federalism. This perspective is intended to provide a basis for better appreciating the issues and common themes discussed in Section III drawing from the three RPERs. Section IV presents some additional issues and conclusions in the context of GoE's renewed emphasis on decentralization. There are two related appendices in Volume II to provide background and analysis to some of the issues discussed in this chapter. 59 Ethiopia: Regionalization Study, World Bank, Report No. 18898-ET, June 2000. The main field work took place in 1998. 60 Ethiopia: Public Expenditure Review, World Bank, Report No. 20810-ET, August 2000. 61 The three RPER reports are being issued separately by the respective lead institutions. The RPER for Oromiya, for which the World Bank is the lead institution is being issued as a third volume in this grey cover series. Appendix 6 deals with tax and expenditure assignments as between the federal and regional levels while Appendix 7 explains the federal budget grant formula and the budget offset mechanism. II. FISCAL FEDERALISM - MAIN FEATURES AND DIMENSIONS 5.5 Ethiopia has a unique federal structure and one which, in practice, is still evolving. The Bank's Regionalization Study characterized decentralization in Ethiopia as "a bold and thoughtful process.... supported by a widely shared consensus over both development strategy and objectives, and very large transfers of untied resources from the federal government to the regions." Successive PERs have also focused on aspects of federal-regional fiscal relations (e.g. its implications for aid management as reflected in the offset system). The operation of decentralization at woreda level is the subject of additional World Bank research (the forthcoming Woreda Studies), which is providing insights that can complement the RPER findings. 5.6 Without attempting a full description of the federal system, it is important to draw attention to its main distinctive features and their implications for public expenditure management. Key features include the heterogeneity of the regions, the high degree of regional autonomy, large vertical and horizontal imbalances, and the continuing evolution of the system with its impetus for decentralization within regions. All are reviewed below. Subsequently, this section also notes two further aspects of special interest for the PER and RPERs - the relevance of decentralization to poverty reduction and to aid management. The Framework of Decentralization Heterogeneity of Regions 5.7 Ethiopia's 1994 Constitution recognizes nine region-states.62 In addition, there are two special city administrations - the capital city of Addis Ababa and the second largest city of Dire Dawa - both of which enjoy the status of regions. The regions are not demarcated on criteria of administrative convenience, but along nationality and ethnic lines. As a result they vary enormously in land area and population (Table 5.1). Diversity across regions is such that any generalizations need to be made with extreme care. Not all the regions are themselves homogeneous: for example, SNNP, Gambella and Benishangul-Gumuz are ethnically very diverse (see 'Peoples' in Table 5.1). Afar, Somali, Benishangul-Gumuz and Gambella are referred to as the 'emerging regions', having historically lagged behind other areas in social development and administrative capacity. 62 The English version of the 1994 Constitution refers to the sub-national entities as 'states'. However, they are popularly referred to as 'regions' and this report follows the popular usage. 76 Table 5.1: Overview of Ethiopia's Regions Regions HQ Populatiorr 31 Land Area Population Density Peoples 4 No. of Zones No. of Woredas Regtons ktQ ~~~~~~~~~ ~~(million) (km') (persons/ kin) 16 Region I` Tigray Mekele 3.9 80,000 48.8 3 5 35 71 Region 2 Afar Ayssaita 1.3 270,000 4.7 1 5 29 Region3 Ambara BahirDar 17.2 171,000 100.6 4 11 106 Region 4 Oromia Addis Ababa 21 23.7 367,000 64.6 1 12 180 Region 5 Somali Jijiga 3.9 1 I 9 44 Region 6 Benishangul-Gumutz Asossa 0.6 50,381 11.3 5 3 20 Region7-11 SNNP 11 Awassa 13.3 118,000 112.6 45 9 77 Region 12 Garnbella Gambella 0.2 4 3 4 8 Region 13 Ilareri Harar 0.2 305 557.4 1 - Region 14 Addis Ababa Addis Ababa 2.6 - 1 6 28 Region 15 Dire Dawa Dire Dawa 0.3 1,977 172,0 - - I Total 67.1 1,130,000 59.4 65 63 52S [I] Southem Nations, Nationalities and Peoples Region, formed by a merger of 5 contiguous regions. [21 Also known as Finfinnee in Oromnifa. HQ to be shifted to Nazareth in course of time P31 Source: The Federal Budget Grant Formula in Ethiopia, MEDAC, March 2000. Estimates for 2001/02. [4] The number of nations, nationalities and peoples listed for each region in Proclamation #7/1992. Number of languages spoken in a region may be greater, The non-Tigrinya speaking groups *n Region 1, and the non-Amharic speaking groups in Region 4, are small minorities who have Special WoredawZone status. Dire Dawa became an administrative council subsequently. (5J Bahir Dar (Region 3) counts as both a zone and a woreda. In Region I, Mekele also has zonal status. There are no Zones or Woredas in Hareri and no zones in Dire Dawa. 16] 5 of the woredas in SNNPR have special status, and therefore do not report via any zonal administration. [17 Putiher divided into over 600 localities (tabiya), wbich are sub-divided into homesteads * Official data not available. Degree of Regional Autonomy 5.8 The Regions have been given extensive legislative and executive powers, and the Constitution expressly provides that: "All powers not given separately to the Federal Government, or powers not given expressly and concurrently to the States and the Federal State, are reserved to the States."63 Furthermore, the Constitution provides an explicit right to self-determination and secession from the Federation, and sets out a procedure for exercising this right.64 The financing of regional expenditures by an unearmarked block grant (as discussed in Appendix 7, Vol. II) may be seen as a further reinforcement of regional autonomy. At the same time, the relationship between federal and regional levels is complex, capable of varying interpretations and still evolving. 5.9 The 1994 Constitution was a major break with the centralism of previous regimes, and the line relationships that used to apply between national sector ministries and the sector agencies of government at sub-national levels were severed. However, the 1992 Proclamation under which the regions were first established made it clear that the sub-national governments were expected to operate within and implement not only the legislation but the 'policies' of the central government. The 1994 Constitution similarly balances Regional States' powers by those reserved to the Federal Government and by National Policy Directives' (Chapter Ten of the Constitution) that apply equally to the Federal and Regional States. The Constitution also provides that the Federal State may delegate its powers to the Regional States, and vice versa.65 5.10 The Constitution does not incorporate all the detailed provisions of the 1992 Proclamation concerning dual accountability at each level of government, but these still appear to be reflected in current practice of planning and budgeting.66 The Regionalization Study points out that the potential for divergence between federal and regional levels may be masked by the fact that, at present, there is a high degree of consensus between federal and regional governments. This consensus is reinforced by the fact that the same political parties exercise power at both levels. It is still possible that the defacto balance of responsibilities between the federal government and the regions may change over time. For example, much may depend on how the interpretation of Article 51, ¶2 Constitution evolves. This prescribes that the federal government should "formulate the country's policies in respect to overall economic and social development; it shall draw up and implement plans and strategies of development."67 Consequently, current and future work on RPERs as well as other work on decentralization will need to inquire into how the balance between federal and regional responsibilities operates now and may evolve in fuiture. An Evolving System 5.11 Regions are administratively subdivided into zones and woredas. The woreda (district) level, with a typical population of around 100,000, was the major decentralizing innovation of the present constitutional framework. Partly because this level of administration had been dispensed with by the previous regime, it is at this level that the most acute 63 Constitution Article 52, 11. 64 This is spelt out in Article 39 of the Constitution. Under Article 47, nations, nationalities and peoples within the existing Regional States have a similar right to establish their own States. 65 Constitution Article 50, ¶9. 66 Under dual accountability, each level is held responsible to the level above, as well as to its own electorate. For example, the 1992 Proclamation was quite explicit that "National/Regional Councils are responsible to the people which elected them and to the Council of Representatives of the Central Transitional Government." See Implementing Sector Development Programmes in Ethiopia, Stephen Lister, December 1998 (Chapter 2) for a more detailed review of the constitutional arrangements and the demarcation of responsibilities between federal and regional levels. The dual accountability provisions are there detailed in Annex 2A. 67 The Constitution itself sets some important policy guidelines. For example: to the extent the country's resources permit, policies shall aim to provide all Ethiopians access to health and education, clean water, housing, food and social security. (Article 90, 11.) 