Report No. 29338-ET Ethiopia Public Expenditure Review The Emerging Challenge (In Two Volumes) Volume I: Public Spending in the Social Sectors 2000-2020 June 2004 Poverty Reduction and Economic Management 2(AFTP2) Country Department for Ethiopia Africa Region Document of the World Bank Government Fiscal Year: July-June Ethiopian Year Gregorian (European year Equivalent) 1989 1996197 1990 1997198 1991 1998199 1992 199912000 1993 2000/01 1994 200 1102 1995 2002103 , 1996 2003104 Currency Equivalent Currency unit: Ethiopian Birr (ETB) US$ 1 = 8.66 (2002) ACRONYMS AND ABBREVIATIONS AAU Addis Ababa Universitv JICA Japan International Cooperation Agency AAWSA Addis Ababa Water & Sewerage Authority MDG Millennium Development Goal ~ ARI Acute respiratory Infection MEDAC Ministry of Economic Development & Cooperation ARM Annual Review Mission MEFF Medium Term Expenditure & Fiscal Framework ARV Anti Retro Viral MMR Maternal Mortality Rate CFAA Country Financial Accountability Assessment MoFED Ministry of Finance & Economic Development COMESA The Common market for Eastern & Southern Africa MoH Ministry of Health CPR ContraceptivePrevalence Rate Mt Metric ton CSA Central Statistical Authority MWR Ministry of Water Resources CSR Corporate Social Responsibility NEA National Examinations CYP Couple-yearprotection NGO Non-Govemntal organization DAG Donor Assistance Group NPPE National Population Policy DALY Disability Adjusted Life year OECD Organisation for Economic Cooperation & Development DFID Department for International Development O&M costs Operation & Maintenance Costs DHS Demographic &Health Survey PASS Pharmaceutical Administration & Supply Services DOTS Directly Observed Treatment PEP Public Expenditure Program DPPC Disaster Preparedness & Preparation Commission PER Public Expenditure Review DPT Diphtheria, Polio and Tetanus PIP Public Investment Program EFAA Education for All PSCAP Public Service Delivery Capacity Building Program EFY Ethiopian Fiscal Year PRGF Poverty Reduction & Growth Facility EGS Employment generation Scheme PRSP Poverty Reduction Strategy paper EPI Extended Program of Immunization PTR PupiliTeacber Ratio EPPD Economic Policy & Planning Department RWFP Reproductive Health & Family Planning ESDP Education Sector Development Program SCF (VK) Save the Children Fund (UK) ESRDF Ethiopian Social Rehabilitation & Development Fund SDPRP Sustainable Development & Poverty Reduction Program ETB Ethiopian Birr SNNP Southern Nations Nationalities and Peoples (Region) EU European Union SPS Special Pharmacies EY Ethiopian year SSA Sub-Saharan Africa FFW Food for Work STD Sexually Transmitted Disease FP Family Planning STR Studentrreacher Ratio GAVI Global Alliance for Vaccines & Immunization TB Tuberculosis GDP Gross Domestic Product TFR Total Fertility Rate GER Gros s EnrollmentRatio TVET Technical & Vocational Education &Training GoE Government of Ethiopia ucs Unit Cost Study GY Gregorian year USMR Under-Five Mortality Rate HAPCO HIV/AIDS Prevention Control Office UNDP United National Development Program HEP Health Extension Package UNEP United Nations Environment Program HEW Health Extension Worker UNFPA United Nations Fund for Population Activities HPC Heavily Indebted Poor Countries UNICEF United Nations Children’s Fund HIV/AIDS Human ImmunodeficiencyVirus/Acquire Immuno Deficiency Syn. WE Universal Primary Education HMIS Health Management Information System USAID United States Agency for International Development HSDP Health Sector Development Program USHEW United States Department of Health, Education & Welfare ICT Information & CommunicationTechnology World Food Program IDA International development Agency WFP World Health organization IEC Information, Education Communication WHO Welfare Monitoring Survey IMP Interim Master Plan WMS Water Resources Development Fund IMR Infant Mortality Rate WRDF Water Supply & Sanitation IUD Intrauterine Device wss Water Supply & SanitationDevelopment Plan WSSDP 1 Vice President: Country Director: Sector Manager: Task Team Leader: Callisto Madavo Ishac Diwan Kathie Krumm William James Smith TABLE OF CONTENTS ACKNOWLEDGEMENTS ..................................................................................................... i EXECUTIVE SUMMARY ........................................................................................... i Part I: Social Sector Spending - The Emerging Challenge .................................................... i Part 11: Overall Public Expenditure (Volume 11)............................................................... xxii Part 111: Summary Listing o f Recommendations Drawn from the Review ...................... m i x INTRODUCTION .................................................................................................................... 1 . 1 OVERVIEW OF SOCIAL SECTOR SPENDING, FUTURE NEEDS, AND POLICY IMPLICATIONS ...................................................................................................... 4 A. A Brief Overview o f Social Sector Spending and Outcomes ..................................... 4 B. Projections o f Future Social Sector Spending Needs................................................ 10 C. Some Policy Implications and Implementation Considerations ............................... 19 D. Locating Social Sector Spending in the Context o f Wider Development and Poverty- Reduction Strategy ................................................................................................................ 26 2 . EDUCATION EXPENDITURE IN ETHIOPIA .......................................................... 30 A. Introduction............................................................................................................... 30 B. An Overview o f Total Public Spending on Education.............................................. 32 C. Public Expenditure on Primary and Secondary Education ....................................... 35 Trends and Composition ........................................................................................... 35 Issues o f Adequacy. Appropriateness. and Effectiveness o f Spending .................... 39 What Would It Cost to Achieve Some Fundamental Policy Objectives? ................. 43 D. Capital Expenditure on Education ............................................................................ 44 Trends and Composition ........................................................................................... 44 The Role o f Aid - Domestic vs . Extemally-Financed Capital Spending ................. 46 Utilization o f the Capital Budget .............................................................................. 47 . E Other Expenditure-Related Issues................................................................................. 49 Equity and the Link with Poverty - Who Benefits from Education Spending? .......49 Link with Decentralization ....................................................................................... 50 Private Provision o f General Education.................................................................... 52 F. Loolung Forward: Projected Expenditure Requirements for Primary and Secondary Education .............................................................................................................................. 57 3 . HIGHER EDUCATION ................................................................................................. 66 A. Technical & Vocational Education and Training ..................................................... 66 B. University Education ................................................................................................ 72 4 . PUBLIC HEALTH EXPENDITURE AND HEALTH IN ETHIOPIA ........... 81 A. Introduction............................................................................................................... 81 B . Context ...................................................................................................................... 81 C . Public Health System. Services and Outputs ............................................................ 85 D. Public spending and its Allocation ........................................................................... 87 E. Options and Projections .............................................................................................. 101 F. Financial Sources ........................................................................................................ 109 G. Conclusions ............................................................................................................. 112 5 . POPULATION AND FAMILY PLANNING ............................................................. 113 A . Background .The Population Challenge ................................................................ 113 B. Methods and Data ................................................................................................... 115 C. Findings .ProjectedRequirements and Costs........................................................ 117 D. Observations on Challenges and Constraints in Achieving the Objective of Providing Adequate Family Planning Services .................................................................. 122 . 6 FOOD SECURITY AND SAFETY NETS - PUBLIC SPENDING ON TRANSFERS TO THE POOR ..................................................................................................................... 127 A. Poverty and Safety Net Needs ................................................................................ 127 B. How MuchI s Ethiopia Spending on Safety Net Programs? - Current Public Expenditures ....................................................................................................................... 132 C. What Could Ethiopia Spend on A Safety Net Program? .Potential Public ExpenditureNeeds .............................................................................................................. 135 D. What Should Ethiopia Spend on A Safety Net Program?....................................... 140 7 . WATER SUPPLY AND SANITATION ..................................................................... 143 A. Introduction............................................................................................................. 143 B. Analysis of existing pattern of expenditure ............................................................ 145 C. Projections of expenditure requirements................................................................. 153 D. Observations on policy implications and trade.offs, and the link with a wider development strategy .......................................................................................................... 158 List o f Boxes & Tables BOXES: Box 1: Summary o f Three Scenarios Used to Project Costs o f Social Sector Policy ...................... 11 Box 2: Costing The MDGs Alone .................................................................................................. 19 Box 6.1: Three Broad Cost Scenarios for Safety N e t Spending................................................... 139 TABLES: Table 1.1: Estimated Aggregate 200 1/02 Social Sector Spending................................................... 4 Table 1.2: Selected Social Sector Outcomes - 1990-2000............................................................... 6 Table 1.3: The MillenniumDevelopment Goals in Ethiopia............................................................ 7 Table 2.1: Ethiopia - Evolution o f the Education System 1980-2002 (EY 1972-1994)................31 Table 2.2: Selected Indicators o f Educational Outputs and Outcomes........................................... 31 Table 2.3: Ethiopia - Education as a Share o f the Budget - Past 16 Years .................................... 32 Table 2.4: Total Education Spending US$ per Capita.................................................................... 32 Table 2.5: The Changing Composition o f Recurrent Education Expenditure ................................ 33 Table 2.6: Estimated Expenditure per Student at Different Levels o f Education (2001/02) .......... 34 Table 2.7: Comparative Recurrent Education Spending - .............................................................. 34 Table 2.8: Budgeted Expenditure on Education: Federal and Regional Levels ............................. 35 Table 2.9: General Education Recurrent Budget............................................................................. 36 Table 2.10: Recurrent Budget Per Student 1997/98-2001/02........................................................ 36 Table 2.1 1: Composition o f Recurrent spending on General Education ........................................ 37 Table 2.12: Levels o f Spending on Some Key Non-Wage Recurrent Items .................................. 38 Table 2.13 Regional Variation in General Education Spending (Selected Regions 2001/02) .......38 Table 2.14: Evolution o f Student: Teacher Ratios 1992/93- - 2001/02 .......................................... 41 Table 2.15: Comparative Primary Teacher Salaries - Selected Countries ..................................... 42 Table 2.16: Evolution o f Education o f Federal and Regional Capital Budgets............................... 45 Table 2.17: Budgeted Capital Expenditure on Education - Cumulative 1996/97-2000/01 ........... 46 Table 2.18: Budgeted Education Capital Expenditure By Level o f Government........................... 47 Table 2.19: Actual vs . Budgeted Education Capital Expenditure .................................................. 48 Table 2.20: Proportion o f Education Capital Budget Actually Spent - by Sub-sector ...................48 Table 2.21: Gross Enrollment Ratio by Consumption Quintile...................................................... 50 Table 2.22: Structure o f the Non-Government School Sector in Ethiopia (2001/02) ...................... 53 Table 2.23: Non-Government Schools and Enrollments - 2001102 (EY 1994) .............................. 53 Table 2.24: Average fees Charged b y Sample o f Non-Government Schools - December 2002 ....54 Table 2.25: Approximate Savings to Government Budget Due to Increased Shift to Private Education................................................................................................................................ 55 Table 2.26: Main Characteristics o f the Projection Scenarios ........................................................ 59 Table 2.27: Annual Recurrent Expenditure Requirements............................................................. 59 Table 2.28: Estimated Annual Capital Costs o f General Education Scenarios ............................... 60 Table 2.29: Composition o f Annual Costs o f ‘Extended PRSP’ Scenario 2015 ............................. 60 Table 2.30: Annual savings over 2015 Base Case Scenario........................................................... 62 Table 3.1: Recurrent Unit Cost for Selected Trades, 10+1 and lot-2 Programmes ......................... 67 Table 3.2: Projected Annual Technical and Vocational Education Costs ...................................... 69 Table 3.3: Evolution o f University Spending - Budgets 1997197-2002/03.................................... 72 Table 3.4: Recurrent University Spending by Main Institutions - 2001/02 Budget........................ 74 Table 3.5: Actual vs. Budgeted Expenditure - University Sector, 2000/01 (EY 1992) ...................75 Table 3.6: Results o f University Projections - Full-time Enrollments and Annual Costs............... 79 Table 4.1 : Selected Health Status Targets for Ethiopia ................................................................... 82 Table 4.2 : National Health Accounts Data for 2001 (WHO Estimates) ......................................... 83 Table 4.3: Public Spending on Health in Ethiopia, EFY1989-94 (1996/97-2001/02) .................... 88 Table 4.4: Execution of the Capital Budget.................................................................................... 90 Table 4.5: Adjusted odds-ratios for children under 5 years old, EFY1993 (2000) .......................... 91 Table 4.6: Composition o f Recurrent Budget, EFY 1989-1994..................................................... 92 Table 4.7: Funding for Drugs and Medical Supplies in Public Health Sector ................................ 98 Table 4.8: HIV/AIDS Programs. Future Scenarios ...................................................................... 101 Table 4.9: Selected Health Care Financing Targets ...................................................................... 102 Table 4.10: Selected Indicators for Expenditure Projection Scenarios.......................................... 105 Table 4.11 : Public Health Spending under different scenarios ..................................................... 105 Table 5.1 : Projected Total Fertility and Contraceptive Prevalence Rates Under Different Scenarios Ethiopia, 2000-2020 ............................................................................................................ 117 Table 5.2: Contraceptive Usage Among All Currently Married Women by ................................ 118 Table 5.3 : Contraceptive Commodity Requirements Under Different Scenarios ........................ 120 Table 5.4: Cost o f Contraceptive Commodities, According to Three Different Projection Scenarios, 2000 - 2020 ........................................................................................................ 121 Table 5.5: Rough Estimate o f Total Costs o f Family Planning Services According to Different . . Scenarios of Projection......................................................................................................... 122 Table 6.1 : Estimated Levels o f Safety Net Spending ................................................................... 132 Table 6.2: Some Estimates o f Transfer Costs - Selected Programs .............................................. 134 Table 6.3: Approximate Unit Costs o f Large-Scale Transfer Programs ........................................ 137 Table 6.4: Estimated Costs o f Safety Net Scenarios .................................................................... 138 Table 7.1 : External Financing InBudget for Water Supply EFY 1995-97 ................................... 148 Table 7.2: Regional Government Spending on Water Supply - Four Largest Regions ................ 148 Table 7.3: Estimated Cost Per Capita For Different Types O f Urban Water Supply Systems..... 151 Table 7.4: Sample Costs o f Town Water Supply Systems ........................................................... 151 Table 7.5: Estimated Unit costs o f Rural Water Supplies ............................................................ 152 Table 7.6: Actual vs. Planned Water Supply Spending ................................................................ 152 Table 7.7: Projected Investment Needs for Water Supply - ‘MDG Plus ’ Scenario .................. 155 Table 7.8: Sanitation Coverage in Addis Ababa .......................................................................... 155 Table 7.9: Projected Investment Needs for Water Supply- Moderate ‘ MDG’ Scenario ........... 156 Table 7.10: Projected Investment Needs for Water Supply - L o w Scenario ............................... 157 Table 7.1 1: Expected Reduction in morbidity/mortality from improved water and sanitation .... 161 FIGURES Figure 1.1: Selected indicators o f Benefit-Incidence o f Social Services in Ethiopia ....................... 8 Figure 1.2: Projected Social Sector Expenditure Needs. 2005-2020 .............................................. 14 Figure 1.3: Recurrent and Capital Breakdown o f Social Program Costs under ‘Extended PRSP’ Scenario. and Likely Resources Envelope ............................................................................. 17 Figure 1.4: Federal-Regional Breakdown o f Funding Requirements under ‘Extended PRSP’ Scenario, and Likely Resource Envelope ............................................................................... 18 Figure 2.1: Composition o f Education Recurrent Budget (2001/02 EFY 1994) ............................ 33 Figure 2.2: Federal -Regional Breakdown o f Education Budget - 2003/04 ................................... 35 Figure 2.3: Evolution o f Recurrent Spending on General Education ............................................. 36 Figure 2.4: Composition o f Recurrent Spending on General Education (2001/02 Budget) ........... 37 Figure 2.5: Average Maintenance Budget - Past 5 Years .............................................................. 39 Figure 2.6: Federal and Regional Capital Budgets for Education 2001-2003/04 ........................... 45 Figure 2.7: Composition o f Capital Spending in Education 1996/97 - 1999/00 ............................. 46 Figure 2.8: School-Aged Population Profile 2000 and 2020 (millions) .......................................... 57 Figure 3.1 : Composition o f University Recurrent Budgets- 2002/03 .............................................. 73 Figure 4.1: Composition o f Health Spending - EFY1993 (2000/01) .............................................. 84 Figure 4.2: Public Health Expenditure (percent GDP), EFY1983-1994 ........................................ 88 Figure 4.3: Public Health Expenditure (US $ Per Capita), ETY1983-1994 .................................... 88 Figure 4.4: Government Spending as Share o f GDP ( percent), 1993............................................. 89 Figure 4.5: Budget Execution (Share o f Health Budget Spent: EFY) ............................................. 90 Figure 4.6: Population per Nurse and Health Assistant by Region, EFY1993 ................................ 93 Figure 4.7: Annual Change in Human Resources, Facilities and Population by Region, EY89-94 (19972002) ............................................................................................................................. 94 Figure 4.8: Composition o f Ethiopian Government Health Expenditure, EFY 1993 ...................... 95 Figure 4.9: Composition o f Ethiopian Government Recurrent Health Expenditure, 1993............. 95 Figure 4.10: Composition o f Regional Recurrent Expenditures...................................................... 96 Figure 6.1: Per Capita Food Production 1970-2000...................................................................... 128 Figure 6.2: Approximate Distribution o f Funding o f Transfer Programs 1996-2000 .................. 133 Figure 6.3: Projected Number o f Persons Below the Food Poverty Line ..................................... 136 ACKNOWLEDGEMENTS This report represents the collaborative efforts o f a number o f individuals and donors. The main team consisted o f W. James Smith (team leader and principal author, World Bank); William Savedoff (WHO) wrote the chapter on health expenditures; Dieter Orlowslu (financed by Gtz) undertook much o f the analysis o f expenditure trends and developments underpinning Chapter 8, with substantial data assistance from Mesfin Girma Bezawagaw (World Bank). Jutta Franz wrote the background paper on Technical and Vocational Education and Training, financed by Gtz. Ferry Philipsen and Abebaw Tiruneh undertook a background study on unit costs in education, financed by Irish Aid. Rafiqul Chaudhury o f U N F P A wrote the background paper on public spending on population which i s the core o f Chapter 5, in association with Benoit Kalasa (UNFPA). Maude Svensson (World Bank) prepared background papers on foreign aid flows and HIV/AIDS spending. Catherine Revels (World Bank) wrote most o f Chapter 7, o n water supply. Michael Latham and Laroque undertook a background study o n private provision o f education in Ethiopia, financed by DfID. Kalanidhi Subbarao and W. James Smith (World Bank) prepared the background paper on safety nets and food security issues, drawing on preliminary work by Tassew Woldehanna. Finally, Jemal Mohammed (World Bank) provided extremely useful guidance and liaison with the government in Addis throughout. The Public Expenditure Review was conducted under the guidance o f a Government PER Steering Committee including At0 Demelash Alem, At0 Melaku Kifle, A t o Getachew Adem, At0 Berhanu Legessie, Ato Teferi Demeke, At0 Sytotaw Yemer (MoE), and Dr. Girma Azene (MoH) and under the general auspices o f the State Minister H.E. At0 Getachew Gebre in the Ministry o f Finance and Economic Development. Valuable inputs and comments were provided by Jee-Peng Tan, Nazmul Chaudhury, Gary Theisen, Trina Haque, Agnes Soucat, Julie McLaughlin, Shekhar Bonu, Son N a m Nguyen, William Saint, Christine Pena, Grebreselassie Okubagzhi, Anwar Bach-Baouab, Getahun Gebru, and Andrew Dabalen o f the World Bank. Sibel Kulaksiz helped with analysis and presentation o f the results. Saba TeMe, Momar Gueye, Roselynd Goudiaby, Patrick Mamboleo, and Arlette Sourou painstakingly processed the drafts, under the able guidance o f Tanisha McGill. Fredrick Kilby and Robert Blake served as Sector Managers during the PER period, and Ishac Diwan was the Country Director. Peer Review was provided by Allister Moon, K e i t h Hinchliffe, Vinaya Swaroop, and Trina Haque. I EXECUTIVE SUMMARY PART I:SOCIAL SECTOR SPENDING - THEEMERGINGCHALLENGE Ethiopia faces challenges that are almost unique: human development needs are arguably the greatest o f any country in the world, while at the same time the capacity to finance those needs i s probably lower than anywhere else. T h i s fundamental contradiction conditions much o f the approach to questions o f social sector spending and policy choices. The expenditure requirements confronting the govemment, to provide even minimum levels o f health and education services, are enormous. This i s partly because the current human development status o f the population i s so poor: 59% o f children are malnourished to some degree, the average Ethiopian has less than 1 year o f schooling, women face a 5% probability o f dying in childbirth, and at least 5-6 million people are so chronically short o f food that they face potential starvation each year, and only survive on the basis o f massive inflows o f food aid. Ethiopia i s also starting from an exceptionally l o w base in terms o f the coverage o f existing services. The secondary school system only has capacity for about 8% o f high- school-aged children; only 18% o f the family planning needs o f married women are being met; and water supply coverage, at 24%, i s the lowest in sub-Saharan Africa. As a consequence there i s a massive unserved population: over 12 million children are not in school, 1.9 million additional children need to be immunized each year, and 5 1 million people are estimated to be without access to clean water.’ However the capacity to finance these services i s far lower than elsewhere. At US$lOO, Ethiopia has a per capita income half that o f most o f the worlds’ very poorest countries, and a correspondingly lower capacity to generate revenue domestically. For example govemment revenue i s equivalent to about $19 per capita, compared to about $40 in Malawi. At the same time Ethiopia receives proportionally less foreign aid than other very poor countries. (In 2001, the last year for which comparable numbers are available, Ethiopia received official development assistance amounting to $16 per capita, compared to about $33 per capita in Burkina Faso, Rwanda, or Ghana).3 The expenditure requirements confronting Ethiopia are compounded by rapid population growth. The number o f people requiring basic services has increased by 25 million over the past two decades, straining coverage and quality. Massive further increases will take place over the coming two decades. The population i s currently growing at 2 million persons per year, and by 2020 i s projected to reach 106 million, adding a further 40 million people for whom health care, education, adequate food, and water supply will be ’ Based o n Census and D H S data, maternal mortality o f 1,800/100,000, and fertility o f 5.9. 2 Based o n net enrolment rates o f 52.2% (primary) and 7.4% (secondary); effective health facility coverage o f 62% (MoH); DPT3 coverage o f 47% (HSDPI Review), and water supply coverage o f 24% (WHO). African Development Indicators, 2003; World Bank, excluding emergency relief. i needed. Mounting a wide-reaching family planning program can have a major impact, and i s one o f the most important things that government can do. Within this overall context, the government and donors are rightly pursuing an agenda o f aggressively expanding social service coverage. However no-one has sat down to add up the costs o f the strategies being proposed, and to assess how they can be made to fit within a realistic fiscal framework. The expenditure consequences are significant, because social programs are intensive in their demands o n recurrent resources, so the expansions being committed to today could overwhelm budget obligations in 20 years time, crowding out the capacity to fund other investments for growth. Furthermore, because o f the very limited capacity to finance these needs domestically, they will heavily influence foreign aid requirements over the next two decades. Finally, because the cost o f these programs fall primarily on the Regional and local governments, they imply the need for a major increase in the share o f resources transferred to lower level governments. The object o f this Public Expenditure Review i s to put the numbers on the table to inform this debate. Current spending on education, health, population, food security transfers, and water supply i s examined, along with an assessment o f i t s adequacy, effectiveness, and absorptive capacity. The PER then projects the costs o f alternative coverage targets over the next 20 years, including the costs o f reaching the Millennium Development Goals (MDGs), and examines the implications for financing and sectoral policy choices. Three cost scenarios are forecast for each sector: a ‘business-as-usual’ scenario that shows the cost o f just keeping up with population growth; an ‘Extended PRSP’ scenario, that projects the costs o f continuing with moderately ambitious targets developed over the past few years; and finally an ‘MDGs Plus’ scenario that forecasts the full costs o f implementing the most ambitious plans that government i s currently proposing in various policy documents. Sector Expenditure Reviews and Projections Approximate Annual Share o f Share of US$ per Spending (ETB Budget GDP capita millions) Education 2,316 13.9% 4.5% $4.15 Health (incl. Population and 1,053 4.8%+ 1.6% $1.78 AIDS) Water Supply 456 2.7% 0.9% $0.77 Social Safety Nets 2,280 n.a. 4.5% $3.87 Total public spending in Ethiopia has increased substantially in recent years, from 24% o f GDP in 1997/98 to 35% today. The bulk of these increases have been allocated to education, road building, and agricultural services. Spending on health, population, and Figures are for 2001102, the last year for which reliable actual expenditure data were available. 11 water supply have remained consistently lower - both in terms o f shares o f expenditure, and relative to needs. However this appears broadly rational, given the effectiveness and absorptive capacity constraints discussed subsequently. Judging the impact o f spending on outcomes i s difficult, because past spending has taken place in a context o f almost continuous crises - revolution, civil war, and external conflict for much o f the past 30 years, combined with severe droughts, have resulted in very l o w economic growth and conditions under which securing improvements in human development indicators has been difficult. Nonetheless evidence suggests some moderate progress in intermediate outputs (most notably in school enrollments, which have doubled) - and to a lesser extent in outcomes: literacy increased from about 30% in 1990 to 40% in 2000; and under-5 mortality declined from 190 to 167 per thousand. Nonetheless, improvements at this rate will not be adequate to bring Ethiopia up to the levels o f human development to which it aspires, and outcomes in some areas remain particularly bad: for example maternal mortality i s among the worst in the world, with complications o f pregnancy and child birth accounting for 25% o f all adult female deaths. Given the difficult circumstances o f the past two decades, the progress that has been made i s in fact notable, suggesting that public spending has had an impact, and if conditions were better would likely have had more. Furthermore, given that most o f the expansion in education and health spending has been quite recent, we should expect to see more substantial results in terms o f outcomes in the next 5 to 10 years. Government’s plans for the immediate future are laid out in the 2001 Sustainable Development and Poverty Reduction Program (SDPRP), which involves ambitious plans in all the social sectors through 2004/05 - not all o f which are affordable. Given the importance o f long-run recurrent budget implications, and the impact o f population growth, this PER generally focuses on a longer-term horizon - over the next 10-20 years - but with outcomes that are consistent with the shorter-term SDPRP . The Millenium Development Goals will be difficult to achieve, except perhaps in education. The total cost o f the inputs needed to reach the MDGs, in terms o f social sector budgets alone, would be very roughly ETB 12 billion (US$1.4 billion) annually, or about four times recent annual spending o n basic education, health, and water supply. The beneJits o f social sector spending are fairly evenly spread in Ethiopia, across geographic areas and across income classes - at least compared to many countries - although the poorest 20% tend to be l e f t behind, partly because they live in the least accessible areas, while urban dwellers tend to benefit disportionately. Water supply spending in particular has been heavily biased towards urban areas. More work i s needed using household data to better understand the gender balance in spending: in education about 40% o f students are female at both the primary and secondary levels. ... 111 Education Public Expenditure 0 Total education sector spending in 2001/02 was ETB 2.9 billion (US$ 333 million), or 14% o f total public expenditure.’ Two-thirds i s spent o n primary and secondary schooling, and the balance o n technical and university education. 0 Budgets increased by over 50% in real terms in the past five years, reflecting a conscious effort to redirect resources and expand access to education. 0 Spending on education as a share o f GDP (4.5%) i s now comparable to other low-income countries, but in absolute terms, at about US$3.50 equivalent per capita, recurrent spending i s below that in most other very-low-income countries. 0 The composition o f education spending appears broadly appropriate. There has been a shift in the last few years towards higher education, and an increase in the proportion spent by federal (as opposed to regional) governments - an outcome o f the government’s push to build higher-level capacity. This balance should remain a topic for debate; but the over-riding issue i s the insufficiency o f spending at all levels. General (Primary and Secondary) Education Spending 0 Primary school enrollments almost doubled, from 4.5 million students in 1996/97 to 8.1 million in 200 1/02, consistent with government’s education-for-all policy. Gross enrollment i s currently about 62%. The expansion has placed tremendous strains on the system, and quality has suffered. 0 Total recurrent spending on general education in 2001/02 was ETB 1.4 billion ($ 159 million); about 67 % o f it at the primary level and 33% at secondary level.6 0 Despite increases in education budgets, recurrent spending per student had declined by about 20% over the past five years, as the result o f massive enrollment increases. 0 Educational outcomes are difficult to assess in the absence o f standardized testing. Intermediate indicators show improvement in getting students enrolled in school, but less success in keeping them there. High student-teacher ratios and drop out rates are limiting the effectiveness o f money that i s currently being spent. 0 The average number o f children in a class has deteriorated to 73:l at the primary level, and 8O:l in secondary school. These averages mask large variations, and it i s not uncommon to come across 120 students or more in a single class. I t i s Throughout the analysis draws o n 2001102 expenditures, the last year for which reliable detailed spending data were available. Budgets for both health and education have been increased since, but not all the budgeted increases have been spent, and the broad policy findings are not affected. 6 Note however it i s difficult to reliably separate primary and secondary spending o n salaries, because teachers are classified by qualification rather than by the level they actually teach at. iv almost impossible for effective teaching or learning to take place under these circumstances. 0 Salaries consume 93% o f recurrent spending’, leaving little for non-wage operating costs. Some items are dramatically under-funded - for example budgets for building maintenance have declined over the past five years, while the stock o f schools has increased. Textbook availability, while improved, i s s t i l l a major problem, with an average o f only about one textbook to every three children. The costs o f correcting some o f these problems are not huge. T o provide minimum non-wage recurrent inputs i s estimated to cost ETB 200 million p.a. ($23 million); to provide a full set o f textbooks at the primary level i s estimated to cost ETB 178 million ($21 m.) p.a. 0 The largest problem in a fiscal sense i s the deficit o f teachers’: to restore the student-teacher ratio to a reasonable level, even without allowing for any growth, would cost an additional ETB 333 million ($39 m.) annually, requiring an immediate 33% increase in current budgets. Relatively small amounts o f non-wage recurrent funding, if predictable and regular, could be extremely useful. The current education strategy proposes such grants to schools, but the Regions lack the resources to finance them. In the absence o f good survey data it i s difficult to judge the efficiency o f spending. Service delivery systems seem to work fairly well - for example teacher absenteeism does not appear to be the problem that it i s in many countries - but obviously the deficit o f textbooks, classrooms and above all, teachers, limit the effectiveness o f spending. Efficiency i s reduced by high drop out rates: 29% o f students drop out in grade 1, and over 55% leave before the end o f grade 3; for these children little effective education i s taking place. 0 Expenditure Projections. Chapter 2 shows what would be required to meet low, moderate, and ambitious targets for coverage o f the system, and to restore quality. 0 Under the baseline ‘extended PRSP’ scenario, to achieve close to universal primary enrollment by 2015, reasonable improvements in quality, and a modest increase in secondary enrollments (to 23% at the junior secondary level, and 14 % at the senior level) would require a four-fold increase in spending in real terms, to ETB 6.9 billion in 2015 (US$743 million annually, compared to a level o f about US$232 million today). This would represent about 21% o f total public ’ 97% at the primary level, and 81% at the secondary level - although there are some data problems with the classification of teachers between primary and secondary, making it difficult to be precise by level. 8 The deficit i s estimated at 40,000 today, even without allowing for increasing coverage or population growth V spending. The final section o f this Executive Summary presents cumulative cost requirements for the social sectors and compares them with resource availability. 0 The problem i s not just that Ethiopia i s starting from far behind in terms o f coverage, and that quality needs to be restored following the deterioration o f the recent past, but that there will be massive growth in the school-aged population in the coming two decades - by over 10 million children. O f the required expenditure increases almost a third i s accounted for by population growth. 0 The greatest share o f the projected costs are associated with expanding enrollments at the grade 5-8 level, because current enrollments are relatively l o w (30%), and because unit costs are higher at this level. 0 There are a number o f ways o f reducing costs: (i) using cheaper construction techniques, ( ii)greater reliance on lower-paid, community-employed teachers, especially at the lower grade levels; and ( iii) more private sector provision. The policy choice with the greatest savings impact would be to establish grade 5 or 6 level as the terminal level o f education for much o f the population in the medium-term. While each o f these measures helps, cumulatively they do not outweigh the overwhelming need for spending increases implied by the sheer numbers involved: the number o f teachers needed by 2015 for example, under the baseline scenario, i s 424,000, compared to 141,000 today. 0 The objective under the education MDG i s to achieve 100% primary school completion by 2015. While it will be increasingly hard to amact and retain marginal students, with a concerted effort it should be possible to come close to the goal. The public expenditure requirements for retaining children in school - better student-teacher ratios, adequate materials, and decent classrooms - are the same as those projected above for system expansion and restoration o f quality. 0 Finally, we perceive a risk, moving forward, that necessary spending on education may be squeezed even more unless it i s explicitly taken into account in adjusting transfers to Regions and woredas. The pressing need to re-establish effective student-teacher ratios has massive financing requirements, and cannot be accommodated within current transfers or block grant^.^ 0 Secondury Education Spending. Ethiopia spends relatively little o n secondary education (20% o f total recurrent education spending in 2001/02)'0, primarily due to the l o w coverage ratio. With the attention paid to primary education in recent years by donors and government, the emerging crisis in secondary education has not been confronted. 0 Pressure for the expansion o f secondary spending in the coming two decades i s going to be immense, due to three factors: ( i)the demographic bulge; (ii) the In Tigray, for example, an extra ETB 75-90 million p.a. i s estimated to be needed to improve student-teacher ratios today, but the total amount o f block grants to woredas, for all functions, i s only i s only ETB 267 million. lo As noted earlier there i s a problem classifying salaries accurately - as a result o f which the true share i s likely higher. vi much larger proportion o f children now passing through primary school; and (iii) the very l o w initial conditions (with a 7.4 % net enrollment ratio, 92.6% of secondary-aged children are not currently in school). T o reach a secondary enrollment rate o f only 25% in the next 15 years would require a real increase o f about five-fold in spending from current levels. W h i l e this may seem inconsistent with the policy emphasis on primary education, it needs to be recognized that no country has made the kind o f development transformation that Ethiopia needs to make with only 3 % o f it’s population having secondary education. However while an increase in skilled people i s undoubtedly required, for the majority o f Ethiopians who will remain subsistence farmers for the foreseeable future, between 5 and 8 years o f education i s probably the most that i s required. Given the high costs o f secondary expansion, solutions l i e in: (i) judicious choice o f exit points (the current system o f 8, 10, 12 i s not bad - as noted above, a grade 5 or 6 exit point may also make sense) and (ii) some greater reliance o n private financing for secondary education - probably mostly through private provision for the better off. Secondary school fees are an option but we have serious concerns regarding access and equity implications. Increased spending on primary, secondary, and even university education should not be seen as competing with each other - all are needed. Role o f the Private Sector in General Education. The private school sector i s very small - accounting for only 2.4% o f secondary school enrollments and 5.3% o f primary enrollments (and o f these the majority are NGO, faith-based, or ‘public’ schools - the number o f truly private schools i s very small indeed.). Government’s policy framework i s open to private provision, but there are limits on effective demand due to l o w incomes. Simulations show that private schooling, combined with some cost-recovery in secondary schools, could reduce total public budgetary requirements for secondary education by about 10-20% over the next two decades. There i s growing community funding o f teachers, and contributions to school construction. Data i s weak, but the scope i s probably greatest at the lower primary levels. At the grade 5-8 levels, where the bulk o f h t u r e costs lie, the scope for hiring unqualified teachers locally will be less. Public Expenditure on Technical and Vocational Education and Training (TVET) Current public spending o n TVET (training in skills such as tailoring, hairdressing, mechanics, and basic accountancy) i s about ETB 92 m i l l i o n ($12 million p.a.), triple the level o f just two years ago. In addition government i s engaged in an ETB 1.8 billion ($210 million) investment program over 3 years, mostly funded out o f its own resources. vii The unit costs o f TVE are almost 5 times higher than regular secondary education (ETB 2,252 per student compared to ETB 500 per student per year) - due to the high cost o f consumables and equipment, and the lower student- teacher ratios required. A major expansion o f in-school technical training i s underway - including building new institutions, and converting some existing secondary schools (and teachers) to TVE - the objective i s to increase capacity to 130,000 students by 2004105, compared to about 38,000 in 2001102. While the government 's concern with adding technical capacity and improving the employability o f youth i s entirely legitimate, there i s substantial evidence from cross-country experience that in-school TVE and government training institutions are often not cost-effective (usually because the costs are so high, and the relevance and quality o f the training are not suited to the needs o f employers). Alternatives include equipping labour market entrants instead with good, practical general education; greater reliance on on-the-job and employer training; alternate financing mechanisms; and lower cost models o f public provision than those currently being used. The projected costs o f the stated policy o f absorbing 40% o f grade 10 leavers i s estimated at about ETB 1.1 billion annually ($123 million) by 2015, larger than the total primary education budget in 2001102. Even a l o w scenario, with substantially slower approach to the current enrollment target o f 130,000 would s t i l l require spending o f about ETB 488 million ($57 m.) annually by 2015. e While it i s impossible to assess the effectiveness o f Ethiopia's system yet, there i s reason for concern, and careful tracer studies need to be initiated. Commitments to a major expansion should be slowed down until the effectiveness o f the training provided can be confirmed. University Public Spending The university system i s extremely small by international standards, with an enrollment o f 35,000 full-time students, representing less than 0.8% o f the university-aged population. Total budget for the universities in 2002103 was ETB 681 million (US$79 million), representing 3.4% o f public expenditure, and 19% o f education spending. Government i s engaged in a major push to double intake capacity in the next three years. A significant building program i s underway, mostly financed from domestic sources. The 2002103 (EY95) budget represents a 73% increase over the previous year, mostly accounted for by capital investments, but recurrent spending per student has also increased. e The review did not detect major distortions in university spending. 15% o f the budget i s absorbed by food and student services (significant, but less than in ... Vlll some other countries); the authorities are committed to increasing cost recovery in this area, but to do so will take time. 0 Projections show the ‘Extended PRSP’ scenario (based on current plans to double enrollment over the next 5 years, followed by growth thereafter that would result in an enrollment rate o f about 2% o f the university-aged population) 11 would cost about ETB 1.3 billion ($153 million) p.a. by 2015, compared to spending o f about 700 million ($80 m.) today. The more ambitious ‘MDG Plus’ scenario i s based on reaching enrollment o f 3% o f the university-aged population by 2020 would cost about ETB 2.3 billion ($267 million) p.a. 0 Undoubtedly the expansion o f university education i s justified given Ethiopia’s development and capacity needs. However the planned expansion will be expensive, and it i s important to examine ways o f doing it at lower cost to the budget. 0 Given the substantial private returns to higher education, there i s a case for greater cost-recovery. The government i s currently considering a ‘graduate tax’. Simulations suggest that it will not yield as much revenue as the government has projected, and it i s recommended that consideration also be given to introduction o f tuition fees or measures to shift the cost o f food and accommodation to students. Such proposals are politically sensitive, they would need to be phased in over time, and accompanied by a program o f waivers and scholarships for the poor. Health Public Expenditure 0 Health expenditure in Ethiopia, both public and private, i s very low. However, the obstacles to effectively delivering health services are large -- the population i s largely illiterate and dispersed in rural areas with difficult access, fertility i s high, and sanitation i s limited. Nevertheless the leading causes o f mortality and morbidity are related to infectious diseases, most o f which could be substantially reduced with cost-effective public interventions. 0 Health services are severely limited. Ethiopia has one o f the world’s highest ratios o f population per doctor and per nurse; and there are only 384 health centres serving a population o f over 67 million. 0 Coverage has increased in the past 5 years; the proportion o f the population within 10 km. o f a health facility rose from 52% to 61%; but utilization remains low, and lower-tier facilities are often by-passed because they lack staff or drugs. 0 The decline in immunization coverage i s a cause for concern, given that immunization i s one o f the most cost-effective health interventions available. The highest estimates show current coverage o f DPT3 o f 51%, well below the rate o f 67% estimated to have been reached in EY 1989 (1996/97). “ Equivalent to absorbing about a quarter o f grade 12 graduates. ix Current Health Spending. Over the past decade health spending increased from 0.9% to 1.5% o f GDP; but the bulk o f this increase took place almost 10 years ago, and recurrent per capita spending has been fairly stagnant since. Subsequent increases have taken place mostly on the capital budget; they have been volatile from year to year, and substantially under-spent in some years. On average, total spending (including capital) increased only f i o m about US$1.20 per capita in 1996/97 to $1.40 in 2001/02.'2 Ethiopia spends a lower than average share o f the budget on health (5%, compared to, for example, 10% in Tanzania). The fact that recurrent spending has remained constant in real per capita terms while capital spending increased lends support to concems that facilities may be being built that cannot be properly staffed and operated, l3 and also suggests there are operational constraints to raising recurrent spending. According to MoFED data, execution o f the budget was only 80% in 2000/01, mostly due to lags in capital spending. Problems in implementing investment plans, and difficulties in utilizing donor funds are the main reasons for l o w exec~tion.'~ Public spending i s somewhat biased towards hospitals and urban services, although the extent o f the bias i s not as extreme as in some other countries. In 2001/02 38% o f recurrent expenditure took place at the hospital level, and 30% at the health post and centre level". The composition o f recurrent spending i s reasonable: approximately 61% o f the recurrent budget pays for salaries, while 26% goes to medical supplies. Regional differences are quite large; and the share o f operation and maintenance expenditures i s l o w in most o f the regions. Availability o f essential drugs has improved recently, but shortages are s t i l l common in public facilities. Funding for drugs has increased five-fold from 1996/97 (EFY89) t o 2001/02 (EFY94); a substantial part o f the increase came from the government's own budget. There i s a concem with capital bias. The historical balance o f 40% capital spending and 60% recurrent i s high in a sector where outcomes depend mostly o n the adequacy o f routine service delivery. While there is justification for capital Health budgets have been increased in the past two years, but indications are there has been trouble spending the money. l3 Although note that as in other sectors there are some data problems in assessing expenditure composition - donor finds tend to be concentrated in the capital budget, even though they may finance recurrent inputs (such as drugs); and increases in government recurrent spending may have been intentionally curtailed in some cases because recurrent inputs were being provided off- budget by donors (again, drugs i s a good example) l4Note however there i s also a data problem, as aid-financed expenditures are over-estimated in the budget, and under-reported in execution. l5 Although note that in Ethiopia District hospitals provide a fairly basic level o f care. X investment given the l o w levels o f coverage, annual reviews o f the health program have consistently raised the issue o f under-utilized, under-staffed, and un-equipped facilities at the same time that an aggressive program o f constructing new facilities has been pursued. 0 Evidence on who benefits from public health spending i s mixed: Welfare Monitoring System data suggest that the poor use public facilities as much, or more than the non-poor, possibly because they use rural facilities which are bypassed by the better-off. The Demographic and Health Survey however shows that the poor are much less likely than the better-off to benefit from immunization and other specific public health interventions. 0 Increasing the effectiveness o f spending depends o n improving staffing. Current plans for a major expansion o f health workers would put Ethiopia on a path to achieving coverage targets in terms o f aggregate supply. The issue however i s not just (or mostly) the supply o f staff, but their utilization, posting, management, supervision and incentives. 0 In the short run, as long as the public sector cannot provide the conditions necessary to retain highly skilled health professionals, the emphasis should be on hiring and gradually upgrading lower-skilled community outreach workers, and paramedic staff such as midwives and nurses. 0 Donorflnancing. There i s uncertainty regarding aid levels, but it i s estimated about U S $ 100 million equivalent16 was programmed to be spent by donors o n health-related activities in 2001/02, o f which only 10-15% was being channeled through the budget. 0 The government i s concemed that aid agencies continue to pursue activities that do not match the agreed-upon health sector program; and donors are concerned that weak information systems makes it hard to measure execution and impact.17 Donors have strongly indicated that if they could demonstrate effective use o f health sector funds, more would be forthcoming. The Projected Costs o f Reaching Policy Targets 0 'Extended PRSP' Projection. T o reach the targets govemment has set for 2005 and continue expansion at the same pace through 20 15, would require spending o f about ETB 4.3 billion (US$500 million) p.a. by 2015, or something like 4-5% o f GDP (compared to 1.6% in 2001/02). This i s high for public spending on health in Africa, but represents only US$5.41 per capita - well below the l6 This i s a high estimate, and probably includes some non-health elements associated with relief assistance. l7And a concern that the decentralized woreda spending system may make it more difficult to track use o f funds or associate them with outcomes. xi estimated cost o f an essential package o f care for developing countries (about $12 per capita)." 0 An alternative higher 'MDG Plus' projection - based o n reaching policy targets more quickly, and with a major shift in emphasis to outreach services, i s estimated to cost about ETB 9.3 billion (US$ 1.1 billion) p.a. by 2015 - more than 10 times current health spending in real terms, and about 9% o f projected GDP. Delivering such a package o f care would cost $11.50 per person; while reasonable by international standards, this i s likely unaffordable in the case o f Ethiopia, even with robust growth. 0 Given the Ethiopian government's extremely limited resources -- both financial and institutional -- the argument for orienting i t s approach toward community outreach, and concentrating on a few high priority illnesses with cost-effective interventions i s very strong. The four priorities for health services that emerge from such considerations would be: vaccine-preventable diseases, malaria, HIVIAIDS, and reproductive health. 0 As in other countries, some o f the most significant determinants o f health status lie outside the realm o f the health service (eg. hygiene, water supply, food consumption). Interventions in these areas, including accelerating growth and female education - and measures to reduce fertility and improve infant nutrition - will have as great, or greater effect on health status as anything that can be done through the health service. 0 However in Ethiopia there are such fundamental health interventions that are inadequately funded: the delivery o f immunization and malaria programs, and the staffing and equipping o f basic outreach services, that we believe there i s also justification for expanding health service funding. However this increased funding needs to be focused on these priority activities. This in turn has implications for building a more binding consensus between government and the donor community o n the balance o f spending on different interventions, and between capital and recurrent spending (especially if shifting to more untied budget support). 0 Finally, any expansion in spending needs to be accompanied by measures to strengthen management, manpower allocation and incentives, and assure full capacity utilization. 0 Private expenditure and scope for private jinancing. The Welfare Monitoring Survey (WMS) and WHO data suggest private spending on health i s very l o w (about $2 per capita), consistent with very limited cash incomes. 0 There i s some scope for increased cost recovery and encouraging private provision at the curative level among better-off urban consumers. '' This scenario would involve potential coverage o f about 85% o f the population (compared to about 62% today); but effective coverage would remain much lower, due to gaps in quality, staffing, and recurrent inputs. xii 0 Government’s policy environment i s open to private providers, and there has been substantial growth o f private provision in urban areas. To have a significant impact on public expenditure requirements however would require a supply- response by the private sector in rural areas.lg The purchasing power o f rural households i s so much lower than in urban areas that private providers are likely to continue to focus their attention o n the larger effective demand in urban areas. 0 While private financing will help relieve the public system o f catering to some high-end beneficiaries, a strategy that relies predominantly o n private market responses i s not likely to generate substantial improvements in the availability o f basic health services for the majority o f the Ethiopian population over the next 10, or even 20 years 0 Fees can play a role in the public health system, particularly ‘special pharmacy’ revolving funds. Although such fees can help assure maintenance and drug supplies, the government should probably not expect to mobilize more than a relatively small proportion o f total recurrent costs. Also note the risks o f excluding the poor are significant - even for drugs, and even with exemptions. Population and Family Planning Expenditures 0 Much o f the present crisis in social sector spending i s due to rapid expansion o f the population. Furthermore, expected population growth over the coming two decades accounts for about half o f the massive increase in spending o n social services required to approach the MDGs. 0 One o f the most effective things the government can do to reduce the long-term burden o n social sector budgets i s to reduce the rate o f population growth. N o t only will this result in substantial fiscal savings, but it will also improve the effectiveness o f services, by reducing over-crowding, as well as appreciably increasing the poverty-reducing impact o f growth, and improving food security and maternal and child health outcomes. 