78 problems of capacity arise. All woredas have an elected woreda council. Woredas are further subdivided into communities (kebeles). There are around 30,000 of these neighborhood administrative units across the country. The government's goal of further decentralization is clearly stated in the I-PRSP - Federalism will be enriched further by decentralization within the federal states to the district level, in the medium to long tern period. However, the I- PRSP also emphasizes the capacity building effort that this will require. In his recent statement to the parliament on the government's development strategy (October 2001), the Prime Minister said that they will deepen the process of decentralization and capacity building at the sub-regional level.68 5.12 The zonal level of administration is not explicitly provided for in the Constitution but regions have enabling power to establish administrative hierarchies as they see fit. Most regions have seen zones as a necessary administrative bridge between region and woreda, but without a separate elected body. Thus the Executive Committees for most zones are appointed by the Regional Council. However, SNNPR, with its exceptional ethnic diversity, has elected councils for all zones; and in other regions there are some Special Zones which do have elected councils as a means of ensuring self-government for minority peoples within the regions. There are also a number of Special Woredas which enjoy the status of zones and report directly to the regional level. 5.13 There are significant variations in the way different regions have approached intra- regional decentralization. More generally, decentralization is part of a broader process of reform. It is inter-twined with sector development programs and with the Civil Service Reform Program (CSRP). The Expenditure Management and Control Program (EMCP), a component of the CSRP, seeks to address regional as well as federal systems. It is important to assess the interaction between federal and regional programs, and to be aware of the differences in regions' capacity and in their approaches to decentralization of planning and public expenditure management. As will be noted in Section III of this chapter, these are issues to which all the RPERs paid particular attention. Vertical and Horizontal Imbalances Horizontal Imbalance 5.14 In any federation, the component states differ in their capacity to raise revenue, and their revenue-raising capacity is typically poorly correlated with their expenditure obligations. This gives rise to horizontal imbalance. Horizontal imbalance in Ethiopia is exacerbated by the diversity across regions, not only in area and population but also in their economic circumstances and administrative capacities. Severe disparities exist between war and drought affected regions as well as between large and small regions. For example, the problems of a large region like Oromia with a population density of 64.6 per sq. km and largely settled agriculture are markedly different from those of Afar with a population density of 4.7 per sq. km. and a pastoralist economy. 5.15 Table 5.2 presents some economic and fiscal parameters to demonstrate the extent of horizontal imbalance. The poverty ratio (percentage of population below poverty line) varies from a low of 24.6 percent in Dire Dawa to a high of 57.9 percent in Tigray. As would be expected, fiscal capacity, as measured by per capita own revenue, has a significant negative correlation (-0.69) with poverty ratio. There is wide dispersion in per capita expenditures from a high of Birr 659.9 in Gambella to a low of Birr 65.3 in Amhara. Regional fiscal imbalance, measured as a ratio of a region's own revenue to its expenditure, varies significantly across regions from 5.4 percent in Gambella to 37.7 percent in Dire Dawa as per FY01 budget estimates. Interestingly, although the correlation between poverty ratio and fiscal imbalance is negative as expected, it is not significant. However, larger regions are in general characterized by lower fiscal imbalance than smaller regions. Addressing the problem 68 Some important features of this renewed focus on decentralization are outlined in Section IV of this chapter. 79 of horizontal imbalance has been a continuing and complex problem even in advanced federations. Table 5.2: Horizontal Imbalance Across Regions Regions 1/ Population Poverty Ratio Per capita (00/01) Regional fiscal (1997/98) 3/ (Birr) imbalance 5/ Own Federal Expenditure 1997/98 6/ 2000/01 revenues subsidy (budget) 4/ Tigray 3.90 57.9 20.8 92.4 113.1 23.3 18.4 Afar 1.27 51.8 12.7 187.0 199.7 4.4 6.4 Amhara 17.21 56.7 11.3 54.0 65.3 18.8 17.3 Oromia 2/ 23.70 34.7 15.2 51.3 72.7 25.6 20.9 Somali 3.90 34.6 8.2 90.5 98.7 12.8 8.3 Benishangul-Gunuz 0.57 47.6 24.2 314.1 338.3 4.5 7.2 SNNP 13.29 56.5 11.5 58.7 70.2 18.5 16.3 Gambella 0.22 41.8 35.6 624.2 659.9 4.8 5.4 Hareri 0.17 29.1 54.0 394.6 448.6 8.9 12.0 DireDawa 0.34 24.6 71.7 118.6 190.4 33.6 37.7 [I ] Addis Ababa (not included in the table) had a per capita own revenue of Birr 580.1 in FY00 and Birr 397.8 in FY01, and receives no federal subsidy.. 12) in the case of Oromia, an amount of Birr 147.44 million (cash balance) is included in the total budget but not in its own revenue because it was unused amount from the previous years [3] Source: Federal Budget Grant Formula in Ethiopia, MEDAC, March 2000. [4] The expenditure (budget) refers to the budget ceiling of regions as determined by the Federal Govemment after considering the overall resource envelope. Hence figures of per capita expenditure do not match with those in Table 5.5.. [5] Ratio of a region's own revenue to its total expenditure. [6] Source for 1997/98 regional fiscal imbalance figures is the Ethiopia Regionalization Study, June, 2000 Between Poverty Ratio and Per Capita Own Revenue (-) 0.69 Between Poverty Ratio and Regional Fiscal Imbalance (1997/98) (-) 0.21 Between Poverty Ratio and Regional Fiscal hnbalance in (2000/01) (-) 0.34 Vertical Imbalance 5.16 Another typical characteristic of virtually all federations is that the aggregate revenue-raising capacity of the component states fails to match their expenditure responsibilities, while the central government is able to raise more Tevenue than is needed to finance its own direct expenditures. Vertical imbalance in Ethiopia, as measured by the difference between the regions' own revenue and their expenditure needs, is not only high by the standards of other federations but growing.69 As calculations in Table 5.3 show, the vertical imbalance moved from 0.58 in FY97 to 0.46 in FYOI. A declining fiscal imbalance is, in general, an indicator of efficiency as it implies that regions are financing an increasing share of their expenditures from their own revenues and thereby internalizing the costs of service provision. Examination of the figures in Table 5.3 shows, however, that the decline in vertical imbalance in Ethiopia over the period 1997-2001 owes less to rising revenue shares rather than falling expenditure shares, occasioned by pre-emptive defense commitments of the federal government. Also, the figures may understate the extent of vertical imbalance in the latter years as all expenditures out of external loans and assistance are attributed to the federal government since disaggregated accounts are not yet available. 69 The values of vertical imbalance in some other federations are: Argentina 0.03 (1995); Australia 0.24 (1996); Bolivia 0.66 (1996); Germany 0.15 (1996); India 0.28 (1995); Mexico 0.36 (1994) and South Africa 0.88 (1995). Data from Ethiopia: Regionalization Study, World Bank, June 2000. 80 Table 5.3 Shift in Vertical Imbalances Over Time, 1997 - 20011 1996/972 1997/98 1998/99 1999/00 2000/01 2000/01 Category (Program) (Prel. Actual) Regions' share of revenue (°/e) 18.0 19.7 18.0 18.3 18.0 18.0 Regions' share of expenditure (%) 3/ 42.5 39.6 30.5 23.3 27.9 33.4 Vertical imbalance 0.58 0.50 0.41 0.21 0.36 0.46 1] Vertical Imbalance is computed as: [1-{r/e}] where r = Regions' share of revenue and e = Regions' share of expenditure. [2] Figures for 1996/97 through 1999/00 are preliminary actual expenditures. [3] As disaggregated figures are not available, regions! capital expenditures do not include external loan and assistance for FY98, FY99, FYOO and FYO. To this extent, vertical imbalance may be underestimated. 5.17 This vertical imbalance necessitates a system of fiscal transfers from center to the states. Tables 5.4 and 5.5 demonstrate the extent and the nature of the vertical imbalance in Ethiopia. As Table 5.4 shows, the federal share of total revenues has consistently exceeded 80 percent. The assignment of revenues between federal and regional governments has some unusual features in Ethiopia (see Appendix 6, Vol. II) but it is not exceptional for a federal government's revenue sources to yield much more those of the states. It often makes sense for high-yielding taxes to be set and collected at federal level for reasons of economic and administrative efficiency. This gives rise to a potential problem, frequently pointed out by students of fiscal federalism, if the bodies which are responsible for determining expenditure are not the same ones which have to raise the taxes to pay for them. This could lead to upward pressure on public expenditures and inefficient use of resources. 5.18 Vertical imbalance depends on the assignment of expenditures as well as the assignment of revenues. Expenditure assigmnent is govemed by a different set of considerations, including in particular the case for local management of local services. In Ethiopia's case, social development objectives and the importance of equity are explicitly stated in the Constitution,'° and this is reflected both in the pattern of federal and regional expenditures and in the mechanisms for transferring revenue to the regions. Table 5.5 summarizes the division of recurrent expenditures by sector between federal and regional governments, while Table 5.6 gives a breakdown of regional per capita expenditures. A more detailed analysis of trends in sectoral expenditure has been provided earlier in Chapter 2 of this report, but Table 5.5 is sufficient to demonstrate the predominant role of the regions in delivering health, education and agricultural services: in each case 80 percent or more of recurrent expenditure takes place at regional level. 5.19 As Table 5.6 shows, vertical imbalance in a number of the smaller regions is magnified by high per capita levels of public expenditure. There is wide variation in per capita expenditures of regions, ranging from a high of Birr 659.9 in Gambella to Birr 65.9 in Amhara and an average of Birr 102.0 across all regions."' The wide dispersion is reflected across all functional categories and is an indicator, among other actors, of the widely differing unit costs of service provision.72 70 Notably in Constitution Article 41, 13 and 14: Every Ethiopian citizen has the right to equal access to publiclyfunded social services. The State has the obligation to allocate increasing resources to provide health, education and other social services. 71 However, the picture in this table is incomplete, since it does not capture aid flows outside the budget, which are significant and differ in their regional and sectoral incidence. n For a more detailed discussion of regional revenue sources and tax assignments see Appendix 6. 81 5.20 Ethiopia's system of fiscal federalism is unusual in the w; At deals with these vertical and horizontal imbalances. The main instrument involved is an annual unearmarked block grant to the regions.73 Each year the federal govermnent determines an overall envelope for public expenditure and takes a view on the appropriate division of expenditures between federal and regional levels.74 This determines the aggregate federal subsidy, which is then divided among the regions according to a forrnula which takes account of each region's population, expenditure needs and revenue effort. The federal grant formula has been modified recently, and now also includes an explicit poverty variable.75 All regions (except Addis Ababa) are heavily dependent on the federal subsidy to finance their expenditures (see the comparisons of own revenue and federal subsidy in Table 5.2). Sub-national borrowing would require federal government approval and has not been used.76 The federal subsidy mechanism allows the federal authorities to maintain overall fiscal discipline, by determining the total public expenditure envelope, but does not explicitly influence the composition of regional expenditures in the way that sector-specific grants might do.7 73 This federal subvention is universally referred to as the federal subsidy'. 74 A consultative process is involved, and the ultimate arbiter of the revenue sharing arrangements is the Parliament's Federal Council, whose composition reflects the interests of the regions. Since federal recurrent expenditures are treated as a first call on resources, there is a direct trade-off between federal capital expenditure and the subvention to the regions. 75 Its current structure is reviewed in Appendix 7. 76 See Appendix 6. " However, the Road Fund does operate as a sector-specific grant. (See Mokoro Limited, Tracking of Public Expenditures for Poverty Reduction, PER 2001 Discussion Paper, August 2001.) 82 Table 5:4: Shares of Federal and Regional Governments in Revenues 1997 - 2001 1996/97" 1997/99 199199 1999/00 2M00O1 2000/01 (Budget) (PER Esfisate) Category Federal Regional Total (min. Federal Regional Total Federal Regional Total Federal Regional Total Federal Regional Total Federal Regional Total share share Birr) share share (mmn. Biff) share share (tbin. Biff) share share (tbin. Birr) share share (mmn. Biff) share share (sbns. Birr) Tax Revenue 81.9 16.1 53593 79.5 20.5 5268.5 80.7 193 5591.4 80.5 19.5 6484.2 22.1 17.9 7961A 81.0 19.0 7391.9 Direct taxes 61.2 38.8 1906.4 54.4 45.6 1869.3 56.4 43.6 2008.5 58.2 41.8 2367.0 62.1 37.9 3093.3 59.5 40.5 2819.9 Incomeandprofittax 66.7 33.3 174S.S 61.4 38.6 1655.9 61.8 3S.2 133.3 63.5 36.5 2168.8 69.9 30.1 2747.5 65.5 34.5 2561.1 Indirecttax 82.1 17.9 1289.4 80.6 19.4 1180.9 83.0 17.0 1204.3 81.0 19.0 1441.4 85.3 14.7 1696.8 82.1 17.9 1465.9 Foreigntradetaxes 100.0 0.0 2163.5 100.0 0.0 2218.4 100.0 0.0 2378.5 100.0 0.0 2675.8 100.0 0.0 3171.3 100.0 0.0 3106.1 Non-Tax Revenue 82.2 17.8 2176.2 81.9 18.1 2831.8 84.5 15.5 3061.6 84.2 15.8 3016.0 81.1 18.9 2766.6 81.5 18.5 2570.9 Total Revenue 82.0 18.0 7535.5 80.3 19.7 8100.4 82.0 18.0 8653.1 81.7 18.3 9500.2 81.9 I.I 10728.0 81.1 18.9 9962.8 11] Figures for 1996/97 through 1999/00 are prelimiinay actual expenditures. Table 5.5: Federal and Regional Shares in Recurrent Expenditure, 1997 - 2001 1996/97 1997/98 1998199 1999/00 2000/01 (Prel. Actual) (Prel. Actual) (Prel. Actual) (Prel. Actual) (Prel. Actual) Category Federal Regional Federal Regional Federal Regiona Federal Regional Federal Regional Share Share Share Share Share I Share Share Share Share Share General administration 66.2 33.8 77.5 22.5 83.8 16.2 89.4 10.6 79.4 20.6 o/w Defense 100.0 0.0 100.0 0.0 100.0 0.0 100.0 0.0 100.0 0.0 Economic infatucture 72.8 27.2 24.8 75.2 57.9 42.1 54.6 45.4 60.4 39.6 o/w Road construction 1/ 76.4 23.6 25.2 74.8 28.5 71.5 27.0 73.0 33.3 66.7 Economic services 19.9 80.1 22.4 77.6 23.5 76.5 28.4 71.6 25.9 74.1 o/wAgricult. & nat. res. 13.9 86.1 17.4 82.6 17.0 83.0 21.1 78.9 18.3 81.7 Social services 19.0 81.0 22.2 77.8 20.8 79.2 29.6 70.4 22.3 77.7 o/wEducation 14.3 85.7 14.9 85.1 14.2 85.8 16.7 83.3 15.9 84.1 Health 18.1 81.9 20.0 80.0 19.3 80.7 17.4 82.6 18.9 81.1 Others 87.5 12.5 83.5 16.5 96.6 3.4 97.1 2.9 97.6 2.4 o/winterest&charges 100.0 0.0 100.0 0.0 100.0 0.0 100.0 0.0 100.0 0.0 external assistance 100.0 0.0 100.0 0.0 100.0 0.0 100.0 0.0 100.0 0.0 Total current expenditure 56.4 43.6 60.3 39.7 70.2 29.8 78.4 21.6 66.4 33.6 Shares of non-defense expenditure: 49 51 43 57 49 51 57 43 51 49 [I] Includes urban development. Note: These data are presented in greater detail in Chapter 2. 5.21 In principle the federal grant mechanism is simple. In practice, the main complication is the way aid resources are treated. Donor funds may be passed down to the regions as earmarked project funding, and the allocations of federal subsidy to the regions are adjusted to ensure that the balance among regions is not distorted by the regional preferences of donors (this is the so- called 'offset' mechanism, which is described in Appendix 7, Vol. ID). Certain categories of resource flows (of which food aid is the most substantial) are outside the federal grant formula and not subject to the offset mechanism. Although the regions have constitutionally independent audit responsibilities, the federal government is entitled to audit their use of the federal subsidy. Decentralization and Poverty Reduction 5.22 As already noted, a high proportion of poverty-reducing public expenditure takes place at the regional level. The PRSP process increases the salience of poverty in dialogue between the government and donors, and the I-PRSP stresses the importance of decentralization in relation to poverty reduction: considerable space has been created for effective decentralization and empowerment, which in turn creates room for tackling poverty directly at the grass root level. 5.23 Against this background and the work initiated in this round of RPERs, further work should focus on: - differences in the patterns of poverty across and within regions; - level of awareness at the regional level about the different dimensions of poverty, and the systems in place to monitor poverty; and - how poverty considerations feed into the planning and the evaluation of public expenditures at regional level. Decentralization and Aid Management 5.24 Aid management greatly complicates the fiscal relationships between the federal and regional tiers of government. The federal government has primacy in dealing with aid and aid relations (the federal government is constitutionally responsible for official foreign relations and international agreements and the regions have no powers to borrow abroad) and, as per regulations, all aid flows require federal approval, even if the federal government is not directly 84 involved in the disbursement. Regional shares of the federal subsidy are adjusted to take account of aid to the regions under the offset mechanism, and any earmarking of donor funds has to fit into the context of the unearmarked federal subsidy (see Appendix 7, Vol. II). Therefore a major interest in the RPERs is to see how aid management actually operates at the regional level - the relative importance of aid resources and how they affect the regional processes of planning and resource allocation. 5.25 Specific issues of importance from the standpoint of aid partners include: - the interaction between federal and regional levels in setting expenditure priorities and monitoring expenditure performance (this is especially relevant for HIPC and for other forms of non-project aid); - ability to track expenditures at regional level (the detail and the timeliness of expenditure information available etc); - the quality of public expenditure management at regional level (including the fiduciary standards of expenditure control and reporting, but also the efficiency and effectiveness of resource use); - capacities and capacity building requirements at various levels. 85 Table 5.6: Regional Per Capita Expenditure FYOO/OI' (In Birr) Category Tigray Afar Amhara Oromia Somali Benishangul- SNNP Gambella Hareri Addis Ababa Dire Dawa Total Gumuz General Administration 16.7 42.3 10.6 10.6 21.2 57.6 11.4 187.1 72.2 63.0 42.9 15.6 Economic Infrastructure 8.9 11.8 3.3 5.3 12.6 47.8 3.8 109.4 35.1 86.2 13.4 9.2 o/w Road Construction 2/ 8.9 11.8 0.5 4.9 12.4 47.8 3.8 109.4 35.1 83.9 13.4 8.2 EconomicServices&Development 18.7 32.6 13.8 17.2 19.2 60.2 13.5 83.7 38.2 74.7 20.8 18.9 o/wAgric. &Nat. Res. 16.5 30.1 13.4 16.8 18.7 53.4 13.0 75.2 20.2 72.2 18.1 18.2 Social Services & Development. 71.1 81.9 34.9 39.5 38.3 153.5 44.7 248.4 274.8 225.1 105.0 51.6 o/w Education 42.6 37.3 21.4 25.0 16.5 70.8 27.1 110.8 139.9 113.0 49.3 29.7 Health 23.9 32.9 11.2 12.6 17.7 58.8 17.2 123.0 101.4 45.0 47.3 16.9 Others 8.2 31.2 3.5 2.1 8.6 22.8 2.7 31.1 23.5 67.6 5.5 6.7 Total Expenditure 123.7 199.7 65.9 74.7 99.9 341.8 76.2 659.9 443.8 516.6 187.7 102.0 Ownrevenuepercapita(Table 5.2) 20.8 12.7 11.3 15.2 8.2 24.2 11.5 35.6 54.0 409.5 71.7 1/ As per FY01 regional budgets which are marginally different from the federal expenditure ceilings. Hence per capita expenditure figures are marginally different from those in Table 5.2. 2/ Road construction under recurrent includes urban development. Note: Figures used in this table refer to FY01 budget estimates. III. COMMON THEMES AND ISSUES FROM THE RPERs Approach and Objectives of the RPERs 5.26 The three RPERs were supported by different aid agencies78 but followed common terms of reference. In each case a working group of regional officials was constituted to guide the exercise. A joint workshop for the three regions was held (February 2001) before the main fieldwork to identify common issues and to harmonize the approaches and the proposed format. At the same time there were certain differences in emphasis. For example, the Benishangul- Gumuz study was the only one to pay special attention to gender issues, while fieldwork in Oromiya focused especially on agriculture. Lessons from the RPER process are reviewed at the end of this section. 