0 Because o f the momentum effect o f past population growth, it will take time to have a major effect o n absolute levels o f population, but a start has to be made now in order to have a long-term impact. 0 Unmet demand for contraception i s significant: The DHS found that desired fertility levels are at least one child less per couple than current levels; and that only 18% o f the contraceptive needs o f currently married women are being met. 0 W i t h recent technical developments, it i s possible to have a major impact at relatively l o w cost, without awaiting major complementary service delivery improvements. Depo-Provera provides reliable and safe three month contraception with a simple injection.” Depo i s already the most popular form o f l9 Private providers do offer some services to the rural population, but overwhelmingly related to selling medicines without prescription (with the associated risks o f misuse and resistance). 20 At a cost o f about 86 cents US, or 7 birr, per dose. xiii contraception in Ethiopia, and there i s growing demand, especially among younger women. 0 Experience elsewhere, for example in Bangladesh, has shown that contraceptive prevalence - currently only about 15% in Ethiopia - can be raised to the 30-40% level through supply-side interventions alone, especially in the presence of significant unmet demand. Ethiopian experience confirms substantial potential: fertility has already dropped to 3.3 in urban areas as a whole, and to a remarkable 1.9 in Addis Ababa (DHS). 0 The costs are not prohibitive. The projections show the cost o f a full-scale family planning program o f approximately $25 million p.a. (ETB 215 million) by 2010, rising to $49 million p.a. by 2020. Such a program would raise contraceptive prevalence to about 35% by 2015, reducing fertility to 3.7 (from a current level o f about 5.9 births per woman). 0 Effective family planning also requires a cadre o f female outreach workers to support and advise women clients. I t makes sense to initially focus efforts in areas with the greatest concentration o f population and un-met demand, to maximize cost-effectiveness, but this should be followed quickly by expanded services to less accessible rural areas, where the largest share o f high-fertility women will remain. Public Spending on HIV/AIDS 0 HIV/AIDS prevalence i s estimated at 6.6%, with about 2.3 million people currently infected, and some 1.2 children have already lost one or both parents. High prevalence among 15-24 year olds suggests Ethiopia may be on the cusp o f a more rapid spread o f HIV/AIDS in the next 5 to 10 years unless significant action i s taken. 0 The budget for the HIV/AIDS program was estimated at $37 million equivalent in 2001/02 (EY95), most o f it externally funded. Disbursement o f funds through official channels has been slow, and an increasing amount i s being spent by NGOs and donors without being channeled through HAPCO (the HIV/AIDS Prevention and Control Office). Estimates put the amount at about $20 million equivalent in 2001/02. 0 The costs o f a modest plan that would stabilize prevalence and reduce it slightly to 5.5% by 2020 i s estimated at about $60 million annually. A more comprehensive program costing $140 million p.a. i s projected to reduce prevalence to 5% by 2010, and 3.6% by 2020. 0 These estimates do not include anti-retroviral therapy, which could consume an additional $20 million annually. xiv Public Spending on Food Security/Safety Nets The depth and severity o f poverty in Ethiopia justify some form o f safety net transfer. An estimated 13 million people currently live at levels o f consumption so l o w they are at permanent risk o f destitution or starvation. Even under fairly robust growth, the numbers under the food poverty line are predicted to remain above 15 million over the coming decade. However the limited revenue base makes it difficult to sustain a large-scale transfer program, therefore solutions are needed which: (a) minimize the trade- o f f with spending on other, long-run poverty reducing measures; and (b) maximize the welfare effect per birr, or dollar, spent on transfers. Currently about US$265 million equivalent (2.3 billion birr) i s spent annually o n transfers to the poor, (varying from a l o w o f $120 million in good years, to a high o f $500 million in ‘bad’ drought years), the vast majority in the form o f emergency r e l i e f from donors. In 2003 government and donors convened a Coalition for Food Security, aimed at developing longer-term solutions; and the recent federal budget includes ETB 1.1 billion ($130 million) o f the government’s own funds for a Food Security initiative - mostly to fund measures such as water-harvesting and out-migration in food insecure areas. The current pattern o f spending i s dominated (and distorted) by the preponderance o f donor food aid. Ethiopia i s almost unique in the world in s t i l l receiving substantial amounts o f food aid on a regular basis. W h i l e this has helped avert starvation on the scale seen in 1984, it i s a costly and ineffective way o f providing sustained safety net transfers to the poor. Government’s basic policy i s sound: to allocate the majority o f food through the works-based Employment Generation Scheme. However the EGS suffers from problems o f lack o f counterpart funds, o f implementation capacity, and from the stop-start nature o f both works and transfers; in reality much o f the food i s distributed free outside the EGS. The free food distribution program suffers from unpredictability o f transfers, and amounts are often too small, or arrive too late, to have a significant effect on households. Food aid also distorts prices and incentives, and undermines the development o f functioning markets that could help intermediate food availability across seasons and regions. Despite these drawbacks, the framework that has evolved, consisting o f gratuitous relief, EGS, and foodcash for work programs, has worked w e l l in serving short-term needs and has saved lives in the face of repeated rainfall failures. I t has some laudable features: a functioning early warning system, the involvement o f NGOs, and administrative and community-targeting that has undoubtedly helped food reach the most vulnerable groups. The drawback i s that it i s designed almost exclusively around acute short-term relief, and operates o n an ad hoc basis outside o f the normal functioning o f government. xv 0 Donors and Government generally agree on the desirability o f shifting to domestic purchase o f grain, and (better) to cash transfers; however the transition needs to be phased to (i) take account o f the pace o f development o f food markets; and ( ii) to minimize the loss o f resource transfers currently provided through food aid. Perhaps half, and maybe more, o f the relief aid flowing at the moment would not be available to Ethiopia otherwise. T h i s can be handled by using the non-fungible food-aid resources to finance more continuous transfers under nutrition and food-for-work programs. 0 Existing transfers are concentrated on a limited number o f geographical areas (157 woredas); while this i s understandable given the immediate drought- response orientation o f the process, it leaves out large numbers o f very poor people in non-food-insecure areas. 0 A greater impact could be achieved using the same level o f resources by designing a sustained safety net program. Elements o f the solution include a continuous works-based program; a nation-wide nutrition program; and a shift to more cash-based transfers. 0 The costs o f three safety net scenarios are forecast. An ambitious scenario, that provides support to the majority o f the poorest, i s unaffordable (costing about U S 5 1 4 million annually, or 4.4% o f 2015 GDP); even the more moderate scenario i s only financable if large donor r e l i e f inflows are sustained - which seems unlikely over the long term. 0 The recommended program i s a relatively modest one, costed at about U S $ 150 million p.a. currently; this would provide essential coverage - in varying degrees - to about 15% o f the population, plus a focused national nutrition program. Public Spending on Water Supply and Sanitation 0 Water supply i s central to improving the living conditions o f the poor, but also for longer-run economic growth. Furthermore, improved sanitation and hygiene i s one o f the single greatest public health interventions available to government. 0 According to WHO in 2000 Ethiopia had the lowest level o f water supply coverage in sub-Saharan Africa (24%)*l and the second-lowest level o f sanitation coverage (15%). The level o f current spending i s remarkably low. As in other sectors there are data problems, but it appears total spending - both urban and rural - i s about $53 million p.a. (including donor financing, but excluding much NGO spending). 0 Consistent with the l o w level o f spending, and increasing population pressure, coverage was estimated to have increased by only 2-6% between 1990 and 2000. 21 The Ministry o f Water Resources estimates higher coverage - of about 30% - l i k e l y due to differences in the definition o f coverage used. xvi Spending over the past decade has been concentrated in urban areas. This has partly been justified by the public health and economic development needs o f urban centers; but the result i s very l o w levels o f spending (and coverage) in rural areas, where the vast majority o f the unserved population are. Average unit construction costs are roughly comparable to elsewhere in Africa. The efficiency o f use o f resources i s limited by the operation and maintenance problems o f urban schemes (water losses are estimated to be as high as 50% in some utilities); and the failure o f many rural systems to be kept in functioning condition. e The policy framework i s basically sound. Government i s shifting to a policy o f self-financing for urban areas. However there are going to be continuing large investment needs in cities, and while O&M costs can probably be covered from tariffs, cost-recovery i s unlikely to be sufficient to finance the very large capital requirements in many urban areas. e At any given time 20 to 30% o f rural schemes are reported to be non-functional. This i s a cause for concern, and raises questions as to the wisdom o f allocating additional resources rapidly. W h i l e there i s an increasing move towards community responsibility for operations and maintenance, there has been limited experience in implementing these models in Ethiopia. Future spending requirements are massive. T o reach the MDG target o f halving the proportion o f people without access to safe water by 2015, new service would have to be provided to an estimated 44 million people, while maintaining service to those who already have it. The government’s Water Supply and Sanitation Master Plan proposes raising rural water supply coverage to 71% by 2015, (and 100% by 2025); and would require average spending o f about $143 million (ETB 1,200 million) annually for rural water supply alone over the next 20 years. In addition it i s estimated ETB 8 billion (US$ 950 million) needs to be spent o n investments in urban water supply and sewage. The largest share o f these costs are in Addis Ababa, and will hopefully be financed without recourse to the budget, but there will s t i l l be a substantial public expenditure requirement for smaller towns. e Even to reach the somewhat more modest MDG target o f halving the proportion without water by 2015 (coverage o f 61%), would require rural water supply spending o f about US$ 115 million annually, (about five times estimated current spending), and i s unlikely to be affordable, and an estimated ETB 7.1 billion (US$825 million) would need to be spent on urban water supply capacity - about half o f it in Addis. e There are number o f policy issues moving forward: (i) despite the commitment to cost-recovery in urban areas, there i s a risk that the scale o f need, combined with the political power o f urban xvii dwellers, will exert pressure to spend scarce resources on urban supplies at the expense o f rural coverage. (ii) if rural water supply i s to be financed out o f block grants to regions and woredas, there i s a r i s k that there will be insufficient resources, unless transfers are expanded dramatically; (iii) functioning rural supplies require a substantial investment in supporting and training communities. If this function i s l e f t entirely to woreda governments it seems unlikely that they will have the resources and technical capacity to meet these needs. (iv) Spending o n sanitation and sewerage i s going to be a major issue. I t i s essential that piped water be accompanied by systems to evacuate waste water. Unit costs are high, and willingness to pay i s typically far lower than for water supply. T o date this has not been a major issue in Ethiopia (most users do not have house plumbing) - but it will increasingly be a cost in the future. Consolidated Social Sector Spending Needs and Some Policy Implications What is the aggregate fiscal impact o f the policies being considered in Ethiopia? The table below shows the combined cost o f programs in all the social sectors under the low, medium, and high scenarios. Two-thirds o f these costs are associated with expansion o f the school system - especially at the upper-primary level - and the primary health care system. (See Box on page 11 in chapter 1 for the broad characteristics o f the scenarios.) Aggregate Annual Costs of the Total Social Sector Expenditure Scenarios (Constant 2002 ETB billions) I I I I I Low Scenarios I 6.8 billion I 8.6 10 I I ~ 11.4 30.10% Middle ’Extended PRSP’ 6.8 13.6 18.2 21.9 54.80% plus’l 1 1 1 I 1 ‘Scenarios Ambitious ‘MDG 6.8 24.2 30.6 36 89.40% Scenarios Annual Financing Gap for -4.6 -6.6 -7.4 ‘Extended PRSP’ Scenario /a ($535 million) ($767 m.) ($860 m.) Annual Financing Gap for -14.9 -19 -21.5 ‘MDG Plus’ Scenario ($1.7 billion) ($2.2 b.) ($2.5 b.) The ‘low’ scenarios involve almost n o increase in coverage or quality fkom current levels, and are probably not acceptable to government, donors, or the Ethiopian people, but they illustrate the costs o f just standing s t i l l - which would s t i l l require a 40% real increase in spending. xviii Both the middle ‘Extended PRSP’ scenario’, which involves fairly modest pursuit o f existing strategies in health and education, and the more ambitious ‘MDGplus’ scenario, which models full implementation o f all o f government’s stated policy goals, imply the need for substantial additional financing. The amounts are huge, requiring a doubling o f social sector spending in real terms by 2015 under the middle scenario, and a four-fold real increase for the ‘MDG Plus’ scenario, with a financing gap o f $2.2 billion annually. These financing requirements are fiscally overwhelming. Even the moderate ‘Extended PRSP’ scenario would consume about 55% o f projected total public spending by 2015”. Given the financing constraints, innovative approaches are going to have to be adopted - relying on lower-cost interventions, such as less-qualified teachers and less reliance on facilities-based health care - but even then hard choices will have to be made between programs and sectors. Nonetheless most o f the targets are modest by international standards, and there i s a strong case for trying to accommodate them. The implications - among others - are for: (i) substantially increased foreign aid in support o f an agreed spending program, ( ii)a shift in budget shares towards the social sectors, ( iii)increasing the share o f recurrent as opposed to capital expenditure in the totalz3, and (iv) increasing the share o f the budget transferred to the Regions. (For example the annual expenditure requirements at the federal level increase by 1.4 times by 2015 under the ‘Extended PRSP’ scenario, while those at the regional level increase by almost 4 times). On balance we would argue that there i s justification for increasing spending to achieve these objectives. But there i s also a need to focus on selected interventions. A number o f findings stand out from the sectoral chapters: There exist some obvious ‘best-buys’ , these include: the widespread provision o f injectible contraceptives; revitalizing the immunization program; and provision o f relatively small, but predicable, amounts o f non-wage recurrent financing to schools, and supply o f adequate textbooks. There are some unavoidable major cost items - foremost among them i s expansion o f the teacher cadre to reduce student:teacher ratios. There are areas where large-scale expansion o f spending is questionable, at least until issues o f effectiveness and absorptive capacity can be resolved. They include: technical and vocational education, where confirmation o f effectiveness needs to be confirmed by tracer studies; rural water supply, where the reasons for up to 30% o f systems currently not fimctioning need to be understood; 22 And even this i s based o n Ethiopia sustaining a 5% real growth rate, and doubling aid levels by 2015. 23 Recurrent social sector budget needs increase to ETB 16 billion p.a. by 2015 under the Extended PRSP scenario, if the current mix (45% capital spending, share o f total net o f military) remains the same as today, there would only be about 18 billion for all recurrent activities. xix rapid further capital investment in health facilities (until it can be established they can be staffed and equipped); and, several areas where private-financing or cost-recovery need to accompany increases in public spending, including tertiary health care, university (especially the provision o f room and board), urban water supply, and WET. Finally, there are some areas which while challenging (and possibly expensive) would have major returns: a female outreach service delivery cadre to support family planning; a nationwide nutrition program; expansion o f secondary education capacity; and, investments in urban sanitation. There are clearly strong linkages to be exploited between spending in one sector and outcomes in others. For example, it i s estimated that achieving universal primary female education would reduce under-5 mortality by 15.5 per 1,000; improved rural water supply could reduce it by a further 9.5 per 1,000. Reducing population growth through spending on family planning i s central to relieving the pressure on food insecurity; fairly small amounts spent on family planning now can radically reduce future social spending requirements (very roughly by $100 million annually in the following decade for each 0.5 percentage-point reduction in population growth); and spending o n works-based safety net transfers can create infrastructure that contributes to longer-term poverty reduction. There are also a number o f ways o f reducing costs, including: Lower cost construction techniques for classrooms; Shifting to locally-recruited and employed teachers (this i s already happening, it needs to be accompanied by measures to assure quality, and i s probably mostly applicable at the lower-primary levels - the scope for savings at the grade 5-8 level, where most o f the future costs are, will be more limited); Slowing the rate o f expansion o f public technical and vocational education, and shifting the burden to employers, especially through on- the-j ob training; Reducing the incremental cost o f safety net programs by utilizing money that i s already being spent o n capital works elsewhere in the budget (roads construction and maintenance, classroom construction) to employ the poor during the off-season at l o w wage rates; Using non-fungible food aid to finance a national nutrition program, and a works-based safety net scheme; Establishing a lower exit-point for general education, after about 5 or 6 years; Relying on community health workers, and relatively low-skilled staff to provide outreach health services; Increasing cost-effectiveness by f i r s t focusing efforts on areas o f relatively greater population density (for example in expanding family planning and immunization.) Note even the ‘MDG Plus’ scenarios are not particularly ambitious by international standards. For example they involve aiming for secondary gross enrollment rates o f 30% on average by 2020; implementation o f a full primary health care package, but only limited expansion o f tertiary care. These are not unreasonable, and in many cases the government and donors should be aiming at achieving the ambitious scenarios (subject in some cases - especially health and water supply - to absorptive problems being resolved). But the bottom line i s that the government i s not going to be able to afford to pursue all o f i t s currently-stated sector plans. This fact needs to be confronted; the risk otherwise i s that officials continue to develop sector programs in isolation that cannot in aggregate be financed. The proposals in the Water Sector Master Plan, the ambitious TVET expansion, and health coverage plans, cannot all be accommodated. Perhaps more wonyingly, new initiatives keep being added each year without any central agency keeping track o f the total fiscal implications, and without first protecting essential funding for core recurrent activities. The choice o f service delivery mechanisms, role o f non-governmental actors, and o f private funding, will all be increasingly important in mapping out a strategy for the social sectors in the coming decade if Ethiopia i s to overcome the dual problems o f affordability and limited capacity. Absorptive capacity. In many countries the management o f service delivery i s a binding constraint, and we would be reluctant to advocate increasing social sector spending o n ineffective public services. In Ethiopia we would argue this i s not the case. While there certainly exist some service delivery bottlenecks, by and large systems work compared to elsewhere: for example, teachers generally show up at school and teach, and the need i s for spending to fill the existing deficit in teachers and classrooms - both o f which can consume large amounts o f funding relatively quickly. In health the problems are more significant, but nonetheless there are some basic high-impact interventions that are under- provided (immunization, family planning, and malaria among them), and subject to confirmation that staffing bottlenecks can be overcome, it would seem justified to expand funding for these core recurrent services. Other potential concerns are leakage and the misallocation o f funds. Compared to many low-income countries these are not over-riding problems in Ethiopia. There i s doubtless some mis-use o f public resources, but neither anecdotal evidence, nor macro-level data o n the utilization o f recurrent budgets, suggest that the problem i s widespread. Decentralization provides the potential for better resource allocation, utilization, and mobilization. T o realize these benefits will take time however, and they will not remove the need for substantially higher social sector spending, which i s driven by the sheer magnitude o f the unserved needs. There are also risks in decentralization - especially o f unfunded mandates if responsibilities are pushed down without a commensurate increase in transfers to the lower levels o f government. Private Financing. Government i s pursuing many o f the right policies in terms o f encouraging private provision. At the same time, the scope for private financing and cost-recovery i s more limited than elsewhere, due to the lack o f effective demand, and lack o f ability to pay among the 80% o f population who are subsistence farmers with little or n o involvement in the cash economy. xxi While cost recovery and private provision are certainly part o f the solution in the case o f tertiary level services, the fact remains this will not obviate the need for substantially higher levels o f public spending for basic services in the foreseeable fbture. We demonstrate for example, that even under fairly optimistic assumptions with respect to the growth o f private schools and introduction o f user fees, the net effect would only be to reduce total public budget requirements in general education by about 5% by 2015, (in part because so much o f the spending needs are at the lower levels o f the system). Links with Broader Development Strategy. What matters in terms o f reducing poverty in Ethiopia i s growth: improved productivity o f agriculture, creation o f off-farm employment, and continued stability. But in the context o f a longer-term development strategy, spending on education and population are also o f paramount importance. General education because o f i t s direct productivity impact, to expand the pool o f potential entrepreneurs and lay the base for higher-level training, and because o f the large positive impact on health, hygiene and fertility. Family planning because if the population continues to grow at over 2 million persons per year, interventions in almost all other areas will have a vastly diminished impact. Conclusions Given the basically sound policy priorities o f the govemment; the ability o f the public sector to deliver services; the central importance o f education, population efforts, and basic health, water, and nutrition provision; and the limited potential for private investment to fill these gaps, we would have n o reservations in advocating a substantial increase in extemal aid to support expanded social sector spending in Ethiopia. The amounts will clearly depend on more nuanced costing within the sectors, but if savings and prioritization can be made within the ambitious ‘MDG Plus’ scenario, the need might be in the order o f an additional $1billion annually over the coming decade. However not all social sector spending i s equally effective, nor contributes equally to poverty reduction, and this commitment needs to be accompanied by a consensus between govemment and donors - including lower level tiers o f govemment - on the complexion o f this spending, and choice o f highest-level interventions i t will support. A closely related point: the expansion o f social services requires primarily recurrent spending - o n a large scale, and over a prolonged period - the implications are not just for aggregate levels o f donor support, but also for a shift in i t s composition towards recurrent financing, and in the time horizon and predictability o f donor commitment. PART11: OVERALL PUBLIC EXPENDITURE(VOLUME11). Public spending in Ethiopia has increased substantially, by 80 % in nominal terms since 1997l9gz4. This represents an increase from 24% o f GDP in 1997198 to 35% in 2002103. The increase was driven at first by increased military spending during the war years (1998-2000). Defence spending has since declined, but spending in other areas has been increased by a corresponding amount, leaving total spending at the new, substantially 24 In the low-inflation environment o f recent years there i s not much difference between real and nominal. xxii higher, plateau level. The increase was initially financed by increased domestic borrowing, but since 2001 it has been sustained by greatly increased donor inflows. At 35% o f GDP, in relative terms public spending i s high by sub-Saharan Afiican norms; however because the denominator i s so low, in absolute terms public spending, at only $35 equivalent per capita, i s among the lowest in the world. In macroeconomic terms the recent level o f spending has been consistent with the agreed IMF program, and judged to be broadly sustainable given the level o f aid inflows. The increases in spending have taken place in areas broadly consistent with the poverty- reduction strategy: in education, road construction, and, to a lesser extent, in agriculture and health. The table below shows the broad composition o f the budget over recent years. Changes in Sectoral Spending 1999/2000-2003/04 (Budgets - Current Birr millions) Agriculture & 1,136 1,358 1,228 1,743 2,518 1,382 11.7% Natural Resources Other Infra & Econ. 636 811 917 858 705 69 3.3% Total *** 18,822 18,469 17,661 19,856 21,478 In Ethiopia seven regional governments are responsible for delivery o f primary and secondary services (education, health, agriculture, and some roads) and just over 40% o f public expenditure. Regional recurrent spending has grown continuously since the dip during the war years. Despite increasing absolute levels o f transfers from the federal government to the regions, transfers as a share o f discretionary domestic resources have been stagnant o n average, and have in fact declined over the past two years. Federal spending has been growing proportionally faster in both the capital and recurrent budgets, xxiii and this i s a cause for some concern, given the responsibility o f the Regions for delivering core social services. 25 This i s to some extent being rectified in the 2003/04 budget, in which transfers to the regions from domestic resources are increased by ETB 0.5 billion (although the relative increase in federal spending i s s t i l l larger (ETB 1.7 billion)). Budget execution has generally been sound in Ethiopia. There i s n o evidence o f loss o f expenditure control or over-spending at the aggregate level, and based on the available accounting system (which in our view i s generally sound) funds appear to be spent on their intended uses. The Country Financial and Accounting Assessment completed in 2002 confirms the finding o f earlier PERs that there do not appear to be major problems o f leakage or diversion o f hnds from intended uses. With the exception o f 2002/03, typically the recurrent budget has been 100% spent in any given year. There i s under-spending o n the capital budget, which was 77% spent on average the last four years for which actual data are available. This i s however to some extent a data and reporting problem. The government-financed part o f the capital budget i s largely spent (90% or more), with the under-funding concentrated o n the donor- financed part26. There are three elements to the capital under-spending story: ( i ) firstly, there are legitimate problems o f physical implementation; ( ii)secondly, there are bottlenecks due to donor requirements for procurement and accounting; and ( iii) thirdly, there i s an estimation and recording problem - the budget i s based on donor projections o f spending, which tend to be optimistic, while reporting o n actual aid expenditures i s incomplete, resulting in a large apparent mismatch between budget and actual spending. Unfortunately in the absence o f better data, it i s not possible to disaggregate how much o f the problem relates to reporting gaps, and how much to real implementation difficulties. The 2002/03(EFY 1995) and 2003/03 (EFY96) Budgets A very large increase in expenditure was programmed in the 2002/03 (EY95) budget, 2.2 billion birr ($226 million), or 12.8% above the previous year. Big spending increases were planned for agriculture, roads, and education, especially at the federal level. 27 The preliminary out-turn for 2002/03 shows that the assumed level o f revenue, and increased spending, was overly optimistic. Revenue was at least ETB 1.5 billion below target, and both capital and recurrent spending were about 15% below budgeted levels. Foreign aid levels were also lower than forecast (by about $120 million). These results are some cause for concern, as previously recurrent budgets have generally been fully- funded in Ethiopia; and especially because preliminary results suggest that 2002/03 under-spending was concentrated o n recurrent and regional budgets, which finance most o f essential service delivery. 25 I t i s difficult to generalize about the externally-funded part o f the capital budget, which changes with the level and composition o f aid flows from year to year. 26 The capital budget i s about 50 % domestically financed, and 50% aid-funded. 21 In addition a supplementary budget for ETB 600 million was approved in May, 2003, to fund the government’s infonnation-technology initiative under the capacity-building program (more than half o f the total), and restocking o f food reserves. The supplementary budget was f i a n c e d with funds generated by the fuel stabilization fund and DfID budget support. xxiv Developments in the 2003/04 (EFY 1996) Budget Notable features in the 2003/04 budget included: an increase in federal spending o f ETB 1.7 billion ($197 million, or 15%), mostly o n the federal capital budget, which i s up 41% from the previous year; major new spending a national food security program (ETB 1 billion or $116 million - mostly for water-harvesting, out-migration, and the like in food-insecure areas), road construction, agricultural services, and higher education. M u c h o f the increase i s from the government’s own resources; a large increases in the recurrent budget for health at the regional level (increased by 100%); ETB 3.0 billion ($350 million) i s budgeted for defence. The level i s consistent with the agreed IMF program, and represents about 15% o f total public spending; the 2003/04 budget includes budget support financing o f ETB 2.7 billion ($315 million); and HIPC relief amounting to about ETB 819 million ($95 million); transfers to the regions remain the same as the previous year (ETB 6 billion; about 32% of total federal expenditure), but with a substantially larger transfer from domestic resources (up by ETB 0.5 billion), offset by lower projected aid-funded transfers. The over-riding factor however i s the large programmed increase in total expenditure (up by ETB 3 billion, $355 million, or 15% above last year’s budget). However, the experience o f the past year, o f shortfalls in projected revenue and aid, and the more general difficulty in spending large capital budget increases, raises questions as to whether such large increases in the budget will materialize in actual spending. Aid Flows and Aid-Related Issues I t i s difficult to quantify aid inflows precisely, because substantial amounts are off- budget, and reporting on aid spending i s incomplete. Within these constraints we estimate that Ethiopia currently receives approximately US$ 1 billion equivalent in development assistance. In addition there are major inflows o f emergency food relief, which vary from year to year, but have averaged US$ 265 m i l l i o n annually over the past 7 years. Approximately 65% o f development aid i s reflected in the budget. Using different data sources, we estimate about $350-450 m i l l i o n equivalent o f aid i s received off-budget each year (not counting food aid). By 2002 the level o f development aid had doubled, from an average o f about $400-500 million p.a. in the mid 1990s. T h i s i s due to the return o f bilateral grant donors following the end o f the border conflict, and the re-introduction o f budget support and quick- disbursing policy-based credits from the EU, IDA and the African Development Bank. Aid finances about 37 % o f total public expenditure. Nonetheless, the absolute level o f assistance i s l o w relative to other low-income countries (at about US$ 16 per capita excluding emergency relief, compared to, for example, about $33 per capita for Burluna Faso, Rwanda, or Ghana) XXV Aid Process Issues Incomplete information on expected aid inflows creates problems for the government in (a) forecasting the amounts o f financing available in aggregate, and in determining transfers to the regions; and (b) in determining how much o f i t s own resources to allocate to sectors and expenditure items, such as rural water supply, textbooks, or drug procurement. At the same time, incomplete reporting o n aid expenditures makes it difficult for policy-makers to assess the true level o f actual expenditure and i t s composition. There i s an urgent need to strengthen systems o f aid forecasting, as well as reporting on actual spending. Progress has been made in the last year towards a fi-amework for improved donor coordination. Ethiopia i s a pilot country for the Donor Harmonization exercise; and a Donor Budget Support group has been formed to agree on a common framework for budgetary support. I t remains to be seen at a practical level how the actual coordination o f aid will be affected - much aid s t i l l takes the form o f enclave projects, and there are uncertainties about how the allocation o f sector-related donor support can be linked with the increasingly decentralized spending autonomy o f Regions and woredas. Progress on Budget System Reforms T w o major developments have taken place since the last PER that affect budget processes: the merger o f the former Ministries o f Finance and MEDAC into the Ministry o f Finance and Economic Development (MoFED); and rapid implementation o f the second phase o f fiscal decentralization. The government has been pursuing a sound process o f budget reform over the past several years, aimed at improving expenditure management and shifting to a medium- term planning framework. Reforms in the accounting system and chart o f accounts, covered in previous PERs, have been implemented at the federal level, and are now being rolled out to the regions. The shift towards longer-term planning has three components: establishment o f the MEFF (Medium-Term Expenditure and Fiscal Framework, primarily a tool for projecting the resource envelope), a Public Expenditure Plan, intended to better link expenditure planning with policy; and a fixed budget calendar. The MEFF was used in setting the 2002103 and 2003104 budgets, although with delays. The merger o f M E D A C and M o F paves the way for better co-ordination o f capital and recurrent budgets, and more policy- based expenditure planning, however implementation o f the PEP approach has been delayed by the amalgamation o f the two Ministries, but also because it i s not clear how i t will work in the context o f the shift to woredu decentralization. There i s a concem that new initiatives keep being added without any central system keeping track o f the total costs o f commitments being made; and without any mechanism in place to screen proposed new initiatives, nor to assess their affordability. Furthermore, it appears commitments are being made to new initiatives and capital investments without first assuring adequate financing for essential recurrent activities. Some version o f the earlier-proposed PEP process, or a more detailed MEFF, would appear to be essential to managing expenditure commitments. xxvi Adherence to the fixed budget calendar i s important to allow sufficient lead-time for debate on budget priorities, and to give the Regions timely notification o f the level o f transfer they will receive. The calendar has been adopted, but adherence to it was delayed in 2002103. The change in systems, combined with a realignment o f responsibilities, led to a deterioration in budget consolidation and reporting in the past year. I t was not until ten months into the fiscal year that a consolidated budget for the country as a whole was available for 2002l03. A similar problem applies to expenditure reporting. W h i l e there has always been a problem with delays in expenditure reporting from the regions, it has been aggravated with the move to woreda-level decentralization. T h i s i s due partly to logistics and remoteness, and partly to the absence o f systems to handle data generated at the lower levels. W h i l e these problems may be largely transitional, clearly it i s important for policy makers to have good information on overall expenditure levels, composition, and budget execution. This takes on added importance with the shift to budget support. Strengthening the systems o f consolidation and handling reporting data i s an immediate priority. Fiscal Decentralization and Federal-RegionalIssues As noted earlier, the level o f transfer to the regions increased substantially in the FY2002103 budget. Nonetheless, the share o f the budget transferred to the Regions i s not commensurate with their responsibilities for the highest-priority services. Since 2002 there has been a rapid decentralization o f expenditure responsibilities to the woreda level (there are 550 woredas with an average population o f about 100,000). Within the four largest regions, 45% o f the regional budget has been transferred to woredas in the form o f untied block grants. Out o f this, woredas are expected to finance most primary and secondary level health and education service delivery, agricultural extension, and basic rural infrastructure. The initial allocations to woredas, based o n a formula similar to that used to allocate funds fkom the center to the regions, resulted in mismatches between financing and existing expenditure obligations. The regions have handled these pragmatically by using a range o f ad hoc arrangements and contingency funds, but the fact remains that in many cases the transfers are just sufficient to cover recurrent obligations (particularly salaries), limiting the scope for discretionary expenditure management by local government. Initial findings suggest the impact in the social sectors has been mixed: for example in SNNPR officials felt that flexibility in making decisions on minor expenditures at the local level had improved responsiveness to local needs. On the other hand the limited absolute amounts available, and the capacity constraints in the woredas, have resulted in additional strains on the system. While the immediate problems are being handled in a practical manner, there are a number o f longer-term issues: xxvii 0 There i s a risk o f unfunded mandates: that a smaller share o f the budget will be transferred down to the woreda level than would be justified by their relative responsibility for delivering services, and a disportionate share retained by federal, and regional, agencies for their own priority activities. The problem i s likely to worsen as the focus shifts from just sustaining the existing systems to competing for the (very large) incremental resources needed to expand basic services. 0 There i s a need to accelerate ‘partial’ reporting to get real-time results on budget execution. From what we have seen o f raw data generated in the field this should be possible. 0 There remains a major issue o f how conformity with national policy goals will be achieved with the new system - especially in terms o f composition o f expenditure. I t i s not clear how much o f a problem this will become; in the past the decentralized system has managed to pursue agreed priorities on a consensus basis. However there are a number o f ways o f this might be reinforced: through setting benchmarks for certain levels o f service (eg. student-teacher ratios), setting a ‘floor’ for spending on certain key activities, or conditional grants for capital spending. 0 The costs o f additional staff and infrastructure t o build up the capacity o f hundreds o f woredas i s substantial. There i s a risk that this will place a first call on incremental resources at the local level at the expense o f increased spending for recurrent operations o f core services, such as education or family planning. Initial quantification o f these costs suggests they could be massive. Preliminary estimates for Oromiya for example suggest the administrative costs could equal the total amount o f block grants to woredas for all other purposes. Specialized Topics: Extent of the Debt-Relief Dividend Inflows o f HIPC funds allowed an increase o f about ETB 779 million in the 2002/03 budget. On average over the next decade HIPC relief will average a similar amount, about US$90 million p.a., equivalent to about a 10% increase in foreign aid levels. While this represents only a 4% increase in total public expenditure, in absolute terms, if focused on priority activities, the amount can be very useful. For example the ETB 779 million i s double the amount needed to restore the student-teacher ratio in general education. Impact of Shocks Given the vulnerability o f Ethiopia to shocks - both recurrent drought and terms-of-trade shocks - an effort was made to assess the impact o n public expenditures. T h e conclusion i s that the effect i s not large. This i s because subsistence agriculture - which suffers most from droughts - does not contribute much to public revenues, while the additional expenditures o n relief food are mostly financed directly by donors outside the budget framework. The impact o f petroleum price changes i s limited, because taxes on petroleum products account for less than 1% o f revenue, and fuel i s not systematically xxviii subsidized. Exports are mostly not taxed, so the impact o f commodity price changes (mostly coffee and hides) i s also limited. The major fluctuations in spending over the past two decades have been attributable to instability (revenues decreasing during the political transition in the early 1990s; and military expenditure crowding out other spending during the war years); and to volatility in aid flows. F u t u r e o f the PER Process Formal public expenditure reviews have been undertaken in Ethiopia for the past 8 years. Consistent with practice worldwide, there i s agreement among all the actors that that the PER process should move towards a more continuous one o f in-country support, integrated with the budget cycle. The current proposal, under consideration by the government and donors, i s that the tracking o f expenditure and process issues should be accommodated in a rolling process, linked with strengthening o f expenditure monitoring and reporting within government; while a ‘light’ external review would probably s t i l l be undertaken each year to satisfy the requirements o f donor fiduciary accountability and continued budget support. PART HI: SUMMARY LISTING OF RECOMMENDATIONSDRAWN FROM THE REVIEW Be prepared to raise spending o n primary and secondary education to about ETB 4 billion ($465 million) annually by 2010, if significant increases in quality and enrollment are to be achieved.28 Provide additional financing for textbooks and transfers to schools for non-wage recurrent costs, equivalent to a minimum level o f about ETB 25 per student annually (primary) and ETB 75 per student at secondary level. Provide small amounts o f direct grants to schools to finance such costs. Undertake a study to measure the extent o f community-employed teachers, their wage levels, and effectiveness. Estimate realistic scope for increased utilization o f community-employed teachers and impact on public budget requirements. Increase spending to reduce student-teacher ratios in primary and secondary schools immediately - increase transfers to Regions and woredas as necessary to do so. (estimated requirement about 3 3% increase in recurrent education funding from 2001/02 levels). 28In constant 2002 ETB, based o n the Extended PRSP’ scenario o f close to universal primary education, and 80% grade 5-8 enrollment by 2015; and assuming 30% savings due to reduced cost measures. xxix e Quantify donor spending on textbooks, and unify planning and implementation o f donor and government textbook financing. e Consider establishing a grade 5 or 6 exit point for a large proportion o f students. e Aim toward a major expansion o f secondary education; make adequate provision of funding to Regions to finance (costs specified in chapter 2 projections). Undertake a tracer study o f employment and income effects o f training on recent TVET trainees. Slow down investment in expanding government-provided technical and vocational education. Develop systems for on-the-job, apprenticeship, and private-sector technical training. e Continue investment in expanding university education. e Model the effects o f the proposed ‘graduate tax’ before finalizing; consider introducing fees to cover costs o f room and board. e Increase health spending, but tightly focused on immunization, malaria, and staffing and equipping basic outreach services. Review actual staffing and utilization o f newly-constructed health facilities; restrict new investments until satisfactory operations and staffing can be demonstrated. Review need for sustained Regional or zonal involvement in supporting immunization, family planning, and malaria programs (and the school inspectorate function), with a view to retaining management and funding above the woreda level if needed. Concentrate on recruiting, training, and retaining lower-level paramedic staff for outreach health services. Spend about $25 million (ETB 215 million) per year o n a major family planning program, with an emphasis on injectibles. e Donors should assign priority to financing such a program. e Establish a cadre o f village-level female workers to support implementation o f this family planning program. e Develop a country-wide program to employ the poor on public works in the off- season; set wages to be self-targeting. e U s e existing budgets for rural works to employ the poor under the same conditions. e Establish a national nutrition program for under-30 month old children, subject to analysis o f costs and design options. U s e non-fungible food aid to finance the nutrition and public works programs. Re-examine the affordability o f the Water Supply Master Plan Undertake a specialized study o f the reasons for rural water supplies not functioning. Increase funding for rural water supply - recognize the need to raise transfers to Regional and woreda government to do this. Assess capacity o f local governments to support community O&M. Quantify donor funding for rural water supply; establish a systematic process for jointly planning all donor, government, and NGO water supply funding in a common framework. Develop a system for training and supporting communities to maintain rural water supplies, from the Regional, zonal, or woreda level. Strictly enforce cost-recovery policies in urban water supply systems. Develop more refined cost estimates for providing adequate urban sanitation, and a strategy for financing it. Establish a system within the public expenditure management process to track new expenditure commitments being made for major program initiatives. Establish a system for review and screening o f such proposals, tahng account o f their affordability. Apply priority to ensuring core recurrent activities are adequately funded before committing to new initiatives. Revisit the scope for, and possible design of, the Public Expenditure Plan process in the context o f the model o f Regional and woreda-level decentralization. Undertake a sample survey o f actual donor-funded projects for a recent year, and compare spending to budgets, to determine the extent to which under-spending o n the development budget i s a reporting as opposed to an implementation problem, and to design a mechanism for correcting estimated spending. Develop a system for systematic reporting o f donor spending by sector and activity through Ministries and Regions to MOFED, by sector, activity, and type o f expenditure. Undertake specialized initiatives to quantify actual levels o f aid spending in health, education, rural water supply, and family planning in the past year; to identify spending by item and activity; to better inform government budget planning. H o l d an intensive one-day session with donors prior to the budget each year to better estimate expected aid flows by sector and Region. xxxi Provide earlier notification to Regions o f the expected level o f budget transfer for the coming year. 0 Develop systems for quicker partial reporting or estimating o f budget expenditure from woredas and Regions, by main sub-sector, o n a quarterly basis. 0 Consider adjusting the offset to reflect actual aid in the year before. 0 Quantify the administrative costs o f implementing woreda decentralization, hold an open debate around what i s affordable, the options, and trade-offs with funding delivery o f services. 0 Introduce a simple system o f direct reporting on expenditures by off-budget donor- funded projects (see Annex 8). 0 Adjust fiscal transfers to Regions, and woredas, to allow adequate funding for (i) restoring student-teacher ratios, (ii) financing recurrent outreach health services, and ( iii)delivering the family planning program - assuming these are to remain the responsibility o f these levels. xxxii INTRODUCTION This year’s Public Expenditure Review examines the difficult question o f how large-scale social sector programs - which place intensive demands on recurrent budgets - can be adequately funded in the environment o f a US$lOO per capita income country. The primary objective i s to look at what the long-run fiscal costs o f various social strategies are in Ethiopia, and what the implications might be for policy choices being made now. The PER essentially asks three questions: How much is Ethiopia spending on the social sectors? Documenting current spending and how it has evolved; and providing observations on the adequacy, efficiency, and appropriateness o f the existing package o f expenditure. How much could Ethiopia spend on the social sectors? That is, what would the costs be o f pursuing a ‘reasonable’ set o f social sector policies and targets? The object i s to lay out for decision-makers the fiscal implications o f the broad policy choices they are pursuing; and to examine the aggregate costs o f various levels o f coverage and service, loolung forward 5, 10, and 15 years. How much should Ethiopia spend on the social sectors? Finally the PER attempts to provide some prescriptive advice o n the levels and composition o f social sector spending over the coming decade and beyond; by looking at the needs together with: (i) some assessment o f efficiency and cost-effectiveness; (ii) the likely resource envelope; (iii) the potential role for private financing and delivery; (iv) absorptive capacity - that is, the extent to which Ministries and Regions can in fact spend additional resources effectively; and finally, (v) the relative importance o f spending on education, health services, and safety net transfers in the context o f a wider development strategy. It i s hoped that examining social sector spending in this way will help shed light on policy choices, and o n the ‘right’ selection o f programs for the medium term. I t will provide a framework for consideration o f both donor and government financing for the social sectors, as well as a more informed context for overall fiscal planning over the coming decade - since the recurrent cost obligations o f expanding education and health systems will dominate future budget demands. The review covers five sectors which are directly relevant to poverty reduction, and in which the government has a clear role to play as a major provider for the foreseeable future: education, health care and HIV/AIDS; population, safety net programs, and water supply. Together these sectors account for 40 percent o f non-military spending in Ethiopia. 1 Links with Other Work The twin themes o f delivering social services and decentralization are central to Ethiopia’s Sustainable Development and Poverty Reduction Program (SDPRP). The findings o f the PER are important for informing subsequent development o f the SDPRP strategy in the social sectors, as well as feeding into the next phase o f the Sector Development Programs in health and education. Service delivery issues are touched upon in the PER, but are being addressed in more depth in two companion pieces by the World Bank - the forthcoming Country Status Reports on Education and Health. The link with fiscal decentralization - which i s critical, and rapidly evolving - i s being looked at in more depth in ‘on-going technical assistance work under the capacity buildinginitiative. M u c h o f the richness o f this year’s PER lies in the background studies. While the principal findings o f these are incorporated in this report, they include substantially more detail, which should provide valuable input into policy dialogue in the sectors. Among others there are specialized studies on education unit costs and financing, family planning needs, TVET, water supply, and safety net and food security issues.29 The PER Process. The 2002/03 PER differs from usual Public Expenditure Reviews in that it i s intended primarily as a longer-term reference piece o n social sector spending needs, designed to inform the dialogue on funding and expenditure choices over the next decade and beyond. I t i s also probably the last in the line o f traditional annual World Bank-style PERs in Ethiopia. For some time it has been recognized that it i s preferable to shift towards a more continuous, in-country process o f expenditure review, that i s integrated with the government’s budget process, and this i s being pursued between the Government and donor community in Ethiopia (see discussion on the future o f the PER in Volume 11). Structure o f the Report The report consists o f two main parts: This volume deals with social sector spending; individual chapters cover general education (Chapter 2), higher education (Chapter 3); health and HIV/AIDS (Chapter 4), population (Chapter 5), safety nets (Chapter 6), and water supply expenditure (Chapter 7). In each case there i s an analysis o f current spending, followed by projections o f what would be required to achieve a range o f policy outcomes - generally ranging from the costs o f a minimum ‘ ~ u s ~ ~ ~ s s - ~ sapproach, -usu~~~ to an ambitious version that would involve attaining all o f the government’s stated policy objectives. The overview in Chapter 1 brings together the projected expenditure requirements developed in the sector chapters in the context o f likely macroeconomic affordability, and explores policy implications. Volume I1 presents the standard elements o f a public expenditure review: an update on recent developments in public spending, the broad fiscal framework, and budget execution over the past few years. I t also includes an account o f process developments since the last PER: budget management reforms, developments in fiscal decentralization, and aid coordination - although these are covered only briefly, as they are dealt with in 29 A full listing i s provided in the bibliography. 2 depth in other recent documents or on-going pieces o f work. (In particular, the CFAA, and on-going fiscal decentralization and capacity-building initiatives.) As in other years, the Public Expenditure Review has been a joint effort, with substantial input this year from UNFPA, gtz, Irish Aid, DflD, and WHO. Methodological Notes 3 1. OVERVIEW OF SOCIAL SECTOR SPENDING, FUTURE NEEDS, AND POLICY IMPLICATIONS A. A BRIEF OVERVIEW OF SOCIAL SECTOR SPENDING AND OUTCOMES 1.1 This chapter consolidates the fiscal projections and affordability implications o f the work presented in the six subsequent sectoral chapters. Except for some broad observations on priorities, i t does not repeat the findings o f the sectoral analysis, for which readers are referred to the Executive Summary and to the individual chapters. 1.2 I t i s difficult to measure current public social sector spending precisely, because large amounts o f aid are not reported to the budget authorities, but roughly, public spending on the social sectors in 2001/0230amounted to about ETB 6.7 billion (US$ 780 million), representing 13% o f GDP, and about 21% o f the total budget (40% o f spending net o f interest, pensions, and defence). Table 1.1: Estimated Aggregate 2001/02 Social Sector Spending - Estimated Share of Share of US$ per Total \1 Spending GDP Capita ( E m millions) Education Primary & Secondary 1,587 9.5% 3.1% $2.70 Technical and University - 729 4.4% 1.4% $1.24 Sub-total Education 2,316 13.9% 4.5% $4.15 Health 795 4.8% 1.6% $1.35 Water Supplv 12 456 _. 2.7% 0.9% Sub-Total 3,567 21.4% 7.0% Largely Of-Budget Expenditures SafetyNets 13 2,280 da 4.5% $3.87 HIVIAIDS - 258 n/a 0.5% $0.43 Total 6,707 da 13.1% $1 1.40 1/ Based on 2001/02 budget pre-actuals , recurrent plus capital - the last year for which reliable sectoral breakdown o f spending was available at the time o f the review. 2/ Water supply spending, which i s spread among different budget categories and levels of government, i s a best estimate for 2001/02. 3/ Safety net spending varies substantially from year to year; US$256 million (ETB 2.28 billion) p.a. i s the estimated average o f past seven years; share o f budget i s not estimated since most i s financed in the form of extra-budgetary food aid. 1.3 The share o f the budget and o f GDP spent on the social sectors are roughly comparable with other l o w income countries, but because the base level o f per capita GDP i s so low, absolute levels o f spending are extremely low, at, for example, about $ 3.00 per capita on general education, and $1.35 per capita on health. The sectoral chapters describe in detail the levels, composition, and issues associated with current spending. 30 The most recent year for which reliable actual expenditures were available at the time o f the review. Budgets have increased since, but it i s not clear whether they have been spent. 4 1.4 Total public spending in Ethiopia has increased substantially in recent years (from 24% o f GDP in 1997198 to 35% in 2002103); resources have been re-allocated from military spending, and aid inflows have increased (see Volume 1 1. Defacto the ) government has demonstrated clear priorities in allocating this new spending - with relatively large increases for education, road construction, and agriculture, and, within the social sectors, consistently lower allocations to health, population, and rural water supply* 1.5 As described subsequently, very large increases in social sector spending would be required - o f the order o f five times current spending in real terms - to implement all o f the currently proposed social programs. 1.6 Government’s goals for the medium term are laid out in the Sustainable Development and Poverty Reduction Program (SDPRP) o f 2001, which aggregates the sector programs in education, health, and water supply. Fairly ambitious plans are proposed in all the social sectors, initially unconstrained by affordability. The broad targets in the SDPRP- described in more detail in the sectoral chapters - are to increase primary enrollments to 65% between 2001102 and 2004105, to double university intake; to increase technical and vocational training capacity by four-fold; to increase basic health service coverage to from 52% o f the population to 65%, and water supply coverage to 39%.31 T h e requirements involve a doubling o f spending in real terms during the period 2001102-04105, with a substantial financing gap. Recognizing that this i s unlikely to be affordable, the SDPRP also proposes alternate scenarios for somewhat lower resource availability. 1.7 While ambitious, even the SDPRP targets do not start to bring Ethiopia up to nthe following sections levels o f coverage attained in most other l o w income countries. I we show longer-run costs o f both the extension o f these current policy directions, and o f a more ambitious program designed to bring Ethiopia closer to international norms over the next two decades, including achieving the MDGs. Social Outcomes and Spending 1.8 O f course what policy-makers are ultimately concerned with i s not spending, but outcomes. Assessing the relationship between public spending and social indicators i s difficult under any circumstances. I n Ethiopia judging the impact o f past spending on outcomes i s further complicated by the fact that the country has been in almost continuous turmoil from the mid-1970s until the late 1 9 9 0 ~ ~ ~ , as a result both and economic growth, and improvements in living conditions have been limited. Furthermore, many o f the gains that have been made have been eroded by rapid population growth and the periodic impact o f severe droughts. Nonetheless discernable progress has been made in some areas. 3 1 More recently, a major initiative i s being launched to try to eliminate food insecurity within the next five years, through a combination o f agricultural productivity and market improvements, migration, and safety net measures. 32 Including a revolutionary change o f government in the 1970s, followed by sustained internal struggle during the 1980s, leading to another change o f government in 1991, and a border war with Eritrea during 1998-2000. 5 Table 1.2: Selected Social Sector Outcomes - 1990-2000 ~ 1990 2000 Gross Primary School Enrollment Ratio 33% 61% Under 5 Mortality Rate (per 1,000) 190 167 Total Fertility Rate 6.4 5.9 Illiteracy Rate 71% 60% Proportion o f population Below Poverty Line 48% 44% Immunization Coverage (DPT3) 67% 47% 1.9 However the bottom line i s that per capita incomes were about US$lOO in 1970, and are only just above $100 today. Within this framework it i s difficult to affect very significant improvements in the quality o f life for a large share o f the population. In fact, given the l o w level o f growth, the progress that has been made i s quite impressive. These indicators are substantially better than would be predicted on the basis o f correlations with per capita income alone, suggesting that it i s indeed possible to make an impact through public intervention in the Ethiopian context, even under difficult circumstances. The continuous disruption has made it hard for service delivery systems to effectively transform spending into outputs. Despite this, substantial improvements have taken place in some areas, most notably the increase in school enrollments. If greater stability had prevailed, it i s almost certain that public spending would have had a greater impact. 1.10 The relationship between public spending, outputs, and outcomes i s a complex one, determined in part by the levels o f spending, in part by the response o f beneficiaries, and in part by the efficiency with which funds are utilized by the service delivery system. These are explored in the individual sectoral chapters, but by and large the story on the effectiveness o f spending has been mixed. In the case o f education, the service delivery system appears to function reasonably well, although the recent success in raising enrollments has resulted in a deterioration in quality due to over-crowding. I n health, basic interventions operate at a very l o w level o f coverage (for example immunization and contraceptive provision are both among the lowest in the world) and service delivery i s hampered by staffing and human resource constraints. Similarly, a significant proportion o f water supply systems do not function; and safety net programs, while succeeding in forestalling short-term starvation, have not addressed the longer-run problems o f chronic malnutrition and food shortages. 1.11 The Millennium Development Goals provide an agreed set o f development targets for the year 2015. While there i s not a one-to-one mapping with the sectors covered in this PER (for example the MDGs do not cover secondary education, university, family planning programs, or social safety nets), there i s a large area o f overlap, and the MDGs provide a basis for comparing Ethiopia’s progress, and requirements, with those o f other countries. Table 1.3 below summarizes the MDGs for Ethiopia. Clearly there i s a major challenge ahead, and in many cases meeting the specific targets will be difficult, although substantial progress can be made through judicious interventions. T h e costs associated with meeting the MDGs are calculated in the sector chapters, and summarized later in this overview. 6 Table 1.3: The MillenniumDevelopment Goals in Ethiopia Base Level Implied MDG Most Recent Goal (by 2015) (1990) Target by Measured 2015 Level 1. Reduce by half number o f people living below 48% 24% 44% poverty line 2. Halve the number o f people who suffer from hunger. 51% 25.5% 42% 3. Achieve universal primary education 32% 100% 61% 4. Eliminate gender disparity in education 50% 41% 5. Reduce under-5 mortality rate by two-thirds 190 63 167 6. Reduce maternal mortality rate by two-thirds n.a. n.a. 500-700 7. Halt and reverse spread o f HIV/AIDS n.a. <6.6% 6.6% 8. Halt and reverse incidence o f malaria and other major n.a. n.a. n.a. diseases 9. Halve the proportion o f people without access to safe 76% 38% 70-76% drinking water. 1.12 This Public Expenditure Review by definition focuses on expenditure requirements. What i s required to reach the MDGs in terms o f public policy, and the probability o f achieving them, are touched upon, but are covered in more depth in other documents.33 In the PER we have estimated the direct costs o f public actions related to the MDGs; however it needs to be recognized that many o f these outcomes will depend on factors beyond public spending - including the pace o f economic growth, performance o f agriculture, and the responsiveness o f households to programs (for example, improved health outcomes depend as much on changes in food consumption, access to transport, and hygiene and nutritional behaviour, as o n government spending o n the health services.) In each sector, however, we have attempted to cost the direct inputs required to maximize the probability o f reaching the goal. 1.13 Clearly, there exist substantial linkages and externalities in social sector spending. For example, to achieve the health outcomes depends critically on spending to increase female education levels and to reduce fertility; while spending on population programs dramatically reduces the future requirements for education and water supply expenditure. These linkages are explored in the individual sectoral chapters, and brought together in the final section of this overview, which attempts to place social sector spending in the framework o f a broader development strategy. Who Benefits from Public Spending in Ethiopia? 1.14 Benefit-incidence analysis indicates that primary education i s slightly pro-rich; with a somewhat higher than average probability o f children from the richest quintile (20%) going to school in rural areas, and a somewhat lower than average probability for children in the poorest 20% o f households - but with children in the middle 60% o f the population being more-or-less equally likely to benefit from primary education spending. 33 For example the UNDP: MDGs Challenges and Prospects for Ethiopia (2002), the forthcoming Country Status Reports o n health and education, and planned World Bank-IMF paper o n the MDGs and aid requirements. 7 This i s almost certainly because poor households - for a number o f reasons related to parental education, resources, and the opportunity cost o f labour - are less likely to send their children to school. I t i s also to some extent explained by the fact that poorer families tend to live in more remote areas with poorer transport infrastructure. Figure 1.1: Selected indicators of Benefit-Incidence of Social Services in Ethiopia34 Net Enrollment Rndo I I Net Enrollment Ratio SecondaryEducation Primaty Education Q I (PwratZO%) 0 131 Q I (Poorst20%) 0 326 Q2 Q2 Q3 Q3 Q4 Q4 Q5 (Richest 20%) Q5 (Richst20%) 0 321 0000 0 100 0200 0300 0400 OS00 0600 0000 0050 0 100 0 IS0 0200 0250 0300 0350 Use of Public Health Facllrhes Access to Safe Water Q I (Poorest20%) Q I (Poorest20%) 92 Q2 Q3 Q3 3 Q4 Q4 0 321 I I Q5 (Richest 20%) Q5 (Richest 20%) 0000 0050 0 100 0 IS0 0200 0250 0300 0 01 02 03 04 05 06 Sources: Welfare Monitoring System data, as reported in Govemment of Ethiopia (MEDAC, 2002). 1.15 N o t surprisingly, urban primary education expenditures more equally benefit the poor (because many more of them are in school in urban areas), while, equally unsurprisingly, secondary education spending benefits the better-off much more than the poor (with the odds o f participating in secondary schooling being three times higher for children from the richest quintile than from the poorest). 1.16 In health, the data suggest that spending on lower-level facilities (clinics and health posts) disportionately benefit the poor. In fact according to one study households from the poorest quintile use health centers 10% more than the national average. T h i s i s almost certainly because people in rural areas mostly frequent clinics and health centers, whereas urban dwellers by-pass them and go to hospitals. Note however that, as shown in Chapter 4, the DHS data show that the poor are much less likely to be immunized, or benefit from a range o f other specific health interventions than are the non-poor. Water supply spending benefits the rich far more than the poor - with those in the top 20% more 34Note that this data i s drawn from two earlier studies (Seifu,2002) and the government’s poverty assessment. More recent analysis by the World Bank suggests somewhat different results (with generally more egalitarian distribution o f benefits, partly because o f the larger sizes o f poor households), these were not finalized in time to include here. 8 than twice as likely to have access to safe water from those in the poorest 20%. T h i s i s a direct consequence o f the concentration o f water supply spending in urban areas, noted in Chapter 7. 1.17 Geographical Imbalances. I n many large, diverse countries, poverty and poor human development indicators are concentrated in a relatively small number o f areas, so that focusing incremental spending efforts in a few areas would go a long way towards improving outcomes. (For example, recent work in India suggests that just 20% o f villages account for over half o f child deaths). In Ethiopia this i s less the case, probably because poverty - and remoteness - are more evenly spread across the country. T h e three least-developed regions (Afar, Somali, and Beneshangul-Gumuz) almost certainly lag the rest o f the country, but they constitute less than 10% o f the population; and conditions in them are rivaled by those in many poor areas o f the larger ‘more developed’ Regions. The four large regions (Oromiya, SNNP, Ahmara, and Tigray) account for 80% o f the population, and on average health and education indicators in these four Regions equal the (very low) national averages. The greatest distributional differences - whether in poverty, immunization rates, enrollments, or student-teacher ratios - are within Regions, which vary greatly internally, depending on remoteness, degree o f development, and economic productivity and potential. 1.18 Unfortunately data available at the time o f the PER did not allow good disaggregation o f indicators and sectoral spending within Regions. What i s needed i s a more detailed analysis separately for each o f the main Regions, at the woreda and kabele level. Such work i s currently on-going in Oromiya, and should be undertaken in all o f the Regions. However some indicators available (for example school enrollment, and student-teacher ratios) suggest that neither spending shortfalls, nor poor indicators are heavily concentrated in only a few least-developed areas. 1.19 Instead, to the extent imbalances exist, they are largely urban-rural. For example, per capita spending o n education i s 100 % higher in Addis than in Oromiya; and water supply spending has been heavily concentrated in urban areas (see Chapter 7). T o some extent there are reasons for this: for example the need for functioning urban water supplies to avoid epidemics; the high levels o f per capita education spending in Addis reflect the fact that more people go to school there; and urban areas provide more expensive services that are used by rural dwellers (for example hospitals and secondary schools).35 1.20 The strategic implications are likely that the greatest impact in expanding services, and concentrating additional money, in the next phase (2005-15) will probably come from focusing o n ‘accessible poor’ areas - where demand i s already evident (for example through the very high enrollments and student-teacher ratios); and accessibility i s sufficiently good that services can be effectively delivered (for example staff will accept postings). 1.21 Infrastructure i s o f course critical for achieving social sector outcomes. Road density in Ethiopia, at 500 km. per million people, i s about the lowest in the world, and 35 And in some cases the converse i s true - spending per capita i s higher in smaller, poorer areas, where the denominator i s very low. 9 one-fifth o f the average for Africa36. Evidence from elsewhere also suggests that transport accessibility i s closely correlated with many social outcomes (for example, research in Kenya found that distance from a paved road was the single most important variable in explaining maternal mortality). But i t needs to be recognized that under any scenario large parts o f Ethiopia are going to remain relatively isolated for a long time. And while many o f these are the poorest, they are also often the most sparsely populated, and incrementally more difficult, and expensive, to reach; so given the very large number o f areas and people who are already moderately accessible, i t probably makes sense to concentrate efforts on them in the next round. B. Projections of Future Social Sector Spending Needs 1.22 This section brings together the projected spending requirements developed in the sectoral chapters, and compares their total costs with likely resource availability. We illustrate the fiscal consequences o f the policies currently committed to or being considered, and the long-run macro-fiscal implications in terms o f affordability and financing requirements. We then work back to show what this means for policy choices facing the government and donors today and in the medium-term. 1.23 More speculatively, the last section looks at trade-offs between sectors, and some possible priorities in the context o f a long-run development strategy for Ethiopia; as well as the link with growth and poverty outcomes. Consolidated Expenditure Requirements and Macroeconomic Affordability 1.24 A range o f policy options are costed in the sector reviews, building on the strategies laid out in the SDPRP and related policy document^.^' These constitute the base-case ‘extended PRSP’ cost projections. A higher scenario shows the costs o f implementation o f all o f the government’s most ambitious plans, plus full achievement o f the MDGs - this i s called the ‘MDGs Plus’ scenario.38 A lower case scenario shows the cost o f just keeping up with population growth. The box below shows the attributes o f the scenarios at a glance. 36 African Development Indicators, 2003 - data for 1996. 37 For example the National Population Policy, the Water Supply and Sanitation Master Plan, and HSDP I1 and ESDP I1 consultations. When no target was specified in government strategy (for example in the case o f social safety nets, or long-run targets for university enrolments) a range o f reasonable high and l o w assumptions has been made. 38 Recognizing that the two are not always mutually exclusive; there exist some overlaps - for example in some cases the MDGs alone would involve less spending than the full sector strategy (for example education - which includes spending for university and TVE), and in some cases more (where the full range o f MDG inputs costs more than government’s current expansion plans -eg. health). 10 Box 1: Summary of Three Scenarios Used t o Project Costs o f Social Sector Policy Low, Business as Usual Scenario School enrollments, primary health care, and water supply keep pace with population growth, with only limited quality improvements. Current levels o f higher education (TVET and university) coverage are maintained; slow expansion o f HIV/AIDS and population programs; minimal safety net transfers for the poor. M i d d l e ‘Extended PRSP’ Scenario Close to universal primary education through grade 6 by 2020; gross secondary school enrollments o f 20%; with improvements in teacher-student ratios and education materials. Implementation o f current expansion programs in higher education and TVET, and then keep pace with population growth. Improve primary health care quality, and reach 85% coverage by 2020. Stabilize incidence o f HIV/AIDS and reduce slightly by 2020. Expanded population program raises contraceptive prevalence to 41%, reducing fertility to 3.4, from a current level o f 6.0. Moderate food security transfers for poorest 5% o f the population (and an additional 510% in bad years). Achievement o f 64% water supply coverage by 2015 (consistent with MDG). Ambitious ‘MDGs Plus’ Scenario By 2020: Universal primary education through grade 8, with significant quality improvements; secondary school enrollments o f about 30%. In TVET, pursue stated target o f absorbing 40% of grade 10 leavers; in university, continued expansion until 3% enrollment o f university-aged population reached by 2020. Aggressive family planning program reduces fertility t o 3.0 by 2020 (CPR o f 46%); H I V / A I D S program to reduce incidence to 3-4% by 2020. Safety net program ensures food security for poorest 510% o f population, and some form o f support to the poorest 20%. Full implementation o f Water Supply Master Plan (which aims for 100% coverage by 2025) 1.25 Table 1.4 shows the overall expenditure requirements for the middle ‘extended PRSP’ scenarios in each o f the main sectors. Total requirements amount to ETB 17.6 billion p.a. by 2015 (US$ 2.0 billion annually) - approximately equivalent to total public spending, net o f pensions and interest, today. The big ticket items are, not surprisingly, expansion o f general education, primary health care, and safety nets for the poor. 11 Table 1.4: Estimated Annual Expenditure Requirementsunder the - 'Extended PRSP' Scenario 2010-2020 (Recurrent and Capital - Constant 2002 ETB millions annually) Current Memo Item Spending % of 2015 /a 2010 2015 2020 Public Exp. General Education 1,587 5,324 6,849 7,908 21.1% TVE 226 580 772 859 2.4% University 726 1,081 1,315 1,600 4.0% Health 795 2,332 4,329 6,238 13.3% Population n.a. 219 309 419 1.O% HIVlAIDS 258 516 516 516 1.6% Water Supply 456 1,380 1,380 1,380 4.2% Safety Nets " 2 , 6 8 6 2.949 8.3% Total 6,328 13,867 18,156 21,869 55.9% US$ million $ 1,613 $ 2,111 $ 2,543 Memo Item: Projected Expenditure Ceiling for social sectors /b 9,300 11,600 14,500 Shortfall: /c (-4,600) (-6,600) (-7,400) a/ Approximate, based on 2001/02 pre-actuals; safety nets, 7 year average; HIV/AIDS and water supply estimated. b/ At 5% growth, after interest and pensions; based on historical share o f 40% after interest, pensions military. c/ Shortfall relative to historical social sector shares o f total spending. 1.26 Are these projections reasonable? Most are based o n fairly modest ambitions with respect to coverage - for example they involve achieving secondary enrollments o f only 20 percent in the next two decades:' and pursuit o f government's basic targets in health, but would not allow attainment o f the health MDGs. The more ambitious ones are: (i) safety net spending (ETB 2.7 billion in 2015 - but even this corresponds to a fairly limited scale o f coverage, and i s not far o f f the current level o f spending4'); (ii) technical and vocational education (although the moderate forecast i s substantially lower than government's currently-stated targets); and, ( ii ihealth spending in the later years - ) especially the projection o f 6.2 billion in 2020 - which may exceed the absorptive capacity to spend. 1.27 H o w affordabie are these costs? Table 1.4 also shows likely availability o f resources, if Ethiopia i s able to sustain 5 percent p.a. GDP growth over the next 18 years, along with a moderate increase in aid inflows.41 W h i l e there may be some mismatches 39 Gross enrolment rates o f only 15 % at the grade 11-12 level, and 25% at grades 9-10 40 $256 m. pa. average in r e l i e f aid over the past 7 years (Chapter 6), or ETB 2.3 billion p.a . 41 The forecasts assume an increase in revenue to 20.6% o f GDP, a real doubling of aid inflows over the period, to $2 billion equivalent in 2020; debt service payments as in the HIPC projections, and domestic financing as in the current PRGF. 12 between financing and program costs as a result o f simplifying assumptions4*, the broad picture presented i s a realistic one. Under this scenario, by 2015 the total cost o f social sector programs (ETB 17.6 billion annually, in constant 2002 Birr) would equal about 56 percent o f the total resource envelope, net o f military spending. Clearly this i s not feasible given the competing claims on government resources. 1.28 If we compare these costs to the average historical level o f social sector spending - about 40 percent o f spending after interest, pensions, and military spending (last line o f table) - we see there i s a significant financing shortfall, o f about ETB 6.6 billion, or $700 m i l l i o n annually, by 20 15. I t i s worth highlightinga number o f points: 0 These are moderately conservative targets for requirements; the underlying assumptions o f resource availability o n the other hand are optimistic; the growth rate o f 5 percent p.a. i s above the rate has been sustained in the past - and above the level that all but a few countries have consistently maintained over a 15 year period; they also assume that defence spending falls from 15% o f the budget now to 9% by 20 15; and a doubling o f real aid inflows over the period; 0 T o the extent that growth i s faster, andor aid flows higher, the constraint will be relaxed, but it would require a quantum change in orders o f magnitude o f growth and aid to make even these moderate scenarios affordable; and On the other hand, if growth were closer to historic levels, or the projected increased aid flows were not sustained, then even this moderate scenario would become even more unaffordable. 1.29 The graph below, and associated table show the public expenditure implications, and financing gap, for the two main scenarios. 42 The main one being non-fungible food aid associated with financing o f safety net expenditures; which could reduce financing requirements from the budget by about ETB 2 billion p.a. 13 Figure 1.2: Projected Social Sector Expenditure Needs, 2005-2020 Projected costs o f "MDG Plus" Scenario Projected costs of "Extended PRSP" Scenario Likely Resource Envelope- Social Sector Spending I 2005 2010 2015 2020 % of 2015 Approximate Annual Public Financing Gap Expenditure /a ETB billions /b (US$ millions) Medium 'Extended PRSP' Scenario 55.9% 6.1 ($710) High 'MDG Plus' Scenario 90.3% 19.8 ($2,300) ai At 5% growth, after interest and pensions. b/ Shortfall relative to historical social sector allocation o f 40% o f public spending after interest, pensions, and military. 1.30 The conclusion i s that these amounts cannot be accommodated within the likely financing envelope. There are a number o f implications: ( i)substantially more aid i s going to be required; ( ii)ways have to be found o f reducing costs; ( iii) not everything within the package can be afforded, and some choices will have to be made regarding trade-offs within the social sectors; and, (iv) the share o f social sector spending will need to be increased to achieve even relatively modest targets, suggesting the need to confront hard choices in reducing spending elsewhere in the budget. Section C below explores ways in which costs may be reduced, and some possible basis for considering trade-offs. Financing Requirements of the Lower and Higher Scenarios 1.3 1 The estimates suggest that if govemment was to pursue the lower 'business-as- usual' scenarios in all sectors - that is, just keeping up with population growth and barely increasing coverage or quality at all, i t would s t i l l need to spend something like ETB 10 billion ($1.2 billion) annually on the social sectors by 2015, or about 40 percent more in real terms than has been spent in recent years. However this i s not realistic. The aspirations o f both the govemment and donors are for substantial improvements in coverage and quality o f health, education, and water services, and the condition o f the Ethiopian people make such improvements imperative. I n any event, govemment i s 14 already embarked upon expansions well above the minimum ‘low’ scenarios, Nonetheless, this provides an indication o f what would need to be spent just to stand still. 1.32 The total cost o f the more ambitious ‘MDGs plus’ scenarios exceeds likely resource availability by a very substantial margin. Yet the targets underlying this scenario are not particularly ambitious by international standards. They involve aiming for secondary gross enrollment rates o f 30 percent on average by 202043;rural water supply coverage o f about 75 percent by 2020, and implementation o f a substantial primary health care package, but with only limited expansion o f tertiary health care. With the exception o f the ambitious TVET targets, these are not unreasonable, and in many cases the government and donors should be aiming at achieving the ‘MDG plus’ scenarios (subject in some cases - especially health and water supply - to absorptive problems being resolved). 1.33 The projected financing gap for this scenario, even with sustained economic growth and doubling o f real aid levels, i s in the neighborhood o f US$2 billion p.a.. The implication, o f course, i s for a substantially larger mobilization o f aid t h an under the ‘Extended PRSP’ scenario; as well as deeper cuts elsewhere in the budget, and higher transfers to the Regions and woredas, to accommodate the increased recurrent costs. 1.34 The sectoral policy implications and priorities are discussed later, in part C; but we can make some broad initial observations on affordability just based o n relative budget shares (Table 1.5 below). Table 1.5: Sectoral Expenditure Composition of the Moderate ‘Extended PRSP’ and High ‘MDG Plus’ Scenarios Share of GDP (2015) Share of Public Spending ( 2015) Current Extended ‘MDG I Current Extended ‘MDG Share SDPRP Plus’ 1 Share SDPRP Plus’ General Education 3.1% 6.8% 9.0% I 9.5% 21.1% 28.2% TVET 0.4% 0.8% 1.O% 1.4% 2.4% 3.3% University 1.4% 1.3% 2.3% 4.3% 4.0% 7.0% Health 1.6% 4.5% 9.1% 4.8% 13.3% 28.5% Population n.a. 0.3% 0.3% n.a. 1.O% 1.O% HIVlAIDS 0.5% 0.5% 1.1% n.a. 1.6% 3.4% Water Supply 0.9% 1.O% 1.4% 2.7% 3.2% 4 Safety Nets 4.5% /a 2.6% 4.4% 12.0%/a 8.3% 13.6% Total 17.6% 28.7% 54.8% 89.4% 1.35 There i s n o way o f saying what the ‘right’ level o f spending is, but we can make some general observations, based on cross-country averages and norms. The required expansion in general education i s very large - spending would need to go from about 9.5% o f public expenditure today to over 21% under the baseline ‘Extended PRSP’ scenario - while there are some ways o f reducing costs (discussed later) there would appear to be little choice but to significantly expand spending. The expansion o f 43 35 percent grades 9-10,25 percent grades 11-12. 15 university capacity to reach only about 3% o f the population would require spending o f 1.3% o f GDP, or 4% o f public spending, which i s probably reasonable. With respect to health spending, expansion under the middle scenario to 4.3% o f GDP would be a challenge; to expand to the higher ‘MDG plus’ scenario at 9% o f GDP would likely be impossible. For water supply spending, the higher scenarios may be affordable, but would be imply a major shift from elsewhere in the budget. Spending at the projected levels on safety nets i s likely only possible if it continues to be financed by largely additional off-budget donor support. 1.36 The bottom line i s that there i s n o way the government can afford to pursue all o f i t s currently stated sector policies. Having ambitious plans i s laudable, but at some stage the fact needs to be confronted that everything being proposed i s not possible. The risk otherwise i s that officials in individual sectors continue to develop plans - in good faith - that in aggregate cannot be afforded. The proposals in the Water Supply Master Plan, the proposed strategy o f absorbing 40% o f grade 10 leavers in technical education, and the large-scale targets in public health, are in aggregate not affordable. What i s perhaps more worrying i s that new initiatives keep being added, without any central agency keeping track o f the total costs. Examples in the past year or two include the major expansion o f TVET, the information technology, agricultural training, and food security initiatives - each o f which have merit, but there appears to be a very real danger that new commitments are being made without f i r s t financing the essential core recurrent funding increases that are needed to ensure basic services are working, particularly general education, essential health services, and population programs. ADDITIONAL FISCAL A ND AFFORDABILITY ISSUES 1.37 There are two other macro-level considerations that are important. They are the balance between recurrent and capital expenditure demands, and the Regional-Federal division o f spending. 1.38 Recurrent vs. Capital Financing Requirements. Figure 1.3 illustrates the aggregated recurrent and capital cost implications o f the moderate ‘extended PRSP’ scenarios. N o t surprisingly, the major strain i s going to be o n the recurrent budget. If the historical division o f resources i s maintained, the social sectors would absorb about 80 percent o f the recurrent budget by 2020. While ways can be found to reduce some of these costs (section C), the implications are two-fold: (i) that the government i s going to need to re-think the balance between capital and recurrent spending, and if these goals are to be achieved, there probably needs to be a reduction in capital investments; and ( ii) aid support i s going to have to s i f t significantly into recurrent financing. 44 44 To the extent that military spending can be reduced below ETB 3 billion, it would free up additional resources; and if, as noted earlier, non-fungible food aid can be used to fimd safety nets the total recurrent demands would be reduced (by about 18 percent in 2020) 16 Figure 1.3: Recurrent and Capital Breakdown o f Social Program Costs under ‘Extended PRSP’ Scenario, and Likely Resources Envelope 21 # # 18 15 -I 12 : g ’ 6 3 0 2005 2010 2015 2020 1 ~ IC ”Projected Total Recurrent Budget * The expenditure envelope i s based on 5% GDP growth, and the current division o f 55:45 recurrent-to- capital spending, net o f military spending on ETB 3 billion. 1.39 I t i s important to recognize that the labour-intensive demands o f expanding social services imply long-term increases in recurrent budget obligations, and the need to make cuts elsewhere in the budget to accommodate them. The risk o f failing to confront this issue early on i s - as we have seen in other countries - that budgets become over- crowded and increasingly unrealistic, and result in growing, unresolved fiscal pressures. 1.40 Federal and Sub-National Budget Implications. Perhaps the biggest implications, however, are for the balance between federal and regional social sector spending. Figure 1.4 shows the approximate requirements o f the ‘Extended PRSP’ projections at each tier o f government. N o t surprisingly, the weight o f increased expenditure requirements (almost 70 percent) falls o n the Regional and woredu budgets - driven by the expansion o f general education, and in the later years, health. Federal spending requirements increase by 1.4 times in real terms between now and 2020, the Regional by almost 4 times4’. 1.4 1 If the historical division o f the budget between the center and regions maintains, under the base ‘Extended PRSP’ projections the social sectors would absorb about 84 percent o f total Regional spending, leaving almost nothing for all the other functions the Regions are supposed to deliver: agricultural services, roads, administration, and the substantial administrative costs o f woredu decentralization. Note that the same essential algebra applies to the division between Regions and woredus, implying a corresponding need for increasing woreda block grants. 45 Note the demand o n federal resources will be further reduced to the extent that safety net transfers are funded by relief aid. 17 Figure - 1.4: Federal-Regional Breakdown of Funding Requirements under ‘Extended PRSP’ Scenario, and Likely Resource Envelope 20 1 18 rz. 16 2 14 .s 12 10 $ 8 6 4 2 0 2005 2010 2015 2020 - Federal Social Sector Spending Requirements -Regional Social Sector Spending Requirements Expected Federal and Regional Total Expenditure Envelope* Envelope based on current 5050 division, after military spending o f ETB 3 billion, The assumed division o f spending requirements accords to the current assignment o f responsibilities to the Federal and Regional levels. 1.42 Clearly if these basic targets are going to be met, and the division o f responsibility i s going to remain as it is, it implies the need for a major re-thinking o f the relative allocation o f resources between the center and the regions (and woredas). 18 T h e earlier sections have dealt with the aggregate costs o f all social sector needs - including many non- MDG-related interventions; but what would it cost just to achieve the MDGs in Ethiopia?. Clearly attainment o f the MDGs depends o n factors beyond direct public spending in the social sectors - most notably the rate o f economic growth, and development o f complementary infrastructure and service delivery reforms. Nonetheless, we can cost the direct public inputs that are needed to maximize the likelihood o f them being attained. There are seven MDGs that relate directly to social sector spending by the Govemment: (i) ensuring every child can complete a full course o f primary schooling; (ii) eliminating gender disparity in education; (iii ) reducing under-five mortality by two-thirds; (iv) reducing maternal mortality by three-quarters; (v) halting and reversing the spread o f HIV/AIDS; (vi) halting and reversing the spread o f malaria and otheI major diseases; and (vii) halving the proportion o f the population without access to safe water and sanitation. What is involved in achieving each o f these i s discussed i n more depth in the respective chapters, but i n summary, the education goal is potentially achievable with significant additional investments; the health goals are unlikely to be met by 2015, although substantial inroads can be made by focused public expenditures - especially on immunization, malaria control, birth spacing, and outreach maternal and child care; the water supply goal can only be met with a very much intensified effort to expand effective coverage, and the sanitation target i s unlikely to be met. Very broadly the direct sector costs are as s h o w below: Estimated Direct Social Sector Spending Costs of Meeting the MDGs (Constant 2002 ETB Millions) 2005 2010 2015 Total Estimated Public Spending Requirement 462004-2015 Education 2,657 5,045 5,506 50.8 billion ($ 5.9 billion) Health 2,607 8,741 11,633 70.7 billion ($8.2 billion) Water Suuuly 2.760 2.7602.760 33.1 billion ($ 3.9 billion) Total 8,024 16,546 19,899 154.6 billion ($1 8.0 billion) c. SOME POLICY IMPLICATIONS AND IMPLEMENTATION CONSIDERATIONS 1.43 While we would argue strongly for increased assistance to achieve both the MDGs and the wider social sector targets, given the tight conditions government and donors are s t i l l going to need to be selective o f interventions, and look at every opportunity for reducing costs. A number o f factors affect the prioritization o f spending choices: (a) their likely effectiveness; (b) absorptive capacity; (c) maximizing ‘bang-for- the-buck’ - that is, seeking out particularly high-impact interventions that can be implemented at relatively l o w cost; (d) the case for public provision (combined with the capacity o f the private sector to finance the intervention); and (e) the fit with Ethiopia’s long-term development strategy, and the impact o n growth and poverty reduction. This section briefly reviews the expenditure choices in light o f these considerations. 19 1.44 In considering the overall package o f spending, a number o f findings stand out from the sectoral chapters: Firstly, There exist some obvious ‘best-buys’ these are things that are relatively cheap, offer a high return, and in general the system in Ethiopia can deliver them. They include: 0 the widespread provision o f injectible contraceptives; 0 immunization o f children, and, 0 provision o f relatively small, but predicable, amounts o f non-wage recurrent financing to schools; and supply o f adequate textbooks. Secondly, There are some unavoidable major cost items which while not particularly difficult, will require large amounts o f spending - foremost among them i s expansion o f the teacher cadre to reduce studentteacher ratios (which will require additional spending in the neighbourhood o f ETB 2 billion annually by 2015, and about 330 million p.a. immediately). Thirdly, There are a number o f areas where large-scale expansion o f spending is questionable, at least until questions o f effectiveness and absorptive capacity can be resolved. They include: 0 technical and vocational education (where confirmation o f the utilization’ o f the s k i l l s provided, and the effectiveness o f training needs to be confirmed by tracer studies); 0 rural water supply (where the reasons for 20 to 30 percent o f systems currently not functioning need to be understood and rectified); and 0 rapid further capital investment in health facilities (until i t can be established they can be staffed and equipped). Fourthly, There are some areas where private flnancing and/or cost-recovery is an essential element of theflnancing and incentives package, and there i s little justification in going ahead with public spending unless addressed. They include: 0 tertiaryhrative health care; 0 university (especially the provision o f room & board); 0 urban water supply; and 0 TVET. The government has sound policies for cost recovery and private provision in all o f these areas. Yet they are potentially high-risk, in that the initial investments represent big- ticket items, but enforcement o f cost-recovery may be politically difficult. The public pressure to ensure the services are provided, especially from the articulate urban populations, will be significant; and if cost-recovery i s not fully implemented for any reason, these high-cost services will have crowded out spending o n high-priority basic services. And flnally, there are some areas which while difficult, (and possibly expensive) are nonetheless essential 0 a female service delivery cadre to support family planning; 20 0 a nationwide nutrition program; 0 expansion o f secondary education capacity; and, 0 investments in urban sanitation and waste-water removal. Scopefor Savings 1.45 There are a number o f ways o f reducing costs discussed in the sectoral chapters. Foremost among them are: Lower cost construction techniques for classrooms; Shifting to locally-recruited and employed teachers (this i s already happening, it needs to be accompanied by measures to assure quality, and i s probably mostly applicable at the lower-primary levels - the scope for savings at the grade 5-8 level, where most o f the future costs are, will be more limited); Slowing the rate o f expansion o f public technical and vocational education, and shifting the burden to employers, especially through on- the-job training; Reducing the incremental cost o f safety net programs by utilizing money that i s already being spent o n capital works elsewhere in the budget (roads construction and maintenance, classroom construction) to employ the poor during the off-season at l o w wage rates; Using non-fungible food aid to finance a national nutrition program, and a works-based safety net scheme; Establishing a lower exit-point for general education, after about 5 or 6 years; Relying on community health workers, and relatively low-skilled staff to provide outreach health services; Increasing cost-effectiveness by first focusing efforts o n areas o f relatively greater population density (for example in expanding family planning and immunization - although in many cases the distribution o f existing services may already be approaching the limits o f the less- expensive-to-service populations.) Linkages 1.46 O f necessity, the PER examines spending and effects within each sector, but clearly there are a complex web o f linkages between spending in one area (for example female education) and outcomes in another (for example fertility, nutrition, and hygiene). These linkages are difficult to measure quantitatively in a country like Ethiopia, where data i s weak and the traclung systems are just evolving - nonetheless we can make some observations on the basis o f limited data, and experience elsewhere.47 - Water Supply and Health. Spending o n water supply can have a significant effect on reducing the incidence o f disease. This largely comes about through the reduction o f diahrreal diseases. The relationship i s not straightforward however, because it depends on 47 Such linkages are dealt with in more depth in the sector chapters. 21 complimentary changes in behaviour (for example with respect to washing hands and use o f soap), and the data requirements make it difficult to isolate the effect o f water and sanitation from other factors 48 affecting the incidence o f diarrheal diseases. However a simplified projection in the case o f Ethiopia suggests that rural access to safe water could reduce under-5 mortality by 9.5 deaths per 1,000, or about 6%; with the benefits being greatest for households in the poorest consumption quintiles (13.1 per 1,000).49 - Family Planning Impact on Education and Health Expenditures: Spending on family planning services today can save the government very large amounts in terms o f reduced future demand for social services. For example: reducing the population growth rate by half-of- one-percent by 2010 - an outcome easily achievable by raising contraceptive prevalence to 25% - would result in recurrent budget savings alone o f about ETB 1 billion ($1 16 million) in terms o f the costs o f providing lifetime education and health services to just one year’s worth o f avoided population growth. Education, Fertility and Health. Analytical work worldwide has repeatedly confirmed the strong positive impact o f female education in reducing infant mortality and fertility levels. In Ethiopia, the DHS data show that under 5 mortality i s about 25% lower for children o f mothers with primary education compared to those without; and 52% lower among those with secondary education. These results need to be treated with some caution, as women with more education tend to come from households with more and better food, and living in places with better access to medical care - but nonetheless they confirm that the broad relationships observed elsewhere hold in Ethiopia. A basic simulation exercise for Ethiopia indicates that achieving universal primary education for girls would reduce under-5 deaths by about 15.5 per 1,000, or by about 10%. - Higher Education Spending and Social Sector Manpower Requirements. Spending o n secondary and higher education i s an essential input into providing the pool o f skilled manpower needed to expand basic social services over the next two decades. For example, by 2010 the school system will require approximately 37,000 new teachers a year (to keep pace with population growth, attrition, and coverage expansion); and approximately 10,000 health workers annually.” Thus a significant share o f spending o n secondary education i s directly related to providing the trainable people needed to deliver essential primary services. 48 See for example Health Impacts of WatedSanitation Programmes; UNEP, 2003. 49 See Wang (2003) Environmental Determinants o f Child Mortality - Empirical Results from the 2000 Ethiopia DHS. 50 Source: projection models for education and health spending requirements; under ‘moderate’ extended SDPRP scenarios, documented in sector chapters and annexes. 22 Population, Food Security, and Safety Nets - while there i s deep concern now with the safety net requirements o f the 5-6 million people who are unable to feed themselves each year (Chapter 6) - more worryingly, the population i s growing to the rate o f about 1.9 million each year, many o f whom are being added in areas, and under conditions, which make i t likely that they will become part o f this pool o f the chronically poor. (Total fertility, for example, i s only 1.9 births per woman in Addis, but i s 6.4 in rural Oromiya and 5.8 in Tigray.’l); reducing the rate o f population growth i s thus one o f the most effective interventions open to government in addressing food security. - Population and Poverty. More broadly, rapid population increase, in the absence o f corresponding growth in productivity, makes it far more difficult to have an impact in reducing poverty. In the past, additional population was to some extent absorbed in agriculture, and although there i s s t i l l scope for improving agricultural productivity, the sheer numbers being added to the rural population, combined with pressure on the resource base, make it more and more likely that additional population growth will need to be absorbed off-farm. By 2015 the labour force will be growing at about 650,000 per year. T o put this in perspective, total private sector non-farm employment for the whole country was estimated to be only about 2 million in 1998. Fertility and Health. Finally, excessive numbers o f pregnancies and births are a major factor in the poor state o f women’s and children’s health in Ethiopia. A staggering 25% o f female deaths are related to pregnancy and birth complications (DHS, 2000); and under-5 mortality drops dramatically, from 272 per thousand for children born less than 2 years apart, to 96 per thousand for those born 4 or more years apart. Spacing births and reducing the number o f pregnancies by providing family planning services i s thus one o f the principal public health interventions available to the government. Financing Requirements and Foreign Aid 1.47 I t i s difficult to quantify the amount o f aid Ethiopia receives n o w in the social sectors. Based on the TJNDP development assistance report it i s very roughly estimated at US$250 million equivalent per year5’. The amount could be substantially higher, because there have been increases in the past two years (including, for example, the Global Fund resources for health) with which trackmg systems have not yet caught up, and a large part o f social sector aid i s off-budget anyway. Nonetheless, the current amounts are not large compared to needs; and quite a l o t o f this existing aid finances technical assistance, enclave projects, and other activities that do not directly substitute for government spending on service delivery. Also, a surprisingly small share o f aid i s directed towards the social sectors in Ethiopia (for example, the Development Cooperation Report shows 17 percent o f aid going to health and education in 2002). 5’ Demographic and Health Survey, 2000. 52 Developmental assistance, net of emergency relief. Note classifications are ambiguous at times, this should be regarded as a rough order-of-magnitude estimate. 23 1.48 The relatively moderate ‘Extended PRSP’ scenario shows a funding gap for the social sectors o f US$465 million equivalent annually by 2010, and over $700 million by 20 15 (Figure 1.2). Even if costs can be reduced, and programs prioritized, the needs will s t i l l be substantial in areas where private financing i s unlikely to fill the gap: upper primary education, basic health and population services, and safety nets for the poorest. 1.49 Given the basically sound policy priorities o f the government; the ability o f the public sector to deliver services; the central importance o f education, population efforts, and basic health, water, and nutrition provision; and the limited potential for private investment, we would have no reservations in advocating a substantial increase in external aid to support expanded social sector spending. 1.50 However not all social sector spending i s equally effective, nor contributes equally to poverty reduction, and this commitment o f additional aid needs to be accompanied by a consensus between government and donors - including lower level tiers o f government - on the complexion o f spending, and choice o f highest-level interventions it will support. 1.51 A closely related point: the expansion o f social services requires primarily recurrent finance - on a large scale, and over a prolonged period - so the implications are not just for aggregate levels o f donor support, but also for a shift in i t s composition towards recurrent financing, and in the time horizon and predictability o f donor commitment. Untied budget support i s certainly part o f the solution, but not all o f it. Given the need for nuanced service delivery reforms, and to focus assistance on very specific sub-programs and uses, some form o f sector-specific program financing i s going to continue to be needed. Service Delivery and Institutional Issues 1.52 The choice o f service delivery mechanisms, role o f non-governmental actors, and o f private funding, will all be increasingly important in mapping out a strategy for the social sectors in the coming decade if Ethiopia i s to overcome the dual problems o f affordability and limited capacity. These are addressed in the forthcoming Country Status Reports, and in the sector chapters, and are only summarized briefly here. 1.53 Private Financing Options. In almost all the areas being considered there i s a case for private provision or cost-recovery, especially in the case o f tertiary-level services. Selective use o f user fees can also improve incentives (for example by discouraging the by-passing o f lower-level health facilities) and sustainability (as in the case o f drug revolving funds). However the scope for private financing in Ethiopia i s more limited than in many other countries, because o f the limited ability to pay, and l o w effective demand among the 80 percent or so o f the population who have extraordinarily l o w incomes, and in many cases almost n o involvement in the cash economy. There i s also legitimate concern that indiscriminate use o f user fees - especially at lower levels - could have substantial negative effects on access and equity. 1.54 Government i s pursuinggenerally sound policies in terms o f encouraging private provision and cost recovery; and while every opportunity for resource mobilization should be exploited, the fact remains this will not overcome the need for substantially higher levels o f public financing for the foreseeable future. We have illustrated, for example, in the education chapters, the scope for private provision and cost-recovery, and 24 shown that under fairly robust assumptions they might reduce the need for public expenditures at the higher levels (such as high school and TVE) by about 20-30 percent. But because o f the overwhelming weight o f lower-level service requirements, this represents only 5 percent o f total education spending needs by 2015; a useful, but not transformational, contribution. 1.55 Absorptive Capacity. I n many countries the management o f service delivery i s a binding constraint, and we would be reluctant to advocate increasing social sector spending on ineffective public services. I n Ethiopia we would argue this i s not the case. While there certainly exist some service delivery bottlenecks, by and large systems work compared to elsewhere, especially in education, where teachers generally show up at school and teach, and the need i s for spending to fill in the existing deficit in teachers and classrooms - both o f which can consume large amounts o f funding. T h e constraints are mostly the rate at which teachers can be trained and classrooms constructed. I n health the problems are more significant, but nonetheless there are some basic high-impact interventions that are under-provided (EPI, family planning, and malaria among them), and subject to confirmation that staffing bottlenecks can be overcome, at least in accessible areas i t would seem justified to expand funding for core recurrent services. 1.56 The other potential concerns are with respect to leakage, misallocation, and misappropriation o f funds. Again, compared to many low-income countries our sense i s that these are not over-riding problems in Ethiopia; an impression confirmed by the findings o f the recent CFAA, and covered in Volume I1o f this PER. There i s doubtless some theft and mis-use o f public resources at the service delivery level (drug pilferage was a case mentioned during the review), but neither anecdotal evidence, nor macro-level data o n the utilization o f budgets, suggest that the problem i s widespread. 1.57 Service delivery channels. There are three possible reasons for using NGOs or contracting out: to help relieve the pressure o n government capacity; because they may deliver more effective services; or because they can do so at lower cost, or with less leakage. The NGO sector i s in i t s infancy in Ethiopia, and as yet does not have the capacity for large-scale delivery o f public services. There i s some evidence that existing NGOs have been able to deliver high-quality services, and serve as a demonstration o f what i s possible. However they bring relatively little net additional funding themselves (at least compared to the scale o f requirements o f US$1-2 billion equivalent p.a.), and are largely channels for implementing donor-funded activities. The available data does not allow us to assess whether their unit costs, nor cost-effectiveness, are lower or higher than government’s. There i s little evidence that the costs o f government delivery are bloated in Ethiopia (with the exception perhaps o f classroom construction). I n the end the choice o f service delivery mechanism will largely effect who delivers service, it s t i l l needs to be paid for - somehow - out o f public resources. 1.5 8 Decentralization provides the potential for better resource allocation, utilization, and revenue mobilization. T o realize these benefits will take time however, and short- term gains may well be limited. The over-riding public expenditure concern in Ethiopia i s that the absolute volume o f resources available at the local government level i s insufficient to allow funding o f the necessary improvements and expansion o f social services. The problem i s likely to become more acute as the focus shifts from just sustaining the existing health, education, and water system to competing for incremental funds to expand coverage - especially given the other demands o n local government to increase administrative capacity, infrastructure, and agricultural services. 25 1.59 Decentralization carries with it the dual risks o f unfunded mandates (if responsibilities are pushed downward without commensurate funding), and o f capital bias (if local pressures result in building infrastructure - health stations, water supplies - that cannot be staffed, operated, or maintained). It i s too early to tell what the impact o f woreda decentralization on the social sectors will be, but the evolving system needs to be carefully monitored and adapted to manage these concerns. Finally, n o matter how great the gains from decentralization, they will not obviate need for substantially higher social sector spending, which i s driven by population growth, and the very l o w levels o f initial coverage and provision o f basic inputs. D. LOCATING SOCIAL SECTOR SPENDING IN THE CONTEXT OF WIDER DEVELOPMENTAND POVERTY-REDUCTION STRATEGY 1.60 The fiscal constraint taken in isolation would dictate that very l o w levels o f coverage andor quality be adopted. This begs the question o f whether it i s necessary to think outside the envelope in terms o f revisiting what are considered appropriate shares and levels o f public spending on the social sectors. This depends in turn on re- examining: (i) the relative importance o f various social sectors, and sub-sectors, in Ethiopia’s development strategy, along with ( ii)the efficiency with which those expenditures are converted into outcomes, and ( iii) the relative importance o f competing public investments. 1.61 There i s n o analytical methodology for assessing the relative importance o f different types o f spending, and n o ‘right’ level o f social sector spending, but a number o f considerations apply. 1.62 Firstly, i t needs to be recognized that not all social sector spending i s inherently pro-poor. Some (and in the case o f some countries, much) benefits the non-poor; and even in cases where it benefits the poor, spending on the social sectors often does not represent the best way o f using public resources to reduce their poverty. 1.63 Improving the conditions under which Ethiopians live will depend centrally on raising their incomes and consumption, which in turn depends on: - improving the productivity o f subsistence agriculture; - a shift into higher-valued crops; - massive creation o f off-farm employment; and - continued stability. To achieve many o f these depends on interventions outside the social sectors, most notably investments in infrastructure, and deepening o f the reform agenda. 1.64 But two interventions in the social sectors stand out as being essential to such a poverty-reducing growth strategy: (i) education, to improve current productivity and lay the basis for modem sector growth; and (ii) reducing the rate o f population growth, which i s central to increasing the poverty-reducing impact o f whatever growth does take place. 1.65 Secondly, even at high levels o f growth there will remain very large numbers o f absolute poor (see Chapter 6), and for many Ethiopians growth not will come fast enough to have a transformational effect in their lifetimes, nor even the lifetimes o f their children. Spending elsewhere in the social sectors can improve the quality o f life o f the 26 poor at existing income levels: obviously improved health care, providing nutrition services, closer and cleaner water, and safety net transfers fall in this category; but for these the relative importance has to be assessed against the investment needs for growth and raising consumption in the medium-term. 1.66 Given the competing demands for spending in other areas essential for poverty- reducing growth - especially infrastructure - a strategy needs to be pursued which both prioritizes within the social sectors, and minimizes trade-offs. One way o f doing this i s to focus on interventions that have more than one pay-off (for example, general education, which contributes to productivity gains, but also to fertility reduction, improved hygiene and n u t i t i o n behaviors; or family planning services, which have a macro-level poverty reduction impact, but also substantial maternal and child health benefits). Another i s to maximize the extent to which social sector spending can be used to simultaneously finance growth investments, for example using safety net transfers to create infrastructure under works-based employment programs. 1.67 Distributional Considerations. Another way o f prioritizing i s o n the basis o f the distributional impact o f social sector spending. The evidence suggests that spending o n the non-poor i s somewhat less an issue in Ethiopia than it may be elsewhere. The benefit- incidence analysis shows a moderately good distribution o f the benefits o f basic services across income classes. This i s in part due to the very large number o f very poor people; and also because the differences between poor and non-poor are not as stark as el~ewhere.’