5.27 Box 5.1 shows the goals and objectives of the RPERs. In essence the aim has been: * to deepen the understanding of donors and central authorities of regional public expenditure processes, strategies and results; * to assist regional authorities in addressing public expenditure management issues and support strengthening of capacity at the regional level; and * to increase the awareness of regional authorities to their emerging role in the implementation of the PRS and utilization of HIPC resources. 5.28 In the event, all the TORs were addressed by the study teams, albeit with varying degrees of emphasis and detail. The process of inquiry and understanding proved to be more complex and time consuming than originally envisaged. The launch workshop in February 2001 went a long way in helping establish contact and acquaintance with the regional authorities and begin the process of engagement. Even so the regional officials were initially not fully appreciative of the value of the exercise to them, were busy with other commitments and took time to set up the RPER Working Groups to act as counterpart teams. The data provided was inconsistent and typically not up to date and responses to questions were often ambiguous. While the study teams lacked full awareness of the systems, processes and concerns at the regional level, the regional authorities were only vaguely familiar with emerging issues such as PRSP and HIPC and their potentially frontline role in the emerging scenario. 5.29 The yardsticks for judging the quality of the RPERs should be informed by this context, and their value should not be underestimated by the lack of required rigor in reporting or in not providing unambiguous answers. At the least, the RPER exercise helped begin a long and potentially valuable process of engagement and understanding. The regional authorities became more aware and appreciative of the usefulness of this exercise, and more importantly of their growing importance in the PRSP/fIPC context. For the federal government and the donors, the value has been in confirming what they always understood to be constraints to efficient public expenditure management at the regional level and in gaining new insights. Even if there are no clear cut answers, the RPERs have advanced the agenda by helping to ask the right questions. This is a valuable base to build future work on. 7' The EC for Tigray, the ADB for Benishangul-Gumuz and the World Bank for Oromiya. DFID helped to finance the local consultants and workshops. 87 Box 5.1: Goals and Objectives of the Regional Public Expenditure Reviews As stated in the Terms of Reference for the RPERs: The following are the overall goals of the regional focus in the PER: (i) To add depth and relevance to the findings and recommendations of the PER as a whole. (ii) To provide useful advice and support on public expenditure management to the regions themselves. (iii) To assess, from the view point of public finance management, the capacity of the regions to implement the PRSP and to identify measures to strengthen this capacity. (iv) To explore the links between poverty and public expenditures at regional level. The specific objectives for the RPERs are: (i) To consolidate available information on resources and expenditures and on that basis review fiscal performance at the regional level. (ii) To understand the budgetary institutions, systems and processes at the regional level. (iii) To engage key regional officials in the review of regional expenditure allocations and expenditure management processes. (iv) To review reforms undertaken/planned in the area of public finance management and to make practical recommendations for the strengthening of public expenditure management in the regions studied. On the basis of experience from these pilot regional PERs, to make recommendations for future PER work at regional level, in these and other regions. 5.30 As noted in the RPER Terms of Reference, the reports are intended to cater simultaneously to multiple audiences, including (a) officials charged with public expenditure management at the regional level, (b) the federal government, (c) aid partners, and (d) civil society. These audiences are very different. Things that are very familiar to one audience may be new to other readers, and there is value in presenting the same coherent picture to all stakeholders. This section accordingly puts together some common themes and issues drawing from the three RPERs. No quantitative comparisons have been attempted, however, as the fiscal data bases at the regional level are not yet robust enough to support such a comparative analysis. This section begins with a note on the characteristics of the pilot RPER regions. It then highlights the RPERs' key findings concerning: patterns of expenditure; planning and budget formulation at the regional level; decentralization within the regions; budget execution, tracking and accountability; poverty issues at the regional level; implications for aid management; and lessons for future RPERs. Characteristics of the Pilot Regions 5.31 The RPERs were deliberately conducted in three very different regions (shown highlighted in Table 5.1 above). Oromiya is Ethiopia's largest region, with a huge land area and an ethnically homogeneous population larger than most African states. In scale, Amhara and SNNP are comparable to Oromiya, but SNNP is far more diverse, and both Amhara and SNNP have gone further than Oromiya in decentralizing to zonal level and below. By comparison, Tigray is much smaller; its population is comparable to the Somali region's, but in administrative capacity and sophistication it is often bracketed with the three larger regions. Benishangul- Gumuz in turn has only a fraction of Tigray's population but is ethnically much more diverse. It shares some characteristics with the other emerging regions (Somali, Afar, Gambella), but there are also vast differences within this group. Somali and Afar, for example, are ethnically n The Benishangul-Gumuz RPER (12.2. 1) comments: The development issues in every field are overlain with complications arisingfrom the diversity of origins of its citizens. 88 homogeneous but predominantly pastoral economies. Thus the diversity among the RPER regions does not necessarily make them representative of all of Ethiopia's regions. Patterns of Expenditure Recent trends 5.32 Expenditure trends over the past three years (1998/99 - 2000/01)80 have been dominated by the impact of the border conflict, and the RPERs report on how the regions adapted to the reduction in funding that resulted. Initially (FY99) the main effect was a reduction in aid funding for capital expenditure, but as the conflict continued there was also a sharp reduction in the federal subsidy (FY00). In all three regions, capital expenditures bore the brunt of the reduction in funding. Recurrent expenditures were protected, and sectoral shares changed relatively little. Nevertheless, against a background of efforts to expand health and education facilities, there were reductions in per capita recurrent expenditures and even more substantial shortfalls compared with planned levels of funding. Collections of regions' own revenues increased substantially, but, given the regions' dependence on external funding for the bulk of expenditures, they could only partially substitute for the reduction in federal and donor funding. Expenditure projections for FY01 and FY02 show an increase in the federal subsidy and the beginnings of a revival in aid funding, but with regional revenues returning towards previous levels. Budgeted vs. actual expenditure 5.33 The relationship between budgeted expenditures and actual outturn is an important indicator both of implementation capacity and of discipline in expenditure management. The three RPERs find that actual expenditures correspond closely to budgets as far as domestically funded expenditures are concerned, but that there are regularly shortfalls in utilization of budgeted aid. In Oromiya, for example, on average about 95 percent of the recurrent budget has been spent over the past three years and 91 percent of the domestic capital budget, but only 74 percent of the aid-funded budget. Reasons for the difficulty in utilizing aid are further discussed below. The performance in utilizing the domestically-funded budgets suggests that implementation capacity is not yet a binding constraint at this level. The Tigray RPER describes systematic quarterly reviews of budget appropriation and execution levels in order to identify possible transfers of resources from non-performing administrative units to those with unsatisfied absorption capacities (even across budget appropriations and across sectors). Capital vs. recurrent expenditure 5.34 As already noted, capital expenditures absorbed most of the stringency on expenditures that resulted from the border conflict. Benishangul-Gumuz reports a much higher share of capital expenditure than the national average (55 percent of total expenditure over the period 1997/98- 1999/00). Prima facie, such a high share of capital expenditure would raise concerns about the sustainability of the resulting recurrent costs, but this would need to be examined sector by sector. The high apparent level of capital expenditure in Benishangul-Gumuz may partly reflect a larger proportion of federal (as opposed to aid) resources compared with some other regions."' 80 Ethiopian Fiscal Years 1991 - 1993. 811t is not unambiguously clear that Benishangul-Gumuz has higher proportional capital spending relative to other regions. In view of the widespread practice of some donor funds flowing off-budget at the regional level (a practice widespread in Tigray), overall capital spending (on-budget and off-budget) in B-G may be comparable to other regions. Indeed off-budget funding not only erodes budget integrity, but could also create unanticipated recurrent cost obligations in the future. 