~ Nonetheless there are substantial geographical and urban-rural differences in indicators and in coverage. 1.68 Given the difficult geography and weak infrastructure, inequality in the physical distribution o f services i s inevitable; costs per beneficiary are higher in difficult-to-serve rural areas where the population i s much more dispersed. The question in the case o f Ethiopia i s whether the object o f social sector spending in the medium t e r m should be to close the gap - in terms o f reducing disparities - or to make progress where it i s feasible to get the most out o f scarce resources. We would argue that there are very large numbers o f ‘middle-poor’ (who would, it i s worth noting, be among the ultra-poor anywhere else in the world) and who live in modestly accessible rural and peri-urban areas, and that, although not ideal from an equity point o f view, it probably makes sense to focus incremental resources on these groups and areas first. Conclusions 1.69 Education. The target o f universal primary education by 2015 i s one that has been ascribed to under the EFAA and the MDGs, and remains justified, both for growth and to improve the quality o f life o f the poor. I t has to be recognized however that for the majority o f the population who will remain in subsistence agriculture in the medium-term the direct income effects will be limited, and 4 to 6 years o f schooling will probably be sufficient. The policy question facing government i s whether to aim for universal education through grade 8, or at a lower exit point, probably at grade 5 or 6. 53 This i s not surprising: when average incomes are so l o w (33 U S cents a day) there i s not as much room either for people to be far below the mean - they would be dead - nor far above, because there i s l i t t l e surplus to accumulate. 27 1.70 The importance o f expanding secondary education cannot be over-emphasized. No country has made the kind o f economic transformation Ethiopia needs to make with only 3 percent o f its population having secondary education. The expansion o f secondary capacity i s essential both to raise productivity, and to produce the pool o f people for much expanded professional and technical training. 1.71 The issue o f tertiary education i s somewhat more difficult. There i s n o doubt that university education needs to be expanded - a long-run target level o f enrolling 3 percent o f the university-aged population would s t i l l put Ethiopia well behind most other developing countries, and there do not appear to be significant distortions in current university spending. We would therefore argue that significantly increased university spending i s justified, and that this should not be seen as ‘competing’ for the same pool o f funds as primary and secondary education - all three need to be expanded. 1.72 Technical and vocational training can clearly contribute to long term growth - Ethiopia i s critically short o f mechanics, bakers, computer technicians, plumbers, and accountants - the question i s whether the current public system can effectively, and cost- effectively, produce them. Finding a model that can do so, probably with more private and employer-based training, i s a prerequisite for expanding spending. 1.73 Population programs are central to achieving development outcomes. If the population continues to grow at the current rate almost everything being attempted elsewhere in the development strategy will be o f limited consequence. The creation o f jobs, ability to raise incomes in agriculture, availability o f food, and the capacity to fund social services - will all be outstripped by the addition o f more than 3 million persons per year by 2015. Spending o n reproductive health interventions i s therefore central to any development strategy. I t has the added benefit o f being fairly low-cost, and contributing substantially to improved maternal and child health outcomes. Along with education, our view i s that this should be the highest priority for social sector spending, especially given the substantial unmet demand for family planning services. 1.74 Health and HIV/AIDS. The principal developmental (as opposed to welfare) impact o f health spending i s in reducing labour losses due to illness. I t i s difficult to quantify productivity losses due to poor health given the lack o f morbidity and time-use data. I t i s likely that the greatest impact will come from a few interventions: malaria control in the lowlands (which i s essential to laying the base for migration), immunization, diarrhoeal disease control, and improved hygiene. A nutrition program could also have a significant effect, given the long-term damage done to individuals’ intelligence and productive capacity by severe malnutrition in early childhood. 1.75 The direct economic impact o f HIV/AIDS i s not dramatic, according to simulations in the background paper; but (a) it i s possible that Ethiopia i s o n the cusp o f an increase in prevalence, given the concentration o f incidence in younger age groups; (b) the long-run costs o f having large numbers o f orphans -both economically, and to the social fabric - could be very high; and (c) the direct costs o f increased attrition among trained staff, if A I D S were to approach the levels o f elsewhere in sub-Saharan Africa, would be very high. Any expansion o f spending, however, has to be balanced with a careful assessment o f why funds have not been absorbed to date, and their impact on behaviour change. 28 1.76 Water supply. Urban water supply i s obviously essential for growth - the question i s one o f ensuring cost-recovery mechanisms are adequate. The productivity effects o f rural water supply consist mostly o f time savings to women, and the health improvements associated with improved hygiene, although the benefits are much reduced in the absence o f accompanying behaviour changes. In much o f Ethiopia there i s an added productivity effect because o f the widespread ownership o f livestock. However spending on k r a l water supplies only makes sense if they are going to be maintained in working order. 1.77 Safety Nets. Given the competing demands on public expenditure, we would argue for minimizing spending on pure transfers. An exception would be a national child nutrition program, for the reasons cited above. A significant scale o f transfers can still be achieved through capturing existing spending on road construction and maintenance, and other rural works, to provide low-wage employment o f the poor in the off-season. The question remains what i s to be done about those who remain unable to feed themselves each year in the absence o f transfers. A better designed national public works program should be able to cater for a large portion o f these; and the government has just launched a major initiative to try to reduce chronic food insecurity substantially within the next five years. 1.78 Within this framework, we would argue that the highest priorities, in terms o f contributing to a longer-run development strategy, would be education (at the primary, secondary, and tertiary levels), and a nationwide population program. The next highest priorities are almost certainly selected preventative health interventions - especially malaria, immunization, A I D S control, and child nutrition - and rural water supply, provided operations and maintenance issues can be resolved. With respect to safety nets, the challenge i s to develop a sustained national program that can tap other resources to finance transfers. 1.79 De facto the government has already revealed i t s preferences through existing budget allocations I t has substantially increased spending on education - especially higher education and TVE - while spending relatively l o w shares o f the budget on rural water supply and health. Expost these appear to have been rational choices, given the absorption capacity problems that effect both these sectors. The only adjustments we would argue strongly for are for a significant increase in spending o n mid-level general education, and for a population program. 29 2. EDUCATION EXPENDITURE IN ETHIOPIA A. INTRODUCTION 2.1 The object o f this part o f Public Expenditure Review i s to examine expenditure aspects o f education - recent changes in spending, the current composition o f public expenditure and i t s adequacy, and the projected costs o f different policy targets being considered. Broader issues o f sector strategy, and more specialized ones o f service delivery, which are equally critical to understanding the context for expenditure choices, are being addressed in a companion Education Country Status Report - under preparation. Chapter 2 describes trends in total education spending, and covers primary and secondary education in depth. Higher education i s addressed in Chapter 3. The conclusions are stark - that the primary system i s badly under-funded in the wake o f major enrollment increases; and that planning for future needs must confront major growth in the school-aged population, and greater attention to expansion o f the secondary school system. Major increases are needed in transfers to the Regions to adequately fund the system. System Background 2.2 The general education system in Ethiopia consists o f four levels: Primary schooling i s divided into a first cycle (grades 1-4) and a second cycle (grades 5-8) with a somewhat more specialized curriculum. Junior secondary (grades 9-10) i s expected to be the terminal level for most students entering secondary, and upper secondary (grades 11- 12) i s intended primarily for those students going on to university. 2.3 Recently government i s pursuing expansion o f a major in-school technical & vocational education (TVE) system, with one and two-year programs (known as 10+1 and 10+2) intended to absorb a large proportion o f those leaving school after grade 10. A small (by international standards) university system consists o f five universities and a number o f specialized colleges, with total enrollment o f about 35,000 full-time diploma and degree students. 2.4 The basic architecture o f the system in Ethiopian year 1994 (2001/02) was as follows: N o o f Students l a Schools Teachers Primary 8.1 million 12,089 126,882 Secondary 764,000 455 14,09 1 University 35,000 13 2,740 ai note about 5% o f primary and 2.4% o f secondary students are in non-government schools, many o f which however are quasi-public in that they receive public funding; university refers to full-time students in the public system, there are about the same number o f part-time students. 30 2.5 The Government o f Ethiopia has made a major effort to expand education in the past decade. Partially in response, enrollments have skyrocketed, especially at the primary level, where 3.6 million students have been added in the past 5 years.54Despite these gains, coverage i s s t i l l very low, the net enrollment ratio (that i s the proportion o f school-aged children in school) was only 52 percent at the primary level, and just over 7 percent at the secondary level in EFY 1994 (2001/02). The gender balance i s relatively good compared to many countries, with girls constituting about 40% o f both primary and secondary students; and in fact outnumbering boys in urban areas. - Table 2.1: Ethiopia Evolution of the Education System 1980-2002 (EY 1972-1994) 1980 1985 1990 1995 1996197 2000101 2001102 Number of Students Primary n.a. 2.8 million. 2.9 m. 3.8 m. 4.5 m. 7.4 m. 8.1 m Secondary n.a. 292,000 454,000 403,000 468,000 736,000 764,000 Gross Enrollment Ratio Primary 37% ma. 33% n.a. 26% 62% 61% Secondary 9% n.a. 14% n.a. 17% 17% 17% m o t e : definition o f primary education, and national population base, changed between 1980s and 1990s so figures from earlier years are not strictly comparable - although the broad trend i s accurately represented. 2.6 N o t surprisingly, as has happened in many other low-income countries in the recent past, quality has deteriorated following this rapid expansion. In particular, the student- teacher ratio has increased dramatically, from about 32: 1 in the early 1990s to 63: 1 today. I t i s difficult to judge the impact o f spending on educational outcomes in the absence o f standardized testing, but there seems little doubt that literacy levels have been slowly rising over the past decade (from about 29% o f the population in 1990 to 40% in 2000); and equally little doubt that average levels o f education remain very low. The Welfare Monitoring System survey (2000) found that 70.9% o f the population had n o schooling at all, and only 11.5% had completed primary school or higher. A number o f intermediate indicators (such as the dramatically increased student-teacher ratios) suggest that the system i s struggling to keep up with i t s rapid expansion; and many o f the new entrants are not staying in school for very long. Table 2.2: Selected Indicators of Educational Outputs and Outcomes 1995196 2001102 Grade 1 Drop-outs 28.5% 27.5% Average Grade 4-8 Repetition 12.8% 10.5% Share o f Girls in Primary School 38.0% 41.1% 2.7 Government’s broad strategy for the medium-term, as laid out in the SDPRP, i s to: (i) concentrate on expansion and quality improvements in primary education; (ii) place an emphasis on remote and under-served populations to improve equity; ( iii)encourage a greater role for private provision and community involvement, recognizing that government cannot provide all that i s required; (iv) to decentralize management and budgets to improve responsiveness; and (v) undertake a massive expansion o f tertiary and technical education, to build capacity for longer-term economic growth. The policy i s generally sound; 54 Enrollment data need to be treated with some caution: as in many countries, there i s a tendency for schools and principals to over-report - we do not sense this i s a major problem in Ethiopia, but there are many students who register at the beginning o f the year, and then show up only occasionally, and very large drop-outs during the year in grade 1. (see discussion later.) 31 although one can debate the relative emphasis o n secondary versus tertiary education; and o f capital versus recurrent spending - issues that take on added weight in the context o f the extreme resource-constraints facing the sector, and the shift to decentralization. B. PUBLIC SPENDING ON EDUCATION AN OVERVIEW OF TOTAL 2.8 Total planned public spending o n education in Ethiopia, including the tertiary level, was ETB 2.85 billion (US$ $333 million) in the 2001/02 (EY 1994) budget; representing 13.9 percent o f public expenditure, and 4.5 percent o f GDP.55 Although there have been significant nominal increases over the past two decades, these gains have largely been eroded by population growth. 2.9 Budget allocations for education have been increased substantially in the last few years in nominal terms (from ETB1.6 b. in 19961’97 to 2.85 b. in 2001/02). Following a major increase from just over 9 percent o f the budget in the late 1980s, the share o f education stabilized at about 14 percent since the mid-1990s. It has been increased to over 17% in the last two budgets, but complete data i s not yet available o n actual spending, and it appears the total has not been spent. Table 2.3: Ethiopia - Education as a Share of the Budget - Past 16 Years 1985/86 1990/91 1995/96 2000/01 2001/02 2002/03 Recurrent 13.0% 12.1 % 16.5% 14.4% 16.8% 18.7% Capital - 2.7% 2.7% - - 3.0 % 11. 4 % - 8.8% 15.7% Total 9.3% 9.7% 13.8% 13.4% 13.9% 17.7% 2.10 Due to population growth, however, combined with under-spending o f capital budgets, actual expenditure per capita has not increased as appreciably, neither over the past two decades, nor with the increases in budgets o f the past 5 years.56 Table 2.4: Total Education Spending US$ per Capita (Actual) 1996/97 1999/00 2000/01 2001/02 2002103 (est.) Recurrent $ 2.82 $ 2.66 $ 2.85 $ 3.18 $ 3.90 Capital $ 1.15 $ 0.57 $ 1.14 $ 0.97 $ 1.36 Total $ 3.96 $ 3.23 $ 3.99 $ 4.15 $ 5.26 2.1 1 With massive increase in enrollments, real recurrent spending per student has in fact fallen, by about 20 percent since 1996/97 at the primary level, and by 15 percent at the secondary level (see analysis in general education section below). 55 Data i s presented for 2001102, the most recent year for which detailed data i s available; in 2002103 and 03104 the s h f t to block grants to woredas and delays in formulating regional budgets have resulted in a situation in which a detailed breakdown o f expenditure i s not yet available. Also for actual expenditure, there i s a lag due to incomplete reporting o n aid expenditures in pre-actual budget results. Foreign aid accounts for about 15% o f total education budgets. The budget for education had increased to ETB 3.5 billion ($407 million) in 2002103. 56 Note however the data in table 2.4 probably under-state actual capital spending in the past two years, as reporting o n aid expenditures i s incomplete. More generally, all data in this Chapter draw o n official MoFED sources, and there i s a generalized problem o f under-reporting o f donor expenditures, discussed in Chapter 9. 32 Expenditure Composition 2.12 The largest share o f recurrent spending i s allocated to primary education; this i s due in part to the very l o w enrollments at secondary and tertiary levels; as well as to the government’s basic policy stance over the past decade o f assigning first priority to primary education (see fig.2.1 below). Figure 2.1: Composition of Education Recurrent Budget (2001102 EFY 1994) Teacher Training, General \ Curriculum, etc. 5% Secondary 2.13 There has been a shift in the composition o f recurrent expenditure over the past five years (see Table 2.5). Although all sub-sectors have grown, spending o n tertiary education, administration, and other support functions has grown proportionally substantially more. This to some extent reflects the fact that base levels o f spending were relatively lower; but it also reflects conscious policy choices by government to expand tertiary education, and to some extent the balance between Federal and Regional spending (see discussion below). Table 2.5: The Changing Composition of Recurrent Education Expenditure (Current ETB millions) 1996197 2001102 Avg.Annual Growth Primary 529 (52%) 734 (41%) 5% Secondary l a 218 (22%) 358 (20%) 9% Tertiary 127 (13%) 313 (18%) 18% General Administration 85 (8%) 193 (11%) 15% OtherlSupport to General Education /b 44 (4%) 98 (5%) 15% TVET IC 8 (1%) 92 (5%) a/ Breakdown between primary and secondary i s not always reliable, as teacher salaries classified b y qualification rather than school level. b/ Includes teacher training, NEA, curriculum development, etc. c l Some o f TVET recurrent spending included in secondary. Note - actual increase less than appears, due to inclusion o f pensions in sectoral budgets post-1 999 2.14 Table 2.6 below shows the vast differences in spending per student at the different levels o f education. These are driven in part by the inevitably higher unit costs o f providing slulled teaching staff and specialized equipment at the university level, and by the very high costs o f consumables and equipment in secondary technical training (discussed in chapter 3). In addition, recurrent spending per student at the primary and secondary level i s expected to have increased somewhat in the last year as the result o f 33 recent budget increases; and the estimates are rough, due to the difficulty o f dividing out primary and secondary teacher salaries. Nonetheless, they illustrate some fairly substantial differences, with the cost to the government o f $260 per year to educate a student in TVET and $1,381 per year at the university level; as opposed to $11 per year in primary school, and $55 per year in general secondary education. ETB per Student (US$ equivalent) Primary 95 $ 11.05 Secondary 477 $ 55.47 Technical Education 2,252 $ 261.86 University 11,879 $ 1,381.28 2.15 As a share o f GDP and o f the budget education spending in Ethiopia i s comparable to other very-low-income countries (see Table 2.7); however because the denominator i s so l o w (at $100 per capita) absolute levels o f spending have been very low. Although with recent budget increases, per capita recurrent spending o n education in Ethiopia (an estimated $3.90 in 2002/03) i s now more comparable with that in other low-income African countries. Table 2.1: Comparative Recurrent Education Spending - Ethiopia and Other Very-Low Income Countries (1999) YOof Public Spending /a Yo of GDP US$ per capita Nepal 11.1 2.6 $5.46 Bangladesh 9.6 3 .O $10.50 Ethiopia 15.0 2.7 $2.70 Chad 20.9 1.7 $3.91 Malawi 19.8 3.6 $7.20 Tanzania 16.4 1.8 $3.96 a/ Proportion o f current government revenues, excluding grants. Sources: Table A2.1 “Financing education for All: Country Simulations for 33 African Countries. World Bank, 2002; combined with GDP per capita from World Development Indicators 2000. All data for 1999, except Bangladesh and Chad which are for 2000. Federal-Regional Division of spending 2.16 Responsibility for education spending i s split between the federal government (responsible for higher education and overall policy matters) and the Regional governments (responsible for delivering primary and secondary education and some TVET). The division o f spending has historically been about 25 percent Federal and 75 percent Regional; although the share o f actual spending i s closer to 30 percent Federal, 70 percent regional, due to under-spending on the capital budget in the regions.57 57Capital under-spending i s discussed in more depth in section D. Note these shares do not include some o f own-revenue spending by the universities, which would increase the share o f Federal level spending somewhat. 34 Figure 2.2: Federal -Regional Breakdown of Education Budget - 2003104 Federal - Regional - Capital Recurrent 14% Regional - Recurrent 52% 2.17 As Table 2.8 shows, the federal share has been growing more rapidly than the Regional in recent years. The share o f the federal government in total education spending has increased from about 25% in the late 1990s to 34% in the current budget, largely as a result o f (a) the increased emphasis on tertiary education and TVE; and (b) the fact that regional spending i s constrained by the absolute size o f the regional transfers from the federal level, out o f which they must fund education. In fact the changing division o f actual spending has been biased even more heavily towards the federal level, as almost all under-expenditure takes place on the aid-funded part o f the capital budget, which i s concentrated on primary and secondary education, which are regional responsibilities. Table 2.8: Budgeted Expenditure on Education: Federal and Regional Levels (Current ETB millions) 1996197 199912000 2000101 2001102 2002103 2003104 Federal Recurrent 147 273 220 395 501 587 Capital 198 - 189 181 - 324 - 538 762 Total 345 (22%) 462 401 719 1039 1349 (34%) Regional Recurrent 869 1096 1266 1447 1936 2125 Capital - 333 385 - 716 - 652 543 -547 Total 1202 (78%) 1481 1982 2099 2479 2672 (66%) c. ON PRIMARY AND SECONDARY EDUCATION PUBLIC EXPENDITURE Trends and Composition 2.18 Spending o n general education, meaning primary and secondary levels, has been increased appreciably over the past 5 years (Fig 2.3 and table 2.9). 35 Figure 2.3: Evolution of Recurrent Spending on General Education (1996/97-2001/02 - EFY 1989-1994) 1600 I 1 1400 8 .3 1200 .g 3 4 1000 e - w 800 600 400 200 0 1996197 1997198 1998199 1999l2000 2000101 2001102 0 Primary H Secondary 0 Admin, curriculum, teacher training, etc. 1 Table 2.9: General Education Recurrent Budget (Current ETB Millions) 1996/97 1997/98 1998199 1999/2000 2000/01 2001102 Primary 529 581 628 643 687 734 Secondary 218 23 1 256 244 337 358 General Administration 77 99 147 147 145 176 NEA, curriculum, etc. 18 17 20 21 48 53 Teacher Training - 26 - 28 - 38 - 24 - 33 45 Total \a 868 956 1089 1079 1250 1366 al Includes regionalgeneral admin; federal plus regional teacher training, exams, etc 2.19 However, as the result o f rapid increases in enrollments, there has been a substantial decline in real spending per student, as shown in Table 2.10 below. Average spending in the 2001/02 budget was about 95 birr per student (US$11) at the primary level, and 477 birr (US$56) per student at the secondary level. Table 2.10: Recurrent Budget Per Student 1997/98-2001/02 (Current ETB) 1997/98 1998/99 1999/2000 2000/01 2001/02 Primary 122 117 105 98 95 Secondarv 502 502 43 6 468 477 Average la 174 175 154 151 150 a/ Note: Total spending on direct delivery o f general education services, averaged over all primary and secondary students; the adjustment compensates for possible misclassification o f salary expenditures between the primary and secondary levels. 36 Composition of General Education Spending 2.20 The balance o f recurrent spending within general education, shown in Figure 1.4, appears broadly consistent with the government’s stated policy priorities. Figure 2.4: Composition of Recurrent Spending on General Education (2001/02 Budget) Primary 54% Includes regional level general admin, teachers training colleges, and spending on curriculum development and national examinations. Excludes federal-level general admin. Expenditures. 2.21 Salaries dominate overwhelmingly at the primary level. In 2001/02, 97% o f the regional budgets for primary education was for payment o f salaries. At the secondary level, the predominance o f salaries i s somewhat less; representing 81 percent o f secondary education budgets. The subsequent sections discuss the adequacy o f expenditures, and the critical need for non-wage recurrent funding. Table 2.11: Composition o f Recurrent spending on General Education (2001/02 Budget) /a (Current Birr millions) Primary Secondary General Other Direct Education Education Admin. Ed’n Costs Total Wages & Salaries 709 292 108 23 1132 Operations & Mtc. 7 32 26 9 74 o/w Printing 3 10 7 1 21 Utilities 1 1 2 2 6 Bldg. Mtc. 3 2 1 1 7 Materials & Supplies 17 27 11 17 72 o/w Ed’n Materials 14 23 3 4 43 Other /b 1 8 28 43 80 Total 734 359 173 92 1358 Notes: a/ Total across all regions; some data errors occur due to incomplete or inaccurate reporting or classification o f expenditures, but in aggregate these do not, in the judgment o f the mission, detract from the accuracy of the overall picture o f spending presented. The main exception i s the classification o f salary expenditures between primary and secondary levels. In many cases teachers’ salaries are classified by the level o f their qualifications, rather than the level at which they actually teach, so there i s a bias toward over-reporting primary salaries and under-reporting secondary salaries. b/ Mostly consists o f ‘retained revenue’ in the Regional general administration budget, and of NEA (under Direct Education Costs). 2.22 The situation with respect to non-wage recurrent spending at the primary level has worsened in the past 5 years with the expansion o f the system, but it has always been a problem. After a period o f decline, funding for operations and maintenance and supplies have in fact been increased in the last two years (see table 2.12). Note however that the 37 increases have been almost all at the secondary level and in general administration; at the primary level budgets for operations and maintenance have in fact declined. This may be due to a combination of: (i) the squeeze o n non-wage funding due to expansion o f the teacher cadre to keep pace with massive enrollment increases; (ii) changes in classification (for example maintenance may be budgeted under general administration in some regions), and, ( iiipossibly due to reductions in recurrent budgets because o f the substitution o f ) donor-funded inputs - especially textbooks. Table 2.12: Levels of Spending on Some K e y Non-Wage Recurrent Items (General Education - current ETB millions) 96/97 97/98 98/99 99/00 00/01 2001l02 O&M 37 40 48 41 73 79 (o/w Primary) 16 13 14 4 7 7 Materials & Supplies 47 58 81 51 74 132 (o/w Primary) 16 17 17 13 16 17 Regional variations 2.23 Education spending averages just over a third o f total recurrent spending in the largest Regions, although it varies appreciably from Region to Region: from a l o w o f around 23 percent in Benshangul and Gambella to a high o f 44 percent in Oromiya (2001/02 budget). As a general point, it should be noted that in all aspects o f education spending there i s substantial variation between the regions, but that the levels and composition o f expenditure are fairly consistent in the four main regions, which account for the vast majority o f the population (86 percent), o f school enrollments (90 percent), and o f regional-level public expenditure on education (80 percent). Table 2.13 Regional Variation in General Education Spending (Selected Regions 2001/02) (,OOOS) (ETB millions) Oromiya 23.7 million 2,970 225 554 43.6% 173 Ahmara 17.2 m. 1,964 120 336 42.5% 161 Tigray 3.9 m. 606 72 102 34.8% 150 SNNP 13.3 m. 1,784 115 171 38.5% 90 Addis 2.6 m 467 122 138 28.2% 234 Afar * 1.3 m. 31 3 25 18.4% 735 Somali * 3.9 m. 107 5 26 14.6% 232 Benshangul 0.6 m. 102 6 24 22.6% 222 Gambella 0.2 m. - 42 - 2 -19 23.2% & 2 All Regional 67.2 m. 8,144 685 1437 36.8% 163 al source Education Statistics Annual Abstract; p.25 and p.32; Note * data may be incomplete for these regions. 2.24 As shown, average recurrent expenditure per student varies substantially across regions (Although the most extreme outliers, such as Afar and Gambella, are explained by the l o w denominator: very l o w absolute numbers o f students in these Regions result in high per-student expenditure levels compared to the rest o f the country.) In the four largest regions recurrent expenditure on primary education averages about 90 Birr per student 38 (US$ 10.50 ).’* There i s greater variation in spending at the secondary level; ranging from B 3 10 per student (Tigray), to Birr 746 (Oromiya) (again, leaving aside the outliers in very small regions). These differentials are accounted for almost entirely by differences in the student-teacher ratio, which varies widely at the secondary level, even among the large regions, from 44: 1 in Ahmara, to 85: 1 in Tigray. Issues of Adequacy, Appropriateness, and Effectiveness of Spending Adequacy of Expenditure 2.25 As noted earlier, non-wage recurrent spending i s critically l o w at all levels. Some key recurrent inputs are dramatically under-funded. Building maintenance averages 240 Birr per school at the primary level (US$30 p.a.); and 4,290 Birr ($500) per school at the secondary level. Furthermore, budgets for building maintenance have been reduced substantially in recent years; while the stock o f buildings as increased (see table 2.13). Field visits consistently confirm that there i s insufficient maintenance o f existing stock. As the recent Annual Review Mission (ARM) noted: “ i n general, facilities in existing buildings urgently need repair. Maintenance o f school facilities appears almost absent.” Figure 2.5: Average Maintenance Budget - Past 5 Years 13000 800 700 12500 600 12000 500 -No.of Schools 11500 400 +Average Mtc. Budget 300 1lo00 200 10500 100 10000 0 I 97/98 98/99 99/00 00IO1 2001/02 I 2.26 Textbooks: I t i s almost impossible from existing budget data to t e l l how much i s spent on textbooks in Ethiopia each year. Classification i s different across regions”, some spending i s included in the capital budget, and information o n donor funding o f textbooks i s incomplete. Empirically, however, textbook availability i s clearly s t i l l a problem. 58 I t i s not clear why per student expenditure in SNN’PR i s so much less than the average - it suggests there may be an error in reporting or classification. 59 Some regions show large amounts as printing charges, some as education material - textbooks are not distinguished within either. 39 2.27 The situation has improved from 1996f97, when the average ratio was 1 book to every 5 students.60 The Facilities Survey will shed further light on actual availability, but field visits reveal ratios o f between 1:3 and 15 in many areas. (ARM field report). Problems relate not just to the absolute numbers procured, but also to the longevity o f books - printing on low-grade paper results in books deteriorating quickly - and bottlenecks in distribution, as well as in-school management o f textbook stocks. 2.28 The Unit Cost Study estimates the cost o f a full set textbooks at between 12 and 18 birr at the primary levels, and at 60 and 95 birr per student at the lower and upper secondary levels respectively. On this basis, to provide a complete set o f textbooks to each student currently enrolled would cost approximately ETB 175 million p.a. (equivalent to U S $ 21 million; o f which $ 14 million would be at the primary level, and $7 million at secondary). As far as one can ascertain from budget data , current spending i s about ETB 64 million (although this includes spending on teachers guides and stationary by Regional education bureaux; and likely excludes significant amounts o f donor-funded textbooks.) 2.29 We do not know precisely what an ‘adequate’ level o f non-wage recurrent expenditure would be; but the Unit Cost Study (UCS) arrives at estimated requirements o f 10 Birr per student per year at the grade 1-4 level; 15 birr for grades 5-8; 20 birr (grades 9- 10) and 50 birr (grades 11-12), and considers that these might be on the l o w side. T o quote the UCS: “it is estimated that it would be practically impossible to manage a school with operating costs less than 75 percent o f the minimum requirements defined’). We do know that providing even small amounts o f non-wage recurrent funding - if they are assured and predictable - can have quite a significant effect o n school performance.6’ I nthe projections section, we illustrate the costs o f providing different levels o f funding to all schools. 2.30 The levels cited in the UCS correspond to a ‘school block grant’ outlined in the new education financing strategy ( “Education, Organization, Community Participation & Finance Guidelines (MoE, 2002)). The strategy proposes that such grants be made to all schools o n the basis of enrollments, to be supplemented by performance-based funding from woreda offices; however, none o f the Regions have actually adopted the strategy yet, and there i s in most cases substantial uncertainty about how, and when, it will be implemented in practice. Our view i s that such a guaranteed transfer, even if at lower level, could be extremely useful. At the moment however it would be impossible for most woredas to accommodate these new transfers within the fixed block grant they receive from Regional government, which are entirely consumed by existing obligations, especially salaries. Effectiveness and Efficiency 2.3 1 In assessing the link between educational expenditure and educational outcomes, it needs to be recognized that the relationship i s not straightforward. Among other things, analytical work in different countries has shown that many aspects o f outcomes depend primarily on the attributes o f the child entering the system - above all the level o f his or her parents’ education - so assigning the impact o f public spending i s difficult, especially in an 6o According to the ESDP I1PAP (p.16) 20 million textbooks were estimated to be in circulation in 200/01, but this implies 2 textbooks per student, and does not seem credible; other sources suggest a ratio textbook-student ratio o f 1~2.5. 61 See for example, Primary education in India; World Bank, 1997. 40 environment where we do not have good data to control for household and child characteristics 62. However, we do know some factors determined by public spending are fundamental - among these the ratio o f students to teachers, and the quality o f teachers, are paramount. 2.32 Teacher-related issues. In Ethiopia the student-teacher ratio has deteriorated dramatically in the past 5 years, as the result o f rapid expansion o f the system. The table below illustrates the magnitude o f this collapse: Table 2.14: Evolution of Student: Teacher Ratios 1992193- - 2001102 1992193 1994195 1997198 199912000 2001102 Primary la 25:l 32: 1 47: 1 56: 1 63:l Secondary 33: 1 33:l 38:l 43:l 49: 1 a/ Reported for grades 1-6 prior to 1995/96 2.33 The problem does not appear to be (as it often i s elsewhere) primarily one of teacher allocation. Although there are undoubtedly problems of attracting and retaining teachers in rural areas, o f occasional over-concentration o f teachers in more attractive urban postings, and o f variance in student-teacher ratios between areas; the over-riding problem i s one o f insufficient numbers of teachers relative to the number o f students. Nonetheless, geographical imbalances across regions cannot be ignored, not least because they suggest different problems o f service delivery and therefore different calls for policy. The large regions (Amhara, Oromiya, SNNP, Tigray) and the isolated regions (Afar, Somali) have radically different pupil-teacher ratios, very high for the former and very l o w for the latter. The high student teacher ratios in the bigger regions are an unintended consequence o f one form o f success - enrollment promotion - while the far-flung states show the failure o f the same. While the high PTR regions may suffer from the problem o f absolute shortage, the l o w PTR regions do not, and may require demand-side interventions to improve coverage. 2.34 Both field visits and school-level data confirm the magnitude o f the problem. Although the average reported student-teacher ratio in 2001102 was 63:1, in reality it i s not uncommon to come across classes o f 100, 120, or more students. Similarly, detailed school-level data for SNNPR show a range o f student-teacher ratios, but in many cases with over 100, and in some cases over 200, students per teacher. (Conversely, there are some schools where the ratio i s in the 20s, but these are usually schools with very few students, presumably in remote rural areas.)63 2.35 Effective teaching-learning cannot take place in an environment where there are 80, 100, or more students per class. The greatest single intervention government can make with respect to effectiveness i s almost certainly lowering the student-teacher ratio. A first step i s to design a strategy to alleviate pressure o n teachers with sections o f 70 students or more. There are also likely some gains possible in reducing the relatively early 62See for example: Jencks eta1 ; “Do Teachers Make a Difference?” USHEW . 63 ‘Educational Statistics Annual Abstract SNNF’R 1994 E.C. (2001/02)’. In our view the data need to be treated with some caution, because enrollment data may overstate the number o f children actually attending school, due to high dropouts; but nonetheless, that the overall scale o f the problem conveyed by the data i s legitimate. 41 specialization o f subjects and teachers (at grades 5-S), which might reduce the number o f teachers needed, and in the adjustment o f hours worked per teacher, but in terms o f orders- of-magnitude, these are not sufficient to outweigh the overwhelming needs placed on the system by the teacher deficit. I n aggregate, about 44,000 additional teachers would be needed today to restore student-teacher ratios to the 1997/98 levels, just in primary schools. 2.36 Teacher training and supervision are other factors which are clearly within government’s control. The rapid expansion has resulted in the hiring o f large numbers o f untrained teachers. Government appears to be tackling the backlog in a fairly pragmatic way, although the capacity to train teachers will be strained with the requirements o f the coming decade (see projections in section F). School supervision i s variable, especially in outlying areas, and will be stretched further with the limited capacity o f woreda offices, both in terms o f recurrent budgets and experienced school inspectors. 2.37 Average teacher salaries are about ETB 7,000 p.a. at the primary level, and 15,000 p.a. at the secondary level. In relation to per capita GDP (a commonly used measure for assessing teacher salary levels) these are relatively high compared to elsewhere, but as with so many such indicators, the very l o w denominator in the case o f Ethiopia makes proportional comparisons less meaningful. In absolute terms, salaries compare roughly with those in other very poor countries (table 2.15). By comparison, better-off urban private schools pay teachers about 23,000 Birr p.a., and the lower-end ones pay about 10,000 Birr p.a ; suggesting government salaries are within the market range. Still, there are scope for savings (explored further in para 2.103) : for example, communities are already hiring 10 and 12 grade completers without full teaching qualifications at substantially lower salaries than certified teachers. Table 2.15: Comparative Primary Teacher Salaries - Selected Countries Per capita Multiple Implicit GDP (US$) of GDP Salary Level (US$) Bangladesh $350 5.3 $1,850 Burkina Faso $240 8.0 $1,950 Ethopia $100 8.1 $ 806 Malawi $200 4.0 $ 800 Tanzania $220 3.6 $ 792 2.38 The efficiency o f public spending on education depends on two factors: h o w well money spent i s translated into outputs, and o n the flow o f students through the system. With respect to the first, the evidence suggests that by and large the service delivery system works well in Ethiopia compared to many other low-income countries. W h i l e there are problems o f textbooks and buildings, and - as graphically illustrated above - aggregate funding and numbers o f teachers have not been able to keep up with the rapid expansion, nonetheless teachers show up at work and teach. One does not see, for example, the widespread phenomenon o f teacher absenteeism or non-performance that are common elsewhere; nor do we have the sense that there i s major leakage or mis-use o f funds in the system. 2.39 Drop-out rates are extremely high; o f the students entering the school system, only 38 percent remain to grade 5. W e h o w that l i t t l e effective education i s taking place for the 55 percent who leave before the end o f the third year (and none at all for the 27 percent 42 who drop out in grade 1 each year). Keeping more children in school i s obviously one way o f using existing spending to improve outcomes. 64Why children leave i s not always within the control o f govemment, but to the extent that it i s related to ineffective teaching and learning, due to excessively over-crowded schools and under-trained teachers, it i s within the government’s power to influence retention via increased spending. The other intervention i s development o f informal schools, and those especially adapted to pastoral and nomadic peoples, both o f which the government i s pursuing. What Would I t Cost to Achieve Some Fundamental Policy Objectives? 2.40 This section illustrates the broad costs associated with achieving some immediate policy outcomes. The longer-run costs associated with system expansion, and future budget requirements are dealt with in more depth in Part F. To Improve Quality 2.41 T o improve quality at current levels o f coverage we examine the costs of: ( i) providing adequate non-wage recurrent inputs (supplies, maintenance, and supervision); ( iio f providing sufficient textbooks; and then ( ) iii) o f lowering the student-teacher ratio. 2.42 The unit cost study (UCS) estimates the approximate annual cost per student o f providing basic inputs as follows: Non-wage Recurrent Costs Textbooks Primary Gr. 1-4 10 Birr 12 Birr per student Gr. 5-8 15 Birr 18 Birr Secondary Gr. 9-10 20 Birr 60 Birr Gr. 11-12 50 Birr 95 Birr 0 T o achieve minimum improvement in the availability o f non-wage recurrent inputs, at current enrollment levels, we estimate would cost something like 200 million Birr per annum ($23 million) ~ . a . ~ ’ .(This would not provide a full set o f textbooks, but would represent a significant increase over current levels). This compares with current non-wage recurrent spending at the Regional level o f 150 million Birr annually ($17 million; 2001/02 budget, including general administrative overheads, but excluding funds spent by donors under the capital budget on textbooks or other inputs). 0 T o provide afull set of textbooks for all students currently enrolled, based on the UCS estimates, would cost something like 178 million Birr annually (about U S $ 21 million; $14 million at the primary level, and $7 million at the secondary level). I t i s almost impossible from existing data to determine how much i s currently spent on textbooks; but clearly these levels o f spending are not overwhelming, when compared to the total education budget or levels o f aid. 64 Although in pure efficiency terms, i t i s not clear that there i s a l o t o f scope for increased capacity utilization in the system, given the high levels o f over-crowding at the moment. ‘’ Talung somewhat arbitrary minimum levels o f spending (for non-wage recurrent spending including textbooks) o f Birr 15 per student (GR.l-4); 25 Birr (Gr.5-8); 65 Birr (Gr.9-10) and 100 Birr (Gr.11-12); using an average o f B.18 per student in primary school, and estimated enrollment o f 8.1 million; and an average o f B.75 per student at secondary level, and enrollment o f 764,000. 43 2.43 Thus providing minimum adequate levels o f non-wage recurrent funding, and adequate text books, turn out to be relatively affordable. The real fiscal challenge in improving quality i s in reducing the student teacher ratios to more manageable levels. To Restore Adequate Student-Teacher Ratios T o restore student-teacher ratios to a moderately acceptable level on the basis o f current enrollments would cost in the order o f an additional 334 million Birr p.a. (US$ 39 million annually); the vast majority o f which i s required at the primary level. This compares with a current teacher salary bill o f Birr 1 billion annually ; so would represent a 33 percent increase in the salary bill. Clearly, to the extent that these teachers can be employed at a lower wage rate - for example through community engagement, or reduced standards or qualifications - the total costs will be lower; but the sheer numbers involved mean that expenditure requirements are going to be large under any scenario (see discussion in part F.) To Keep Up with Population Growth Just to keep up with population growth - with absolutely no improvement in enrollment ratios, and only modest improvements in student-teacher ratios 66 - would s t i l l require a real increase o f about 60 percent in recurrent spending on education over the next decade (an additional ETB 800 million p.a.); and capital investment o f about ETB 1 billion annually. To Raise Enrollment Ratios 0 T o reach Universal Primary Education by 2015, and achieve only a modest increase in enrollment rates at the secondary level over a longer time frame (to 15 percent GER in grades 11-12, and 25 percent in grades 9-10) would cost about ETB 3.9 billion ($453 million) p.a. in recurrent spending by 2015 (three times current spending, in real terms), and about another ETB 2.0 billion ($232 m.) on average annually in capital investments. Detailed projections for a number o f scenarios are presented in section F. D. CAPITAL EXPENDITURE ON EDUCATION6’ Trends and Composition 2.44 Capital spending has increased dramatically over the past two decades, both in absolute terms, and as a share o f total education spending. This i s due to the increased emphasis the government has placed o n expanding access to education, as well as the relatively expensive increase in providing higher education, and to the shift o f donor resources into the education sector. (It should be noted that all donor-funded activities are classified under the capital budget, as a consequence i t includes some spending o n inputs 66 Restoring the S:T ratio to 60: 1 at the primary level ,and to 45: 1 at higher levels. 67 The discussion in this section covers aggregate capital spending o n all levels o f education. Note in addition to spending through the Ministry o f Education and the Regions, there has been substantial expenditure in the last two years through the Ministry o f Capacity Building, and o n technical training in agriculture, which i s not included in these estimates. 44 such as textbooks and training, which can be seen as capital initially, but are more properly recurrent expenditures. Unfortunately from the available classification o f accounts it i s impossible to break these out from truly capital expenditures.) Increasing Share of Capital in the Education Budget - Selected Years 1985186 1990191 1995196 2000101 2001102 2003104 10% 8% 29% 28% 23% 33% 2.45 Federal-Regional Split: The federal share o f capital spending (about 58 percent) i s higher than in the recurrent budget (about 20 percent), and higher than would be expected given the relative division o f responsibilities for delivering the different levels o f education. This i s explained in part by the higher costs o f tertiary facilities; but also suggests a tendency to ‘hold back’ funding in the budget process by f i r s t assigning it to priority federal investments. There also appears to be a substitution taking place to some extent, with aid-funding, which i s concentrated at the primary and secondary levels in the Regions, freeing up domestic funding, which i s then being used to finance federal, tertiary- level, investments (see table 2.16 and figure 2.6). Table 2.16: Evolution of Education of Federal and Regional Capital Budgets (Current ETB millions) 1997198 1998199 199912000 2000101 2001102 2002103 2003104 Federal 149(29%) 284 189 181 324 538 762(58%) Regional 367(71%) 400 - 385 - 716 - 652 - 543 547(42%) Total 516 684 574 897 976 1081 1309 ~~~ ~~~~ Figure 2.6: Federal and Regional Capital Budgets for Education 2001-2003104 900 800 700 ’’ v) E 2 600 500 2 400 300 5 200 100 0 2000/01 200 1102 2002103 2003104 45 2.46 The chart below shows the sectoral composition o f capital spending over the last three years for which actuals are available. Note that these almost certainly over-state the share o f higher education, because o f under-reporting o f donor spending for primary and secondary education. Figure 2.7: Compositiono f Capital Spending in Education 1996/97 - 1999/00 Primary Hiaher 27% " - Education 44% Secondary 11% Other Teacher 8% Curriculum 10% [Note: this chart based on cumulative reported -expenditures 1996197-199/2000] The Role o f Aid - Domestic vs. Externally-FinancedCapital Spending 2.47 Foreign aid financed about 38 percent o f capital spending in the education sector over the most recent period for which data i s available (and about 11 percent o f total education spending). A note o f caution i s in order here, estimates throughout this section are based o n budget data; significant amounts o f aid are not recorded in the budget; at the same time, there i s substantial under-spending in the aid-funded part o f the capital budget, while the domestically-financed part o f the budget i s generally spent, and reported on, so budgets only imperfectly represent actual spending. Table 2.17: BudgetedCapital Expenditure on Education - Cumulative 1996/97-2000/01 (ETB Millions) (US% millions) Aid-Funded 1111 $146 Domestically Financed 1780 $244 Total 2891 $390 2.48 O f the $146 million in aid resources accounted for in the budget during this period, about $ 113 million consisted o f concessional loans, and about $33 million o f grants. The most recent available figures for actual spending cover the period through 2000/01; a period in which aid levels were suppressed due to the withdrawal o f some donor support during the war years. More recent aid levels are almost certainly higher, although the level o f domestically-financed capital spending has also increased substantially. 2.49 Over the past two years government has been substantially increasing expenditure o f i t s own resources on capital investments in education (see table 2.18). There has been almost a tripling o f the allocation to the federal budget for the expansion o f TVET and universities (from ETB 123 million in 2001 to 316 million in the 2002/03 budget), and a substantial increase in capital spending through the Ministry o f Capacity Building - much o f it for upgrading tertiary-level institutions. T o some extent there i s a substitution taking 46 place - the inflow o f donor financing for capital in primary and secondary i s releasing domestic funds to finance higher education investments at the federal level. Table 2.18: Budgeted Education Capital Expenditure By Level o f Government (current ETB millions) Federal 1999/2000 2000/01 2001/02 20021’03 Domestic 149 121 194 316 Aid 40 59 131 22 1 Regional Domestic 106 215 171 198 Aid 126 413 485 345 2.50 So how much aid does Ethiopia receive in the education sector? I t i s difficult to say with certainty. Flows included in the budget undoubtedly represent an under-statement o f the amount o f aid actually received, while those based on donor pledges equally certainly represent an over-statement. 2.5 1 The UNDP Development Cooperation Report for 2002 estimates an average o f $57 million equivalent p.a. in assistance went to the education sector over the years 1997-2001, and a level o f $109 million (ETB 937 million equivalent) in 2002; representing about 8 percent o f total aid inflows. Budget documents show aid financing o f about ETB 470 million ($5 8 m.) in 2000/0 1, the last year for which a reliable breakdown in available. 2.52 The donor mapping exercise by the education donors’ group and the Ministry o f Education estimates a total o f 3.4 billion birr over the next three years (EY1995-97). This would represent an average o f $128 million annually; equivalent to 40 percent o f public spending o n education. But this estimate i s based o n a combination o f pledges and commitments, not all o f which will materialize as disbursements; and more importantly, includes large amounts o f technical assistance and other forms o f project support, which while useful, do not substitute for direct expenditure on education service delivery by the government.68 Utilization o f the Capital Budget 2.53 The capital budget has been consistently under-spent, by significant amounts in some years (Fig. 2.19). This i s particularly true in years when it has been expanded rapidly, suggesting - not surprisingly - that the capacity to absorb sudden increases in capital spending i s limited. However, much o f the apparent under-spending i s on the aid- funded part o f the capital budget; and it i s difficult to t e l l how much o f this i s due to under-reporting o f spending (combined with excessively optimistic forecasts o f planned donor spending), as opposed to problems in actually implementing the program. 6a I t also includes such things as commodity aid from WFP for school feeding; capacity building, and training operations. 47 Table 2.19: Actual vs. Budgeted Education Capital Expenditure (Current Birr millions) 1996197 1997198 1998199 199912000 20001102 2002103 Capital Budget 530 487 66 1 419 900 987 Actual Expenditure 420 - 396 - 44 1 - 242 606 _. - 540 Actual as % o f Budget 79% 81% 67% 58% 67% 55% 2.54 More generally, under-spending o n the capital budget has always been a problem, for several reasons, including : (i) procurement bottlenecks; (ii) difficulty in complying with donor requirements for administering funds and contracting; ( iii)unpredictability, and late notification and release o f o f capital funding (as a result o f which works cannot be committed to in time to implement them within the budget year); (iv) contracting delays and insufficient contractor capacity; (v) problems o f transport and remoteness o f sites, and (vi) seasonality, which severely hinders the execution o f works during the rainy season. 2.55 The majority o f under-spending, however, has been on the aid budget. Historical data suggest that the while on average 82 percent o f the domestically-financed part o f the capital budget has been spent, only about 30 percent o f the projected aid-funded budget was spent, during the period 1996197-2000101. As noted elsewhere, this i s to some extent a forecasting problem: both donors and the Ministries are overly optimistic in projecting how much will actually be spent, so amounts included in the budget are over-estimates. Some o f the apparent divergence from budgets i s also due to reporting problems. The Regions, Bureaux, and MoFED do not have access to reliable data o n how much has actually been spent o n aid-funded projects, and reporting on actual aid expenditures i s substantially delayed. 2.56 More generally, the federal-level capital budget has typically been spent more completely than have Regional level budgets, and as a consequence actual capital spending has been more biased towards tertiary education than the budget numbers would suggest. (This i s also a corollary o f the aid-funded implementation problem referred to above: since aid tends to finance primary and secondary education in the Regions, where there i s under- spending, whereas domestic resources are concentrated more on capital investments for higher education at the federal level). The table below illustrates reported capital budget execution by educational level. Table 2.20: Proportion o f Education Capital Budget Actually Spent - b y Sub-Sector (Average 1996197-2OOOlO1) Budget Actual Actually Spent (Current ETB millions) as YOof Budget Primary 611 405 66.3% Secondary 195 172 88.2% Teacher Education 114 82 71.9% Curriculum Dev’t 132 71 53.8% Other Services 306 118 38.6% Higher Education 730 65 1 89.2% Government’s Short-term Plans 2.57 The government’s plans for the next few years include a continued expansion o f primary and secondary schooling, as laid out in the SDPRP and phase I1o f the Education 48 Sector Development Program; a new initiative to link secondary schools by satellite for video-teaching (SchoolNet), and a major expansion o f technical education and university capacity, discussed in Chapter 3. ESDP I1 sets targets for -the next three years o f the education program (ESDP 1 1Program Action Plan; Ministry o f Education, 2002). The physical targets ,while ambitious, are broadly consistent with putting Ethiopia on a track to reach the EFAA goals. The proposed expenditure increases, in the short term, however, are probably not feasible - with planned expenditures for the sector as a whole (including higher education) o f ETB 5 billion p.a. over the next few years, compared with current spending o f about ETB 3.5 billion. While such increases are almost certainly required over the medium term (see section F below), they are unlikely to be realistic in the short term. 2.58 The School Net initiative has been added since the ESDP program was formulated, and proposes an ambitious program o f putting satellite-linked television-broadcast teaching in secondary schools to supplement teachers. Costs o f the program are not known with certainty, but various review reports mention estimates between $100 million (ETB 860 million) and $210 million (ETB 1.8 billion)69. The proposal i s being reviewed, and while it has merit, there are a number o f concerns: one i s that the proposed pace o f implementation may exceed capacity to execute it; the second i s that such a system relies on electricity and the capacity to support complex technical systems in rural areas; the third i s that the recurrent costs (including systems upkeep and replacement) would need to be borne by the education system. Reviews o f the program have suggested that the approach should be tested and phased, before committing to large-scale expenditures. E. ISSUES OTHEREXPENDITURE-RELATED Equity and the Link with Poverty - Who Benefits from Education Spending? 2.59 There are a number o f ways education expenditure affects poverty and distributional outcomes. The first and most obvious i s through the direct impact on incomes o f the poor, as returns to their labour increase - either in agriculture or off-farm. Secondly, education i s clearly also the main investment, along with infrastructure, in laying the basis for growth more broadly, and driving the economic and employment expansion that i s central to reducing poverty in Ethiopia. Thirdly, even relatively l o w levels o f education improve the quality o f life o f the poor at existing income levels - through their impact on hygiene, fertility, and empowerment. 2.60 One way o f assessing equity impact i s to examine the incidence o f public spending on education across income groups and geographical areas. Matching school participation with household consumption using the Welfare Monitoring Survey data (Table 2.2 1) shows that primary enrollment rates are slightly higher for those in the richest 20% - but relatively evenly spread among the rest o f the population, with a l o t of the difference being accounted for by rural-urban differentials. In secondary education, not surprisingly, the benefits are concentrated more clearly in the upper income quintile; with children from the richest 20% being more than twice as likely to attend secondary school as those from the poorest 20%. 69 Findings o f World Bank and DfID missions, March, April, 2003. 49 Table 2.21: Gross Enrollment Ratio by Consumption Quintile (1999/200 WMS) Consumption Primary Primary Secondary Secondary Quintile GER - GER-Addis GER -GER - Addis Ethiopia Ababa Ethiopia 1 (Poorest 20%) 61.8% 112.3% 17.0% 67.2% 2 63.4 109.4 19.6% 64.6% 3 64.0 110.7 22.8% 70.9% 4 69.3 110.8 24.8% 66.6% 5 (Richest 20%) 78.7 106.9 44.8% 70.0% Source: MoFED; Development and Poverty Profile o f Ethiopia, 2002 2.61 Two points are worth malung: one i s that because poor households have more children on average, and more over-aged children in school, the actual distribution o f primary education benefits across income groups i s more equal than that suggested by the enrollment ratios alone. (In fact at the primary level, almost equal shares o f total student numbers come from each o f the five income quintiles). The other i s that comparing changes across time, the distribution o f benefits appears to have been improving somewhat, suggesting that the middle-poor, i f not the poorest, are benefiting more than the rich from increased public spending on education. (This i s not surprising, since in rural areas so many more o f the poor are not in school, and thus represent more o f the marginal new students as the system expands; and because - at least in urban areas - the richest are those that opt out o f the public system to go to private schools). 2.62 Geographically, enrollment rates vary significantly from a l o w o f 13% in the less- developed and pastoral regions such as Afar and Somali, to a high o f loo%+ in Addis. I n the four largest regions, however, the rates are close to the national average o f about 60%. The disparities have more to do with variations within the large Regions, between more and less remote areas. As shown earlier (table 2.13) spending levels very substantially from Region to Region, and this i s a cause for some concern. Reported recurrent education spending per capita averages around ETB 20($2.30) in most o f the large regions, but i s dramatically higher in Addis (ETB 53, or $ 6.16 per capita), and lower in S N N P (ETB 13, $1.51) and Somali (ETB 7, $0.81). While some o f this may be explained by reporting and classification errors, (including missing aid-funded spending), it suggests that there i s likely to be a public policy issue in terms o f the geographic distribution o f spending b e n e f i t ~ . ~ ' Link with Decentralization 2.63 As noted earlier, the bulk o f education service delivery, including delivery o f all primary and secondary schooling, and much o f TVE, i s the responsibility o f the Regions, while the Federal government i s responsible for overall policy and for university-level education. The division o f education spending has been about 23 percent federal, and 77 percent regional over the past 5 years - although the relative share o f the federal government has been increasing in recent years (to 34% in the 2003/04 budget), and the 'O As noted in table 2.13 and para. 2.24, spending per student i s in fact highest in some o f the smaller and less-developed Regions, but this i s because o f the very much smaller number o f students going to school i n those Regions, thus lowering the denominator and raising the average expenditure per student. 50 division o f discretionary domestic resources has been even more heavily weighted towards the federal level. 2.64 The most significant change in the past year has been the shift to block grants to woredas in the four biggest regions (see Chapter IO). While i t i s too early to tell what the implications are, anecdotally there i s already some evidence o f greater flexibility and responsiveness - for example in interviews woreda education officials in S N ” said they found it easier to respond to the immediate needs o f schools, without referring decisions back to the zonal level. However, the already bad situation with respect to availability o f recurrent funding has been exacerbated, at least in the short run, as in most cases the amount transferred to woredas has been barely adequate to pay the existing salary bill, making i t all but impossible to consider hiring large numbers o f additional teachers within the tightly-constrained block grant. According to the Annual Review Mission o f the ESDP (April, 2003), “In all woredas the allocation for non-salary wage recurrent budget is very small. As a result woreda education of$ces either have allocated very small amounts, or did not allocate budget at all to schools.” 2.65 We perceive a risk, moving forward, that necessary spending o n education will be squeezed even more, unless it i s explicitly taken into account in adjusting transfers to Regions and to woredas. The main reason for this i s the pressing need to expand the teacher cadre in order to re-establish meaningful student-teacher ratios. Just illustratively, 71 to achieve this in Tigray would require an extra ETB 75-90 million p.a., whereas the total value o f the block grant to woredas, for all purposes, i s only ETB 267 million. In the projections section we discuss ways that this may be done at lower cost; but even if substantial savings are possible due to decentralization, it will not obviate the need for much greater funding for recurrent education than i s possible under the current level o f woreda grants. 2.66 There are a number o f other potential concerns, discussed in the section o n fiscal decentralization: these include the risk o f capital bias in spending (already there are reported cases o f local authorities wanting to build junior secondary schools that cannot be staffed or financed); and the difficulty o f managing optimal school investment strategies if decisions on school size and location are taken independently within individual woredas. These can likely be managed, but it i s important the process be carefully monitored, and compensating adjustments made. 2.67 The other concern i s whether woredas - with an average population o f about 100,000 (and some much smaller) - are in fact a large enough unit to manage school systems, and support adequate professional school and teacher supervision functions. (School boards in many countries, for example, cover substantially larger populations). There needs to be a debate, based o n sound analysis and education experience, o f what the appropriate level o f discretion i s over educational matters at the woreda level. 71 Based o n current STRs o f 69: 1 (primary) and 85: 1 (secondary), enrollments o f 606,000 primary and 72,000 secondary; and average salaries as in the education projections, and target STRs o f 50- 60: 1 primary and 50: 1 secondary. 51 Private Provision o f General Education 2.68 One way o f reducing the pressure on the government budget i s clearly through private financing o f schooling. Such financing can take the form o f children attending private, rather than government, schools; o f cost-recovery in the form o f fees at public schools; or o f community contributions to the capital or recurrent costs o f schooling. In this section, we briefly explore the current shape o f private education in Ethiopia, and forecast the orders-of-magnitude o f savings in public spending that might result under different mixes o f private financing, and the associated policy issues.72 2.69 Non-government schooling played an important role in the education sector prior to the mid-l970s, accounting for some 30 percent o f institutional capacity in 1975. In that year, the military government outlawed the establishment o f non-government schools, and nationalized existing ones. I n recent years, and particularly since the beginning o f ESDP, the role o f the private sector has been seen in a more favourable light. This i s exemplified by the Government’s 1994 Education and Training Policy, which commits to: “[creatind the necessary conditions to encourage and give support to private investors to open schools and establish educational and training institutions.” 2.70 One needs to be c a r e l l in discussing non-government education in Ethiopia, as there i s some definitional ambiguity, and a specialized set o f terminology that i s used only in Ethiopia. Although there are significant numbers o f non-government schools, the number o f truly ‘private’ schools - that is, privately-owned and funded - i s very small. 2.71 ‘Non-government’ schools consist o f three types: (i) what are called ‘public’ schools (formerly private schools that were nationalized during the Derg regime, and now operate with substantial public financing), ( ii) NGO/faith schools, and ( iii)private schools. O f the 1,862 non-government schools operating in 2001/02 (with total enrollments o f 560,000), over 76 percent were ‘public’ or NGO/faith-based; and only 17 percent were and o f these, a large share were kindergartens, which account for almost actually pri~ate’~, two-thirds o f all non-government schools. So in 2001/02 non-government schools accounted for about 6 percent o f total enrollments; but within the non-government sector, only about a fifth o f the schools are truly private, in the sense that would be understood elsewhere. l2 A more detailed note i s available as part of the PER Background Paper series: “Private Participation in Basic Education in the Federal Republic o f Ethiopia” ; Latham and LaRocque, 2003; financed by DfID. 73 The remaining 6 percent being international schools catering to the foreign community - these are excluded from subsequent analysis. 52 Table 2.22: Structure of the Non-Government School Sector in Ethiopia (2001/02) Number of Schools NGO/faith Schools (41%) 260 Operated by NGOs, missions, and religious organizations; operating o n the basis o f donations from domestic and overseas NGOs and faith-based organizations, as well as fees. ‘Public’ Schools (35%) 223 Formerly private schools, nationalized in the 1970s; financed via fees and some government assistance in the form o f teacher salaries, training, and curriculum and materials. Private Schools (17%) 109 Operated by private entrepreneurs; h l l y financed by tuition and other fees; receive no government subsidy. ‘Community’ Schools (6%) 39 Serving specialized groups o f foreigners; fee-based. Source: Latham & LaRocque ,2002 2.72 The sector i s quite small, and despite recent growth, remains below the relative importance it attained in the mid-1970s. Primary non-government school enrollments stayed in the 300,000-350,000 range through the 1980s, and fell in the 1990s. The non- government market share declined almost continuously from 1967/68 to the mid-1990s. More recently, non-government school enrollments and market share have begun to recover somewhat, and by 2002/02 accounted for just over 5 percent o f enrollments. (See table 2.23). The non-government secondary education market share fell from a high o f 12.4 percent in 1967/68 (when the public sector was presumably very small), and has fluctuated in the 1.5 to 3 percent range ever since. Table 2.23: Non-Government Schools and Enrollments - 2001/02 (EY 1994) Schools Enrollments Share of Total Enrollments Kindergarten 1,189 119,000 100% Primary 619 429,000 5.3% Secondary 54 18,400 2.4% 2.73 Non-government schools make up an important proportion o f total enrollments in primary education level in certain areas - mostly urban (one-half in Addis, over one-quarter in Dire Dawa; 10 percent in Harari). The recent increase in non-government sector enrollments in Addis appears to be continuing. For example the Addis Ababa Regional Education Bureau reports receiving an average o f around three applications per month at the primary level, and about five per month at the kindergarten level. 2.74 The regulatory environment for provision o f private education i s basically good. For example, there are n o artificial restrictions o n tuition fees, and incentives are available for private operators in the form o f free land, and exemptions from import duties o n educational equipment. The main difficulties faced by investors are rather ones common to all small businesses in Ethiopia: bureaucratic delays in processing investment incentives, delays in customs duties exemptions, and long delays in obtaining land. 53 2.75 Despite this growth, it i s clear that one o f the main constraints in the expansion o f non-government education in Ethiopia i s the inability o f much o f the population to afford the fees. For example, a survey carried out in the mid-1990s found that: 0 an inability to pay fees was the main reason why students dropped out of private schools; and 0 non-government schools who felt there was a limited demand for their service identified an inability to pay as one o f the main (though not the most important) reasons for l o w level o f demand. 2.76 Table 2.24 below shows the range o f fees charged by private schools recently. Later in this chapter, we forecast the potential role o f private financing in reducing the public expenditure requirements for general education. Table 2.24: Average fees Charged by Sample of Non-Government Schools - December 2002 ~ School Type Grade Level Average Fee (Range) Private Grades 1-4 1,368 (410-2,970) Grades 5-8 1,424 (410-2,700) Grades 9-10 1,910 (410-3.080) Public Grades 1-4 240 (100-355) Grades 5-8 235 (120-400) Grades 9- 10 317 ( 170-600) Mission (faith-based-NGO) Grades 1-4 542 (235-735) Grades 5-8 1,240 (550-2,820) Grades 9-10 1,604 (550-3.245) Memo item: Per capita GDP is approximately 900 Birr annually. 2.77 Non-government schools are free to set their own salaries, which vary considerably depending on the school’s ability to pay. For example, some private schools catering to lower-income students paid teachers only ETB 800-900 per month, (this compares to government salaries ranging from about 600 ETB for primary teachers up to 1,270 ETB at the secondary level). Private schools serving wealthier students paid as much as ETB 1,900-2,000 per month.74 2.78 There are increasing numbers o f community-funded schools and teachers - generally taking the form o f community contributions to construction or upgrading, or communities employing teachers to work at govemment schools. W h i l e the latter practice i s reportedly becoming widespread, there i s n o reliable data o n the numbers involved, nor on the salaries paid or quality o f teachers (who are presumably unqualified). Encouraging community contributions i s a formal part o f the government’s education strategy; and while specific norms are not set, most regions seem to be operating o n an assumption o f a minimum 5-10 percent community contribution. This i s an area where potential public expenditure savings are possible, especially if community-employed teachers can be retained at lower salaries - a phenomenon that i s likely to be more feasible at the lower grade levels. However, there are also concerns with respect to quality-control, training and supervision and a risk that reliance on community funding either places an excessive Note l4 teachers in private schools are typically paid 10 months a year, as opposed to 12 months for government teachers, ; and may often work a longer day. 54 burden on poor communities or households, or results in dualist systems - with high- quality provision for those who are better off. A first step i s to get more systematic data o n the scale o f the phenomenon, and to launch an assessment o f the effectiveness and quality o f schooling under these informal arrangements. 2.79 So what role can privatefinancingplay? Obviously there are two broad options: private provision for fee-paying students, which reduces both the capital and recurrent expenditure burden on government; and secondly there i s the potential for charging fees at the secondary level, thus effectively reducingthe recurrent cost burden on government. 2.80 The table below illustrates the possible scope for savings due to private provision; o n the assumption that Ethiopia were to start to approach levels o f private schooling that prevail in some other parts o f Africa (note however there i s very wide variation from country to country). Somewhat arbitrarily it i s based on levels o f 10.5 percent o f primary enrollments, and 9 percent o f ~econdary’~ in private schooling. Table 2.25: Approximate Savings to Government Budget Due to Increased Shift to Private Education Primary Secondary Private Market No. o f Est. Public Private Market No. o f Est. Public Share Students Budget Saving Share Students Budget Saving 2005 7.5% 810,000 ETB 75 million 5% 48,500 41 million 2010 9.0% 1.2 m. 188 m. 7% 87,500 112 m. 2015 10.5% 2.1 m. 466 m. 9% 159,300 312 m. 2.81 Under this scenario, the burden on public spending would be reduced by some 778 million Birr in 2015 - this compares with projected public expenditure requirements o f ETB 6.2 billion under the ‘moderate’ growth scenario - suggesting that while increased reliance on the private sector may have an impact in reducing the financing requirements o f the education sector, the impact i s likely to be in the order o f 10-12 percent, and not significant enough to make a transformational difference. 2.82 These represent a fairly modest set of projections, compared to rates o f private provision in other countries. A more aggressive set o f projections (with 12.5 percent o f primary and 15 percent o f secondary enrollments being accounted for by private schools by 2015) shows potential savings o f ETB 1 billion by 2015. However, there are a number o f reasons to suspect that it may not be so easy to achieve these levels o f private penetration in Ethiopia, and some policy concerns associated with attempting to do so. 2.83 The first i s ability to pay. Despite substantial expansion, an accommodating policy environment, and interest in urban areas, it should be noted that the non-government sector (which includes all faith-based, NGO, and ‘public’ schools as well as private) still accounts for less than 3 percent o f secondary school enrollments - suggesting that the pool o f families who can pay for private secondary education i s very limited. 2.84 With average per capita incomes o f about $100 equivalent annually, the vast majority o f which i s consumed paying for food and shelter, it i s clear that only a small share o f the population are going to be have the disposable income to pay for private 15 See Latham & LaRocque (2003) for details. 55 education. At average fees ranging from ETB 926 ($108) for kindergarten to 1,910 Birr ($220) at the secondary level, the effective demand for private schooling i s going to be limited to the highest income groups, and while there will be a substantial market in urban areas, the view o f education professionals in Addis Ababa i s that this market will likely reach saturation quite quickly. 2.85 There are a number o f other issues associated with widespread increase in reliance o n private education. One i s the question o f quality. W h i l e non-government institutions can serve as models o f higher efficiency and quality, it i s also true that the rapid increase in unregulated private schools can result (and has done so elsewhere) in the establishment o f schools o f questionable quality, with un-inspected and unqualified teachers. This can be handled through the strengthening o f inspectorate and regulatory functions for private schools - but will also tend to restrict the growth to those which can afford to provide the required quality. 2.86 The other r i s k i s o f the emergence o f a ‘dual’ system, under which the better-off and better educated abandon the public school system, reducing the quality further and - perhaps more importantly - reducing the public pressure from the politically influential urban population for improvements to the public system. Clearly there are equity implications to such a two-tier educational system, whereby the better-off are assured a good quality education, and the poor generally consigned to a poor-quality public system. This i s to some extent inevitable, as more affluent urban consumers will make choices to shift to the highest quality service. None o f this i s to say that the option o f private provision should not be exploited - in the absence o f adequate public funding Ethiopia needs every source o f financing it can tap - but rather that the government needs to be aware o f the equity trade-off in designing public policy responses. 2.87 Consultants working o n the private financing options have also projected the possible budgetary impact o f introducing cost-recovery fees at the secondary level. They show potential savings o f between ETB 250 and 500 million annually by 2015 (4 -8 percent o f the education budget), associated with a range o f options varying from 7.5 percent cost recovery fees for all secondary students, up to 35 percent cost recovery in grades 11-12. 2.88 W h i l e there i s an argument that the private returns to secondary education are substantial, ‘and that some households are willing and able to pay fees, there are nonetheless concerns o f equity and access with introducing fees in the public system. Evidence from elsewhere suggests that fees have a significant effect on access and equity, discouraging or preventing poorer families from sending promising children to school, even where waiver and scholarship programs are in place. In the case o f Ethiopia, where enrollment rates are so low, where the vast majority o f households have little or n o cash income, and a major push i s being made to expand participation in the school system, it i s not clear that introducing fees o n any significant scale would be advisable in the near future. 56 F. LOOKING FORWARD: PROJECTED EXPENDITURE REQUIREMENTSFOR PRIMARY AND SECONDARY EDUCATION 2.89 The overwhelming fact o f the future o f education spending in Ethiopia i s the very large demographic bulge facing the country. N o t only are over 45 percent o f the population below the age o f 14, presaging a massive growth in the secondary-school aged population, but as this current age group approaches fertility, major increases will continue to take place in the next cohort o f children. The figure below illustrates graphically the situation for the next two decades. Figure 2.8: School-Aged Population Profile 2000 and 2020 (millions) 0.0 5.0 10.0 15.0 15-19 Yr. -olds 10-14 Yr. -olds 5-9 Yr. -olds 0-4 Year-olds 0.0 5.0 10.0 15.0 2.90 The number o f school-aged children i s projected to increase to over 30 million by 2020 from 20 million today; the number o f secondary-aged children (aged 15-18) will increase from 6 million to about 9.3 million.76 Taken together with the fact that only about 76 Based o n the CSA national population projections; the projections for broad age cohorts for the period as a whole appear reasonable, and are in line with the official UN and World Bank projections , although there appear to be errors in the single-year-age-specific cohort projections for some years. 57 58 percent o f children are currently in primary school, and only 7 percent in secondary, the pressures on the system, i f Ethiopia i s to attain universal primary education, and even modestly increase the proportion o f the population with secondary education, are immense. The following section explores the implications for education spending requirements. What Would it Cost to Achieve Various Outcomes in General Education? 2.91 In this section we project what Ethiopia would have to spend to achieve a range o f different outcomes, with respect to coverage and quality, in the general education sector.77 In the overview Chapter these projections are combined with those for other sectors to highlight trade-offs that the government and donors need to consider, and the financing needs if Ethiopia i s to achieve the desired outcomes. 2.92 We note the very good work that has already been done in costing using the EFAA model. These projections, which build on that work, differ only in that they cover a longer time frame (through 2020), and a wider range o f possible outcomes; and are methodologically consistent with the projections used elsewhere in the PER. As with all projections, the purpose i s largely pedagogical: to demonstrate the broad costs o f policy options. The intention i s thus to show orders-of-magnitude o f costs, rather than to produce definitive forecasts. A more refined version o f the EFAA model i s being adapted by the Ministry o f Education, to model the effects o f repetition and drop-out rates. Over the long- term covered here however, at close to 100% GERs, the total costs o f inputs i s not expected to differ dramatically between the two models. The Scenarios W e forecast three basic scenarios: 0 A Low, or ‘Business-as-Usual’ scenario: that envisages a slow approach to Universal Primary Education (UPE) over the next 15-20 years, and a very modest increase in secondary education coverage. Unit expenditures improve only slowly from their current levels, with coverage increasing to keep up with population growth. Appreciable improvements in student-teacher ratios are 78 achieved only very slowly. A moderate ‘Extended PRSP’ scenario: corresponding to moderate GovernmenuEFA targets: achievement o f UPE by 2015; some spending increases to restore quality, especially in reducing student-teacher ratios; and incremental improvements in non-wage recurrent spending; this would bring Ethiopia close to the primary education MDG, coupled with a modest increase in secondary enrollments (to 25 percent GER for junior secondary, and 15 percent for senior secondary, by 2020). A more ambitious ‘MDGs Plus’ scenario: full implementation o f government’s targets, and significant spending increases to improve quality. This scenario involves more rapid achievement o f UPE, and a more aggressive expansion o f secondary education - especially at lower levels. (Note however that the targets for secondary education: GER o f 35 percent for grades 9-10, 77 This section covers primary and secondary education; projections for TVET and university education are presented in Chapter 3. 78 All scenarios involve restoration o f reasonable student-teacher ratios, but t h s one approaches it much more slowly. 58 and o f 20 percent for 11-12, are not excessively ambitious compared to what other low-income countries aspire to). In addition we also show what it would cost just to achieve the MDGs (which do not cover secondary education). 2.93 Consistent with projections elsewhere in the PER, they cover the period 2002- 2020, and constant 2002 Birr are used throughout. Detailed assumptions are shown in the annex tables, but some key features are summarized below: Table 2.26: M a i n Characteristics of the Projection Scenarios Targets for 2020 (EFY13) Grades 1-4 Grades 5-8 Grades 9-10 Grades 11-12 BasicLow Scenario GER (in percent) 100 85 20 10 Student-Teacher Ratio 60: 1 45: 1 42: 1 36: 1 Moderate ‘Extended PRSP’ Scenario GER (inpercent) 110 95 25 15 Student-Teacher Ratio 55: 1 38:l 32: 1 28: 1 Ambitious ‘MDGs Plus’ Scenario GER (in percent) 105 100 35 20 Student-Teacher Ratio 45:l 35:l 23: 1 22: 1 Memo Item: Current Levels (2001/02) GER(in percen2) 86 30 17 10 Student-Teacher Ratio 79:1 48:l 46: I 49:5 2.94 Table 2.27 shows the recurrent cost requirements o f the four scenarios. M u c h o f the cost i s incurred in expanding Grades 5-8; and, under the higher scenarios, in secondary education. This i s driven by the fact that the enrollment ratio in grades 5-8 i s currently very l o w (only about 30 percent), combined with the fact that there i s going to be very large growth in the upper-primary and lower-secondary-aged population over the next 15 years. Table 2.27: Annual Recurrent Expenditure Requirements (Constant ETB millions) 2010 2015 2020 BasicLow Scenario 2699 3515 4218 ($314 m.) ($409 m.) ($490 m.) ‘Extended PRSP’ Scenario 3009 4433 5755 ($350 m.) ($515 m.) ($669 m.) ‘MDG Plus’ Scenario 3485 5673 7520 ($405 m.) ($659 m.) ($874 m.) Memo Item: Recurrent spending (2001102 Budget) on general education: approximately ETB 1,090; or US$ 127 million equivalent. 2.95 The bottom line i s that to achieve UPE it i s going to cost something like 4 billion birr annually in recurrent costs by 2015 (compared to about 1 billion today, implying a growth rate in spending o f about 11% pa. in real terms), and over ETB 5 billion annually by 2020 if there i s to be even a moderate expansion o f secondary education. 2.96 Capital costs are high under all the scenarios - driven mostly by the large number o f new classrooms required to accommodate the growing population, and to raise the enrollment ratio in primary school - especially in grades 5-8. Capital requirements vary 59 between a l o w o f ETB 1 billion ($1 16 million) annually in the near term, rising to about 2.4 billion birr ($280 million) p.a. by 2010. Under the more ambitious scenarios, with a significant increase in secondary enrollment ratios, the capital costs rise to a peak o f about Birr 3.4 billion ($400 million) annually by 20 15. 2.97 Under all scenarios, the capital costs o f expansion drop o f f after about 2015 (EY 2007) as the rate o f growth o f the school-aged population slows, and as enrollment targets are reached. Table 2.28: Estimated Annual Capital Costs of General Education Scenarios (Constant ETB millions) 2005 2010 2015 2020 BasicLow Scenario 1121 1993 2007 1778 ($130 m.) ($232 m.) ($233 m.) ($207 m.) ‘Extended PRSP’ Scenario 1289 2355 2473 2214 ($150 m.) ($274 m.) ($288 m.) ($25 7 m.) ‘MDG Plus’ Scenario 1317 2641 3387 2116 ($153 m.) ($307 m.) ($394 m.) ($246 m.) Memo Item: Current capital spending (2001102 Budget) on general education: approximately ETB 630; or US$ 73 million equivalent. 2.98 In the overview chapter we compare these costs with those o f other policies and sector requirements; and with the overall fiscal envelope. Details o f the moderate ‘Extended PRSP’ projection, and some ways o f reducing i t s costs, are explored below. Details of the Middle ‘Extended PRSP’ Cost Projections 2.99 Even under the moderate projections, the costs o f expanding general education are massive - amounting to about ETB 6.9 billion p.a. by 2015 (compared to ETB 1.7 billion in the 2001/02 budget). As noted above, these costs are driven primarily by the expansion o f the second cycle o f primary. The impact o f secondary-school expansion i s relatively small under the moderate scenario - because even though unit costs are high, the numbers involved are much smaller at these modest enrollment targets. The approximate composition o f the cost requirements i s illustrated by the following table: Table 2.29: Composition o f Annual Costs of ‘Extended PRSP’ Scenario 2015 (Constant 2002 ETB millions) Total Primary 1-4 Primary 5-8 Secondary - Wage bill 1374 1880 71 1 3965 Non-wage recurrent 171 133 164 468 Capital -530 1399 - 544 2473 Total 2075 3412 1419 6906 2.100 The requirements compare with total expenditure in 2001/02 o n general education o f about ETB 1.7 billion ($200 million), indicating the need for about a four-fold real increase in education spending to meet the needs o f the ‘moderate’ scenario. This increase i s accounted for in roughly equal parts by population growth, and by increasing coverage (35 percent each), and the remainder by the costs o f restoring quality (reducing student- teacher and student-section ratios, and providing adequate non-wage recurrent inputs - although note that the targets in this area are not excessively ambitious, with long-term 60 student-teacher ratios o f 50:l at the basic primary level, and about 40:l at the higher levels). 2.101 The number o f teachers required to achieve even these fairly modest targets i s massive: under the moderate scenario 424,000 teachers are required by 2015 (compared to about 145,000 today). 20,000 new teachers would need to be produced each year to cater for growth, and, assuming only 4 percent p.a. attrition, another 17,000 per year would need to be trained just to replace those retiring or leaving. This compares with a current capacity to produce teachers o f only about 6,000 p.a. 2.102 Similarly, to absorb new entrants classrooms would need to be constructed at the rate o f approximately 15,000 each year by 2015 (and approximately 9,500 p.a. in 2006). At present approximately 5,000 p.a. are being constructed annually, suggesting that capacity constraints may restrict the rate at which the physical goals can be achieved. Possible Areas of Savings 2.103 Even this moderate ‘Extended PRSP’ scenario i s going to be difficult to finance. The estimated costs amount to about 21% o f projected total public expenditure in 2015, compared to about 10% today. The following section explores, briefly, ways o f reducing these costs. Table 2.31 illustrates the likely cost-savings associated with a number o f policy choices, relative to the base-case projections. M u c h o f the focus i s on changes that affect the upper primary cycle (grades 5-8), which i s where the greatest impacts can be achieved, because under the middle scenario the incremental costs o f expansion in lower primary and secondary are far smaller than in the upper-primary. Reduce targets for grade 5-8 enrollment. The projections have assumed that achieving UPE would involve a high level o f completion o f the second cycle o f primary school (GER o f 80 percent by 2015, and o f 95 percent by 2020). Table 2.30 shows the impact o f significantly reducing the target - to 45 percent in 2015 and 50 percent in 2020. T h i s has the greatest single impact on costs (about a 27 percent saving annually by 2015). Reduce student-teacher Ratios in Grades 5-8. Holding the student- teacher ratio in grades 5-8 to 46: 1 would only marginally affect total costs, reducing them by 3%. (S:T ratios at other levels are already quite conservative; and the scope for lowering them further may be limited.) Allow larger section sizes. The forecasts assume a long-run target o f 50 students per section - the government’s target level. Increasing this to 60: 1 in the two primary cycles would have some impact in terms o f reducing the numbers o f classrooms that need to be built to absorb new students, but would reduce total expenditure requirements by only about 5 percent p.a. by 2015. Reduce Costs o f the Wage Bill. Teachers’ salaries are relatively high in proportional terms compared to those in other developing countries. One option i s to allow the recruitment o f locally-based and financed teachers (employed by either communities or local governments) at lower salaries. This i s already taking place through community financing o f teachers under informal arrangements; and there i s anecdotal evidence that some 61 teachers are employed by NGOs at salaries substantially lower than government levels. I t i s likely that this option i s available more at the junior primary level than at others, and the lower scales will only apply to some incremental proportion o f newly-recruited teachers. We have projected the impacts o f (a) a 25 percent decrease in the average salary bill for lower primary (1 -4) teachers; and (b) a 25 percent average decrease for grades 1-4, plus a 20 percent decrease for grades 5-8. The impacts are useful, but not enough to dramatically affect total costs (a 5 percent p.a. saving for option (a), and 10 percent for option (b)). (5) Lower construction costs. Given the large number o f classrooms required, the impact o f lowering unit costs can be substantial. Both standards and costs for classrooms are high in Ethiopia. Recent work by World Bank missions suggests it i s possible, through a combination o f alternative construction techniques and materials, to reduce costs substantially from the current official model. Table 2.30 shows the effect o f reducing the average costs o f a furnished and equipped primary classroom by about half, to ETB 70,00079. ESDP I1proposes a substantial proportion o f ‘low-cost’ schools can be built at 25 percent o f current costs, although it i s difficult to confirm this assertion; to the extent greater cost- reductions are possible, commensurately larger savings will be realized. Table 2.30: Annual savings over 2015 Base Case Scenario (ETB millions p.a. by 2015) Recurrent Capital Total Savings (in percent) (1) Reduce GER Target for Grades 5-8 88 1 948 1829 27 (2) Higher S:T Ratio Grades 5-8 208 23 228 3 (3) Allow Higher Section Size Targets 309 309 5 (4a) Reduce Average Salary Bill Grades 1-4 3 12 312 5 (4b) Reduce Average Salary Bill Gr.1-4 & 5-8 688 688 10 (5) Reduce Construction costs 1616 1616 24 2.104 The Combined Effect o f Savings. Clearly savings need to be made wherever feasible, and although the government may not want to sacrifice targets in all o f these areas (nor would it be advisable to do so- especially with respect to coverage and quality improvements), the combined effects o f savings can be considerable. Combining a range o f ‘reasonable’ assumptions regarding cost reductions and adjusted targets suggests that the total annual expenditure requirements could be reduced to around ETB 4.1 billion ($ 476 million) annually by 2015, as opposed to a base case cost o f 6.9 billion birr. 2.105 The most significant choice facing government i s that o f the exit point for the majority o f the population. As the foregoing analysis demonstrates, the greatest savings come from slowing the expansion o f the second cycle o f primary (grades 5-8). While i t i s a matter for policy debate whether Ethiopia wants to adopt such substantially lower objectives over what is, after all, a fairly long term, it should also be recognized that in most countries primary education extends only through grade 6. Accepting some such 79 From ETB 110,000, 140,000, and 250,000 for a filly-equipped and furnished classroom for grades 1-4, 5-8, and 9-12 respectively, to 70,000, 100,000, and 200,000 respectively. 62 intermediate exit point i s probably worth considering i f resources will not allow full coverage. As a next step, it would be useful to more precisely model the education costs o f the year 8 system as opposed to the year 6 system. What Would I t Cost Just to Attain the MillenniumDevelopment Goals? 2.106 The PER projections cover the costs o f the integrated Ethiopian general education system; however from a policy point o f view, and to allow comparisons with other countries, it i s also useful to know what it would cost just to achieve the MDGs in education, the cost implications o f which are largely confined to the primary cycle. There are two education MDGs: the first i s to ensure that all children can complete a full course of primary schooling; the second i s to eliminate gender disparity at all levels o f education by 2015 2.107 Goal 1: Ensure all children can complete primary education by 2015. The achievement o f close to 100% primary school enrollment by 2015 should be feasible. The government’s plans would allow it; and recent progress - with substantial enrollment increases in the past 5 years - suggests that Ethiopia i s potentially o n track to achieve the goal over time; although it will require very substantial increases in education budgets. The constraint will come in ensuring that families actually choose t o have children attend school, and also that health and nutrition and time-use barriers can be overcome. Clearly the government can increase the probability o f students enrolling and remaining in school through public spending, but there i s a major challenge in reducing drop-out rates, and it will be increasingly difficult to attract the last marginal children to school - especially in less developed regions, and from poorer households. 2.108 Cost: The costs o f providing full primary coverage, with adequate student-teacher ratios and complementary inputs are estimated at ETB 5 billion p.a. by 2010; and the aggregate public expenditure requirement between n o w and 2015 i s estimated at ETB 50.8 billion, or US$5.9 billion equivalent, in constant 2002 prices. 2.109 Caveats: There i s a complication in the case o f Ethiopia in that primary education consists o f grades 1 through 8, with cycles o f 1-4 and 5-8; whereas in most countries primary education i s defined as the first five or six years o f school. The MDGs do not provide a definition, but the assumption i s that six years i s being referred to. Achieving universal coverage o f grades 1-4 would be much easier (and more affordable), but falls somewhat short o f the proposed attainment goal. On the other hand providing the full 8 years - the current Ethiopian definition - i s correspondingly more expensive, and significantly more ambitious, than the MDG as being pursued elsewhere. 2.110 Secondly, there i s a difference between the costs o f 100% gross enrollment and 100% completion - the difference being accounted for mostly by repetition rates. Repetition rates are not excessively high in Ethiopia, but drop-out rates are very high*’; although as the system approaches loo%+ GERs the costs - in terms o f inuuts required start to converge. Clearly a major effort i s needed to reduce drop-out rates in order to achieve Based o n the costs o f teachers and facilities to provide capacity for 100% in grades 1-4, and 50% in grades 5-8; plus substantial increases in recurrent inputs, up to the levels specified in the ‘MDG plus’ scenario. Although enumeration problems tend to categorize some repeaters as drop-outs. 63 the MDG, however from a public expenditure forecasting point o f view, the inputs required to reach 100% GER - adequate numbers o f teachers, classroom capacity, and provision o f quality materials - are the same as those required to improve quality to reduce dropouts and repetition. (The exception i s probably the need for specialized schools for nomadic populations - which are catered for in government’s ESDPII plans, and while very important, do not amount to very large costs relative to the total amounts required.) Bearing in mind that the purpose o f these projections i s to illustrate the order-of-magnitude o f fiscal needs to meet broad policy targets in the long term, these estimates are more than adequate. A more refined model that actually simulates drop-out and repetitions i s being further refined by the Ministry o f Education to simulate different scenarios, but we do not anticipate that total long-run costs will differ significantly. 2.11 1 Goal: Eliminate gender disparity by 2015. Current trends suggest that the Ethiopia i s potentially in a position to reach this goal; currently about 40% o f students are female in both primary and secondary school. At the university level, where only about a quarter o f students in public institutions are currently female, the challenge i s greater. 2.112 Costs: The public expenditure implications are not large. T w o o f the most important factors in attracting girls to school are to have female teachers, and access to appropriate sanitary facilities. The constraints to increasing the proportion o f female teachers are not the cost, but more the limitations on the pool o f potential trainees, and cultural and social values with respect to women working and being posted away from their families. T o the extent that female teachers - especially locally-recruited ones - are lower cost than males, the public expenditure requirements are actually lower. A potential additional cost - although it has not been costed, nor as yet formally proposed by government - would be if a decision were made to institute a program o f scholarships or fee exemptions for girls at the secondary and tertiary level to encourage greater proportional enrollments. 2.113 Caveats: I t i s difficult to cost this goal explicitly, without a specialized exercise. As with the other education goal, full achievement l i e s to some extent outside the government’s control, and certainly outside the realm o f public expenditure; depending as it does on the decisions o f families to send girls to school. The Link with Development Strategy and Other Social Outcomes 2.114 I t i s difficult to measure the returns to education in a subsistence economy precisely. T w o studies have attempted to do so in the case o f Ethiopia (Amha (1996) and Krishnan (1997)). Both found relatively l o w returns to primary education (2-3 percent), and highest returns to junior secondary and diploma-level training. (5-8 percent). These findings are not surprising, and are consistent with those from other similar countries. In the current economic environment returns to formal education may not be very high - both studies conclude that parents are really aiming for their children to gain access to (higher return) secondary education, and that the lack o f secondary capacity may be a disincentive to parents keeping children in primary school. Nonetheless, work in other countries has also shown that 4-5 years o f schooling can have a substantial effect on productivity in subsistence agricultural (see Jamison and Lau). 2.115 Furthermore, social externalities, in terms o f the impact o n health, fertility, and nutrition are huge. More broadly, widespread primary education i s an essential investment in laying the base for future growth. Analytical work worldwide has repeatedly confirmed 64 the strong positive impact o f female education in reducing infant mortality and fertility levels. In Ethiopia, the D H S data show that under 5 mortality i s about 25% lower for children o f mothers with primary education, compared to those without; and 52% lower among those with secondary education. These results need to be treated with some caution, as women with more education tend to come from households with more and better food, and living in places with better access to medical care - but nonetheless they confirm that the broad relationships observed elsewhere hold in Ethiopia. A basic simulation exercise for Ethiopia indicates that universal primary education for girls would reduce under-5 deaths by about 15.5 per 1,000, or by about 10%. 2.116 Although formal retums to education are low, private retums to secondary education may even decline in the medium-term, as larger numbers o f secondary-educated youth may exceed the creation o f modem sector employment opportunities.82 At the same time there are substantial externalities in terms o f creating a pool o f potential entrepreneurs, and an educated population that will contribute to, and increase pressure for, the responsiveness o f public systems, especially in rural areas. 2.117 More fundamentally, no country has made the kind o f economic transition that Ethiopia needs to make to have an impact on poverty reduction with only 3 percent o f its population having secondary education. Furthermore, expanding secondary education i s a necessary step in creating the essential pool for higher levels o f education and capacity building - in’for example, producing the raw material for teachers, health, and supervisory staff. By 2010 the public system will require approximately 37,000 new teachers a year (to keep pace with population growth, attrition, and coverage expansion); and approximately 10,000 health workers annually83- together representing over 15% o f projected senior secondary school throughput; thus a substantial share o f spending o n secondary education i s directly related to providing the trainable people needed to deliver essential primary services. 2.118 Finally, general education i s one o f the most cost-effective uses o f public funds available to govemment. I t i s one o f the few areas in which there i s little doubt about the choice o f strategy needed for effectively delivering outcomes, it i s relatively simple to implement, and impacts are guaranteed (unlike, for example, govemment expenditures in agriculture, industrial development, or even areas o f infrastructure and provision o f finance - where the track-record in public provision has been mixed at best). 82 Based on the experience o f other countries, it i s quite likely that the pool o f unemployed educated youth w i l l increase in the medium-term (for example, by 2015, under the medium projections there w i l l be about 400,000 secondary school graduates annually). 83 Source: projection models for education and health spending requirements; under ‘moderate’ extended SDPRP scenarios, documented in sector chapters and annexes. 65 3. HIGHER EDUCATION 3.1 This chapter covers public spending on technical and vocational training, and on universities. It does not go into as much depth as the previous chapter, as the TVET sector i s covered in a more detailed background paper (Franz, 2003) the findings o f which are summarized here; and issues in the university sub-sector are being dealt with in an ongoing World Bank study (Higher Education Development for Ethiopia - draft). A. AND TRAINING TECHNICAL& VOCATIONAL EDUCATION 3.2 Since EFY 1993 (2000101) the government has placed a major emphasis on expanding technical and vocational training. This i s seen as an essential investment in building mid-level capacity and technical skills to lay the basis for longer-term growth, and as a way o f addressing the expectations o f the increasing number o f students coming out o f junior secondary school (grade 10) who cannot be accommodated in higher education 3.3 The program involves a rapid, ten-fold increase in public TVET over four years, with the construction o f 71 new TVET institutions and facilities, including 14 new colleges (at an expected capital cost o f ETB 1.8 billion (US$ 210 million)84and the establishment o f additional TVET capacity at existing secondary schools - including re-training secondary teachers and converting classroom capacity for technical training, in areas as diverse as hair-dressing, baking, and construction trades. In addition, a parallel program i s being undertaken for mass training in agriculture. 3.4 In EFY 94 (2001/02) there were about 38,000 students enrolled in 10+1 and 10+2; an increase o f more than three times from the level just two years earlier. The current TVE enrollment i s estimated to represent 25 percent o f the previous year’s grade 10 leavers. Public Spending on TVET 3.5 The recurrent budget for TVE was about 92 million Birr ($12 million) in 2001/02. This i s only 5 percent o f recurrent education spending, but there has been a major increase in recent years (the 2001/02 budget represents an increase o f almost 4 times from previous year’s) . There has also been a very significant increase in capital spending, with introduction o f an allocation o f ETB 134 million for TVE at the Federal level in the 2002/03 budgetg5 3.6 Furthermore, a massive increase in recurrent spending on TVE i s to be expected in the coming years, in concert with the planned major increase in enrollments. The short-term targets are to have enrolled 130,000 students by 2004/05, (with an intake capacity o f 55,000 84ESDP I1 Program Action Plan document, Ministry o f Education, 2002. 85As in other sectors, a detailed breakdown o f spending i s not available the most recent two budget years. 66 p.a.); which would imply recurrent spending needs o f about ETB 315 million p.a. (equivalent to about US$36 million p.a.). 3.7 In the medium-term, government’s policy goal i s to expand TVE to absorb 40 percent o f grade 10 leavers by EY 2004 (2012) - equivalent to about 200,000 students. We explore the cost implications o f this strategy, along with other policy options, below, but very roughly, this would imply spending something like ETB 450 million p.a. in recurrent costs by 2010, and require another 300 million annually in capital investment, or seven times recent annual spending. 3.8 Recurrent unit costs in TVE average about 2,250 Birr per student annually (ranging from 1,388 Birr for junior accounting, to a high o f 5,125 Birr p.a. for machine technology). This compares with current expenditure o f 500 Birr per student for general secondary education. 3.9 The reasons for the very high unit costs o f TVE are threefold (and these are common across all countries): (1) the number o f students per teacher i s substantially lower than in general education; (2) recurrent costs for consumable inputs are substantial (representing the largest share o f recurrent costs); and (3) equipment i s expensive - although this i s a capital cost, and i s not even reflected in the above unit cost estimates, it needs to be remembered that this equipment depreciates, and needs to be replaced every 7 years (or less, in many cases.). Table 3.1: Recurrent Unit Costs6for Selected Trades, 10+1 and 10+2 Programmes (ETB per student per year) Training Trade material Teachers Other cost Total Beginner Accountant 264 403 72 1 1,388 Junior Secretary 283 403 72 1 1,407 Information Technology 476 883 72 1 2,080 Auto mechanics 1,301 883 72 1 2,905 Machine Technology 3,550 883 72 1 5,154 General mechanics 1,960 883 72 1 3,564 Electronics 1,143 883 72 1 2,747 Woodwork 907 883 72 1 2,511 Surveying 963 883 72 1 2,567 Dressmaker 493 883 721 2,097 Bakery and confectionary 1,556 883 72 1 3,160 Cook 1,429 883 72 1 3,033 Hotel Service provider 23 1 883 72 1 1,835 Hair dresser 460 883 72 1 2,064 Weighted average87 2,252 3.10 Costing o f capital and equipment was initially estimated in the government program at 14,000 Birr per student capacity for 10+2, and 33,200 B. for 10+3. I t ’ s 86 Recurrent cost per student per year 87The weighted average i s based o n EY94 enrolment figures and includes all trades offered at the current moment. 67 generally recognized that these initial estimates were based on excessively ambitious equipment specifications; using more modest costs derived from the Ethiopia-German TVET program suggest a cost o f about 9,300 Birr per student. Effectiveness and Efficiency 3.11 It i s too early to assess the effectiveness o f TVET in Ethiopia in the absence o f good tracer studies that assess what happens to graduates several years later, the extent to which they are engaged in employment that uses their skills, and impact o n their incomes. Nonetheless based on experience elsewhere, and what we have seen in Ethiopia during the PER missions, some issues are worthy o f note. 3.12 Firstly, the experience in many countries has been that publicly-provided TVE i s often .of limited relevance to the job-market. Tracer studies often find workers are not employed in the areas in which they have been trained; and employers often prefer to recruit young people with a good general education, and train them in skills themselves. 3.13 Secondly, the costs o f providing TVE within public school systems are very high, relative to those o f producing generally-educated students. Technical teachers are often o f a lower skill level, and/or have insufficient experience o f real-world work requirements, to impart lmowledge that i s usable, especially relative to master craftsmen or senior employees who can provide apprenticeship or on-the-job training. A compounding problem i s the very high level o f attrition among TVE teachers in Ethiopia. It i s reported that at a minimum 10- 15 percent leave each year, in part because those who are effectively skilled can do better in the job market than teaching. 3.14 The desire to expand TVE i s understandable, but the constraints - in terms o f shortages of money, trained and motivated teachers, workshops that are in appalling condition, and the high cost o f specialized equipment - all suggest making use o f the world of work, through apprenticeship training or some variant, would potentially be cheaper, more relevant, and more cost-effective. 3.15 The conversion o f high schools to TVE institutes risks squeezing out resources that are critically needed to expand general secondary education. (For example, the PER missions visited several secondary schools where the recurrent budget has been entirely absorbed by the consumable requirements o f newly-added TVE courses). General secondary teachers are being ‘re-tooled’ to become TVE teachers - depriving the system o f experienced secondary teachers, and adding not-very-useful technical teachers (the mission has come across grade 9-12 teachers who were n o w assigned to teach hairdressing, and tailoring.). In addition, the expansion o f public TVE can crowd out the fragile private TVET-provider sector; this can be observed particularly in Addis Ababa . Projected Costs of TVET Policy Options 3.16 Given the potentially large expenditure requirements o f a major expansion o f TVET, we have projected the costs o f two basic scenarios, to explore the fiscal 68 implications, and provided a basis for policy debate o n the trade-offs with other forms o f expenditure.88 Scenario 1 - L o w Scenario - This scenario illustrates the costs o f a modest continued expansion o f TVE, somewhat faster than the growth o f the TVET-age population; it assumes the ESDP I1goal o f 130,000 enrolled in TVE, but achieving i t 5 years later (in 2010) and thereafter an approximately constant share o f grade 10 leavers continue to participate in TVE. Scenario 2 - Ambitious. This scenario assumes full implementation o f the government’s stated policy goal o f absorbing 40 percent o f grade 10 leavers in TVET, and the inclusion o f a o f a modest version o f the proposed Basic and Junior TVET, for earlier school-leavers. 3.17 Note in the overview volume we also present the approximate costs o f the ‘Extended SDPRP’ scenario, which falls between these two, and corresponds to the big push currently planned under ESDP I1 to increase TVET capacity to 130,000 students by 2005, and then growing just slightly faster than population growth. 3.18 The table below illustrates the basic characteristics o f the l o w and high scenarios, and the rough costs projected to be associated with each o f them. W e explore later the way that these costs can be reduced, by increased reliance on private provision, and cost-sharing in the public sector. Table 3.2: Projected Annual Technical and Vocational Education Costs ____ (annual requirements in constant 2002 ETB millions) 2005 2010 2015 2020 Proiected No. o f Students L o w Scenario 79,000 130.000 147,000 166,000 H i g h Scenario 158,800 196;500 262;OOO 3 171000 Projected Costs - H i g h Scenario Recurrent 358 456 606 737 CaDital - 3 72 - 295 451 424 Total 73 1 752 1058 1161 ($85 m.) ($87 m.) ($123 mJ ($135m.) Projected Costs - L o w Scenario Recurrent 176 290 330 3 74 Capital -78 - 119 - 157 183 Total 254 409 488 557 ($30m.) ($48m.) ($57 m.) ($65m.) 3.19 Note even the low scenario involves a doubling o f enrollments from the 2001/02 level, and a substantial increase on present levels o f spending - with the recurrent budget requirement doubling by 2005 and tripling in real terms by 2015. T o put this in perspective, the amounts involved (about ETB 409 million, or $48 million annually in 2010) are more than h a l f o f today’s recurrent spending o n primary education. However given current expansion plans already underway (which are based on achievement o f the 130,000 p.a. target much more quickly), this represents a minimum spending requirement facing the public sector in the near term. For purposes o f comparability the overview chapter also presents a third ‘middle’ scenario, that falls halfway between the two; corresponding to faster achievement o f the ESDPII goals, but less than the full 40% target in the high scenario. 69 3.20 The ambitious scenario shows the costs o f achieving the stated policy goal o f 40 percent o f grade 10 leavers being absorbed into TVE. The costs o f this rise quickly, to the equivalent o f $87 million (ETB 752 m.) annually by 2010 - approximately equal to the total recurrent primary education budget today. These levels are almost certainly unaffordable. The section below discusses ways the budget impact may be reduced - by greater reliance on the private sector, and cost-recovery - but under any scenario these magnitudes are probably unattainable. Furthermore, there are good policy reasons why it might not make sense to try to achieve them. 3.21 Even a more moderate strategy, lying mid-way between these two, and achieving the stated goal o f 130,000 TVE enrollees by 2005’06 (EY98) would imply costs o f something like ETB 400 million p.a. - equivalent to more than half o f the recurrent primary education budget today. Financing Options and The Role of the Private Sector 3.22 Clearly one potential solution to the overwhelming financing requirements i s greater involvement o f the private sector; and some form o f on-the-job or apprenticeship training i s prevalent in most countries. 3.23 Experience elsewhere has generally been that employers provide more marketable, and market-related, training than do public institutions, and have a stronger incentive (and more resources) to provide quality training. The other advantage o f course i s that this shifts the burden o f financing to employers. 3.24 Other modes o f private financing are the development o f private colleges, and the charging o f fees at public TVE institutions. Below we explore the potential cost-savings to the public budget o f various alternatives. 3.25 The drawbacks in Ethiopia are that the formal private-sector i s relatively small and under-developed, limiting the scope for apprenticeship and on-the-job employer training; while effective demand, in terms o f ability to pay, limits the scope for private colleges, and also places limits on the potential for cost-recovery at public institutions. Nonetheless there i s potential in all three areas. 3.26 The most promising area o f budgetary savings i s through private provision o f TVE at private fee-based institutions. At present only 4.5 percent o f TVE students are enrolled in private institutions - although the numbers have been growing - especially in Addis, in areas such as computer skills and office training. If this could be raised to 15 percent, savings to the government might be in the order o f ETB 60 million p.a. (at the 130,000 enrollment level). The large number o f unaccredited private providers at the moment, and experience in other countries, suggest such a target might be reasonable over the medium- term. A more ambitious scenario - o f 25 percent enrollment in private institutions - would save the government roughly ETB 100 million p.a., however that the scope for reaching these higher levels may well be limited by the limited number o f families able to pay commercial fees. 3.27 The policy framework for private provision o f W E i s fairly open - the barriers relate to high equipment standards and costs, and the curriculum-based accreditation system, as well as the crowding-out effect o f public provision. If these constraints can be relaxed, it should be possible to reach the level o f 15 percent private provision. 