89 Wage and non-wage recurrent expenditures 5.35 All three RPERs report systematic efforts to give priority to essential recurrent expenditures (see below on budget prioritization criteria). At the same time there is clear evidence that the pressure on resources has led to under-funding of operating costs. In health and education, salaries have been a large and growing share of recurrent expenditures, and it is reported that some critical non-wage items are under-funded. One response to this pressure has been to divert capital resources for items that are strictly recurrent. In Oromiya, it is estimated that between 8-16 percent of the capital budget is used for recurrent costs.82 The Tigray RPER expresses concern that the tendency for many donor funds not to appear in the budget may undermine efforts to ensure that capital expenditures remain consistent with the region's capacity to meet recurrent costs in the longer term. Aid, budgeted and unbudgeted resources 5.36 All public expenditure analyses within regions (including comparisons over time) and all comparisons across regions need to pay special attention to the impact of various aid flows, which can be substantial but are not uniformly captured in regional budgets. Principal findings from the three RPERs were: * Oromiya: Reported foreign financing is a small part of the overall budget (less than 10 percent over the period from 1998/99-1999/00, though projected to increase significantly in 2000/01). The team was unable to get an estimate of the foreign resources not captured by the budget, as information on these flows is not kept by any single agency. The aid-funded part of the capital budget had fluctuated considerably, and disbursements regularly fell short of budgeted amounts. * Tigray: Donors and NGOs are especially active in this region, but, again, financial data on donor and regional activities are not systematically collected and monitored at regional level (even though activities are physically monitored and coordinated at zone and woreda level) The RPER made estimates of resource flows not captured within the regional budget, which in Tigray's case include very substantial quantities of food aid, and concluded that off-budgetary resource flows (off-budgetary donor and NGO flows together with relief food aid) nearly doubled total resource flows to the region in the past three years. The existence of these flows has an influence on the pattern of expenditure that is reflected in the regional budget. For example, water supply has been mainly implemented through various NGO and off-budget donor projects, with the result that the regional budget figures do not fully reflect the priority accorded to this program. * Benishangul-Gumuz: The RPER identifies about ETB 70 million of extra- budgetary resource flows to the region over the past five years. This figure (although it may be understated) amounts to less than one year's budgeted capital expenditure. This is proportionally far less than the equivalent flows to Tigray, and may help to explain why a higher proportion of Benishangul-Gumuz's budgeted expenditures are for capital projects. Like other regions, Benishangul- Gumuz reports that treasury funds are much preferred to aid funding because they are more reliable and less demanding to access. S2 From the available data it has not been possible to ascertain how this is accounted for ex-post - under capital or recurrent expenditure. Given the close alignment of budget figures and the actual outturn, there may also be some mis-accounting. This is an issue that needs to be further investigated. 90 Budget Formulation at Regional Level 5.37 Quite expectedly, the Benishangul-Gumuz RPER is the one that puts the most stress on weaknesses in planning and management capacity, but there are strong similarities in the basic procedures for planning and budget formulation in all three regions. Recurrent and capital budgets are formulated separately, but the regions' five-year plans provide strategic guidelines especially for capital expenditure. The five-year plans indicate priorities, but not within a definite medium-term projection of resource envelope. 5.38 There are bottom-up inputs in indicating priorities, but final allocations are made at upper levels of the system. Clear expenditure ceilings are not available until late in the budget preparation process. This is mainly because, although regions can forecast their own revenues for the coming year, information on the federal subsidy and on aid is received only in the later stages of budget preparation. Once the overall ceiling is known, the regional finance and planning bureaus work with the sectoral bureaus to adjust budget proposals to fit within the ceilings. The broad order of priorities is: (a) to meet non-discretionary recurrent costs; (b) to meet discretionary recurrent expenditure commitments (including those that arise from the completion of new projects); (c) to complete ongoing capital projects; and finally (d) to fund new projects. Although this process does not prevent operating costs from being severely squeezed, systematic failure to fund the recurrent costs of projects does not appear to be a major problem. Decentralization Within Regions 5.39 The Tigray RPER (§5.2.1) assesses the prioritization process as follows: "As public resources are very lirnited compared to identified needs, trade-offs can not be avoided. These are made at several levels. Firstly, communities are reportedly well informed about the regional development priorities and strategies and are guided by the administration in the identification of their constraints, priorities and needs: although a 'blue sky' attitude might still prevail to some extent, the process is undertaken within an established framework with inherent trade-offs between priorities. Secondly, at all planning levels (woreda, zone, region), account is taken of existing and potential NGO and off-budget donor activities in each area: these are factored into the decision-making process, such as to obtain an equitable distribution of total resources (budgetary plus off- budgetary) in each geographical area. Thirdly the planning and budgetary processes involve both the administrative non-elected structure and the directly elected members of the councils: even though in last instance it is the Regional Council which decides on resource allocation by approving the regional budget, all communities through their elected representatives have a voice in this Council. Fourthly, all expenditures are planned according to a priority order. On the whole then, the current process appears to produce a trade-off between the satisfaction of local priorities (relief of locally identified constraints) and regional priorities for development which is probably more in favor of the latter but where the former are taken account of to the extent of available resources." 5.40 This approach is probably at its most effective in Tigray. In Benishangul-Gumuz, its smaller scale and capacity constraints mean that procedures are not so systematically articulated while, conversely, Oromiya's larger scale means that zones and woredas are more remote from the regional headquarters and it is probably harder to reconcile prioritization within sectors with inter-sectoral prioritization across geographical areas. Even in Tigray, the RPER's assessment of the degree of decentralization is ambiguous. It notes that, although the commitment to bottom-up planning is genuine, there is no delegation of budget formulation to the zonal level or below and comments: 91 "It is not clear from the information available to the mission's team whether the current level of decentralization (both in terms of organizational structure and budget planning and execution) entails any disproportionate costs in terms of efficient use of human resources. From observation it seemed that a lot of effort is spent at woreda level for information collection which is transmitted to zone then to region, but that the actual analysis of this information at the level where it is collected is extremely limited (possibly even non existent) while at the same time the informnation at regional level becomes so aggregated that it loses its pertinence for informing woreda level decision-making." 5.41 The Oromiya RPER similarly concludes that the system is more decentralized on paper than it is in practice. There is consultation from below, but decisions come from above. This is reinforced by the very limited discretion accorded to zonal and woreda levels in budget implementation (see budget execution, below). Oromiya is considering moving to a system of block grants to the zones (a system already partly operational in Amhara and SNNP regions). Benishangul-Gumuz officials also reported that such an approach was under consideration (in the context of moves towards a Public Investment Program or Public Expenditure Program) but noted concerns about the capacity of zones and woredas to formulate budgets, and the need for an easily workable formula to allocate resources to sub-regional units. Further decentralization within regions has many ramifications which are discussed in the final section of this chapter. Budget Execution, Tracking and Accountability 5.42 Delegation of responsibility for budget execution is generally pragmatic: funds for budget execution pass down to the level judged most able to handle disbursement effectively. In practice this means that woreda-level responsibilities are very limited and almost all capital projects are implemented at zone or regional level. Actual transfers of funds to lower levels are made when expenditure requirements exceed revenue collections, but this is a system of cash flow management - it does not mean that a woreda, for example, has discretion in how to use the revenue it collects. Detailed line item budgets are approved from above, and implementing agencies have very little discretion in expenditure allocation. There are strict rules on virement, with approvals required from above. Internal audits, unsystematic as they are, serve as an additional check that expenditure authorizations are in line with approved budgets. As described in the Tigray RPER: "Budget tracking of each individual administrative unit is possible at any moment in time since transfer requests at each level are done on the basis of documents retracing the initial budget appropriation, any modifications to it, the budget already claimed, the balance available and the new claim (all detailed by budget line and functional classification). In addition the monthly statements of revenue and expenditure give the consolidated picture per administrative layer. [However] the consolidated reports are routinely used only for control purposes." 