70 3.28 The recent Draft Financing Strategy for TVET proposes the introduction o f training fees for formal TVE programs, and a policy to widen the income-generating activities o f public TVE schools. The strategy foresees potential cost-recovery o f 30 percent o f operating costs from each o f these sources by EFY2000 (2007/08). 3.29 The targets seem very ambitious, and are well in excess o f the types o f cost- recovery ratios achieved elsewhere in Africa. Income-generating activities can certainly produce revenue, but the amounts implied here would amount to about 350,000 Birr per school annually - which would require a robust market, high-quality services, and a very entrepreneurial school management; furthermore, the scope for generating such incomes in rural areas o f Ethiopia i s extremely limited. 3.30 Cost-recovery through fees should doubtlessly be part o f the strategy, but again the levels projected seem ambitious. T o cover 30 percent o f recurrent costs per student would imply fees o f about 675 Birr per student ($80 equivalent), which may be feasible for a proportion o f urban households, but would become a binding constraint if - as the coverage targets imply - government’s intention i s to extend TVE to a much wider proportion o f the population. (Noting also that the higher-income households would be more likely to have children remaining in the academic secondary schools; and that if relatively high fees are charged, many o f those who could afford them may migrate to the private TVE colleges). A more promising avenue i s to cultivate the provision o f on-the-job training. 3.31 We suspect that an optimistic estimate may be that about one-third o f projected public sector costs could be saved by combination o f these cost-recovery and private financing options. At same time quality and relevance o f training would be increased. The most promising area however i s probably on-the-job and apprenticeship-style training. This i s an area that i s in i t s infancy in Ethiopia, and it i s recognized that the small formal private sector limits scope, but to the extent such mechanisms can be developed government will save major investment costs, and trainees will receive skills that it i s more likely they will be able to use. This i s an area that needs to be pursued aggressively. Issues/Conclusions We are concerned that pushing ahead with large-scale public investment may in fact reduce the pressure to move quickly o n private, employer-based training. Despite best intentions, assumptions with respect to cost-recovery are often not fulfilled at the level hoped for, and making very significant investments, on the basis o f overly-optimistic assumptions about how much will be recouped in fees, may lock the government into high recurrent cost obligations, which will then be unavoidable. There i s a risk that policy decisions are being made in absence o f full information on cost implications, and without good information about effectiveness. Certainly the government should proceed with TVE, but adopt a more cautious expansion path. I t i s important to be careful with the meager public resources available. I t i s possible to construct and equip workshops to lower standards and at lower costs. N o t only would this save money, i t would also improve relevance. (For example, new TVE centers have been equipped with ice-cream making 71 machines, imported at great expense, which are not used commercially in Ethiopia). B. UNIVERSITY EDUCATION 3.32 The university system i s extremely small by international standards, with 35,000 full-time degree and diploma students - representing only 0.8 percent o f the university- aged population, compared to an average o f about 4% in Africa. There are a further 39,000 evening students; and about 20, 000 in private colleges. Under ESDPII the government i s engaged in a major push to double capacity, to about 70,000 full-time students, within the next three years. This implies a need to expand capacities for management and leadership within the higher education system, to bring about greater efficiency in light o f restrictions on the recurrent budget and to begin a process o f cost recovery to lower the cost burden on government. 3.33 Total university spending in the EF95 (2002/03) budget was planned at ETB 681 million ($79 million), representing 3.4 percent o f total public expenditure, and about 19 percent o f education spending. (Table 3.3). The fairly large share o f spending allocated to the capital budget i s a result o f the substantial building program for the expansion referred to above. 3.34 The recurrent budget for universities has grown rapidly in the last few years - approximately doubling since 2000 (EY92). This i s to some extent due to enrollment increases, and to the incorporation o f new institutions in the education budget (for example the Black L i o n hospital, now included with Addis Ababa University, with an annual running cost o f ETB 24 million (about $3 million)); but it also represents a real increase in spending per student. Table 3.3: Evolution of University Spending - Budgets 1997197-2002103 (Current ETB millions) 1997198 1998199 199912000 2000101 2001102 2002103 (EY89) (EY90) (EY92) (EY93) (EY94) (EY95) Recurrent 146 170 240 249 324 438 Capital 148 - 284 189 _. - 180 - 243 - 243 Total 295 454 429 430 567 68 1 Note: N e t o f TVET Spending, covered in previous section 3.35 We do not note major mis-allocations in the composition o f university spending (See Figure 3.1). The share o f food and accommodation i s large, but not excessive compared to some other African countries. The university authorities realize that having to feed and house students i s a major tax on their budgets, i s essentially inequitable (the state does not for example feed and house the other 99.2 percent o f - generally much poorer - youths who are not in university), and limits the capacity to increase intake. Furthermore, the management burden on university administrators i s enormous, by the estimate o f University presidents taking up to 75% o f management time. The authorities are committed to trying to introduce cost-recovery in this area; although as in other countries, the issue i s a sensitive one, and will likely take some time to phase in. (see discussion o f financing below). 72 3.36 One other possible area for rationalization includes the shift of large numbers of managerial staff positions into teaching ones; but o n the face o f it the distortions associated with this do not appear very large compared to total recurrent spending. Figure 3.1: Composition o f University Recurrent Budgets- 2002/03 Capital in etc. 3.37 The majority o f recurrent spending - over 40 percent - i s accounted for by Addis Ababa University (Table 3.4); note however that the many o f the other institutions are former technical institutions that have been converted only recently into universities, and deliver substantially less resource-intensive programs. Also, the fact that expenditure-per- student at AAU i s double the average needs to be tempered by a recognition that most graduate students are concentrated at the Addis uni~ersity.~’It also includes a teaching hospital, recently transferred from the Ministry o f Health budget. Nonetheless, the large differential in spending per student with other universities bears further investigation. 89 The budget cannot be separated by graduate and under-graduate activities. 73 Table 3.4: Recurrent University Spending by Main Institutions - 2001/02 Budget (ETB millions; budget per student in ETB) Recurrent Budget Enrollment Total Average Budget Share Share Enrollment per Student Addis Ababa University 135.0 41% 19% 5,237 25,914 Alemaya University 22.2 7% 10% 2,724 8,139 Bahir Dar University 24.7 8% 11% 3,108 7,937 Mekelle University 27.1 8% 10% 2,791 9,714 Debub University 29.2 9% 14% 3,811 7,661 Jimma University 34.0 10% 13% 3,720 , 9,127 Gonder College o f Medicine 17.9 5% 4% 1,211 14,795 Arba Minch Water Tech Inst. 13.1 4% 6% 1,554 8,407 Nazaareth Technical College 7.3 2% 3% 848 8,547 Addis Ababa College o f Commerce 10.0 3% 7% 1,871 5,348 Ambo College o f Agriculture - 7.4 2% - - 3y o 777 9,567 Total 328,480 100% 100% 27,652 11,879 Note - enrollments show are regular diploma and degree students; there are significant numbers of part-time and evening students as well; the full-time numbers give a reasonable measure o f relative size, however. 3.38 There also does not appear to be any significant mismatch in the distribution o f effort across disciplines. By and large students are concentrated in areas which are central to Ethiopia’s development, and one does not see the heavy concentration o f students in more esoteric subjects common in some countries, suggesting that well-off youth are not ‘parking’ themselves in the university system. The box below shows the approximate distribution o f full-time degree plus diploma holders in the public university system: Social Sciences 10% Commerce 13% Medicine 12% Science 8% Education 23% EngineeringtTech. 16% Agriculture 14% Other 2% 3.39 Although not revealed in the aggregated data, the gender bias in enrollment in public institutions i s quite pronounced: only about 25% o f the undergraduate enrollments are female. Since higher education is, at the present, free to those admitted, a significant inequity exists in the use o f public funds. Part-time enrolments reveal that females are highly interested in pursuing higher education, nearly 75% o f the evening, part-time students are female, and are self-paying. Both the demand for services and the ability to pay for higher education exist and need to be reflected in redirected public subsidies, funding formulas, and recruitment strategies. Capital Spending 3.40 Expansion o f the system under the SDPRP and ESDPII goals calls for the introduction o f new faculties and expansion o f existing ones in line with national 74 development objectives and perceived demand for technical skills. Govemment has embarked o n a three phase infrastructure investment program to provide additional space for increased enrollment. 3.41 As projected, the capital investment required (about US$129 million equivalent) i s planned as a one-time outlay to “catch up” on under-investment in recent years. The recurrent cost implications o f the expansion have not been carefully calculated, nor the additional capital outlays that may be required to further develop and update existing facilities. These figures also do not include the cost o f the ambitious ICT upgrading that i s scheduled for 2003 -2004. Although some computer equipment has already been purchased by the Ministry, the additional capital costs required for establish an information network and for adapting the technology to administrative and classroom use will be additional financial burdens. Budget Execution 3.42 As in other sectors there i s an element o f under-spending . Table 3.5 shows the outturn for most recent year for which data i s available. The under-spending on the non- wage recurrent budget i s a somewhat surprising result, and bears further investigation. The under-spending on the capital budget i s - as noted elsewhere- to some extent a reporting problem, to the extent that aid resources were either over-optimistically projected in the budget, or under-reported once spent. However the majority o f university capital investment i s financed by the government, for which reporting should be complete. Table 3.5: Actual vs. Budgeted Expenditure - University Sector, 2000/01 (EY1992) (ETB millions) Budget Spent Share Salaries 137 142 104% Non-Wage Recurrent - 160 - 112 73% Recurrent 297 254 85% Capital 393 280 71% University Budgeting and Financing Issues 3.43 The Ministry o f Education proposes to introduce a funding formula that will provide incentives for institutional change in the pursuit o f the reform goals. At present, institutional funding i s conducted on a negotiated line item basis, with annual increments based on inflation and revenue availability. The shift to a funding formula i s intended to lay the foundation for the introduction o f block grants as the future means o f funding the universities. K e y decisions about funding formulae, revisions to the process o f allocating resources to institutions and encouraging greater individual institutional autonomy are currently being made and may have a significant impact o n the way resources are generated and disbursed. 3.44 The Ministry has done a great deal o f work in preparing i t s formula funding model. The chosen variables are: (a) estimated unit costs for major disciplines; (b) student enrollments; (c) number o f graduates; (d) amount and quality o f the institution’s research and community services; (e) the proportion o f women and minorities; and ( f ) the amount o f income generated internally. The conceptual approach i s o f good quality. The purposes 75 and component parts o f the formula have generally been well defined and appropriately operationalized. The entire funding formula should probably be progressively applied to the allocation o f public resources to the universities over 4 to 6 years, rather than phasing in one variable at a time as i s currently propo~ed.~’ 3.45 There i s almost universal agreement that the judicious use o f formula funding, guided by associated performance indicators, leads to the more efficient and effective use o f public resources. In short, better value i s obtained with the government’s money. In order for govemment to receive the full value o f formula funding, however, adequate levels o f transparency, accountability and autonomy are required by institutions in the operation o f devolved budgets. It i s also important that those institutions not meeting the established performance targets will be penalized by receiving fewer financial resources. Cost-Recovery 3.46 Govemment has indicated that it aspires to recover approximately 30 percent o f higher education costs from students in the form o f a graduate “tax”- at least that i s the preferred recovery mechanism at the moment. This sum i s derived in part from the estimated percentage o f the total amount per student that i s currently spent o n room and board (free to all full-time students). The cost o f adding dormitory space i s nearly equal to the cost o f the additional classrooms planned in the current investment period. Privatizing the service delivery o f food and the management o f dorms would result in both a cost savings to the system and would provide the ability to refocus efforts from logistics management to instructional support. At present, nearly one-half o f university faculty are civil servants - all o f them belonging to the management cadre o f the university system. Faculty all serve on a contract basis and are not civil servants. 3.47 Removing the food and lodging service from the responsibility o f the university would not only allow many staff whose sole function i s to manage these services to be reallocated to faculty roles (or at least their budgets shifted to this purpose), it would shift a substantial portion o f the cost o f university education to the students, who would ordinarily be responsible for their own maintenance. 3.48 Graduate Tax. The govemment goal o f achieving a 30% recovery o f tertiary education costs i s laudable but may not be feasible through a post hoc recovery scheme. Additionally, as noted above, the overall financing plan for the higher education expansion and reform effort may require higher levels o f government budgetary support over time than are currently expected. Considerable financial outlays will be needed over the next five to ten years in order to enable the ambitious expansion o f enrollments n o w underway. Yet, the graduate tax revenues will only begin to trickle in to the treasury five years after the tax i s announced. At that point, it will represent a rather small (but as yet uncalculated) share o f total higher education costs. In the meantime, a substantial building program of student dormitories and cafeterias will make it increasingly difficult for government to consider the option o f withdrawing from the direct management o f these facilities at a later date. In i t s worse case scenario, the potential risk i s that an unexpectedly (and The latter approach, although easier to implement due to the ready availability o f the necessary statistical information, w i l l temporarily distort the desired effects o f the formula until all o f the variables come into play during the fifth year o f implementation, possibly creating confusion and possible negative reactions among university leaders. 76 unsustainably) high budget for higher education may be generated that becomes “locked in” with little flexibility for downward adjustment. 3.49 I t i s unclear exactly how much the graduate tax i s likely to contribute to the government’s revenue diversification efforts for higher education. In some cases, critical information remains outstanding (e.g., definition o f the salary thresholds for different tax levels; the total numbers o f proposed waivers, and the criteria for their award). In other cases, the reality o f assumptions would benefit from testing (e.g., likely repayment rate; maximum graduate tax rate on top o f income and other taxes). As a result, it would appear important, and indeed urgent, for the Ministry to combine these plans with available data into a simple financial model that would permit a more accurate estimation o f graduate tax receipts and the amount o f financing likely to be required from other sources. 3.50 If our concerns are confirmed by the financial modeling exercise, then a significant need for additional financing will be evident. In this case, the Ministry may wish to revisit the question o f student fee charges. For example, specific fees might be introduced for the maintenance and upkeep o f dormitories. Fees such as these or others (e.g., privately sponsored students) could help to make an identified financing gap more manageable for government. Availability o f staff and recurrent cost implications. 3.51 Perhaps the most daunting challenge to the implementation o f Ethiopia’s higher education reform i s the sheer physical numbers o f academic staff that will need to be recruited and trained over the coming years. For example, one estimate shows that if the system i s to reach public enrollments o f 130,000 students by 2007’l, some 3,600 new academic staff will be required. This represents a 125 percent increase in just five years, and would mean 1,082 new PhD holders and 1,804 new Master’s degree recipients would need to be hired. At present, just 40 Ph.D. students are registered at Addis Ababa University. 3.52 Many would see this as an impossible task. I n addition, the supply o f Bachelor’s degree-holders from which to recruit these numbers may not be sufficient. The public higher education system i s currently producing 3,959 Bachelor’s degree-holders a year. Of these, only 250 appear to have specialized in education, and 150 in technical education. Assuming that roughly 2,600 new qualified secondary school teachers and 1,100 qualified technical education teachers would have to be recruited in order to reach ESDP I1 enrollment goals for secondary and technical education, this would mean that secondary education teacher recruitment needs would absorb 30% o f higher education graduates over the next three years. If this occurred, then roughly 35% o f the remaining degree graduates would need to be recruited as higher education academic staff over the next three years in order to get them started on the necessary graduate training programs. In other words, the Ministry o f Education alone would have to hire 55% o f the country’s total degree-holder output from public institutions during the ESDP I1time period. T h i s seems highly improbable. ’’ Full-time and part-time. 77 Private Sector Institutions 3.53 Private tertiary institutions in Ethiopia are a relatively new phenomenon. Virtually all o f them have been established within the past five years. Today some 37 private colleges and institutes (not all o f them accredited) enroll more than 20% o f the nation’s higher education students. They all offer diploma programs, a half dozen o f them have mounted degree programs, and one has initiated a Master’s degree program. These private institutions provide diverse educational programs often not available in the public institutions, enable a significant expansion o f tertiary enrollments at almost no cost to government, provide client-oriented instruction focused on the shifting needs o f the job market, and attract a high proportion o f women students (almost 50%). Their tuition fees generally range from Birr 2,500 to 3,500 a year. Recently, some o f their graduates have staffed a new software services firm that earns foreign exchange through the international sale o f i t s services. 3.54 Private tertiary institutions face various obstacles. One i s the lack o f clarity regarding the import tax exemption they are reportedly entitled to for the importation o f instructional materials, chiefly books. Another i s the difficulty o f obtaining loan financing for business expansion. Because these institutions possess l i t t l e in the way o f tangible collateral, they are rarely able to obtain bank loans. In an effort to address these obstacles, private tertiary institutions have recently formed their own association, called the “Ethiopian Private Colleges Forum”. The Forum’s member institutions hope that it will permit a more coordinated response to common problems, encourage the sharing o f information and experience, and facilitate dialogue with government. Projections: 3.55 As in other sectors, we project the costs over the next two decades o f three scenarios for expanding university education. T h e low scenario corresponds to an expansion in the next 5 years that i s faster than in the past (10 percent p.a. enrollment increase), but slower than the doubling planned under ESDP; and just doing a bit better than keeping up with population growth thereafter. Unit spending i s kept at the current levels. T h e moderate ‘extended PRSP’ scenario - includes full implementation o f the ESDP plans to double enrollmentsg2 in the next 5 years; and thereafter continues to expand at 4 percent p.a.; to the point at which about 2 percent o f university-aged population i s in school by 2020. Unit recurrent spending per student i s increased by 10 percent in real terms, to allow for quality improvements. T h e ‘Ambitious’ Scenario - includes full implementation o f the ESDP plans to double enrollments now, and then increases to the point where 25 percent o f upper-secondary graduates can be absorbed into university. (Total enrollment rises to 2 16,000 by 2020, representing 3 percent o f the university-aged population.) Recurrent spending per student i s increased 20 percent in real terms. 92 In fact intake capacity i s doubled in three years under the ESDP scenario and this one, but it takes about 5 years for total enrollment to double. 78 Table 3.6: Results o f University Projections - Full-time Enrollments and Annual Costs (Constant 2002 ETB millions) 2010 2015 2020 Low Scenario Enrollment 54,464 63,138 73,195 Recurrent Cost 517 600 695 Cauital Cost 66 77 89 Total 584 677 784 Middle ‘Extended PRSP’ Scenario Enrollment 89,636 109,055 132,682 Recurrent Cost 937 1,140 1,387 Cauital Cost 144 175 213 Total 1,081 1,315 1,600 ‘Ambitious ’ Scenario Enrollment 98,003 157,076 2 16,149 Recurrent Cost 1117 1791 2464 Cauital Cost - 494 - 494 494 Total 1,611 2,285 2,958 3.56 The moderate ‘Extended PRSP’ scenario best reflects government’s current plans, which involve a big push to double intake in the next few years. (There are not formal longer-term targets; in discussions with officials, they tend to see the current push as moving the system to a new plateau level, which would then presumably be sustained for some time). The costs o f this scenario would imply increasing recurrent budgets for the universities to about ETB 1.1 billion p.a. ($127 million p.a.) by 2015, or about three times the current level in real terms. 3.57 The ‘ambitious’ scenario i s not really very ambitious by international standards. Although it involves a target o f 216,000 university students by 2020 (EY2013), this would still only take Ethiopia to the point where 3 percent o f the university-aged population went to university, and would cost ETB 2.3 billion ($263 million) annually by 2015, about equal to what i s currently spent on primary and secondary education combined. 3.58 Undoubtedly the expansion o f tertiary education i s justified given Ethiopia’s development and capacity needs; and a scenario somewhere between the ‘moderate’ and ‘ambitious’ scenarios i s probably reasonable. However the planned expansion will be expensive, and it i s important to examine ways o f doing it at lower cost. 3.59 Details will need to await the findings o f on-going sector work o n higher education, but options include lowering the level o f non-educational spending (at present the government spends about ETB 2,000 per student annually o n food and accommodation), or increasing the efficiency o f spending, for example by adjusting the ratio o f teaching to non-teaching staff. 3.60 Given the substantial private returns to higher education, there i s a case for greater cost-recovery, either through fees, and/or a system o f student loans. The govemment i s currently considering a ‘graduate tax’ designed to recover 30 percent o f university costs. Preliminary simulations suggest i t i s unlikely the tax would generate these hnds o f revenues. But to the extent cost-recovery can be increased, the additional resources will 79 either reduce demands on the budget, or can be used to increase quality further. I t i s recognized that such proposals are politically sensitive, and would need to be phased in over time; and that careful attention would need to be paid to access implications - including considering the scope for waivers and scholarships. 3.6 1 The need for cost-recovery not withstanding, there i s s t i l l substantial justification for public spending on university education. These issues will be dealt with in more depth in the forthcoming sector study on higher education, but they include the need for a much- expanded cadre o f educated professionals in both managerial and technical aspects o f development ; and the desirability o f public intervention to ensure that women, those from poor households, and from less-developed Regions, are not denied the opportunity for higher education. There i s not good data that would allow us to assess the ‘right’ level o f university output, but i t would seem that in an economy o f almost 80 million people producing the 20,000 or so graduates a year projected under the ‘Extended PRSP’ scenario by 2010 would not be excessive. However linlung university throughput more explicitly with labour market demand over time will be important in reviewing the scale and pace o f expansion. 80 4. PUBLIC HEALTH EXPENDITURE AND HEALTH IN ETHIOPIA93 A. INTRODUCTION 4.1 The purpose o f this chapter i s to analyze expenditures by the Ethiopian government on i t s health services and public health activities. I t will present recent trends in this expenditure, evaluate its effectiveness, and project financing requirements necessary to reach health targets set by the government. The scope o f this chapter i s narrower than in many other reviews because family planning and nutrition programs are dealt with separately in succeeding chapters. Also, a more detailed analysis o f the health sector i s being prepared in a Health Country Status Report, allowing this chapter to focus specifically on government health expenditure and resource allocation. B. CONTEXT 4.2 Health status in Ethiopia i s extremely poor. L i f e expectancy i s about 47 years at birth. Infant, child, and maternal mortality rates are among the highest in the world. The major causes o f death are infectious diseases (TB, respiratory illnesses, malaria, gastro-intestinal infections, meningitis, A I D S ) . Infectious diseases and parasites also account for the largest share o f morbidities, among which malaria, helminthiasis, respiratory tract infections, skin and gastrointestinal infections, sexually transmitted diseases, and TB are the most common. 4.3 Health status i s improving in Ethiopia, but slowly. At the current pace o f improvement, Ethiopia will not be able to meet the MDG targets for health. Meeting the targets would require reducing: 0 Child Mortality from about 160 per 1000 live births to under 70 per 1000 live births; 0 Maternal Mortality from between 500-700 per 100,000 births to under 200 per 100,000 births; 0 Infant Mortality from nearly 100 per 1000 live births to under 35 per 1000 live births; and 0 Stabilizing HIV/AIDS prevalence at 7.3% and Malarial prevalence at 6.2 per 1000.94 Some o f the targets for the medium term, laid out in the Health sector Development Program(HSDP) and re-stated in the SDPRP, are summarized in Table 4.1. 93This chapter was prepared by WSavedoff o f WHO, in partnership with W.J.Smith o f the World Bank. 94 Note that many o f these baseline figures are imprecise or contested. For example, estimates o f Matemal Mortality range as high as 1400 per 100,000 births in 1990. 81 Table 4.1: Selected Health Status Targets for Ethiopia Millenium Development Goals H S D P I I Base SDPRPI & Related Targets Year HSDPII for (2000/01) 2004105 I By2015 Reduce child USMR(per 1000) 167 1 160 ( 0 70 mortality by 2/3 IMR (per 1000 live births) 97 1990-20 15 % under 1 immunized against measles 314 Reduction in MMR (pe~O0,OOO) 500-700 MMR B~rths with skilled assistance (inpercent) 16 I 1 I 25 I Halt and reverse HIV prevalence among 15-24 year HIV/AIDS olds (in percent) 18.7 Condom use prevalence rate D 2 1 Number o f children orphaned by AIDS Halt and reverse Prevalence & death associated with 7.7/1000 malaria and other malaria major diseases % o f population in malarial risk areas using preventive measures and receiving treatment Prevalence & death rate from TB TB cases detected and cured under TB *TI3 DOTS (in percent) services services 75 -- 60 Access to Health Services(in percent) 52 65 Health Budget Share o f Govt. 5.2 8.2 Spending (in percent) 4.4 Many factors contribute to these conditions o f poor health - poverty, malnutrition, l o w levels o f education, high fertility, limited sanitation and inadequate health services. Other chapters o f this expenditure review address strategies for dealing with high fertility rates, malnourishment, l o w educational attainment, and limited access to potable water and sanitation, while the Sustainable Development and Poverty Reductions Strategy Paper (SDPRP) addresses poverty. This chapter focuses on the role played by government expenditure o n health services within this setting. 4.5 The supply o f health care services i s extremely limited in Ethiopia by any measure. Ethiopia has 1,888 doctors and 12,838 nurses for a population o f close to 70 million. The supply i s also unevenly distributed, with most health care professionals and facilities clustered in major urban areas. The government operates most o f the health care facilities in the country, particularly at the secondary and tertiary level, and in rural areas. However private and NGO health care services, which were not permitted until recently, have expanded in the last few years. Private health services are concentrated in the country's major urban centers (where public health staff are permitted to complement their incomes with private practices), while non-governmental (NGO) health services are also found on a limited scale in the country's towns and rural areas. 4.6 Overall, Ethiopians have very limited contact with health care services. The government reports that there were 0.27 consultations per capita in EFYl993. According to the most recent 82 Welfare Monitoring Survey (WMS, EFY1993), about one-fifth o f the population had at least one episode o f illness in the previous two months. O f these, only about half sought some kind o f treatment, the majority o f whom went to government facilities. Another 25 percent sought care through pharmacies, traditional healers, and unspecified caregivers. I n order o f importance, distance, quality o f care, and cost were cited as the main reasons for deciding whether or not to seek health care and where to go to seek it. 4.7 The limited number o f operational facilities forces people to travel long distances to get care. Average travel time to get care i s about 2.3 hours, but when people arrive they have to wait an average o f 7 hours at a public facility to get care (compared to an average wait o f 2.3 hours in private facilities and 6.2 hours in NGOs).” People react to the perceived quality o f health services, with 53 percent o f those surveyed bypassing a closer facility to reach one that they believe i s better. Sometimes bypassed facilities could have provided the necessary services, but in many cases the closer facility regularly lacked drugs, supplies, or staff so that people preferred to travel further. In addition to the inconvenience and cost o f transportation, this leads to public expenditures that are greater than necessary due to l o w rates o f capacity utilization in bypassed facilities. 4.8 Overall, Ethiopia spent about 1.9 billion birr in EFY1994(2001) o n health care services, o f which government expenditures accounted for 40% (See Table 4.2 and Figure 4.1). Out-of- pocket spending by households accounted for another 52%. The remainder was spent by NGOs and non-profit institutions (about 7%) and public enterprises (about 2%). More recent estimates o f national health expenditure are being prepared, and it i s likely that they will reveal a higher level o f spending than indicated here. The figures on public expenditure, which are the focus o f this chapter, are not likely to change substantially. However, private spending by employers in the formal sector and by NGOs and other agencies outside o f governmental channels may be substantially higher.96 Table 4.2 : National Health Accounts Data for 2001 (WHO Estimates) Mi11ions Share of Total Birr US% Birr (in percent) per person per person Total expenditure on health 1,942 100 29.79 3.50 ... General government expenditure on health 771 40 11.77 1.38 ..... .... Central govemment 299 1 4.60 0.54 ..... .... State, regional, provincial authorities 466 24 7.16 0.84 ... Private expenditure on health 1,171 60 18.02 2.12 ...... Public enterprises 36 2 0.56 0.07 .. .... NGO’s and non-profit institutions 135 7 2.08 0.24 .... .. Net out-of-pocket spending on health 1,000 52 15.38 1.81 Sources: Various sources including Min. o f Health, Min. of Finance, IMF Statistics, and World Bank. Note: Figures on govemment spending differ somewhat from those in later tables due to different sources and definitions. A more recent study has estimated substantially more federal and regional spending, as well as ODA, see footnote 44. 95 See Fairbank, 2001. 96 N e w data became available after this study was completed in a draft National Health Accounts study conducted by the Health Care Financing Secretariat o f the Ministry o f Health. Contrary to expectations, the estimates o f private spending were not substantially higher. However the study did estimate more spending at the federal level (for a total o f 672 million birr) partly as the result o f spending through the military and Ministry o f Education, and by donors outside o f government channels (255 million birr). As a result total health spending was estimated at $5.60 per capita instead o f $3.50 as indicated in Table 4.2. 83 Figure 4.1: Composition of Health Spending- EFY1993 (2000/01) NGOs and Out o f Pocket 52% Government 15% I would be quite different. The NHA estimates more spending by government and external agencies; as a result although the amount of out-of-pocket spending i s similar, the share appears to be substantially smaller, closer to 36%. 4.9 The foregoing national averages mask a great deal o f geographic variation. The epidemiological profile o f the country's highland areas differ from the lowlands. Urban areas are better supplied with health services and potable water than rural areas. Government health services, spending, and policies also vary across Regional, zonal and woreda divisions. 4.10 Most o f the following analysis will express this geographic variation in terms o f differences across the country's 14 Regions since this i s the form in which data are readily available. This i s not problematic for analyzing public policy since the Regions account for 87 percent o f government health spending. However, it i s problematic for analyzing epidemiological conditions and healthcare-seeking behaviors because in some cases highlandlowland or urbadrural differences are more significant or relevant than Regional differences. For example, among children under five, Regional differences are more significant than whether place o f residence was urban or rural in explaining differences in vaccine rates, malnutrition and the likelihood o f being treated for ARI. By contrast, place o f residence was a more significant factor in explaining access to antenatal care and trained birth a ~ s i s t a n c e . ~ ~ 4.1 1 The variation across Regions must be evaluated in light o f the large differences in geographic area, population, and urban character. Two o f the "Regions", Addis Ababa and Dire Dawa, are formally denoted ''City Administrations" because they are essentially large urban areas.98The remaining Regions are largely rural. They include three Regions, Amhara, Oromia, and SNNPR, with more than 10 million people each. Tigray and Somali have almost 4 million people each and Benishangul, Gambella, and Harari have fewer than one million people combined. Thus, conclusions from average statistics must be qualified by recognizing the big differences in size and population density. 97 CSR, forthcoming, based on odds-ratios calculatedwith DHS 2000. 98For simplicity, the rest of this chapter will include Addis Ababa and Dire Dawa as "Regions" although formally they are Administrative Units. T h i s is in line with the Ethiopian govemments own presentationo f subnational data. 84 c. PUBLIC HEALTH SYSTEM, SERVICES AND OUTPUTS T h e Public Health Service System 4.12 Ethiopia i s currently consolidating its governmental health services into a four-tier system characterized by primary care health units, district hospitals, zonal hospitals, and referralhpecialized hospitals. The model for organization o f primary care involves a health center complemented by an average o f 5 satellite health posts which should provide comprehensive and integrated primary health care services and some surgical treatment. The District Hospitals are supposed to provide comprehensive outpatient and in-patient services; while Zonal Hospitals and Specialized Hospitals should provide specialized services, along with clinical training for nurses, health officers, generalists and specialists. In practice, establishing this model has proceeded slowly. I t has depended not only on new construction and staffing but also on rehabilitating and restructuring facilities that do not fit in the current scheme. 4.13 Responsibilities for health care services and different health system functions are divided among the different levels o f government as a consequence o f the country's overall process o f decentralization. After the fall o f the Derg in 1991, most government responsibilities were transferred to 14 Regions and Administrative Units. The larger Regions, in turn, established administrative units called Zones. In EFYl994, the government began to implement a second wave o f decentralization which transferred most responsibilities to local districts called woredas. 4.14 As a result o f this latest decentralization policy, almost all responsibilities for primary health care delivery are assigned to the woreda and i t s Woreda Health Office. In practice, responsibilities continue to be shared among different administrative levels. The Woredas that have assumed direct responsibility for health services fully manage the primary level o f health care. Although these woredas have formal authority to hire, fire, and manage personnel, they are s t i l l dependent upon zonal, Regional, and national level authorities for allocation o f graduates from the training programs, and all contractual terms are uniformly set by the same c i v i l service code. The Woreda Health Offices also rely upon the Regions and Zones for support related to continuing training, distribution o f drugs, and supplies. 4.15 The remaining responsibilities are divided among Zonal, Regional and Federal authorities. The Zonal Health Departments manage Zonal, Referral/Specialized, and District Hospitals. The Regional Health Bureaus are responsible for operating Regional Hospitals and for training health care workers (with the exception o f doctors whose training i s the responsibility o f the Ministry o f Education). The Ministry o f Health retains responsibility for formulating policies and supervising their implementation; setting standards o f care; issuing licenses; establishing standards for research and training; and coordinating external loans and grants. 4.16 This decentralization presents opportunities and challenges. I t i s hoped that increasing local responsibility will bring health service management closer to the client and make it more effective and responsive to local needs. However, decentralization also increases reliance on local capacity to manage services, which i s generally weak. In the health sector, decentralization raises difficulties for coordinating services that extend over several woredas or even across Regional boundaries, such as hospitals, referrals, maintaining the cold chain for vaccines, or supply distribution. I t also raises the need for increasing local capacity to manage health facilities, personnel, and finances in each woreda. 4.17 I t appears that the Regions are approaching this new wave o f decentralization pragmatically, with continuing interaction, support, and adjustment to assure continuity in existing health services. Within Regions, financial reporting appears to be adequate, although budget execution appears to be suffering over recent years (See Figure 4.5). 85 4.18 Regional Health Bureaus (or Zonal Health Bureaus in the larger Regions) will have to retain a significant managerial role with regard to public health services, regardless o f the amount o f authority and responsibility being devolved to the woredas. This i s true because o f the nature o f the supply chain (bulk purchasing and distribution o f pharmaceuticals, capacity to maintain equipment); human resource management (training, incentives, allocation o f staff, etc.); and specificities o f programs such as EPI and malaria control. Public Health Service Performance 4.19 It i s difficult to evaluate public health service performance in terms o f outputs and outcomes, partly because it i s not clear how much can be attributed to the action o f government providers and how much to other kinds o f health service providers. Nevertheless, government health services are the predominant source o f care in the country and other providers have only recently emerged as significant health care providers. 4.20 Another difficulty in evaluating public health service performance i s that data reported from woredas and regions to the national level regarding outpatient utilization, E P I coverage, and other service-related indicators are inadequate. This hinders the Ministry's ability to supervise the sector and creates difficulties with the external agencies that provide fimding. This latter issue was raised several times at the 5" Annual Review Meeting in April 2003. Therefore, the following information should be evaluated with these caveats in mind. 4.21 In some cases, health services appear to be making progress. For example, contraceptive prevalence reportedly rose in the last five years, from 9 percent in EFY1989 (96/97) to almost 15 percent in EFY1994 (2001), yet this i s still one o f the lowest rates in Africa. Over the same period, geographic coverage o f health services also appears to have improved. The government works with the concept o f "potential health service coverage" since there are n o reliable figures o n the population's actual ability to access health services. This i s defined as the share o f the population living within 10 kilometres o f a health center or health post. According to the HSDPI Evaluation, this "potential coverage" has increased from 52 percent to 61 percent o f the population in the last five years. 4.22 In other cases, health services seem to have just kept pace with population growth. For example, health service utilization as measured by annual visits per capita has remained almost unchanged - at 0.25 in EFY1989, and 0.23 in EFY1994 (2001/02).99 As a further example, only 30 percent o f pregnant women received any antenatal care from a skilled health worker in EFY1994, largely unchanged from five years earlier. 4.23 Finally, some services seem to be in decline. One o f the most worrisome trends has been a decline in immunization rates. Based on the 2000 DHS, D P T l coverage i s only 41 percent; DPT3 21 percent; and measles 27 percent. Other estimates based on administrative data show the current coverage o f DPT3 ranging from 47 percent (HSDPI Evaluation"') to 5 1 percent of win. Health), but in any event this coverage i s far below the rate o f 67 percent estimated to have been reached in EFY1989."' 4.24 Ethiopia has established targets related to many o f these indicators as part o f i t s health sector development planning, which includes i t s own Health Sector Development Plan (HSDP) 99The information in this paragraph and the remainder o f this section are largely taken from the HSDPI Evaluation. The loo formal title o f this document i s "Report on the Final Evaluation of HSDPI", presented at the 5th Annual Review Meeting, Nazareth, Ethiopia, April, 2003. For brevity, it will be referred to in the text as "the HSDPI Evaluation". The SDPRP (p. 99) reports that immunization coverage rose from 20% to 42%. I t i s not clear where these figures came from or which years they applied to since the HSDPI review was not completed until after the SDPRP was produced. 86 and Sustainable Development and Poverty Reduction Program (SDPRP). More recently, the Millennium Development Goals (MDGs) have been incorporated in planning exercises. I n all instances, the government aims to achieve significant progress toward improving health status over the coming 20 years with a substantial increase in financial support to the health sector. D. PUBLIC SPENDING AND ITS ALLOCATION'" Public Spending on Health 4.25 Ethiopia's public sector accounts for about 40% o f total health expenditure (See Figure 4.1). I t spent 765 million birr o n health in EFYl994, approximately 12 birr per capita (See Table 4.3). This represents a significant increase over the last ten years relative to the countryls national income. Public health expenditure represented less than 0.9 percent o f GDP in EFY1983 (1990/91) and reached approximately 1.5 percent in EFY1990 (1998/1999). As a share o f GDP, public health spending has remained at the same level since then, with the exception of 1999/2000 when i t dropped to 1.1 percent due to the war. 4.26 n nominal terms, government spending o n health rose steadily to 8.5 birr per capita in I EFY89 (1996/97), and then to more than 11 birr per capita by EY 91. In EFY94 (2001/02), government spending on health was about 12 birr per capita, only slightly more than the level reached in EFY91. Using the prevailing foreign exchange rate, public health expenditure increased from about US$1.25 per capita in EY89 to about US$1.40 per capita in EY94 (See figure 4.2). Thus, government spending on health clearly increased until EY90 (1997/98) after which it appears to have remained roughly the same. I t i s programmed to rise very substantially in the current (2003/04) budget, although it remains to be seen how rapidly actual spending will increase. Note also that the official data only captures the portion o f donor funding that i s channeled through the government. Since grants and loans to the health sector have increased in the past two years, these figures may underestimate the total increase in "public" funds dedicated to health. 4.27 Over this period, government health expenditure was affected by two major events: the war with Eritrea and the implementation o f decentralization. The impact o f the war on health spending appears to have fallen almost entirely on the capital budget. This was in part an intentional reallocation to finance the war effort, and partly a consequence o f decisions by external donors to withhold assistance. Capital expenditure plunged by one-third between E Y 9 1 and EY92, at a time when recurrent expenditure f e l l by only 12.5 percent. Capital expenditures also appear to account for most o f the subsequent increase in health expenditure (See Figure 4.2 and Figure 4.3). The fact that recurrent spending has remained constant in real terms while capital spending increased lends support to concerns expressed at Annual Review Meetings (ARM)that facilities are being built without being properly staffed and operated. lo2 N e w data became available after t h s study was completed. I t showed some important changes in the composition o f spending in 2000, but did not indicate the need to alter the conclusions regarding trends or priorities, see footnote 95. 87 Table 4.3: Public Spending on Health in Ethiopia, EFY1989-94 (1996/97-2001/02) EY89 EY90 EY91 EY92 EY93* EY94* EY95* Spending in millions o f birr Recurrent 34 1 390 459 402 468 522 526 Capital -154 - 243 252 - 167 - 297 - 273 298 - Total 495 634 71 1 569 765 796 824 Spending in birr per capita Recurrent 5.87 6.52 1.44 6.33 7.16 7.78 7.83 Capital - 2.65 - 4.06 - 4.08 - 2.63 4.55 4.07 4.44 - Total 8.5 1 10.58 11.52 8.96 11.71 11.85 12.28 Source: MOFED. Note: All figures are actual expenditure except for 1993-95 which are pre-actual estimates. Note programmed spending in the EY96 (2003/04) budget rises to ETB 24 per capita, double the level o f recent years. Figure 4.2: Public Health Expenditure ( percent GDP), EN1983-1994 1.8 1 I 1.6 1 1 IPublic Health Capital (% o f G D P ) I ~ 1983 1 1 I 1990/91 1991/92 1992/93 1993/94 1994/95 1995196 1996/97 1997/98 1998/99 1999/00 2000/01 2001/02 1984 1985 1986 1987 I I 1988 1989 1990 1 1 I I I 1991 1992 1993 1994 Figure 4.3: Public Health Expenditure (US $ Per Capita), ETY1983-1994 1.60 , I 1 1995196 1988 1 1996197 1989 ~ I997198 1990 1 1998199 1991 1 1999100 1992 1 2000101 1993 1 2001102 1994 1 4.28 The main reason for Ethiopia's very l o w per capita expenditure on health is its l o w national income. However, Ethiopia also spends a significantly smaller than average share o f i t s public budget on health (See Figure 4.4). Ethiopia has a public budget equal to more than 30 percent o f GDP but spends only 5 percent o f that budget on health. As a result, public health expenditure i s only 1.5 percent of GDP. By contrast, Tanzania has a public budget equal to about 25 percent o f GDP but spends about 10 percent of it on health. Consequently, public health 88 expenditure in Tanzania i s about 2.8 percent o f GDP. Therefore, Ethiopia's l o w level of government health spending i s due in a significant part to spending a smaller than average share o f the budget on health. 4.29 The government recognizes that the share o f public spending on health care services needs to grow. Both the SDPRP and HSDP aim to increase the health share o f public spending from i t s EFY level o f 5.2% to 8.2% by EFY98 (2004/05); and in the current (2003/04)budget it i s programmed to rise to just over 7%. Regional Allocation 4.30 Regional governments are responsible for most government spending o n health and this share has increased. The Regions were responsible for 82 percent o f total public health spending in EY89. In EY94 (2001/02), this share had grown to 87 percent. Regional spending ranged from a l o w o f between 6 birr and 7 birr per capita in Amhara, Oromia and SNNPR, to more than 20 birr in Addis Ababa, Dire Dawa, and Benishangul-Gumuz, and a high o f over 50 birr in the smallest Regions, Gambella and Harari. I t i s difficult to judge the effects o f these variations in spending; however, it i s worth noting that the government's estimates o f potential coverage across Regions are highly correlated with their levels o f per capita spending. 4.31 As part o f the broader government policy o f decentralization, a portion o f the transfers from the national government to the Regions i s n o w being passed on directly as transfers to woredas. T o the extent that transfers are related to population, this has led to a reallocation o f resources from woredas that were favored under the previous scheme to those who were not. Although this can be an effective way to redress inequitable spending, it i s also problematic for the health sector because certain woredas are responsible for services that are utilized by people from neighboring areas. I t also creates financial difficulties for woredas with larger fixed investments and staff. Figure 4.4: Government Spending as Share of GDP ( percent), 1993 40.0 30 0 20 0 x)O 00 89 Budget Execution 4.32 Between EFY1989 and EFYl994, according to Ministry o f Finance data, execution o f the health budget declined fi-om near 90 percent to under 80 percent, mostly due to lags in completing capital projects. Data collected for the evaluation o f HSDPI i s more troubling, indicating budget execution below 70 percent in half o f the Regions (See Figure 4.5 )). 4.33 A variety o f reasons are given for this problem o f budget execution. Poor planning and implementation o f investment plans i s a contributing factor that may be exacerbated during the transition toward a decentralized system. Improving budget execution now requires substantial coordination between at least three levels o f govemment. Another contributing factor i s delays in releasing funds by extemal donors. The actual magnitude o f the shortfall in spending relative to the budget i s not known due at least in part to reporting problems - in that donor funds shown in the budget are often not reported when spent - as noted in Volume 1 1. Figure 4.5: Budget Execution (Share of Health Budget Spent: EFY) -Somali - -- Tigray National -- Addis Ababa -Afar - - Amhara BanshanguCGumu; - - Dire Dawa Gambeila -- Harare Oromia 1989 1990 1991 1992 1993 1994 -SNNPR 4.34 As in other sectors, the under-spending takes place primarily o n the capital budget. The health capital budget expanded very rapidly in the years following the war, doubling to ETB 765 million in 2000/01, but the increased funds were under-spent. (see Table 4.4). The capital budget was scaled back in 2001/02, but has been increased again dramatically in the past year, raising concems as to whether this rapid expansion can be absorbed. To a large extent this reflects the increase in aid funding for the health sector (since aid i s shown under the capital budget), and to some extent, as noted in Volume 1 1, it i s a statistical problem, because donor inflows are over- estimated in the budget at the beginning o f the year, and then under-reported when it comes to actual expenditure; so the gap between ‘budget’ projections and ‘actual’ spending appears larger than it is). Unfortunately the data do not allow a clear assessment o f how much o f the problem i s represented by such under-reporting. But there i s nonetheless some cause for concem - in the past three annual reviews o f HSDP concems have been raised about facilities being constructed which cannot be adequately staffed and equipped. In a more general sense govemment and donors want to avoid a relative shift into capital intensive expenditures, since all the analysis suggests that the greatest impact o n health outcomes will likely come fi-om expanding recurrent budget spending on peripheral service-delivery o f basic services, which would imply a greater share o f recurrent spending, at the regional and woreda level. Table 4.4: Execution of the Capital Budget (Current ETB millions) 1997/98 1998/99 1999/00 2000/01 2001/02 2002/03 Budget 312 354 36 1 765 535 73 3 Reported Actual - 243 251 175 297 -273 - 298. Actual as YO of Budget 77.9% 71.0% 48.6% 38.8%* 51.0%* 40.6%* * Recent years almost certainly reflect incomplete reporting o f donor spending. 90 Allocation o f Public Spending on Health 4.35 As in many African countries, Ethiopia's public health spending i s biased toward urban areas, wealthier populations, hospital care, and capital investments. By Residence and Income 4.36 The health status o f wealthier people in Ethiopia i s better on average than the rest o f the population. However, given Ethiopia's l o w levels o f national income and health status, even the "richest" quintile i s quite poor and suffers from i llhealth. For example, child mortality i s 150 per thousand births among the poorest quintile and 114 per thousand births in the richest quintile. Thus, children in the richest quintile have about a 24 percent smaller chance o f dying before the age o f five, but they are s t i l l at higher risk than the average level in most low-income countries o f the world. 4.37 Despite these income-related health status differences, government health spending reaches the upper income groups before it reaches the lower income ones. While W M S data show about equal utilization o f health facilities, the DHS shows children from the top consumption quintile (that is, the better-off 20% o f the population) are about twice as likely to have been vaccinated or received Oral Rehydration Salts (ORS) for diarrhoea than those in the poorest quintile (see Table 4.5). Among women who gave birth, those in the richest quintile were almost four times more likely to have been assisted by someone with training than those in the poorest quintile. The 15 percent o f the population living in urban areas i s also more likely to receive health services they need than the 85 percent living in rural areas. Table 4.5: Adjusted odds-ratios for children under 5 years old, EFY1993 (2000) Wealth Quintile Poorest zUd Middle 2nd Richest Urban Poorest Richest relative to Rural DPTl 1.oo 0.85 1.17 1.36 2.22 1.19 DPT3 1.oo 0.87 1.3 1.55 2.04 1.41 All EPI Vaccines 1.oo 1.19 1.71 1.94 2.16 1.66 Trained birth Assistance 1.oo 1.61 1.40 3.71 3.87 2.63 ARI Treatment 1.oo 0.87 1.18 1.06 1.74 1.40 ORS Treatment 1.oo 1.42 2.37 1.92 2.36 3.15 Source: World Bank Country Sector Report, forthcoming. Note: The figures are adjusted odds ratios calculated from DHS 2000 for children under 5 in Ethiopia. The odds ratios control for Region, urban residence, size o f household, religion, mothers age and education, among other factors. Numbers in bold face are statistically different from the poorest quintile at the 10% significance level. The numbers can be interpreted as the probability o f receiving the respective service relative to the poorest group, after controlling for other factors. 91 Table 4.6: Composition of Recurrent Budget, EFY 1989-1994 (ETB thousands) EY1989 EY1990 EY1991 EY1992 EY1993 Share (1996/97) (1 997/98) (1 998199 (1999/00) j2000/01) (in percent) By Inputs Wages and Salaries 190,023 213,010 235,106 234,658 282,854 60.5 Operation and Maintenance 29,439 3 1,947 80,481 6 1,972 34,318 7.3 Materials and Supplies 114,685 138,767 154,854 101,932 120,739 25.8 Medical Supplies and Equipment 73,721 96,043 107,394 61,344 76,847 Petrol and Lubricants 5,374 7,146 8,477 6,789 7,991 Other Materials and Supplies 35,590 35,577 38,983 33,799 35,902 Grants, Contrib., Other Transfers 29,260 17,684 33,036 27,561 23,772 5.1 Motor Vehicles & Equipment 4,625 6,213 5,627 3,007 5,705 1.2 By Function Administration 65,977 90,863 167,547 131,683 95,984 20.5 Health Centers and Clinics 111,229 118,022 121,775 106,747 142,186 30.4 Hospitals 138,729 157,032 172,561 144,791 177,854 38.1 Other 52,099 41,704 . 47,221 45,909 51,364 11.0 By Government Level Regional 301,761 332,626 404,590 351,177 408,672 87.4 Federal 66,273 74,994 104,5 14 77,953 58,716 12.6 Federal Share (in percent) 18.0 18.4 20.5 18.2 12.6 Total Recurrent Expenditures 368,033 407,621 509,104 429,130 467,387 100.0 Source: MOFED Note totals do not match with estimated total health spending shown jewhere, due to incomplete reporting from Regions, especially in later years, but the composition of spending i s not significantly affected. By Service Level 4.38 Recurrent spending on hospitals i s higher than recurrent spending on primary level care. In EFY1994 (2001/02), the government spent 178 million birr o n recurrent expenditure in the country's 80 public hospitals; while it spent 142 million birr o n the primary health care level. (See Table 4.6). It i s normal for countries to spend more on tertiary and secondary health care than primary care due to the large differences in average costs for inpatient and outpatient care. Typically, countries spend 60 percent or more on tertiary and secondary health care. Nevertheless, given Ethiopia's large burden o f diseases that can be prevented or treated o n an outpatient basis in primary care settings, i t i s likely that additional spending at the primary level will have a larger impact on health status than at the hospital level. The government's budgets since EFY1994 have projected a shift in emphasis toward primary level care that may become apparent in expenditure data as it becomes available. Between Capital and Recurrent Costs 4.3 9 Recurrent spending represents approximately 60 percent o f the government's spending o n health with the remaining 40 percent going to capital investment. This share increased during the war as capital projects slowed or halted, but has since returned to i t s pre-war levels, and in 2002/03 only 47% was budgeted for recurrent spending. This share has been reversed in the current (2003/04) budget, with a large programmed increase in recurrent spending that takes the share to 65%. It i s not easy to determine whether this i s an appropriate level. Ethiopia has clear needs for expanding physical infrastructure, however, this expansion has to occur at a pace that allows health care facilities to be staffed and operated properly. Without a health facility survey, a rigorous evaluation o f whether this balance i s being attained was not possible. 92 4.40 In addition, much donor funding i s recorded as capital expenditure even when it i s spent o n consumables such as medications, vaccines, gasoline or per diems. Donor funding i s significant, but the share that goes through government accounts as capital expenditure i s not known. Therefore, the recurrent share indicated above i s probably an underestimate, but it i s not clear by how much. 4.41 At the Annual Review Meeting o f HSDP in April o f 2003, concerns were raised that health care facility construction after the war was proceeding too quickly relative to the government's ability to staff, operate, equip, and maintain these new facilities. At that meeting, Regional health authorities indicated that investment plans were being redirected toward rehabilitation and equipping o f existing facilities, but adjustment would be slow because o f the lead time for contracting construction. Training and employment o f nursing and paramedical staff appears to have increased rapidly enough to staff the facilities that were constructed (See Figure 4.7); however, without a survey, it i s impossible to determine if these new staff are deployed appropriately to operate the new health centers and health posts. 4.42 An additional potential concern i s that the transfer o f budget and responsibilities to woredas will lead to a greater capital bias as a result o f local politicians seelung to "build" visible achievements. However, the experience in the first year o f transfers was that many woredas were not receiving enough funds to do more than cover their payroll expenses. Thus, it i s not clear whether decentralization will improve or worsen the balance between capital and recurrent expenditures Figure 4.6: Population per Nurse and Health Assistant by Region, EFY1993 - 93 Figure 4.7: Annual Change in Human Resources, Facilities and Population by Region, EY89-94 (19972002) Source: HSDPI Evaluation Recurrent Spending Composition of Recurrent Costs by Inputs 4.43 The composition o f recurrent spending across inputs i s not unusual relative to other countries. Approximately 61 percent o f the recurrent budget pays for salaries, while 26 percent goes to medical supplies and medication. The remainder covers grants, transfers, and operation and maintenance. (See Figure 4.8 and Figure 4.9). Although at the national level the composition o f spending appears reasonably distributed, Regional differences are quite large and in some cases, troubling (See Figure 4.10). For example, the l o w share o f materials and supplies in SN"R i s worrisome, along with the l o w share o f operation and maintenance expenditures in most o f the Regions.lo3 Furthermore, a study o f the composition o f spending required to provide effective basic coverage suggests that the share o f medical supplies and medications should be closer to 40 percent in the context o f Ethiopia's prices and epidemiological profile. This will be discussed in greater detail in the section on expenditure projections below. 103 I t has been noted that many extemal agencies finance drugs and supplies via the capital budget. This could explain the Regional variations. Firm conclusions require more detailed analysis. 94 Figure 4.8: Composition of Ethiopian Government Health Expenditure, EFY 19931°4 Recurrent Expenditure by Function Recurrent Expenditure by Govt Level onal Capital Expenditures by Function* Capital Expenditures by Govt Level* 99% Regional 'Capital expenditures refer to 1992 because more recent expenditure data was unavailable. Figure 4.9: Composition of Ethiopian Government Recurrent Health Expenditure, 1993 Operation and Maintenance Wages and Salaries 61% / 'L Transfers Vehicles and 5% Equipment 1% More recent data from the NHA suggests significantly larger federal health expenditure; inclusion o f the Ministries o f Defence and Education bring the federal share to about 40% in these estimates, as opposed to 13%. 95 Figure 4.10: Compositionof Regional Recurrent Expenditures I Ia Purchase of Motor I Vehicles and I 100% Equipment 0 Grants, 75% Contnbutions and Other Current 50% Transfers 0 Matenals and Supplies 25% Operation and Maintenance 0% Staffing 4.44 About 61 percent o f recurrent health expenditure i s for salaries. Human resources are not only the largest single item in the government's health budget, but are also arguably i t s most critical factor. The number o f health care workers in Ethiopia i s limited and varies dramatically across Regions and woredas. In EFY1993 (2000), there were almost 50,000 Ethiopians per doctor on average, but this ratio was as high as 100,000 to one in Somali and as l o w as 12,000 in Addis Ababa. Similarly, there were more than 8,000 Ethiopians for each nurse, but this ranged from a ratio o f more than 11,000 in Somali, Amhara, and Oromiya, t o under 3,000 in Addis Ababa, Harari, and Benishangul-Gumuz. Urban areas are clearly better supplied with medical personnel than rural ones, and the variation across woredas must be very large. (See figures 4.6 and 4.7). 4.45 The general situation with health professionals i s improving in some ways and worsening in others. The government reached or came close to meeting most o f the HSDPI targets for staffing in the cases o f surgeons, internists, clinical nurses, pharmacists, and laboratory technicians. The pace o f training and placing nurses seems to be quite good according to several sources. Part o f this trend can be accounted for by increasing reliance on junior nurses and other paramedical staff with briefer training courses. By contrast, the increase in staffing levels for doctors has been geographically concentrated. The Regions have been unable to halt the movement o f doctors from their areas to Addis Ababa, Dire Dawa, or even abroad (See Figure 4.7). 4.46 I t i s commonly stated that retention o f health care workers i s a problem, but the available data do not demonstrate this except for doctors. Between EY89 and EY94, the number o f nurses grew by more than 10 percent in every Region except for Addis Ababa, a rate which exceeded the pace o f building facilities everywhere except for Dire Dawa. Nevertheless, the number o f doctors fell by as much as 7 percent in SNNPR and Gambella. Only Addis Ababa and Dire Dawa registered an increase in the number o f doctors. Given this pattern, it i s not surprising that most contacts with health care services are with nurses or health officers - about two-thirds in rural areas and less than half o f those in urban areas. Trained birth assistants play an important role in urban areas, caring for more than one-third o f those who sought care. However, they accounted for less than 10 percent o f the contacts in rural areas (WMS, 2000). 96 4.47 The HSDPI targets for training and employing midwives and public health nurses were not met by EFY94 (2001/02). This is problematic because these health care workers are essential to basic health care and for reaching targets related to infant, child, and matemal health. The HSDPI evaluation also notes that a large share o f midwives are men, which makes them culturally unacceptable to different segments o f the population, especially the country's nomadic groups. 4.48 The HSDPI Evaluation suggests there may be some misallocation in staffing between administrative and health care service functions. Complete information was not available, but examples were given o f high ratios o f administrative staff to health care workers in Addis Ababa and a Zonal hospital in Afar. It also noted that many trained health professionals are being assigned to administrative and managerial functions in the new Woreda Health Offices in order to respond to increasing responsibilities for management at these levels. 4.49 Since Ethiopia has to expand health care services massively, training and hiring more staff appears to be fundamental. In the short run, as long as the public sector cannot provide the conditions necessary to retain highly skilled professionals such as doctors, the country should concentrate o n training, hiring, and gradually upgrading less skilled professionals such as midwives and nurses, and relying more heavily on community health workers and other frontline staff. The current ratio o f other health care staff per nurse i s close to 2.5 to 1. Based on estimates o f the staffing required to provide basic services in Ethiopia, this ratio should be closer to 3.5 to 1. The fact that HSDPIIprojects deploying more than 10,000 frontline health care workers, most o f whom would be "Health Extension Workers" (HEWs), represents a good response to the priority for delivering basic services given their relatively l o w cost o f training and the greater difficulty o f retaining highly skilled workers. 4.50 The Ministry envisions these new workers as being trained only for health promotion activities; whereas many o f the Regional Health Bureaus argue that such community workers must have training and capacity to offer some curative care -ifthey are to have credibility with the communities. One way or another, this new cadre o f staff could make a large difference by putting people to work on health promotion in dispersed communities. H o w much training the HEWs will get, their required s l u l l levels, scope o f action, and requirements for supplies and support need to be resolved as part o f any effort to scale up public health activities. Drugs 4.51 Drugs account for the largest share o f recurrent spending after salaries, about 14 percent o f total recurrent spending.lo5 Government spending on essential drugs has increased over the last few years -- from 60 million birr in EFY1989 to 245 m i l l i o n birr in EFY1994 (See Table 4.7). This figure rises to 258 million birr when vaccines are included. A substantial part o f that increase came from spending out o f the government's own revenues which grew from 60 million birr in EFY1989 t o 104 m i l l i o n birr in 1994. At the same time, donor spending o n drugs increased even faster, reaching 141 million birr in EY94 due to the disbursement o f IDA funds and donations through the Pharmaceutical Administration and Supply Services (PASS).lo6 4.52 The availability o f essential medications has subsequently improved; however, shortages o f drugs are still common in public facilities. Improving drug availability i s important not only for its role in better treatment, but also because, according to the Ethiopian Health Care lo5 T h i s estimate i s derived from MOFED reports. The HSDPI evaluation in GY2003 estimated this share was significantly higher, 19 percent. As noted above, there i s donor financing that does not pass through the budget, which i s likely to be very substantial in the case o f drugs, but details are not available. 97 Delineation Study (HCFMOH), lack o f drugs i s the major factor behind l o w utilization rates o f lower tier facilities. Table 4.7: Funding for Drugs and Medical Supplies in Public Health Sector (millions o f birr) EY89 EY91 EY92 EY93 I EY94 Share EFY (96/97) (98199) (99/00) (00/01) I (01/02) (EY94) GOE Drug - Budget - 60 85 70 82 I 104 40% IDA Expenditure for pharmaceuticals 0 0 38 46 48 19% and medical equipment Donations through PASS n.a. 99 93 36% Vaccines 20 10 20 16 13 5% Total for Drugs and Vaccines 80 95 128 243 I 258 100% Share o f total public health budget 17% 18% 17% 18% I 19% Source: FMOH, as reported in HSDPI Evaluation, Table 2.1 1. Note: These figures were reported in the 2003 HSDPI evaluation and do not correspond exactly with MOFED reports. 4.53 Regional differences in availability o f essential drugs are quite large. According to the final evaluation o f HSDPI, availability o f essential drugs was fair to good in 6 o f the Regions -- Addis Ababa, Amhara, Dire Dawa, SNNPR, Harari, and Tigray. In at least 3 o f those areas, the evaluation indicated that funding for drugs had increased. By contrast, drug availability was considered poor or below 90 percent in Afar, Benishangul-Gumuz, Oromiya, Gambella, and S0ma1i.l'~ For two o f these Regions, Oromiya and SNNPR, the report said that drug spending per capita had declined; while in Benishangul-Gumuz, spending stayed the same. Information o n drug spending was not provided in the report for the other Regions. In all o f the Regions except Addis Ababa, storage and inventory control was judged to be poor. 4.54 Sources o f problems varied. Some cases demonstrated imbalances -- for example, in Benishangul-Gumuz, facilities reported excess supplies o f ORS and antacids at the same time that they had shortages o f antibiotics and anti-malarials. In other cases, timely disbursement o f funds appeared to be the problem. For example, in Oromiya, none o f the woredas in two different zones that were visited by the HSDPI Evaluation team in 2003 had procured any drugs in the last 6 months because they said that budgeted funds had not been released. Basic management, administrative and logistical problems were also cited. 4.55 Studies on private expenditures demonstrate that a large part o f out-of-pocket spending, about two-thirds, purchases drugs. This implies that private expenditure on drugs i s about four times higher than public expenditures. The introduction o f Special Pharmacies (revolving drug funds at public facilities) has coincided with a stabilization in the number o f private pharmacies in the country. The Special Pharmacies have been meeting many o f their objectives in terms o f increasing drug availability, improving quality and reducing drug prices through competition with the private sector. The HSDPI Evaluation notes that some o f these SPs are not managed well and that retention of revenues at the pharmacy level has not yet been implemented. Nevertheless, the experiences are judged successful enough that the HSDPII and SDPRP are expecting to have Special Pharmacies operating in all hospitals and 50 percent o f primary health care units by EFY1 998. Evaluation o f HSDP-1. 2003. Table lo' 2.9, in Section 2. 98 Specific Priorities & Programs 4.56 As discussed earlier, Ethiopia's burden o f disease i s dominated by infectious diseases. Recent estimates show that HIV/AIDS, malaria, and lower respiratory infections accounted for 33 percent o f all deaths in EFY 1993, and a similar share o f DALYs"'. 4.57 Given the Ethiopian govemment's extremely limited resources - both financial and institutional - the argument for orienting i t s health care service system to address these illnesses i s very strong. W h i l e programs in nutrition, family planning, women's education, and reducing poverty are the foundations for improving health status in Ethiopia, equipping health services to address five top priorities - vaccine-preventable diseases, Malaria, and Reproductive Health, TB, and HIV/AIDS - i s an equally important pillar. Vaccines 4.58 Commitment o f resources to basic immunization services appears to have declined over the last years. Funding for vaccines was reported to be 20 million birr in EFY1989 (less than 7 percent o f government recurrent health expenditures) and except for EFY1992 has been lower in every year since (See Table 4.7). In EFY1994, expenditures on vaccines were only 13 million birr, representing less than 3 percent o f the government's recurrent health expenditure. I t may be that vaccines financed by external agencies o n the capital budget have to some extent substituted for domestic funding, but the available data make i t difficult to assess the extent. Based o n this data, the financial commitment to immunization coverage i s weak. A recent study estimates that an effective vaccination program would cost approximately US$12.5 m i l l i o n per year, which i s three times the level o f spending in EFY1994, but s t i l l only 14 percent o f the total public health budget. 4.59 Other than the national campaign to eradicate Polio, which has been making substantial progress, the coverage o f immunization for other diseases has slowed or reversed. The availability o f cold chain equipment and supplies appears to have improved in most Regions. However, scarcity o f recurrent funds and shortages o f personnel continue to limit progress in increasing coverage. In field visits to a few health centers and posts, the availability o f funds for fuel and per diems was frequently cited as the major restriction o n reaching more children with vaccines, and in at least one case, GAVI funds were cited as the source o f funds that had made it possible to improve outreach. In terms o f both inputs and outputs, E P I has not received the priority it requires from the M o H or Regions. Furthermore, most Woreda Health Offices are confronting the challenges o f managing their own local health services for the first time, and concerned primarily with paying staff and maintaining facilities. I n the short run, vaccination programs that depend on National, Regional, and Zonal support, and which are not necessarily seen as priorities at the local level, may suffer as a consequence unless strong actions are taken to reverse the generally l o w priority accorded these programs. Malaria 4.60 More than 40 million people, representing about two-thirds o f the population, live in areas where malaria i s a risk at least some portion o f the year, with particularly high concentrations in Benishangul, Gambella, Amhara, Oromia and SNNPR. Malaria accounts for some 10 percent of deaths and anywhere from 10 percent to 50 percent o f all contacts with public health care providers, depending on the Region. Spending o n malaria control appears to have A D A L Y i s a "Disability Adjusted L i f e Year" -- a summary measure o f health status that accounts for the effect o f 108 morbidities associated with a disease as well as mortality. For an explanation, see World Bank, 1993. World Bank. Ethiopia: National Immunization Program, Costing and Financing Assessment, Draft, January 2002. log 99 peaked at 105 million birr in GY1998-99 o f which 80% was financed from domestic resources. Subsequently, it declined to an estimated 82 million birr in GY1999/00."0 4.6 1 Under the Rollback Malaria initiative, Ethiopia has mobilized substantial donor support and followed a strategy o f integrating malaria control activities with the general health service system. I t s reliance on Community Malaria Workers and Malaria Control Sector Offices appears to have greatly expanded the reach o f malaria treatment and control beyond the capacity and catchment areas o f health posts, health centers, and hospitals. In this regard, malaria control efforts provide lessons in delivering care through community outreach - including the importance o f logistical support, complementary supplies, and adequate financial incentives. Reproductive Health 4.62 Ethiopia has high fertility rates, maternal mortality rates, and STD prevalence. Knowledge about STDs, contraception, and reproductive health i s limited. Practices such as female genital mutilation, abduction and early marriage contribute to high rates o f reproductive- related mortalities and morbidities. 4.63 It i s not possible to estimate total funding going to reproductive health care because it i s and should be an integral part o f basic health service provision. Specific categories o f expenditure, such as contraceptive supplies, can be calculated, but give only a small part o f the picture. 4.64 Studies o n reproductive health in Ethiopia give the same message regarding the limited access afforded by existing infrastructure and staffing, and the need to put stress o n community outreach and innovative approaches if substantial progress i s to be made. This stress on innovation, community involvement and education i s recommended both because reproductive health i s strongly influenced by individual and social behavior, but also because it i s likely to improve health status more for a given level o f expenditure than comparable investments in new facilities. Solving issues related to cultural and social practices that are harmful to health requires broad community involvement and coordination among community-based institutions. Expanding outreach beyond health centers and health posts requires building on, and complementing interventions by other agencies operating in rural areas, and linking with NGOs. Spending o n improved quality o f care in existing facilities would also yield large benefits for modest investments by improving technical competence in post-abortion care, STD detection and treatment, antenatal and postnatal care, and emergency obstetric care. HIV 4.65 Ethiopia has been seriously affected by the HIV/AIDS pandemic since the virus was first reported in the country in GY1984. The disease i s not as widespread as in several countries in Sub-Saharan Africa, but it has already killed some 200,000 people and left an estimated 1.2 million children without one or both parents. About 2.3 million Ethiopians were estimated to be infected with HIVIAIDS in EY94 (2001/02), representing a prevalence rate o f 6.6 percent, though earlier estimates placed it at 7.3 percent. 4.66 The most recent surveys suggest that Ethiopia may be o n the cusp o f a more rapid spread o f HIV/AIDS over the next five to ten years unless significant action i s taken. The highest prevalence rates are reported for 15 to 24 year old individuals, about 15 percent. These 'loAll these figures are from "Rollback Malaria: Study on Malaria Financing and Expenditures in Ethiopia, Final Report" World BanmUAWAB Business Consultants, August 2002. 100 individuals must have become infected relatively recently and will continue to be sexually active for some years t o come, indicating the epidemic i s likely to continue to spread. 4.67 The country's national program to combat HIVIAIDS i s being overseen by the HIV/AIDS Prevention and Control Office (HAPCO) which implements and coordinates intersectoral efforts by different federal government offices, Regional and local governments, external agencies, donors, and c i v i l and religious organizations. The total budget for HIV/AIDS activities has been set at US$37 million and US$43 million for EFY1995 and EFY1996, respectively. Disbursement o f money in H I V / A I D S programs through government channels has been slow, and a large and increasing amount o f money i s being spent by NGOs and donors without being channeled through HAPCO. Based on interviews, i t i s estimated that about US$20 million may have been spent through these other channels in 200 1-2002. 4.68 I t appears that substantial funding i s available. The World Bank has an existing loan o f US$60 million, much o f which has yet to be disbursed, and the Global Fund recently approved a grant o f approximately $55 million for a proposal covering 2 years. Nevertheless, disagreements exist over the causes o f delays in disbursements over the last few years, and whether responsibility lies with donor agency procedures or domestic implementation capacity. 4.69 Based on World Bank calculations, experience in other African countries, and a UN consultation"', the cost o f implementing an effective program o f prevention and treatment was estimated at approximately US$60 million per year for a modest program that would stabilize and begin to reverse incidence over the next decade. A more comprehensive program, that would substantially reduce A I D S prevalence was estimated to cost US$130 million equivalent. An additional US$30 million per year i s estimated to eventually be required for ARV treatment. I f such funds were implemented effectively, Ethiopia could reduce the prevalence o f HIV/AIDS from i t s current level o f 6.6 percent .to 5.0 percent in 2010 and 3.6 percent in 2020. (See Table 4.7). Without effective action, prevalence i s likely to increase. Reference Modest Ambitious EFY - EY98 - EY13 EY03 EY03 EY13 EY03 EY13 (2005) (2010) (E)(2010) (E)(2010) (2020) National Prevalence ( percent) 6.6 7.6 9.4 6.0 5.5 5.0 3.6 Expenditures (US$million annually) 10 30 30 60 62 132 129 Infections prevented (1,000s 159 560 213 870 cumulative) E. O P T I O N S AND P R O J E C T I O N S Public Health Sector Plans & Policies 4.70 In EFY1989 (1996/97), the Ethiopian government developed a twenty-year health sector development program, with a more detailed 5-year initial phase that was called the Health Sector Development Program I(HSDPI), covering the period EFY 1990 to 1994. HSDPI comprised eight components, with the largest share o f the budget dedicated toward service delivery and quality o f care (5 1 percent), facility expansion (27 percent) and pharmaceuticals (14 percent). HSDPI introduced and strengthened a sectoral approach to the health sector. I t included a national process o f policy development, a comprehensive framework for mixing domestic and external resources, efforts to establish medium and long term budget cycles, common management structures, and a joint monitoring and evaluation process. The sectoral approach has United Nations Special Session on HIV/AIDS, June 25-27, 2001, "Declaration o f commitment on HIV/AIDS: - global crisis global action". 101 tried to shift aid relationships away from donor-financed projects to harmonization o f external assistance, direct budget support, and a dialogue between the government and donors focused on broader strategic issues. 4.71 In order to align the HSDP with the Second National Development Plan, the second phase o f the Health Sector Development Program (HSDP 1 1) covers a three-year period (EFY 1995-1997), and constitutes the health sector component o f the SDPRP. HSDPII reaffirms most o f the basic strategies and targets set in HSDPI (See Table 4.1). However, it addresses a number o f new trends and issues. I t aims to coordinate health sector development with a new drug policy, the national HIV/AIDS activities, the SDPRP, the government's policy o n women's issues, and the new civil service reform. It also seeks to link up with new international initiatives including R o l l Back Malaria, the Global Fund for A I D S , TB, and Malaria, GAVI, and the Millenium Development Goals. 4.72 HSDPII sets a variety o f health status targets including reducing child mortality to 160 per 1000 live births, increasing immunization coverage to 65 percent, and halting the spread o f HIV/AIDS. T o achieve these targets, it estimates the need for more than 2 billion birr per year. HSDPII proposes spending from goverrzmerzt Treasury resources o f 1.3 billion birr in EFY96 and 1.6 billion birr in EFY97, compared to a level o f ETB 758 million in the EFY95 (2002/03) budget. Key elements o f this program are an expanded training program for health care workers, hiring new cadres o f health workers, construction o f new facilities, and reforms in management and information systems. Table 4.9: Selected Health Care Financing Targets a/ Budget estimated at ERTB 1.08 billion for 2001/02, ETB 1.4 billion for 2002/03 and 1.7 billion for 2003/04. 4.73 By 1997 (2004), the HSDPII envisions training and deploying an additional 5,000 Senior and Junior Nurses, 1,700 Senior and Junior Public Health Nurses, and about 670 Senior and Junior Midwives. Maintaining this rate would allow the country to reach a ratio o f 5,000 inhabitants per nurse (counting both senior and junior nurses) in a relatively short time, but would not make up for the general scarcity o f public health nurses and midwives. This imbalance would be somewhat offset by the proposed training and deployment o f 10,869 frontline health workers, who would include 290 primary midwives and 2,857 Trained Birth Attendants. 4.74 I t i s proposed that a significant portion o f frontline workers will be employed under a new program -- the Health Extension Package (HEP) -- which aims to assure a full complement o f preventive and health promotion activities for all Ethiopians through training and posting Health Extension Workers (HEWs) in each kebele"'. T h e goal i s to post up to two HEWs per kebele, which would require between 10,000 and 20,000 workers. About one-fourth o f the 'I2 Kebele's are the country's smallest administrative unit, corresponding to villages and other units that combine to make up a Woreda. 102 frontline health workers to be deployed by 1997 (2004/5), that i s 2,400, would be HEWS. It i s envisioned that these workers would have no curative functions, e.g. n o access to medications or ability to prescribe, and would report directly to the local health center. T o post all these frontline workers would cost approximately 36 million birr annually, and might require an additional 60 million birr in complementary supplies, per diems, training, and supervision costs. 4.75 The HSDPII also envisions training an additional 673 general practitioners and postgraduate doctors over the period EFY 1995 to 1997. If Ethiopia maintains this rate o f approximately 220 additional doctors per year, it will not reach i t s goal o f 10,000 inhabitants per doctor over a 25-year period. 4.76 O f course, the national ratios are not in themselves sufficient because geographical distribution, effectiveness, and quality o f services are also important. Alongside the training and placement o f nurses and fi-ont-line health workers, attention has to be given increasingly to the forms o f personnel management that will keep health care workers up-to-date, motivated, and at work. Given the ongoing geographical concentration o f doctors in Addis Ababa and Dire Dawa, other areas will have to design their health care delivery systems to rely heavily on Junior Nurses and community health care workers. Based on recent experience, even the training, recruitment and placement o f midwives and public health nurses may be a difficult human resource challenge. Construction of Facilities 4.77 HSDPII envisions the construction o f 13 district hospitals and 2 special or Regional hospitals over EFY1995-97. If maintained at this roughly 3.5 percent increase per year, Ethiopia would reach i t s stated target o f 625,000 inhabitants per hospital in EFY2018 (2025). The HSDPII also projects construction o f 143 health centers over the next 3 years. This rate i s slightly below the level required to make progress toward the country's target o f 25,000 inhabitants per health center by EFY2018 (2025); but i s about twice the rate at which health centers have been constructed and made operational between EFY92-94. When the importance o f geographical distribution i s taken into consideration, it i s likely that providing real access to health centers will require staffing and constructing a much greater number o f health centers. Comparing the required rate o f increase in health centers that i s needed to reach the government's desired coverage targets (12.8 percent annually) with the required rate o f increase in hospitals (3.8 percent annually) demonstrates that continued emphasis must be given to investment in primary level facilities if public health policy targets are going to be met. Spending on Critical Complementary Inputs 4.78 The foregoing discusses staffing and construction which are the two elements o f HSDPII that are most easily quantified and costed. But the real success o f HSDPII and the government's longer term action in the health sector may actually depend more on other elements o f the program which can raise the productivity o f these expanding human and physical resources. These activities are less easily quantified, but require substantial managerial attention because they are so important to expanding services while mitigating rising expenditures. Some o f these "soft'l investments that are included in the HSDPII and which could have substantial effects include: 0 changing the mix o f staff s h l l s through innovative use o f frontline health workers and less-skilled health care professionals in providing services and promoting health; 0 maintaining good working conditions, technical support, and recognition to sustain staff morale and improve retention; 0 improving supply logistics, with special attention to drugs and vaccines; 0 improving management capacities in the Woreda Health Offices and associated local functions; 103 0 implementing recurrent activities at a pace that fully utilizes existing capacity; 0 better coordinating implementation and expenditure at the Federal, Regional, Zonal, and Woreda levels; 0 facilitating disbursement and effective use o f donor funding. Resource Requirements for the Future: Three Scenarios 4.79 This section presents three scenarios for future public spending needs. The first scenario uses the trends o f public health expenditure, investment and staffing from EFY1991-EFY1993. I t represents a likely outcome if there are n o significant improvements in coordination and management, but rather an extension o f services at current salaries and average spending per facility on supplies and medications. In this scenario, actual coverage would be limited t o less than 50 percent due to insufficient staffing and inadequate complementary drugs and supplies. Some improvements in health outcomes are likely to take place in the case o f child mortality and maternal mortality, but at a relatively slow pace, while HIV/AIDS prevalence would be likely to increase toward the levels found among today's 15 to 24 year olds. 4.80 The second scenario takes current public policy goals as stated in the HSDPII and SDPRP. I t projects resource requirements in terms o f personnel, facilities, and spending that would be necessary to achieve the medium term targets for 2004/05 and continue the same pace o f improvement through 2015. In this scenario, the composition o f inputs i s also improved by increasing recurrent expenditure on drugs, supplies, maintenance, and in-service training. However, most o f the increases in expenditure result from reaching physical expansion targets. 4.81 The third scenario considers what would be required to reach the MDGs by 2015. I t projects national health expenditures based on (1) reaching coverage targets in 10 years, (2) improving salaries, per diems, drugs, supplies, maintenance and in-service training for the full amount required to assure a limited package o f health services for the whole population, emphasizing prevention, control and treatment o f major infectious diseases along with maternal and child care; and (3) limited continuing expansion o f teriary-level and curative care. While this i s an ambitious scenario in terms o f staffing, budget reapportionment, and costs, it i s still insufficient to reach the MDGs by 2015, although it does suggest substantial progress can be made toward those goals. 4.82 The following tables (Table 4.10 and table 4.11) summarize the major assumptions and average costs associated with each scenario. 104 Table 4.10: Selected Indicators for Expenditure Projection Scenarios Item Scenario 1 Scenario 2 Scenario 3 Expansion o f Staff & Facilities Potential Coverage 66% in2015 85% in2015 95% in 20 15 Effective coverage in 2015 32 % 54 % 85 % Other medical staff per nurse 2.5 2.8 3.5 Composition o f Recurrent Expenditure Salaries 60 % 57 % 49 % Drugs & Supplies 25 % 28 % 36 % Maintenance & Other 15 % 15 % 15 % Health Status Goals (by GY2015) U 5 M R (per 1,000 births) 151 125 112 HIV Prevalence 8.5 % 5.8 % 4.3 % Note: Potential coverage, effective coverage and U 5 M R are only meant to be indicative as they are based on simple log-linear extrapolations from very limited data. Table 4.1 1: Public Health Spending under different scenarios Ethiopian Year EFY1993 EFY2003 EFY2008 EFY2013 (2000/01) (2010) (2015) (2020) Millions o f Ethiopian birr (constant 2002) Scenario 1 (Low) 765 1,235 1,690 2,344 Scenario 2 (Extended PRSP) 765 2,337 4,330 6,239 Scenario 3 (MDGs Plus) 765 7,484 9,273 12,054 Ethiopian birr per capita (constant 2002) Scenario 1 11.71 14.86 17.88 22.23 Scenario 2 11.71 28.12 45.81 59.18 Scenario 3 11.71 90.06 98.09 114.33 Note the health budget in 2002/03 was already increased to 1,391 million, suggesting that government may b e o n track to finance the middle scenario; but note that this i s made up o f a very large increase in capital spending, and in the past such optimistic budget amounts have not materialized in spending. 4.83 n each case, the major parameters are the annual rates at which the number o f staff and I facilities increase; and the annual rate o f increased expenditure per staff member and facility due to improvements in quality and productivity. The linkages o f these trends to output and outcome measures are extremely rough judgments based on simple bivariate relationships and should be seen as only suggestive. 4.84 Inthe projections o f public health expenditure below, improvements in services related to HIV/AIDS, TB, Malaria, Reproductive Health, and Vaccine-preventable diseases predominate. The estimates do not incorporate spending on programs related to nutrition; nor do they include the non-health service elements o f the national program for HIV/AIDS prevention and control (e.g. education programs in schools, social support for HIV positive individuals). However, these projections do incorporate the health-service related costs o f HIV prevention, HIV treatment, and family planning, so there i s an element o f double-counting with projections in Chapter 5. The First Scenario: Continuing Past Trends 4.85 The basic premise o f the first scenario i s that recent trends, in which expansion o f the health system has barely kept pace with population growth, continue. The number o f nurses and other medical staff i s assumed to grow faster than the population (about 5 percent annually) because o f the substantial pace of training and employing nurses in the last five years. Hospital capacity grows at 3 percent per year and, due to the government's increasing commitment toward primary levels o f care, primary care centers are assumed to increase about 5 percent per year. Overall, government health spending increases 4.7 percent annually, 4 percent annually for recurrent expenditure and over 5.5 percent for capital expenditure. This represents an increase 105 from 765 million birr in EFY1993 (2000/0 1 actual ) to 1.2 billion birr in EFY2003 (20 10) and 2.3 billion in EFY13 (2020). In terms o f per capita expenditure, government spending on health increases from 11.7 1 birr per capita in EFY1993 to 15 birr per capita in EFY2003 and almost 23 birr per capita in EFY13 (2020). As a share o f GDP, government health spending increases from near 1.5 percent o f GDP to under 2 percent o f GDP in 2020. 4.86 This level o f expansion does not provide any room for improvements in service quality -- whether through paying for better trained staff, improving staffing, or increasing the availability o f drugs and other medical supplies. I t takes current average costs (and quality) as staying approximately at current levels. The ratio o f population per nurse improves and gets close to 5,000 before EFY18 (2025). The number o f facilities improves somewhat faster than the population. However, after 20 years o f expansion, there are still more than 700,000 people per hospital and more than 100,000 people per health center (as opposed to targets o f 625,000 people per hospital and 25,000 people per health center that are mentioned in government policy documents.) As a result, potential coverage (the share o f population living within 10 km. o f a health center) rises from i t s current level o f 5 1 percent to near 60 percent; but effective coverage (share o f population living with 10 km. o f a health center that i s adequately staffed, supplied, and managed) rises from its current level o f less than 30 percent to only 40 percent. 4.87 Under this scenario, health services improve only a little faster than population growth, and consequently, health status improvements are relatively small. Child mortality might be reduced from i t s current level near 167 per 1000 births to about 150 per 1000 births by 2015; however, malarial prevalence would remain roughly the same and HIV/AIDS prevalence would be likely to increase. The Second Scenario: ‘Extended PRSP’ - The HSDPII and SDPRSP Blueprint 4.88 Government policies articulated in the HSDPII and SDPRP are more ambitious than the previous scenario. They project increasing spending by more than 15 percent annually over the 3-year period, reaching a share o f 8.5 percent o f the government budget by 2005. T h i s second scenario takes the planned rate o f expanding infrastructure, staffing, and complementary inputs in the HSDPII and extrapolates them at the same rate for 2005 to 2015 after which it assumes the pace o f expansion will fall to half this pace. Based on the expansion o f these inputs and modest changes in input costs to account for improvements in quality, the scenario projects the level o f expenditures and average costs that would result. 4.89 Under this scenario, construction o f new facilities continues at a faster pace than in the previous scenario, about 12 percent p.a. for health centers and health posts and 3 percent for hospitals. Adequately staffing these facilities would require the number o f health care workers to increase about 12 percent each year, as well, but the mix o f staff will change to include fewer nurses relative to frontline workers. The composition o f spending i s assumed to improve somewhat, with a larger share o f new funds going to drugs, supplies, and other complementary inputs. 4.90 Overall, government health spending needs to increase by nearly 9 percent annually, with recurrent and capital expenditures both increasing at roughly the same rate, to achieve these goals. This means an increase in health spending from 765 million birr in EY93 to 2.3 billion birr in EY03 (2010), 4.3 billion birr in EY08 (2015), and 6.3 billion birr in 2020. Per capita expenditure increases from 12 birr per capita in EY93 (2000/01) to 28 birr in EY03 (2010), 45 birr in EFY2008, and 60 birr in EFY2013. As a share o f GDP, this represents an increase in government health spending from 1.5 percent o f GDP in EY94 (2001/02) to about 5 percent in EY13 (2020). To put these figures in perspective, 13 African governments currently spend more than 3 percent o f GDP o n health. Of course, to the extent that GDP grows faster than the assumed 106 5 percent rate, the relative burden o f projected health spending would be lower; but this illustrates the difficulty in financing even this relatively moderate level o f coverage. 4.91 In this scenario, the capital budget would have to be weighted heavily toward primary facilities. Hospital capacity would continue to expand at 3 percent annually, but health center construction would expand by 10 percent each year. At an average cost o f 20 million birr (approximately US$2.5 million) per hospital and 2 million birr (approximately US$250,000) per health center and associated health posts, this would require a minimum investment o f 0.75 birr (US$0.09) per person per year in hospitals and o f 1.60 birr (US$0.18) per person per year in health center and health post construction. These rough calculations suggest that the public sector should be investing almost 2.13 birr in health centers for every 1.00 birr invested in hospitals. MOFED reported that the government would spend 48.9 million birr on hospital investments and -- 69.5 million birr on investments at the primary level in EFY1994 a ratio o f approximately 1.5 to 1. Capital investments vary greatly from year to year, and in EFY1993, only one year earlier, capital investments in hospitals and primary level centers were roughly equal (73 million birr and 74 million birr, respectively). Therefore, assuring sufficient capital investment at the primary level would imply a sustained shift in priorities. 4.92 Under this scenario, potential coverage could reach 85 percent o f the population by 2020; however, the effective coverage continues to lag due to inadequate staffing and supplies. Effective coverage could rise from i t s current level o f less than 30 percent to near 50 percent since health services improve at a substantially faster pace than population growth. Consequently, health status improvements would be faster. I t i s estimated child mortality could be reduced to near 118 per 1000 births, HIV/AIDS prevalence could be reduced to 5.8 percent and progress could be made against the prevalence o f malaria. The Third Scenario: Aiming for the MDGs"~ 4.93 The third scenario takes the MDGs as its starting point and therefore considers the impact o f accelerating health service expansion, along with making substantial improvements in quality and resource allocation. I t projects the expansion o f staffing and infrastructure required to reach potential coverage o f 95 percent in 2015, increases pay and training, and raises the budget share o f supplies and medications from i t s current level o f 25 percent to about 36 percent by 2015. After this massive expansion, the rate o f increase i s assumed to fall o f to 25 percent o f its original level. 4.94 The expenditures for effective coverage in this scenario draw o n an ongoing study at the World Bank that i s identifying the bottlenecks to providing high priority health services. The study estimates the cost, at actual Ethiopian prices, for the required staff and associated expenses related to training, supervision, travel, materials, supplies, and medications for a limited set o f services. Although that study i s s t i l l being prepared, some o f i t s preliminary findings are used here to project what could be considered the "adequate" level o f health spending, i.e. spending sufficient to meet the population's health needs by orienting health services to address the highest priority illnesses. 'I3 A draft study o n the marginal costs o f improving public health through cost-effective interventions has estimated that substantial improvements in health status can be achieved at an incremental cost o f approximately $3.50 per capita. Although scenario 3 has comparable health status targets, the projections show substantially higher costs than this other study. The difference i s that scenario 3 includes the marginal expenditures o n cost-effective primary interventions along with continued spending and expansion o f other health services, most notably curative care in hospitals and health centres. 107 4.95 The package o f services that are identified in the study include all the necessary and complementary levels o f services, including health extension, health post and health center level services, campaigns, and a few hospital interventions. The package o f interventions includes preventive care for women (e.g. condoms, malarial prophylaxis, skilled birth attendance, TT immunization); preventive care for children (e.g. bednet use, immunizations, supplements); and selected curative services (e.g. oral rehydration therapy, first aid, management o f sexually transmitted infections, and DOTS). It i s adapted to adjust for differences between urban and rural areas, and to adjust for differences in the epidemiological profile and costs o f serving highland and lowland areas. This package would also include services designed to prevent A I D S , including condoms, counseling and some treatment. Note that the package in the World Bank model does not include a wide range o f curative care at primary, secondary, and tertiary levels o f care. The PER expenditure projections are based on the cost o f expanding the range o f these other services in line with population growth as well, while the predominant use o f funds i s dedicated to extending this basic package. 4.96 Under this scenario the required pace o f capital investments in health centers and posts i s double the second scenario (20 percent annually), as i s the expansion o f health care workers (almost 20 percent per year). 4.97 The preliminary findings suggest that delivering such a basic package o f services over a 10 year period for the nation as a whole would require recurrent expenditures o f 68 birr (US$8.20) per person per year and investments in facilities, training, and equipment o f 28 birr per person per year (US$3.30). The sum o f U S $ l l . 5 0 per capita i s comparable to levels reported in other studies that estimate the cost o f basic packages o f health services in developing countries. The World Bank's World Development Report in 1993 estimated the cost o f a basic package at approximately US$12 per capita. The C M H has estimated a basic package cost o f approximately US$22 per capita for the lowest income countries. 4.98 -- Nevertheless, these would s t i l l represent a major challenge to Ethiopia both in terms o f mobilizing the funds and effectively applying them. Current public spending o n health i s approximately 12 birr ($1.40) per capita, so spending an additional 87 birr per person per year would require increasing the level o f current government spending o n health by ten-fold in real terms. This scenario clearly would represent a massive shift in health expenditure for the government. 4.99 As in the second scenario, both the capital and recurrent budgets would have to be weighted heavily toward primary services and facilities. Hospital construction would continue at the same rate, 3 percent annually, but investments in health centers and posts would need to grow at 20 percent annually until EFY2003 (2010), after which they would slow to keep ahead of population growth. This would require 5 birr in investments in primary facilities for every birr invested in hospitals. The number o f midwives, health assistants, and other paramedical staff would have to increase more than proportionally, reaching a ratio o f 3.5 staff per nurse. The composition o f new spending would be more heavily weighted toward complementary inputs, with about 40 percent o f each new birr going to salaries, 30 percent to drugs and supplies, and the final 30 percent to build and maintain facilities and equipment. 4.100 Under t h s scenario, about 85 percent o f the population could have effective access to services aimed at preventing or curing malaria, HIV/AIDS, matemal and child illnesses, and vaccine-preventable diseases by 20 15. As a consequence, vaccine-preventable deaths from measles (which currently represent more than 3 percent o f deaths in Ethiopia) could be largely eliminated; the spread o f HIV/AIDS could be halted and reversed; and the prevalence o f malaria could also be reduced substantially. Nevertheless, this s t i l l would not represent comprehensive health care. Access to treatments for other illnesses and injuries would remain limited, particularly in rural areas where transportation to and capacity in secondary and tertiary facilities 108 would remain limited. The government's share o f hospitals in urban areas, currently over 90 percent, would probably diminish because the public investment in secondary and tertiary capacity would only grow slightly faster than population in this scenario; whereas private sector providers would probably respond to meet much o f the increasing demand. 4.10 1 .Given Ethiopia's national income level and competing development demands, mobilizing the sums required to finance this scenario would seem to be unrealistic, even with substantial extemal support. The implied spending level o f ETB 9 billion- (US$ 1.1 billion) annually would represent about 29% o f projected total public spending in 2015 (compared to 5% today), and 9% of GDP - many times higher than any l o w income country i s currently spending on public health services. Even if externally funded, it would require extensive efforts to scale up domestic activities in ways that could absorb these extemal funds. This demonstrates how urgent i t i s for public policy to be selective, to allocate spending to a limited set key priorities, and to improve the effectiveness o f every birr spent. 4.102 The wide range in costs represented by these three scenarios are largely driven by the numbers o f staff and facilities required to reach Ethiopia's large and growing population. For this reason, any progress in reducing total fertility would reduce the projected funding requirements. Judicious use o f frontline health care workers and innovations to deliver services without relying on highly skilled health professionals can also mitigate these projected costs. Finally, to the extent that recurrent and capital planning i s better balanced, substantial cost savings may be possible. A limited amount o f operational funds can leverage large impacts when it utilizes the existing capacity o f infrastructure. Service coverage can be increased more quickly and at lower cost by making existing facilities functional than by constructing new ones. Such a strategy would make it possible to defer some capital investments and reduce the overall funding requirements. F. FINANCIALSOURCES Government Spending Projections 4.103 HSDPII sets ambitious goals for public spending o n health, with ETB 1.2 billion in the first year (EFY1995), increasing 30 percent to 1.7 billion in the second year (EFY1996) and another 40 percent to 2.3 billion the following year (EFY1997) (See HSDPII, Table 3.3, p. 56). The composition of the projected budget i s similar to that o f HSDPI, with 50.6 percent o f the expenditures dedicated to health service delivery, 23.4 percent for construction and rehabilitation o f facilities, 6.4 percent for pharmaceutical services, and the remainder for education campaigns and support services. The govemment's ability to reach these spending levels i s based o n expectations that extemal loans and grants will finance about 21 percent o f the three-year program, and that facility revenues will bring in another 11 percent. 4.104 Three issues may hinder the government's ability to meet these spending targets as programmed. First, it i s generally difficult for govemments to raise large amounts o f revenue through facility-level charges. A recent study in Ethiopia, argues that the government's proposed changes in fees (retention at facility level, tightening exemption criteria, and adjusting fees) can cover approximately 13 percent o f facility cost^."^ However, the general experience with such charges in other A h c a n countries has shown that only a small portion o f total costs i s generally re~overed."~ In terms o f recovering costs, the most successful experiences were related to assuring availability o f drugs or recovering and retaining a share o f recurrent costs at the facility level. But in these cases, cost recovery s t i l l rarely exceeded a few percent o f total costs. If Fairbank, 200 1. l4 'I5 WHO/EIP, 2003 and Gilson, 1997. 109 Ethiopia were to achieve cost recovery levels above 10 percent o f total fixed and variable costs for a substantial part o f its public health system, it would be unusual. 4.105 Another issue that may hinder the governments ability to meet the HSDPII spending targets i s the difficulty o f scaling up service delivery -- including the logistics o f hiring and placing workers, distributing drugs and supplies, and building management capacity in many decentralized locations. For example the MDGs scenario projections involve a target o f 213,000 . health workers by 2015, compared to some 32,000 today. Even under the lower SDPRPMSDP projections, staff would have to increase by 120,000. Consequently, even if the funds are budgeted and available, the ability to spend the money may lag. Role o f donor support and money 4.106 Bilateral and multilateral agencies have projected substantial contributions to Ethiopia's health sector over the coming years. I t i s estimated that approximately US$lOO million in grants and loans was spent in Ethiopia in EFY1994 (2001/02), o f which an estimated 10 percent to 15 percent was channeled through government budgets. Some o f this external aid may be associated with response to the food emergency (the accounting definitions include some food emergency aid within the health sector). But pledges o f support from the Global Fund alone would represent an additional US$45 million per year over the next five years, and the W o r l d Bank and bilateral donors have all expanded their aid programs since the end o f the war. The effectiveness o f this aid depends upon coordinated action by the federal, Regional, zonal, and woreda governments, along with the major UN agencies, the World Bank, GAVI, the Global Fund, NGOs, bilateral aid agencies, and many civil and religious organizations. 4.107 Among the external actors, several factors may hinder effective disbursement o f funds and implementation o f programs. First, donor priorities vary in focus and strategies. Despite the existing coordination mechanisms, external agencies continue to pursue activities in ways that do not necessarily match the agreed upon health sector program. Second, there may be conditions on external funding that are difficult to meet for administrative, logistical, legal, or strategic reasons. Third, disbursement o f external funds i s often programmed according to the external agency's budget cycle in ways that can make the availability o f funds erratic and unpredictable. 4.108 Among domestic actors, a variety o f problems may also make effective use o f external funds difficult. First, the weakness o f existing information systems makes it difficult to measure the execution o f programs and to report outputs and impact. At the 5th Annual HSDP Review Meeting, representatives from external funding agencies stated that if they could demonstrate effective use o f funds, more would be forthcoming. Second, execution o f programs i s uneven and has often lagged substantially behind projected targets. Finally, under the decentralized system o f governance, it i s quite difficult for the national government to alter the pace and character o f programs and policies at the Regional and woreda levels. Controlling less than 15 percent o f national public spending on health, and without access to timely accurate information, the Ministry does not have the tools with which to act as a coordinator between subnational spending and external aid. 4.109 These problems are well known to the government and external aid community, and a variety o f strategies are being adopted t o address them. One key proposal i s to move toward budget support and away from program-specific assistance. Along with this proposal, efforts are being made to harmonize reporting and information requirements. Institutional arrangements to encourage coordination and sector planning have become stronger with each passing year, The recent 5th Annual Review Meeting demonstrated capacity for external and domestic actors to come together around a single evaluation process. 110 Scope for Private Sector Funding and Provision 4.110 The projections above focus exclusively on the government's own health services. Yet many o f the targets related to access to health services -- such as the ratios o f population to health care workers or health facility -- can be achieved through an expansion o f non-governmental health services. Also, the financial burden on fiscal revenues can be offset i f wealthier people pay for and use non-governmental outpatient services, lifting some o f the burden o f f the public health services which can then focus on serving those o f lesser means. 4.1 11 There are at least three ways in which the private sector could play a positive role 0 Providing basic health services under contract or for fees. 0 Providing curative services for the wealthier, thereby reducing pressure o n the government so that it can concentrate on the delivery o f services to the majority o f the population. 0 Providing distribution networks for commodities such as bednets and condoms, particularly if the public sector can subsidize the cost and assure quality. 4.112 Which strategy i s more likely to work in Ethiopia? The f i r s t would require a strong supply response from the private sector to provide services to Ethiopia's largely rural population. Private service providers do offer services to much o f this rural population, but they are largely related to selling medications without prescription. The purchasing power o f households in rural areas i s so much lower than in urban areas, and the related amenities are so low, that private providers are likely to focus their attention o n the larger effective demand in major urban areas. Those providers who do respond to the l o w effective demand in poor and rural areas often offer l o w quality services that can unfortunately interfere with improving health status (e.g. selling l o w dosage medications without prescription, facilitating growth o f resistance to antibiotics and antimalarials). Thus a strategy that relies predominantly on private market responses i s not likely to generate substantial improvements in the availability o f basic health services over the next 10 or even 20 years. 