5.43 All three RPERs comment favorably on budget discipline and find no evidence of significant misappropriations of funds. Thus in Tigray it was observed that financial control procedures appear extremely detailed and deliberately redundant from one administrative layer to another, causing much cross-checking while actual 'misappropriations' of funds are very limited. The Oromiya team was unable to investigate in detail into financial control and auditing procedures but gathered a general impression that misappropriations and misallocations of funds are rare. 5.44 As regards effectiveness in transmitting funds down to lower levels, the RPER findings are consistent with those emerging from the Bank's woreda studies (draft), which have found that 92 "..generally, the timeliness of payments from the zones to the woredas could be rated as quite acceptable. This is an unusual and creditable result for an LDC which faces the kind of logistic and personnel difficulties that Ethiopia does, and speaks to the high degree of discipline manifest in the civil bureaucracy. That said, it should be noted that the phenomenon is not universal, and that delays in funding flows are a common cause of frustration in donor-financed projects. These arise very often because of the plethora of donor procedural requirements and the burden they impose on the small stock of decision-makers empowered to deal with them at the federal or regional levels." 5.45 At the same time there are common weaknesses in capacity for audit and in the speed with which accounts are finalized: * Oromiya: Expenditures are classified as "actual" when the books have been closed and expenditures have been accounted for. This happens when an expenditure claim form is filled out to document the actual expenditure (as opposed to the commitment). The process of closing the books appears to take a long time: as of March 2001, expenditures for 1998/99 were still "pre-actual" - a lag of two-and-a-half years. [However] differences between pre-actuals and actuals appear to be small. The Regional Audit Bureau audits the accounts every year. The delay in producing audited accounts is even longer - at present, three years. On accountability, the Executive Committee reports annually to the Council, but this generally has not been a forum for any meaningful inquiry into audit lapses or corrective action. * Tigray: Internal audit functions as an additional layer of control rather than analysis. Meanwhile, extemal (independent) audit is undertaken by the Regional Audit Bureau which, due to its very limited financial and human resources, can execute only 10 percent of the required audits each year. Audit reports are not published. * Benishangul-Gumuz: Sensitivity to efficiency of resource use and the need to minimize waste is reflected in a tight system of control through the inspectorate and auditing systems and the various forms introduced by the Finance Bureau. However, capacity constraints tend to limit the effectiveness of the control and auditing systems. Inspectors and auditors are in short supply and there is shortage of qualified accountants thus constraining timely closing of accounts. There is currently a backlog of two years accounts yet to be audited, the last audited accounts being those of FY98 Fiscal Year (EFY1990). The backlog is reportedly due to delays in closing accounts on account of shortage of qualified accountants, late reporting on expenditure by woredas and zones to regional bureaus and information gap on loans and assistance from the federal government. Audit reports are presented to the Regional Council and appropriate sanctions are applied to delinquents but a shortage of attorneys limits adequate follow-up of external audit findings. 5.46 Expenditure reporting systems are oriented towards financial control, and their ability to provide consistent and timely management information is limited. This arises both from design weaknesses in the systems (e.g. in expenditure classification) and from limited capacity to compile and collate the necessary reports. Human resources are stretched and computerization is still very limited Systemic improvements (e.g. in expenditure classifications) are being introduced through the nation-wide budget reform program, but rolling out the reforms will take time. For example, Tigray plans to introduce the new formats in 2002/03, but possibly only in one 93 or two pilot zones, and anticipates practical teething problems."3 For Tigray: The main problems in public finance management reported by officials concern mostly the weakness of human resources [and there is also] strong need and demand for computerization of Regional Government Offices. 5.47 At the same time, embedded systems of performance monitoring do attempt to link expenditures to physical outputs. As described for Oromiya: The systems to track inputs and outputs - expenditures and personnel, and the direct outputs achieved with these inputs - reach far down to the lowest levels of government. Monitoring of expenditures and physical outputs appears to be conducted regularly and seriously. Lower levels of government know the targets they have to achieve, and monitor achievement versus plan regularly. There are monthly reviews of expenditures, quarterly reviews of expenditures and physical achievements, mid-year workshops with zonal staff, and project visits twice a year. These reviews are considered "tough" by regional staff. Zonal level project committees and technical committees check to ensure that the work is done as planned. 5.48 But, again following the Oromiya description: While the systems are in place, there are problems. The numbers often do not match. The team encountered considerable problems in reconciling data from various sources. Data from woredas do not match data from zones, which in turn don't match data from the region, and data from the region do not match data from the federal government. This indicates that, while some systems are in place, they are generally inadequate to provide timely and accurate reporting on expenditures. More worryingly, the data appear to be used to inform decisions much less than they could potentially be. This is partly due to the limited use of computers, which makes transmitting, collecting and analyzing the numbers slow and prone to error. But more importantly, all the information collected on expenditures and outputs is not used much to inform decisions because the scope for decision-making, especially at lower levels of government, is limited. The overall picture is of a massive tracking effort which does not necessarily lead to better allocation of resources because those who have the best perception for what money is buying have no power to decide on allocations. While there are problems with the systems in place, there are also very significant strengths. First, setting targets and monitoring them in a disciplined way is the basis for any good monitoring and evaluation system. The fact that this kind of exercise is second nature to govermment officials down to the lowest levels of government constitutes a basis upon which to build more effective systems; the human capacity is there. If meaningful poverty indicators were tracked, and officials had more power to make decisions based on what they see happening, they would probably increase the effectiveness of what they do. Second, once targets are set for activities and outputs, the impression is that these do get achieved. In other words, the system is able to deliver on what it plans to achieve. Given how limited funds and capacity are, this speaks of the great resourcefulness and dedication of field level functionaries. 83 Such as fitting the larger number of expenditure categories onto a manual report form in advance of computerisation. 94 Poverty Issues at Regional Level 5.49 Each RPER includes a regional profile with a particular focus on poverty. The profiles reveal significant differences between the three regions, but also highlight the limitations of readily available data. As regards poverty characteristics:84 * Oromiya has lower income poverty incidence, depth and severity than the national average (estimates are lower than for the other large regions); but this does not mean that there is little income-poverty in Oromiya; because the region's population is large, it has the second-largest number of poor people among regions in Ethiopia - almost 7 million; other social indicators confirn that, while on average conditions in Oromiya are not worse than elsewhere, there are large discrepancies within the region. * Compared to national averages, Tigray has less extreme poverty and extreme wealth, but lower food intake and worse nutritional status of children. * Benishangul-Gumuz has income poverty similar to national levels but worse social and gender indicators. 5.50 All three RPER teams were frustrated by the scarcity of information that would allow a more thorough profile of poverty and of trends in poverty within the regions. They were particularly concerned about a lack of analysis and use of information that is already (potentially) available, and this is reflected in specific recommendations: v In Tigray monitoring and evaluation of poverty is not yet carried out systematically; data are collected widely but not systematically linked to the planning process. - The education and health sectors do collect systematic data relating to their sectors; however, for the moment, available social sector data focus on the supply and coverage of services available rather than on the quality of services and their impact on poverty and/or on levels of social well-being. - Overall, the first bottleneck is not lack of data, but lack of a systematized system to consolidate and report on socio-economic data and to link them to the planning and budgeting process. It is recommended, also in view of the coming PRSP, to undertake a detailed inventory of what data are existing and needed for a socio-economic and poverty database. The planning bureau appears in favor of setting up such a regional socio- economic database and information network. * The Oromiya RPER finds that: - While there are systems in place that can track inputs and outputs - although with a number of problems - systems to track outcomes and impacts are virtually non-existent at the regional level. Formalized systems exist only in the sectors covered by sector development plans, education and health, and even there the degree to which they function varies by type of activity. The regional administration does not have a regional poverty monitoring system in place: no regional household surveys, qualitative poverty monitoring tools, or analysis of data collected by federal agencies, and no poverty monitoring unit, or equivalent office are in place. 84 Based largely on the 'snapshot' provided by the first round of the Welfare Monitoring Survey. 95 - Indeed there is no need for regions to conduct independent surveys. They can use the data from the national surveys conducted by the federal government to gain regional perspectives. Both the Household Income and Expenditure Surveys and the Welfare Monitoring Surveys cover enough households i- the Oromiya region to make regional analysis possible. Should the region develop capacity to analyze poverty data, and also to develop quantitative and qualitative household data collection instruments that suit its needs? Should it have a poverty monitoring unit? For a region the size of Oromiya, this would be a justified investment. Thus, the regional authorities should request donor assistance to set up systems and build capacity, and donors and the Central Statistical Authority should assist the regions in this endeavor. * The Benishangul-Gumuz RPER also stresses the need for training of planning staff to take poverty criteria into account at the planning stage. 5.51 Against this background, it is not surprising that all three RPERs comment that rather little is known about the impact of regional public expenditures on poverty, although the emphasis on basic health and education services and agriculture indicates a generally pro-poor orientation. 5.52 The Oromiya report, based on insights from a field visit, suggests ways of monitoring poverty impacts more directly in the agriculture sector: "The team's field visits confirmed the lack of systems to monitor what happens on the ground in terms of poverty reduction - very little if any time and effort goes into monitoring poverty-related outcomes and impacts. Expenditures and activities carried out are monitored carefully in the agriculture sector, but the indicators tracked often do not help assess whether farmers are better off. For example, farmers may be receiving the extension package, producing more, and repaying their input loans on time, but they may have had to sell cattle to do so because the price of the grain they produced was low when they sold it, and their income was not sufficient to repay the loan. If farmers have to sell their cattle to repay loans, in reality they are not better off, even though all the monitored indicators look fine. Government officials have no way of knowing from looking at achievements whether the farmers are better off. They also do not track grain prices and volumes moving across the region, so they do not have ways of assessing whether increases in production lead to lower prices in an area, or whether other factors are at play. On the positive side, the team observed during the field visits a strong discipline to track indicators and the capacity to achieve what was planned. So it seems that it would be possible to introduce a small number of indicators more closely related to the well-being of farmers. For example, Development Agents could also check if farmers who receive the input package have improved their homes in the previous year, have more or less cattle, are sending all their children to school, and so on." Implications for Aid Management 5.53 The RPERs support a number of observations about aid management (most of which are familiar from previous PERs and from reviews of the sector development programs). Thus: * The earmarking of donor funds to specific projects has a high cost, which is reflected in extra demands on regions' scarce administrative capacity. 96 Implementation rates for donor-funded projects are consistently lower than for similar treasury-funded projects, and regions have a rational preference for treasury funds rather than aid.85 * Despite the federal government's policy of accounting for all aid within its consolidated fund, substantial volumes of aid to the regions which are disbursed through Channels 2 and 3 are still not captured in federal or regional budgets. Regions nevertheless try to ensure that aid is used for their planned priorities, and to adapt their budgeted expenditure to complement aid-funded projects.86 This is made more difficult by the unpredictability of aid flows (this is a factor for both off-budget and on-budget aid). Thus (Oromiya RPER): the uncertainties surrounding aid inflows pose a major challenge for good budget planning and management, and donors and the federal and regional governments should find ways to make disbursements more predictable. * There are weaknesses in reporting and coordination of aid at regional level. The Oromiya RPER comments on the need for better communication between regional authorities and MEDAC over aid expectations and flows. The Tigray RPER notes that budget reporting is done separately for treasury and donor resources, the latter being done largely through the sector bureaus rather than the fnance or the planning bureaus. It comments that aid management appears to be stronger at sector level than at the regional planning level; there is little consolidated information available on donor activities across sectors. For NGO activities, management and information collection is mostly done at woreda and zone level, with very little consolidated information available at the regional level. Ironically, regional authorities in Benishangul-Gumuz appear better informed about Channel 3 than about Channel 1 and Channel 2 funds. * The Ethiopian system of public expenditure management gives important responsibilities to the Planning Bureaus, which are expected to have an overview of resources and of development priorities and to lead the processes of strategic prioritization and performance review. The way aid and aid projects are managed can undercut this role. This point is made clearly by the Tigray RPER: - Aid management seems to be quite fragmented and weak in Tigray. There is no centralized database of all ongoing and planned development activities in the region. In the absence of a consolidated picture of donor aid, no analysis can be undertaken of donor flows and their implications for public finance management, monitoring and planning. Aid management and coordination appears to be stronger at sector level with most sector Bureaux being well informed about what is going on in their sector. However, again sector Bureaux do not seem particularly well aware of the distinctions made between budgetary and off-budgetary resources, which renders data collection particularly difficult. BOPED is best placed to co-ordinate all development aid. However, to do this BOPED needs to set up a system for regularly monitoring all development activities, to establish a good database and to increase its 85 See discussion of offset in Appendix 7, Vol II. 86 The Tigray RPER states: at all planning levels (woreda, zone, region), account is taken of existing and potential NGO and off-budget donor activities in each area; these are factored into the decision-making process, such as to obtain an equitable distribution of total resources (budgetary plus off-budgetary) in each geographical area. 97 efforts to collect the information. On the other side the donors should improve their willingness and discipline to keep BOPED informed of planned activities and realized expenditure. - [Also] the mission questions whether the Disaster Prevention and Preparedness Bureau (DPPB) is the right organ to coordinate the NGOs. It seems that this is a heritage of the past when most of the NGOs were relief aid organizations. BOPED seems to be a more appropriate organ to coordinate NGO activities, because BOPED has the responsibility to coordinate all development activities in Tigray. For undertaking the task to co-ordinate NGO activities, BOPED needs to be strengthened. One of the first challenges of BOPED will be to establish a sound system for monitoring all donor aid (including NGO aid). Lessons for Future RPERs 5.54 As regards the preparation process, all three RPERs reported good cooperation from regional officials, but none of them succeeded in formally convening the RPER Working Groups. Partly for this reason, they felt there was less direct involvement of regional officials than was desirable. (This view was echoed in subsequent regional comments, which also emphasized the haste with which the RPER exercise was launched, lack of awareness of the exercise at Regional Council level, and the competing demands on officials' time.) Preoccupation with data collection limited the depth of analysis that was possible. 5.55 The value of these pilot RPERs, both to the regions themselves and to wider stakeholders, was an important topic at the September 2001 PER workshop.87 There was consensus that the RPERs served a useful purpose in a variety of ways. They improved the understanding of both the federal government and the donors of regional level issues while raising awareness of development priorities as well as analytical capacity at the regional level. Several suggestions were made for improving the process: consulting with the regions before firming up the TORs so that the coverage is demand driven; explicit effort at knowledge transfer, translation of the report to local languages, dissemination through workshops.88 Regional representatives felt that extending the exercise to other regions should be a higher priority than studying the same regions in greater depth, but Oromiya did wish to see their exercise repeated with more in-depth study and greater involvement of regional officials in the next round. IV. ADDITIONAL ISSUES AND CONCLUSIONS 5.56 The mission work for the RPERs and the drafting of the reports were completed by end August 2001. Subsequently, in October 2001, the Prime Minister presented his (re-elected) government's development strategy to the parliament. Among other things, one of the central pillars of this strategy is decentralization. The new initiatives of the government include the following: (i) devolution of untied block grants to woredas to effect a better match between priorities expressed by the communities and expenditure allocations; (ii) constitutional amendment giving exclusive responsibility to the regions for primary education, primary health and rural roads; (iii) rolling over reforms under the CSRP to the regions on an accelerated time table; (iv) tax reform one of whose objectives will be improvement of the regional revenue base; and (v) merging the bureaus of finance and planning into a single entity to effect coordination 87 See the Summary Record ofDiscussion, PER 2001 Discussion Workshop, 17-18 September 2001, circulated on 28 September 2001. 88 Only the Benishangul-Gumuz RPER has already been reviewed at a regional dissemination workshop. It includes a series of recommendations reflecting the views of the participants. A workshop for Oromiya is planned in the next few months. 98 between capital and recurrent budgets.89 This final section accordingly addresses a few topics which extend beyond the direct findings of the RPERs but are relevant to the broader agenda of the Public Expenditure Review. The issues are examined in the context of the renewed focus on fiscal decentralization. Regional Revenues and Predictability 5.57 The RPERs have emphasized the narrow own revenue base of the regions which makes them heavily dependent on federal subsidies and external aid. Such a sharp vertical imbalance militates against healthy and robust fiscal decentralization. The received wisdom is that it is only when sub-national units are responsible for a significant proportion of their overall resources will they act in a fiscally responsible manner. The premise is that there can be no pleasure of spending without the pain of taxation. On the other hand, there is no evidence so far of the regions in Ethiopia behaving in a fiscally irresponsible manner. In fact, regions have been remarkably disciplined in conforming to macroeconomic discipline and the expenditure ceilings imposed on them by the federal government. Regions recognize that resources are limited and systematically priorities their expenditures.90 However, their ability to use resources efficiently is compromised by uncertainty about future resource levels. 5.58 There are two initiatives under way to address these problems. First, the long term solution is to improve the own revenue base of the regions. The proposed Bank study on regional revenues under the Italian Capacity Building Trust Fund will provide the analytical underpinnings for this. Second, the regions need some predictability in their resource envelopes over the medium term so that they can plan ahead. Reforms under way such as implementation of PEP and adherence to the financial calendar should provide the way forward in this regard. Community-Oriented Decentralization 5.59 The RPERs describe the (somewhat limited) extent of fiscal decentralization within the three regions reviewed. There has been some deconcentration of administrative powers, but little tangible devolution of resources to local levels. The same finding is reported by the (forthcoming) Woreda Studies: In practice, the woredas studied used a predominantly top-down planning approach in which the inputs of communities and even of elected officials were subordinated to sector-based planning, with zonal and regional bureau staff the dominant actors. As a result, the involvement of communities and their influence on the planning of official development inputs was marginal, as the study's examination of community-level planning in 1998/99 corroborated. 5.60 The woreda studies draw attention to "a new paradigm for rural development,91 which sees the community, not the bureaucracy, as the primary agent of integration, and focuses on ways to empower communities to manage their own development; under this paradigm, Government agencies become the providers of demand-driven services and operate in a universe in which communities have the power to call on other service delivery actors as well (the private sector, CBOs, NGOs)." 5.61 This paradigm has generated interest in mechanisms for providing resources directly to community levels. The World Bank is, or expects to be, involved in a number of such initiatives: 89 It is expected that the new development strategy will be more cogently articulated in the PRSP now under preparation. 90 Indeed, the fact that the federal subsidy is not earmarked makes regions well aware of the opportunity cost of all their expenditures. 9' Referred to as the Community Action Program (CAP) in World Bank's terminology. 99 these include schemes under the ESRDF which are identified and implemented by communities by matching ESRDF grant with their contribution, the rural water supply pilot project designed to enhance community-level control of and responsibility for the planning, construction and maintenance of water points, and the community grant window under the HIV/AIDS project. Under the proposed Food Security Project, kebeles would be provided with grants for rural development purposes. These grants would build community capacity to plan and implement development programs as well as providing them with untied funds to pursue activities of their own choosing. A number of bilateral donors support grant facilities along similar lines. 5.62 Topping these specific initiatives will be the recent decision of the government to pass on untied block grants to woredas.92 It is as yet unclear what the guidelines will be for allocating these grants, for supervision of the expenditures and for enforcing fiduciary accountability. The PM's statement has, however, indicated an intention to devolve greater responsibility to the communities in managing these expenditures. It is also proposed to give greater prominence to women. Capacity Building and the EMCP 5.63 The RPERs make a number of references to reforms in planning and budgeting that are supported by the Expenditure Management and Control Program, which is coordinated by MEDAC and the Ministry of Finance.93 These reforms are central to capacity building in public expenditure management at the regional level. It is significant in this context that the new development strategy gives primacy to rolling over reforms under EMCP to the regions on an accelerated time-table. It would be very useful for all stakeholders to have a shared appreciation of the planned timetable for carrying through these reforms in each of the regions. In particular, a clear approach and timetable are needed for implementing regions' existing commitments to develop systematic medium-term expenditure planning, and for introducing the reformed chart of accounts. The Limits of Expenditure Discretion 5.64 Finally, in assessing the performance of the regions in public expenditure management, it is important not to lose sight of the principal constraints that affect them. The Tigray RPER draws attention to the region's limited discretion in allocating resources: "In total then, the freedom to allocate budgetary resources is very limited: an increasing share is taken up by what is considered to be 'non-discretionary' recurrent expenditure (56.5 percent of treasury funded expenditure in EFY 1990 up to 74 percent budgeted for in EFY 1993; the share of recurrent expenditure in total budgetary resources including loans and grants hovering around 50 percent). From the residual left over, 18.8 percent in EFY 1992 was pre-decided through the provision of matching funds. If one takes EFY 1992 as an example, only 61 million Birr (or 17 Birr/capita/year) remain after taking out non-discretionary expenditure (recurrent budget, loans and grants plus matching funds) representing 23 percent of the treasury funded resources and 18 percent of total resources (including loans and grants). 5.65 Other regions face a similar constraint. Of course, discretion is less important in the short-run: fine-tuning an annual budget will only have a marginal impact on the composition of expenditures, and non-marginal changes are likely to take a number of years to achieve - hence the importance of clear medium and longer term expenditure priorities to guide the successive 92 The regional constitutions are proposed to be amended for this purpose. 93 Since merged into a single Ministry of Finance and Economic Development. 100 annual budgets. At the same time, actors at each level usually have a good appreciation of what does and does not make a significant difference to their situation. Thus, for example, the woreda studies (draft) noted that woreda officials and civil servants alike were much more concerned with the level of resources than with the allocative process - the sheer shortage of funding was seen as a much more urgent concern. The amounts of government capital funds that trickle down to woreda level are so limited that woredas are more likely to be interested in what is available through ESRDF, food-for-work schemes, aid and NGOs. In similar vein, and to take one example, the dividends from fine-tuning the federal subsidy formula are rather small,94 while the benefits from more effective uptake of aid could be substantial. Indeed, as a general point, levels of per capita expenditure are extremely low. It is right to be concerned about efficient allocation and effective use of existing resources, but the International Development Targets for basic services will not be achieved without a substantial increase in regional public expenditures. 94 See Appendix 7. 101 ANNEX A Task Members Counterpart Government Sub-task Institutions Involved (Mansaing Institution) Officials Review of Fiscal Performance World Bank and IMF Duvvuri Subbarao (WB) Ato Fantahun Belew (MoF) Eyerusalem Fasika (WB) Dominique Guillaume (IMF) Tracking Public Expenditures for DFID Stephen Lister (Lead Consultant - DFID) Ato Demelash Alem (MoF) Poverty Reduction Trish Silkin (Consultant - DFID) and Ato Melaku Kifle Henk Koppen (Consultant - DFID) (MEDaC) Stefan Dercon (Consultant - DFID) Donors Shifting to Sector and European Conimission Kurt Comelis (EC) Ato Worku Yehualashet Budget Support Ivory Yong (Consultant) (MEDaC) and Ato Tadesse Shiferaw(MoF) Regional PERs * Benishangul AJ)B Stephen Olanrewaju (ADB) Ato Mesfin Girma (MoF) * Ben is bangu l ADB George Namakando (ADB) and Ato Tesfaye Kassa Patricia Alexander (Consultant) (MoF) Girma Moges (Consultant) Haile Ganoro (Consultant) * Oromia World Bank Giovanna Prennushi (WB) Eyerusalem Fasika (WB) Mohammed Mussa (Consultant) * Tigray Filiberto.Ceriani-Sebregondi (EC) European Commission Karolyn Thunissen, (Consultant) Martin van der Linde (Consultant) Ato Bekabtu Arwaja (Consultant)