4.1 13 A related strategy would be for the government to contract NGOs or private providers to deliver services, instead o f expanding i t s own public system. Substantial responses to such contracting efforts are possible. NGOs in Ethiopia have expanded their activities rapidly since the mid-1990s and there i s some evidence that they can work quite effectively in a variety o f circumstances throughout the country. Depending on h o w lucrative such contracts are, it might also be possible to encourage health care professionals to offer services in priority areas. However, this use o f the for-profit and non-profit providers only represents a change in who provides the service, not who finances it. The appropriate decision here must be evaluated in terms o f "make or buy" -- which way o f providing services i s likely to reach the population with the right kinds and quality o f services at a lower cost. There are examples in which governments have quickly and effectively expanded basic service access through contracting (e.g. Guatemala in the mid-l990s), but also failures. 4.114 It i s more likely that the private sector can be encouraged to play a positive role by responding to demands for non-basic services in urban areas. T o the extent that the middle and upper classes spend their own money to get access to private outpatient services, it will be easier for the government to concentrate i t s efforts o n the key public health priorities that require basic preventive and curative care. However, this approach i s also not without i t s problems. As the private sector expands in urban areas, it could bid up salaries and the costs o f inputs in ways that make it difficult to retain public sector staff. If significant numbers o f people began to rely o n the private sector for expensive inpatient care, it could also contribute to impoverishment o f families in urban areas. Therefore, the government will have to pay some attention to regulating private providers, and supporting private sector expansion within standards o f appropriate treatment. 111 4.1 15 In sum, the existence o f this private sector, and i t s likely expansion as income rises, can be seen as an opportunity for the government to direct increasing shares o f i t s public revenues toward primary level care and public health functions that address major illnesses related to infectious diseases and reproductive health; while the possibility o f contracting basic services with NGOs or private providers should be considered as one option where they are less costly and more effective than direct provision by the government. G. CONCLUSIONS 4.116 Ethiopia has very l o w income, a very large rural population, and a heavy burden o f infectious diseases and malnutrition. With limited funds, technically-trained staff, and managerial capacity, the only way to accelerate progress on population health status i s by a sharp focus on key priorities: preventive and curative services that address malaria, HIV/AIDS, TB, vaccine preventable diseases, and maternal and child health. Given the limitations on the reach o f Ethiopia's current public health system and on private provision, the following recommendations are indicated. 4.1 17 The government currently spends a small share o f public revenues o n health compared to other African countries. This share should be raised. 4.118 The government should put emphasis on l o w cost opportunities to increase the effectiveness o f i t s spending.,Such opportunities include: 0 increasing reliance on health assistants, midwives, frontline, and community health workers, while utilizing nurses, doctors and health officers for support and referral; 0 Improving working conditions, technical support, and recognition to staff in order to sustain morale and improve retention; and 0 Reorienting health care services to tackle the bottlenecks that hinder expanding vaccination coverage, contraceptive prevalence, and similar cost-effective measures. 4.1 19 Finally, the government should put emphasis on improving the managerial functions that assure full utilization o f existing capacity and reduce wastage, including: 0 better programming o f recurrent and capital expenditures and coordination o f National, Regional, Zonal, and Woreda activities to assure that construction and rehabilitation o f physical facilities occurs only when recurrent funding for staff, operation and maintenance will be available; 0 facilitate timely disbursement o f external funding in collaboration with external agencies; 0 improve monitoring and reporting on budget execution and outputs so as to better manage and justify external assistance; 0 improve management capacities and support to Woreda Health Offices. 4.120 Policies and programs that reduce fertility, improve nutrition, increase hygiene, and raise education levels may have the most significant effects o n improving health status in Ethiopia in the coming years. However, in addition to these factors, many basic health care services that are not readily available in Ethiopia would have significant effects o n reducing mortality and morbidity if they could be provided. Hence, we believe there i s justification for expanding health service funding, so long as i t i s focused on these priority activities and accompanied by measures to improve staffing and quality, and to balance recurrent and capital costs. 112 5. POPULATION AND FAMILY PLANNING A. BACKGROUND - THEPOPULATIONCHALLENGE 5.1 As noted earlier, rapid population growth i s among the single largest obstacles confronting policy-makers in trying to adequately finance social sector services in Ethiopia for the coming decades. Just to stand s t i l l in terms o f enrollment rates would require a 60 percent real increase in education spending by 2015 (Chapter 2). Similarly, in the area o f food security, while there i s deep concern n o w with the safety net requirements o f the 5-6 million who are unable to feed themselves each year (Chapter 6), more worryingly, the population i s growing to the rate o f about 1.9 million each year, many o f whom are being added in areas, and under conditions, which make it likely that they will become part o f this pool o f the chronically poor. (Total fertility, for example, i s only 1.9 in Addis, but i s 6.4 in rural Oromiya and 5.8 in Tigray.'16) 5.2 More broadly, rapid population increase, in the absence o f corresponding growth in productivity, makes it far more difficult to have an impact in reducing poverty. In the past, additional population was to some extent absorbed in agriculture, and although there i s s t i l l scope for improving agricultural productivity, the sheer numbers being added to the rural population, combined with pressure o n the resource base, make i t more and more likely that additional population growth will need to be absorbed off- farm. By 2015 the labour force will be growing at about 650,000 per year. T o put this in perspective, total private sector non-farm employment for the whole country was estimated to be only about 2 million in 1998. 5.3 Finally, excessive numbers o f pregnancies and births are a major factor in the poor state o f women's health in Ethiopia. Spacing births and reducing the number o f pregnancies i s one o f the principal public health interventions available to the government. 5.4 While it will take a substantial time for the effects o f a major family planning program to be felt on aggregate population number^,"^ every birth averted represents substantial savings in terms o f the number o f people for whom government will have to finance education and health services, potentially support with a safety net, and for whom the economy will have to create a job. 5.5 Reducing the pressure o n social services in the future will not only save money, but will also increase the effectiveness o f programs by allowing the concentration o f available resources on fewer beneficiaries. (A good example i s the reduction in the pupil- teacher ratio). Demographic and Health Survey, 2000 'I7Due to the momentum effect o f past population growth, the pattem for this generation i s heavily influenced by the large number o f girls already b o m who are approachmg reproductive age. 113 5.6 Ethiopia i s starting from far behind in family planning service delivery. Contraceptive prevalence, the standard measure o f the family planning implementation 118 , i s only 12 percent, among the lowest in the world, and far below the level that would be needed to have a significant impact on population growth. 5.7 It i s difficult to assemble reliable data on current public spending o n family planning and reproductive health, partly because most expenditures o n commodities are financed by donors, and are not systematically recorded in the budget. However we know the levels are very low, relative to what could potentially be spent, and to what i s needed. As the 2000 DHS reported: “only 18 percent o f the family planning needs o f currently married women are being met. There would be a five-fold increase in contraceptive prevalence if this unmet need were met.” The main elements o f spending at the moment are funding by UNFPA, which amounts to U S $ 5 million annually, the supply o f condoms and support for reproductive health programs, supported by USAID and other bilateral donors, and the government’s spending o n the recurrent operations o f the general health service, which i s responsible for delivery o f family planning (FP), but there i s n o separate family planning cadre or system, and the proportion o f the health service budget that accords to FP delivery i s extremely small. 5.8 This chapter sets out to estimate the long-term requirements for an effective family planning program in Ethiopia, in order to help demonstrate the magnitude o f the need, and the policy implications and financing needs that arise. We also hope that this will help inform the policy debate in the sector, and to mobilize additional financing for this critical effort, as well as to underline the over-riding importance o f demographic factors and population pressure in determining the outlook for affordable social services. I t draws heavily from the background paper on family planning expenditures prepared for the PER collaboratively by the UNFPA and the World Bank. l 9 5.9 In the following sections we: (1) Develop a number o f scenarios for target levels o f Total Fertility Rate (TFR) and Contraceptive Prevalence Rate (CPR) to be attained over time under modest, desirable and ambitious scenarios. (ii) Present projections o f a range o f different method mixes to attain the stipulated TFR and CPR targets; and a rough estimation, based on unit costs, o f the cost o f different methods o f contraception. (iii) Assess contraceptive commodity requirements and estimate the costs over the next 5, 10, 15 and 20 years, based o n various assumptions with respect to coverage and method mix. (iv) Identify constraints that might affect the level o f spending and its effectiveness. ‘18 The Contraceptive Prevalence Rate, CPR, i s the proportion o f married women o f reproductive age currently using any form o f contraception “A Guesstimate o f the Long-Term Requirements for An Effective Family Planning Program in Ethiopia.” R.Chaudhury and B.Kalasa; UNFPA, 2003. 114 5.10 Typically four demand-side factors affect the uptake o f family planning. One i s the degree o f economic development and social security; the second i s the involvement o f women in the labour force; the third i s families' expectations with respect to survival o f children born; and the forth are cultural practices and the ability o f women to make choices with respect to reproductive health. Experience o f the past 40 years has shown, however, that substantial changes in contraception behaviour can take place without a transformation in any or all o f these areas, purely through the widespread provision o f family planning services. This i s because at very high levels o f fertility, women do not want to have as many children as they are having, and will take advantage o f any opportunity provided to safely space children and/or reduce the total number o f pregnancies, if such methods are easily available. We have already seen a massive transformation in urban Ethiopia, where fertility has dropped to below 3.3, and the desired fertility rate i s 2.7. 5.11 Experience elsewhere, for example in Bangladesh, has shown that the contraceptive prevalence rate can be increased from single-digit levels to as much as 40 percent purely through effective, large-scale supply-side interventions. In Ethiopia there i s substantial unmet demand for family planning services. Desired family size according to the 2000 Demographic and Health Survey i s 4.9, as opposed to a current level o f just below 6. 5.12 With technical innovations o f the past two decades - especially the advent o f Depoprevara - i t i s now much easier to provide effective contraception in developing countries. (Depo i s provided by injection every three months, at a cost o f about 85 cents U S per dose; and i s highly effective. I t i s the contraceptive o f choice in Ethiopia and in most developing countries, and accounts for the single largest share o f current contraceptive use in Ethiopia.). Although this reduces the administrative requirements for delivering FP services, experience also shows that there i s s t i l l the need for a widespread cadre o f female outreach workers to advise and support women in using family planning. B. AND DATA METHODS 5.13 The family planning projections were preceded by a set o f population projections. These are prepared to obtain data on the number o f women o f reproductive age, which form the basis for assessing family planning needs. A set o f three population projections were prepared, based on three alternative assumptions o f fertility, and one assumption in regard to mortality. The projections are made for 20 years, 2000-2020.121 The three alternative assumptions that are made with regard to future course o f fertility are: (i)Total Fertility Rate (TFR) will decline linearly from 5.9 children per woman in 2000 to 3.0 children per woman in 2020. This will be called the ambitious variant projection, ( iiTotal Fertility Rate (TFR) will decline linearly from 5.9 children per ) woman in 2000 to 4.0 children per woman in 2015 and 3.4 children per woman in 2020. This assumption i s in conformity with the national population policy goal o f Ethiopia. Standardized DHS surveys are the intemationally accepted method o f measuring current and desired fertility rates. A comprehensive DHS was conducted in Ethiopia in 2000. "' For the purpose o f population projections, the cohort component projection method was employed using DEMPROJ computer software program, developed by Stover and Chao (Stover, 1999). 115 This will be known as the desirable variant projection, and (iii)Total Fertility Rate (TFR) will decline linearly from 5.9 children per woman in 2000 to 4.0 children per woman in 2020. This assumption i s made taking into account the past trend in fertility decline, which shows a modest decline o f about one child per woman over the past 10 to 15 years. This will be called the modest variant projection. 5.14 Mortality i s assumed to decline in all three projections in the same order o f magnitude. The size and age-sex distribution o f the population for the base year o f the projection (i.e. 2000) i s the same as that estimated by Central Statistical Authority (CSA, 1998). Detailed assumptions are presented in the background paper (Chaudhury, 2003). CPR Goals 5.15 Family planning projections are then made for the 20 year period under three different scenarios: ambitious, desirable and modest. The Ambitious variant corresponds with the decline in TFR from 5.9 children per woman in 2000 to 3.0 in 2020. Correspondingly, the Contraceptive Prevalence Rate (CPR) would need to increase from 8.1 percent in 2000 to 26.0 percent in 2010, and to 46.0 percent by 2020. For the desirable variant projection, TFR i s assumed to decline from 5.9 children per woman in 2000 to 3.4 in 2020. Correspondingly, contraceptive prevalence rate needs to increase to 23.4 percent by 2010, and 41.0 percent in 2020. The assumptions o f desirable variant projection are consistent with the goal o f the National Population Policy o f Ethiopia, which envisages achieving a TFR o f 4.0 children per woman and CPR o f 44 percent in 20 15 (National Population Policy o f Ethiopia 1993). 5.16 Under the modest variant projection, (decline in TFR from 5.90 in 2000 to 4.0 by 2020) the CPR would need to increase from 8.1 percent in 2000 to 18.5 percent in 2010 and to 31.4 percent by 2020. The targets under each scenario for each 5-year period are shown in table 5.1122. 5.17 The family planning projection assumes an increase in the use o f modern methods, particularly injectables, with a corresponding decline in the use o f traditional methods. Among modern methods, demand for injectables i s very high, especially among young women, and this demand i s likely to grow further in the hture. I t i s popular because: (a) i t provides a relatively longer-term protection against pregnancy, (b) it i s more convenient to use than other temporary methods as it frees users from the trouble o f daily use o f a method or use o f a method with every sex act, and (c) it provides privacy and can be used without the knowledge o f others. The modest and desirable variant projections assume an increase in use o f injectables, from 38 percent in 2000 to 47 percent in 2020. The ambitious variant projection assumes a higher increase in use o f injectables, from 38 percent in 2000 to 50 percent in 2020. The popularity o f condoms i s also gaining momentum in view o f the fact that it provides double protection: against unwanted pregnancy, as well as against HIV/AIDS infection. B o t h high fertility and HIV/AIDS infections are o f major concerns o f the country. In recognition o f this, a significant increase in use o f condoms i s also envisaged in all the projections. An increase in female sterilization, albeit modestly, i s also foreseen under all projections in view o f its gaining popularity among women with high parity. 122 . For the purposes o f family planning projection the Spectrum Computer Program developed by Stover, et.al, i s employed (Stover et.al, 1999).122 116 5.18 All the family planning projections assume a modest decline in the proportion married among women in the reproductive age, consistent with increasing modernization and urbanization o f the country with the passage o f the time; and a modest decline in post-partum sterility. Detailed assumptions are given in the background technical paper (Chaudhury, 2003). Data used for projections are mostly those provided by 1994 Population and Housing Census and 2000 Ethiopian Demographic and Health Survey. c. - PROJECTED REQUIREMENTSAND COSTS FINDINGS Expected contraceptive prevalence rate and total fertility rate by different scenarios of projection 5.19 Table 5.1 provides a summary o f contraceptive and fertility targets under the three different scenarios. Data show a formidable task ahead to achieve the desired fertility and contraceptive goals. For example, to achieve the goal o f total fertility rate o f 3 - 3.4 children per woman by the end o f 2020 as targeted by ambitious and desirable variants respectively will entail a five to six-fold increase in contraceptive prevalence from the base level o f 8.1 percent. The task will be slightly less formidable, but still substantial, in achieving the demographic goals stipulated in the modest variant o f the projection. Even to achieve the demographic target o f 4.0 children per woman by 2020 stipulated in the modest variant o f the projection would entail a four-fold increase in contraceptive prevalence from the 2000 level. Table 5.1: Projected Total Fertility and Contraceptive Prevalence Rates Under Different Scenarios Ethiopia, 2000-2020 Proiection Scenario 2000 2005 2010 2015 2020 TFR CPR TFR CPR TFR CPR TFR CPR TFR CPR Ambitious 5.9 8.1 5.2 16.0 4.4 25.1 3.7 35.8 3.0 46.0 CPR = 46.0 and TFR = 3.0 by 2020 Desirable 5.9 8.1 5.3 14.8 4.6 23.4 4.0 32.0 3.4 40.8 CPR = 40.85 and TFR = 3.4 b y 2020 Modest 5.9 8.1 5.4 12.4 4.9 18.5 4.5 25.0 4.0 31.4 CPR = 31.44 and TFR = 4.0 b y 2020 TFR, total fertility rate, in births per women, CPR, contraceptive prevalence rate, in percentage o f married women of reproductive age. Estimate and projections o f five yearly and annual users o f modern contraceptives 5.20 Table 5.2 presents quinquennial data on the number o f users o f modern contraceptives needed to achieve the demographic goals set out in different scenarios. For example, the users o f modern methods o f family planning methods will need to increase from less than a million (0.60 million) in 2000 to 7.4,6.5 and 5.1 millions to achieve total fertility rates o f 3.0, 3.4 and 4.0 children per woman respectively by the year 2020. 5.21 The implications for the number o f users o f different methods are shown in the background paper, but just to illustrate, to reach the goal o f 3.4 children per woman and CPR o f 41 percent by 2020 in the desirable variant projection will require an increase in condom, female sterilization, injections, IUD, pill and Norplant users by 1.07, 0.38, 3.14, 0.36, 0.90 and 0.10 millions respectively over the base year (2000) estimates o f these methods. 117 Table 5.2: Contraceptive Usage Among All Currently M a r r i e d W o m e n by Different Scenarios of Projection; Ethiopia, 2000 - 2020 (in Millions) Projection Scenario Five year projection 2000 2005 2010 2015 2020 Ambitious 0.60 1.73 3.12 5.01 7.38 CPR = 46.0 and TFR = 3.0 by 2020 Desirable 0.60 1.60 2.83 4.48 6.54 CPR = 40.85 and TFR = 3.4 by 2020 Modest 0.60 1.34 2.24 3.51 5.06 CPR = 3 1.44 and TFR = 4.0 by 2020 Commodity requirements 5.22 An estimate i s made o f contraceptive commodities, in terms o f millions o f condoms, injection doses, pill cycles, sterilization procedures, I U D insertions and Norplant sets which will be required to meet the contraceptive need o f additional users o f the different methods o f contraception to achieve the desired demographic targets stipulated in the three scenarios. The requirements are assessed using internationally accepted Couple-Year Protection (CYP) norms. 5.23 Table 5.3 gives a five-yearly summary o f contraceptive commodity requirements by projection scenarios. The data show a very large number o f contraceptive commodity requirements to reach the demographic goals stipulated in different scenarios. For example, under the ‘desirable’ variant, injectible requirements increase from about 1.2 million doses p.a. today to 13.7 million by 2020; and condom requirements vary from 102 million pieces in the moderate variant projection to 132 and 145 million pieces in the desirable and ambitious variants respectively. Cost of commodities 5.24 Table 5.4 presents five-yearly estimates o f the cost o f contraceptive commodities, under different scenarios o f projection, i f purchased on the international market, based on prices listed below: e Condom $0.03 e Female Sterilization $10.40 e Injection $0.86 e IUD $1.68 e Pills $0.25 e Norplant $27.00 5.25 The prices o f most o f the commodities, particularly those for condoms, pills and injections are taken from a recent study o n contraceptive forecast in Ethiopia (Ministry o f Health: Ethiopia: Contraceptive Forecasts for the year 2002 - 2004). The price o f Norplant i s obtained from Engendered Health Project, Ethiopia, while the prices for female sterilization and IUD are taken from a U N F P A technical report o n contraceptive use and commodity costs in developing countries, 1994 - 2005 (UNFPA Technical Report No. 18). N o cost escalation has been included in the projections o f these current costs. Also, Government levies o n contraceptive commodities are not included. 118 5.26 The data show a sharp increase in spending o n commodities to achieve the demographic goals stipulated in various projections. For example, the cost o f commodities increases by about eight-fold from $2.4 million in 2000 to $18.2 million in 2020, according to modest variant projection, in which fertility i s assumed to decline from 5.9 children per woman in 2000 to 4.0 children per woman in 2020. However, if fertility were to be reduced to 3.4 and 3.0 children per woman in 2020, as stipulated in the desirable variant and ambitious variant projection respectively, the cost o f contraceptive commodities will increase to about $23 million and $25.8 million respectively in 2020 (see Table 5.4). 5.27 Injectables account for almost half o f the commodity costs, and condoms and pills for another 40 percent between them. The average cost o f commodities per user o f modem methods given the current distribution o f users by method i s about $4 in each variant o f projection. 119 8 l- m 0 N 8 0 m W P 8 8 3 0 N 3 d d 0 N 8 8 m r4 N 0 0 4 4 N 0 0 0 8 8 0 0 N - 0 N 0 N m 3 0 N z 0 N m 0 0 N 0 0 0 hl 0 N - W 0 4 r- 8 N 0 N 2 N m 8 3 N 0 N 8 m 0 0 N - 8 0 3 8 0 0 N - m c) 8 W 5 -e z Table 5.4: Cost of Contraceptive Commodities, According to Three Different Projection Scenarios, 2000 - 2020 (In U S $ millions) Projection Scenario Five year projection 2000 2005 2010 2015 2020 Ambitious 2.47 6.19 11.oo 17.32 25.78 CPR = 46.0 and TFR = 3.0 by 2020 Desirable 2.47 5.76 10.05 15.55 22.97 CPR = 40.85 and TFR = 3.4 by 2020 Modest 2.36 4.83 8.11 12.38 18.17 CPR = 3 1.44 and TFR = 4.0 by 2020 Total costs of family planning 5.28 An attempt was also made to arrive at an estimate o f total costs o f family planning services, talung into account all direct, indirect and capital costs. But data on public expenditure on reproductive health, including family planning services are not available for Ethiopia. In the absence o f country-specific estimate o f costs o f delivering family planning services, a rough estimate was made using the default estimate o f per user/per method cost o f family planning given by the Spectrum Computer Programme for Projecting Family Planning Requirements. These estimates are based o n information on family planning expenditure submitted by governments o f various counties in the 1 9 9 0 ~ . ” ~They almost certainly over-state the costs o f provision in Ethiopia, where labour and service delivery costs are lower than the average. As such, these estimates should be regarded as an upper bound on the likely full costs o f a family planning program. (Note however that the average cost per user that emerges from the final estimates (table 5.6) i s roughly comparable to findings for other African counties, giving us some comfort that the estimates are in the right neighbourhood). 5.29 Table 5.5 presents five-yearly estimates o f expenditure on family planning by projection scenario. Data show a significant increase in the costs o f providing family planning services from about $12 million equivalent in 2000 (ETB 100 million) to $ 54 million, $ 49 million and $ 42 million to achieve the demographic goals stipulated in ambitious variant, desirable variant and modest variant projection respectively. However, as expected, cost per user decreases with the increase in prevalence rate. Data in Table 5.6 show a steady - and significant - decline in average cost per user during every succeeding five years, from an estimated cost per user o f $20 in 2000 to about $8 in 2020 in all variants o f the projections. 5.30 Norplant i s the most expensive method o f contraception, at US$14-15 cost per user, followed by female sterilization, at $9 - $10 cost per user, and injections and condom, at $8-$9 cost per user. The least expensive methods are IUD, followed by pills, at $4-$6, in all variants o f the projections. 123 The default estimate o f expenditure, per user, per method, are shown in Annex 5; for details, see FamPlan, version 4: A computer program for projecting family planning requirements, 1999). 121 Table 5.5: Rough Estimate of Total Costs of Family Planning Services According to Different Scenarios of Projection (US$ millions) Proiection Scenario Five vear projection 2000 2005 2010” 2015 2020 Ambitious 12.23 18.77 27.15 38.93 53.73 CPR = 46.0 and TFR = 3.0 by 2020 Desirable 12.07 17.96 25.45 35.94 48.71 CPR = 40.85 and TFR = 3.4 by 2020 Modest 11.77 16.61 22.52 30.94 4 1.46 CPR = 35.44 and TFR = 4.0 by 2020 Table 5.6: Cost per User of Family Planning Services Under Different Projection Scenarios Ethiopia, 2000 - 2020 Projection Scenario Five Year Projection (US$ per user) 2000 2005 2010 2015 2020 Ambitious $20.28 $10.86 $8.71 $7.77 $7.28 (CPR = 46.0 and TFR = 3.0 by 2020) Desirable $20.02 $11.25 $8.99 $8.05 $7.44 (CPR = 40.85 and TFR = 3.4 by 2020) Modest $19.51 $12.39 $10.04 $8.83 $8.20 (CPR = 3 1.44 and TFR = 4.0 by 2020) D. OBSERVATIONS ON CHALLENGES AND CONSTRAINTS IN ACHIEVING THE OBJECTIVE OF PROVIDING ADEQUATE FAMILYPLANNING SERVICES 5.3 1 The paragraphs below outline key issues that would determine the level o f family planning services as well as sustainability o f the program, especially in terms o f supply. Three main issues can be pointed out: (i) the need for a functional system for contraceptive commodities management; (ii) capacity building o f service providers to ensure quality o f outreach services and make health centers attractive for family planning acceptors and users; and ( iii) implementation o f a national IEC and advocacy strategy in support o f demand creation and resource allocation to reproductive health services in general and family planning in particular. Reduce system loss or wastage by improving logistic management information system 5.32 System loss, i f i t i s large, can cause a deficit in supply o f contraceptives. This i s a major problem in Ethiopia. For example, an inventory o f 76 government facilities in 2001 showed 2.8 million cycles o f oral contraceptives expired or expiring by December 2001. T h e loss i s estimated at around one million U S dollars (Contraceptive Inventory and Logistic System Survey, July 200 1, Ministry o f Health). T h i s i s mostly attributed to weaknesses in the logistics management information system. In the above-mentioned survey, only 41 percent o f facilities visited were sending logistic information to the next higher level o f the system. O f these, only 41 percent were sending accurate information. 122 The practice o f systematic recording, reporting and feedback i s not adequately developed. Moreover, there i s a shortage o f persons trained in logistics management. According to the above study, staff at only 12 percent o f the facilities visited had received proper training on logistics. As noted elsewhere, capacity for effective management at all levels o f the health system remains low. This i s mostly attributed to shortages in human resources and high turnover o f staff in key management positions (Ministry o f Health, Health Sector Development Programme 11: 2002/03-2004/05). 5.33 Wastage o f supplies due to expiry o f shelf-life also arises out o f lack o f cooperation between regional medical stores and service delivery agencies (National Population Policy o f Ethiopia). Reduction o f systems loss through improvement o f the logistic information system should be taken as a priority strategic intervention to improve contraceptive security. Improve storagefacilities 5.34 The capacity o f existing warehouses/stores i s inadequate even for the current volumes o f contraceptives, much less the increased level that will be required to meet the national targets (Ministry o f Health, Contraceptives Requirement and Logistic Management Needs: Action plan for Ethiopia: 1997 - 2000). The procurement o f increasing volume o f contraceptives without increasing the capacity o f existing stores and/or building new stores with large capacity will lead to wastage and increase the risk o f contraceptive insecurity. Improve knowledgdinformation base on reproductive health and family planning services 5.35 About one third o f women with unmet need for contraception (Le. those who want to have no children and/or want to delay having children by at least two years, but do not practice contraception) do not use contraception due to lack o f knowledge and/or access. Poor knowledge o f contraception i s attributed, among other factors, to inadequate IEC and advocacy activities in support o f promotion o f family planning, small family size norm and familiarization o f methods o f contraception. Capacity to plan and implement IEC interventions i s weak in all regions. There i s also a dearth o f adequately trained staff in the field, both at the federal and regional levels. Moreover, support among policy makers and health program managers for I E C activities can be described as lukewarm (Aide-memoire o f Appraisal Mission on Health Sector Development program, June, 1998). Poor access to quality reproductive health andfamily planning (RMFP) services 5.36 Lack o f access to contraceptives i s one o f the key challenges to meet the unmet need for contraception. Government policy o n providing reproductive health services, including family planning, through fixed points &e., public centers) constitutes a major impediment to wider access to RH/FP services. The coverage o f public health facilities i s limited, particularly for the majority o f the population who live in rural areas. As noted in the health expenditure chapter, there i s only one health center for every 171,000 people, and public health facilities in rural areas are located far apart. Only 5 1 percent of the population has access to a health facility within 1Okm. Although new facilities are being constructed to increase the coverage o f health services, many o f these lack trainedskilled personnel. 123 5.37 Moreover, public health facilities in rural areas provide limited RWFP services. For example, many o f these facilities are not equipped with necessary instruments and skilled personnel to perform procedures such as vasectomy and tubectomy. I t i s therefore clear that the most effective family planning methods, with clinical components, cannot be performed outside health centers or by community-based reproductive heath agents. Access to family planning services i s also constrained by failure to optimally use the services o f existing health personnel. For example, public health nurses are not authorized to perform such surgical procedures as tubectomy and vasectomy. But nurses, including paramedics, can be trained to perform these procedures as experience from other countries such as Bangladesh shows. O f equal, and perhaps greater importance i s the need for a cadre o f female outreach workers to support service delivery. (See discussion below) Increase utilization of health services 5.38 As discussed in Chapter 4, utilization o f formal health services i s only around 0.27 visits per capita per year. This i s attributed, among other factors, to: (i) poor physical condition o f the existing health facilities, ( ii)lack o f adequate supplies, ( iii) heavy workload o f service providers, (iv) lack o f adequate trained staff and resources, (v) inadequate budget at facility level, (vi) inability o f rural health facilities to retain trained personnel as reflected in absenteeism and attrition rate o f trained health personnel. Unless these conditions are improved, demand and utilization o f health services, including RWFP services, will remain perpetually low. Improve human resources capacity 5.39 There i s an acute shortage o f trainedskilled health personnel to perform the tasks o f management, formulation, monitoring and evaluation o f population and reproductive health programs and to provide quality services at all levels. The challenge in the area o f human resources development i s not confined to the creation o f a cadre o f trained personnel, but also to attracting and retaining them, particularly in rural health care settings. 5.40 In this regard, the creation o f a cadre o f female outreach workers i s particularly promising. Experience in Indonesia, Bangladesh, and elsewhere has demonstrated the tremendous impact that such workers can have on the acceptance o f FP services. The fact i s that many women - quite understandably - are reluctant to deal with male service providers in relation to contraceptive matters, or even with higher-status, educated female staff at health facilities - with whom they seldom come into contact anyway. Part o f the solution i s to have a network o f low-level, locally-recruited workers - who may work o n a part-time basis - to support and advise women. Such a group o f workers must have a number o f characteristics: (i) they must be female; ( ii)they should come from the same general social and economic background as the women they are serving; and should preferably live in or near the same community; and ( iii)their coverage must be sufficient that they have regular contact with households in their area, and preferably know most o f the users o n a continuing basis. They need not be very highly paid, nor trained, but their numbers have to be sufficient to cover rural areas, and they need to be supported by health post staff. It should be possible, with adequate training and supervision, for the more senior o f these workers to provide delivery o f condoms, pills, and injections. The establishment o f such a cadre has the added benefit o f creating a substantial number o f income-earning opportunities for rural women. 124 5.41 I t makes sense to initially concentrate efforts in areas with the greatest concentration o f population and un-met demand, to maximize cost-effectiveness, but this should be followed quickly by a major push to expand services to women in less accessible rural areas, where the largest share o f high-fertility women will remain, and greatest welfare gains to be made in terms o f maternal and child health. Increase government resource allocation on reproductive health, particularly family planning activities 5.42 The Government allocation to reproductive health, particularly family planning activities, although increased over the years, has been and continues to be very small, confined mostly to salary o f the staff and maintenance o f health facilities. N o t only i s the budget allocation to RHEP low, but taxes are also imposed on reproductive health commodities, including contraceptives, which are procured with foreign assistance. This calls for policy dialogue and advocacy with higher level policy-makers to increase their appreciation o f the importance o f population factors in sustainable development and to mobilize their support to increase government allocations for RWFP activities, including removal of levies on imported RH commodities and contraceptives. Ensure program sustainability 5.43 Currently RH/FP services are provided almost free o f cost. This may prove unsustainable in the long run. Also the lack o f sustainable initiatives on the part o f Government may lead to a decline in foreign flows o f funds. One way to ensure sustainability o f the RWFP program i s to introduce cost-sharingkost-recovery schemes in which clients, particularly those who can afford it, will be charged minimum fees for the services received. Improve monitoring and coordination of WFPprogram 5.44 The success o f multi-sectoral and multi-layer population program depends, among other things, o n effective program coordination and monitoring at all levels. There are several stakeholders o f population and RH programs in Ethiopia. These include, among others, central and regional governments, multilateral and bilateral donors, NGOs and the private sector. The program o f activities ranges from central to regional and further lower levels. This will require formulation and institutionalization o f a comprehensive mechanism o f coordination and monitoring o f multi-sectoral and multi- tier program o f activities to achieve the desired program result through ensuring maximum synergy. But such a comprehensive mechanism o f coordination and monitoring i s yet to be developed and institutionalized. 5.45 The coordination between donors and NGOs, between NGOs and Government, and between central and regional Government need to be streamlined. The existing institutional arrangement which entrusts the National Population Office to coordinate and monitor population activities i s found to be weak because it lacks legal backup and trained persons to undertake the task effectively and efficiently. Cultural constraintsharriers and the Empowerment of Women 5.46 One o f the major barriers to demand generation for family planning and small family size i s the persistence o f practices such as early age marriage. Consistent with 125 cultural practice/expectations, which have historically promoted large family sizes, many Ethiopian girls marry at an early age, and begin early procreation. The median age o f marriage i s 16.5 years, and some girls marry before age 15, the legal age o f marriage. By age 19,40 percent o f women have begun childbearing. Early age o f marriage i s not only a major contributory factor to persistence o f high fertility, but also to high matemal and infant mortality and morbidity. 5.47 Women’s involvement in higher-level policy decision-making i s limited in view o f their poor representation at higher levels. A very few women have reached top-level decision-making positions. This has an inhibiting effect o n women’s empowerment including the fulfillment o f reproductive rights, including family planning. There i s also substantial evidence that increased female participation in the labour force, and control o f resources, has a substantial impact on fertility behaviour, reinforcing the arguments in favour o f increased female education. Increasing women’s education clearly has a tremendous role to play in empowering women and accelerating family planning usage. Spending on girls’ education i s thus as important as spending o n FP service delivery, although the effects take some time to manifest themselves. I t i s worth noting that the major increases in enrollments in the past five years will be reflected in greatly increased demand for FP services in Ethiopia in the next five years. 126 6. FOOD SECURITY AND SAFETY NETS - PUBLIC SPENDING ON TRANSFERS TO THE POOR A. POVERTY AND SAFETY NETNEEDS 6.1 In January 2003, a joint appeal by the Government o f Ethiopia and the UN pointed to “the major humanitarian crisis facing Ethiopia that i s a consequence o f a widespread and serious drought”, and identified 11.3 million people as needing food assistance o f 1.4 million mt. This i s not the first time such appeals have been made; indeed Ethiopia has experienced near-famine situations nearly every year since the mid- 1960s, some years more acute (1984, 2003) than others. The continued susceptibility to drought and pervasive food insecurity o f millions o f households has culminated in the evolution o f a safety net program package comprised o f relief works and free food distribution, funded almost entirely by donors from year to year, depending upon the intensity o f the crisis. The amounts involved, in public expenditure terms, are significant: for example the cost o f the 1.4 million mt. o f food covered in the 2003 appeal i s equivalent to about US$500 million. 6.2 Recent improvements in macro-economic indicators notwithstanding, the vast majority o f Ethiopia’s people - agncultural small holders, laborers and pastoralists - are living close to the margins o f adequate consumption, and are vulnerable to weather shocks and price shocks. As a result, cash and food availability for most rural Ethiopians are highly seasonal and unpredictable. Ethiopia’s children are among the most severely malnourished in Africa: in some Regions (Tigray, Amhara), over 60 percent o f children are stunted and over 35 percent severely stunted, reflecting the long term impacts o f chronic food insecurity and poor health. Exceptional population pressure, diminishing land holdings and livestock, and rapidly diminishing communal forest and grazing lands, have contributed in n o small measure to the very high levels o f food poverty and vulnerability. 6.3 Recognizing the chronic nature o f food insecurity, the Government recently constituted a Coalition for Food Security in Ethiopia, comprised o f line departments and donors. The Coalition has submitted a draft Report which i s currently under discussion. The goal i s ambitious: to develop an action plan, duly costed, to reduce chronic food insecurity in Ethiopia on an accelerated and sustained basis. The key elements o f the initiative include providing improved access to land, transformation o f the agricultural extension system and improving agricultural productivity, and building a safety net that protects and promotes asset accumulation. 6.4 The poverty profile for 2000 provides some quantitative dimensions to the situation o f the poor. In 1999/00,42.2 percent o f the population lived below the national poverty line (calculated on minimum calorie requirement o f 2200 kcal, or 40 cents per adult per day). The profile also provides the proportions o f persons in “extreme poverty”, i.e., those below the food poverty line based o n 1650 kcal. For the country as a whole, 22.5 percent o f households belong to this category o f extremely poor households. 127 These are households virtually on the verge o f starvation or destitution. For most o f the year these ultra-poor households have little or n o cash to purchase food from the market, nor do they have any assets to sell to smooth consumption. They typically have very small landholdings and very few livestock - perhaps a few hens and a couple o f goats - or none at all. 6.5 The recently completed Risk and Vulnerability Assessment (2003), estimated about 50 percent o f the population are considered chronically poor, an additional 27 percent move in and out o f poverty, and 14 percent who are not poor now have a high probability o f falling into poverty in the future with a single large shock. Nearly 5 to 6 million people or 10 percent o f the population are considered to be chronically food insecure, in the sense they cannot feed themselves throughout the year from their own incomes even in good years. In addition at least another 6 million people suffer from extremely l o w and highly variable incomes depending on vagaries o f the annual rains. 6.6 The above analysis, coupled with findings from other available literature, enables one to draw some broad conclusions. (a) Approximately a fifth o f households in the country are extremely vulnerable to serious shortfalls in food consumption. These households, due to extremely l o w and highly variable incomes, are often unable to feed themselves even in good years, and even in regions where the agricultural sector has been registering positive growth rates. (b) n there i s also substantial seasonality associated with food deprivation. I response to a question as to how many months they could live from own harvest, most subsistence farmers throughout the country responded that they could live only up to seven months on average; in some regions this “happy” period i s less than six months. (c) Chronic food insecurity and poverty are closely linked. Despite some improvements in recent years due to increased fertilizer use, l o w and uneven agricultural trends, coupled with high population growth rates, have resulted in a long-term secular decline in per capita food availability. (Figure 6.1 below) Figure 6.1: Per Capita Food Production 1970-2000 i i I10 7 130 m 1970 ~ 197.2 1914 1976 1 9 7 8 1 9 8 0 1982 1 9 8 4 1 9 8 6 1988 1 9 9 0 1FBZ 1994 1 9 9 6 1YS8 ZOO0 o IUG:r s I..\ti 2 W l L 128 6.7 Improvements in farm productivity could help extend the number o f months o f survival for some; yet the probability o f drought occurring remains very high in Ethiopia, largely due to excessive climatic variability. Measures o f variability o f annual rainfall across Ethiopia between 1967 and 2000 provide evidence for this claim: for annual -- rainfall, variability ranged from a l o w o f 43% to a high o f 61% among the highest in the world. (Risk and Vulnerability Assessment, draft, 2003). In addition to the failure o f the rains, health risks (including both malaria and HIV/AIDS) exacerbate the poor's vulnerability, driving millions into poverty trap situations. Thus, notwithstanding some improvements in farm productivity, food insecurity i s bound to remain both widespread and chronic in Ethiopia for the near term. 6.8 These characteristics o f the country - and the interplay o f weather-induced, often region-specific, transitory vulnerability and the chronic, nationwide poverty and food insecurity - render a medium-term response hard to conceptualize and harder to program and implement. Yet the need for fine-tuning a medium-term, well-designed safety net response that opportunistically builds on, but i s not conditioned by, the short-term periodic crisis i s urgent. Until now, the safety net package has been driven very much by the drought-related response. Current policy/program responses 6.9 The safety net policy framework that has evolved over time i s largely driven by considerations o f how to meet the immediate, transitory food needs o f poor households located in the two main zones that are subjected to periodic rainfall shortages. This drought-appeals-food aid mindset does not allow much room for the fact that food security in Ethiopia i s chronic rather than transitory. As a result, little conscious effort has been made to integrate programs with the broader development goals o f bridging the gap between agricultural growth and population growth, integrating grain markets, and developing rural infrastructure. Furthermore, to the extent the programs are driven and funded by drought-induced food aid, there has been a built-in regional asymmetry that protected the consumption o f the poor in the two main food-insecure regions and the urban poor, while bypassing millions o f very poor, chronically food insecure households in the otherwise food-secure regions. 6.10 Transfers to the poor are currently based almost entirely o n the amount o f relief food that can be mobilized in a given year. Each year the Disaster Preparedness and Preparation Commission (DPPC) launches an annual appeal, o n the basis o f an assessment o f the local food production outlook, and the estimated number o f households at risk. Programs are concentrated in 156 identified woredas in four regions: Oromiya, Amhara, Tigray and SNNF'R. The identification o f woredas i s done with an Early Waming System that has evolved fairly efficiently over time. Within each woreda, households eligible for different programs are identified using both administrative and community targeting methods. 6.11 Even at the risk o f repetition, it i s worth stressing that these measures are intended to address transitional food insecurity caused by rainfall shortages. Initiatives aimed at chronic food insecurity, i.e., the problem o f households who suffer f i o m consumption shortfalls even in good years and even in relatively better o f f regions, are sparse on the ground, and have mainly taken the form o f isolated rural/agricultural development programs. 129 Overview of Programs 6.12 The majority o f transfers to the poor are channeled through two large-scale government programs: Gratuitous Relief and the Employment Generation Scheme (EGS). In addition there i s a fairly large-scale Food-for-Work Program run by WFP, and a number o f smaller programs, including school feeding, and a range o f cash and food distribution and works-based transfer pilots run by various NGOs with donor financing. The following paragraphs provide a brief overview o f the programs. 6.13 In distributing relief, the government rightly makes a distinction between two broad vulnerable groups: (a) those that cannot and should not work (children, women with many children, the elderly, the disabled), and (b) the able-bodied poor who should work. Gratuitous Relief refers to the free distribution o f food received as aid, generally donated by donors such as the USA, the EU, WFP, and others, intended primarily for vulnerable groups under (a) above. Targeting i s administrative- and community-based. The Employment Generation Scheme (EGS) i s a program under which food from the same sources i s distributed as payment for work performed; targeted partly o n the basis o f the l o w wage rate, and partly administratively. 6.14 The policy i s that the majority o f food aid (80 percent) should be distributed through EGS with a work requirement, and the remaining 20 percent distributed free to those who cannot work. In fact, while it i s difficult to confirm from the available data what the true distribution is, all evidence suggests that the majority o f food i s distributed free, because o f difficulties in mobilizing EGS works, the lack o f counter-part funds to finance works, and the fact that needs are often so urgent that it i s not practical to await the scaling up o f works. The amounts targeted per person (1.5 kg. o f grain per person per day) are approximately equal under both programs. 6.15 Due to the inconsistency o f data, and lack o f systematic reporting, it i s difficult to assess precisely what proportion o f the poor benefit from these programs, how much they receive on average, and the accuracy o f their targeting. Nonetheless we can make a number o f observations: 0 While there are some inclusion errors, evidence suggests that targeting o f transfers i s not particularly bad in Ethiopia - perhaps in part because the level o f need i s so great that the risk o f major inclusion errors i s reduced; The greater problem i s that the available supplies are diluted by the large numbers needing assistance; localized studies by SCF(UK) and USAID show that food aid i s often given to more people than originally intended, and in some cases to all families within a Peasant Association area, and as a result none receive sufficient amounts; The amount transferred - if restricted to the very poorest - would represent a significant share o f the poverty-line income (about 14 percent according to one study); however the dilution referred to above suggests that many households receive substantially less than intended; In works-based programs the number o f days worked i s often very l o w - with the number o f participants being increased in order t o raise coverage, but at the cost o f reducing the number o f days worked for each participating household; Timeliness o f payment i s often a problem, especially under EGS, with food supplies arriving for distribution well after the most critical period 130 o f need has passed (this i s due in part to weaknesses in distribution and part to late arrival o f food aid inputs); 0 EGS suffers from never having been established as a continuously- operating program, but rather has operated o n a stop-start basis, with whatever (generally very meager) resources can be mobilized from local governments and donors to support implementation; and n o assigned institutional responsibility for implementation, monitoring, and evaluation. 6.16 The program framework comprising gratuitous relief, an employment generation scheme, and foodcash for work program, evolved and continues to function as a means to address short-term acute food shortages. Together with other measures such as establishment o f a well-functioning early warning system, and some national reserves, this policy framework has served the short-term needs and saved lives in the wake o f periodic failures o f the rains. Indeed, this kind o f management o f food aid to overcome critical transitory food shortages i s a major achievement, inasmuch as i t has prevented recurrence o f mortality o f the magnitude that occurred in 1984. 6.17 There are other notable good features o f the current system. For example, the use o f local NGOs to deliver programs with donated food (which reduced the burden on the government), and adoption o f administrative and self-targeting approaches that have undoubtedly helped food reach the most vulnerable groups. T o be sure, there have been some exclusion errors and mistargeting, but this i s not widespread. 6.18 Yet, the progradpolicy framework failed in three critical respects: First, it fails to address the issue o f chronic food insecurity. Secondly the entire framework fell outside the planning process. For example, workfare programs are not integrated with activities at various levels o f government; instead, implementation i s generally left to communities or NGOs; with inadequate resources they cannot effectively create assets nor sustain transfers. Thirdly, although the three main programs were geographically concentrated in the food-insecure woredas, this has not necessarily lead to better regional targeting outcomes. 124 Finally, the policy framework as a whole did not include structural measures such as diversification o f income-generating activities, market development, and population-related efforts. 6.19 If these shortcomings are to be overcome, one needs to look at another, alternative safety net program framework that combines the central function o f r i s k reduction with developmental goals - physical such as agro-infrastructure, and human capital development related programs such as nutrition, education vouchers - and which i s sustained on a more continuous basis. The next sections provide estimates o f current spending on safety nets, and some proiections o f spending under alternative scenarios with respect to coverage, and offer some perspectives o n the role o f the government and donors. See background lZ4 paper o n safety net spending for quantitative evidence. 131 B. H O W MUCH I S ETHIOPIA SPENDING ON SAFETY NETPROGRAMS?- CURRENTPUBLIC EXPENDITURES 6.20 I t i s difficult to ascertain the total level o f public spending on safety nets accurately for a number o f reasons: (1) most transfer programs are funded by relief aid, which fluctuates dramatically from year-to-year, (2) much o f this relief i s off-budget; and (3) the local counterpart portion o f spending (on distribution, administration and the like) i s often buried in other parts o f the overall budget - either o f the government or o f donors and NGOs. 6.21 Within these limitations, we estimate total spending o n transfers to the poor in Ethiopia amounts to, very roughly, about U S $ 265 million equivalent p.a. This amount fluctuates widely, from a l o w o f about $120 million in ‘good’ years, to a high o f about $500 million in ‘bad’, drought years. The figure o f $265 million represents about 13 percent o f total public spending, I2’and about $ 4.00 per capita (but over $40 per beneficiary, for the approximately 6 million people who are estimated to be eligible for transfer programs in one way or another.) Table 6.1: Estimated Levels of Safety Net Spending (US$ millions) 1997 1998 1999 2000 2001 2002 Humanitarian Relief & Related \a 181 152 232 449 193 383 Est.Government & Other hending \b n.a. 20 73 33 n.a. n.a. Total n.a. n.a. 252 522 226 n.a a/ as estimated in DevelopmentCooperationReport, 2002 (UNDP) b/ Mission estimates; including counter-part implementation spending related to humanitarianrelief, and other non-relief transfer programs. These are very preliminary estimates; and do not include direct NGO spending for which consolidated data was not available at the time of the mission. 6.22 The chart below shows the approximate composition o f funding for these transfers. Note that there i s substantial variation from year to year (these are based on a five-year average), and that the valuation o f food aid can differ substantially between donors depending o n how competitively the food i s sourced. Furthermore there i s in some cases double-counting o f aid provided by bilateral donors, but provided through WFP. As a result o f these factors the chart probably under-states the relative contribution o f the EU and bilateral European donors, and over-states that o f the WFP and USA. 125 Defined in this case as budgeted expenditure plus relief inflows from donors. 132 Figure 6.2: Approximate Distribution of Funding of Transfer Programs 1996-2000 Estimated 34% I Sources: WFP & EU as reaorted in W.Tassew (2002): EU assistance val ed at $0.95/Euro; US, Bialteral & NGO r e l i e i based on “Volume of Food Aid Delivered” in “Ethiopia: Grain Marketing - Review of recent Trends”; Paul Harrison, August, 2002) and valued at $300/mt.; GoE & local overheads estimated very roughly at 5 percent of total relief aid. Broad Program Expenditures 6.23 As described earlier, spending on safety nets consists o f three main programs: the established Food-for-Work program, managed mostly by WFP; free distribution o f food provided as r e l i e f aid; and food aid distributed under the Employment Generation Scheme. In addition there are a range o f smaller transfer programs run by various NGOs, and Govemment funds the operations o f a foodgrain buffer stock under the Emergency Food Security Administration. It i s almost impossible to say reliably how much i s spent on each o f these: in theory 80 percent o f relief aid i s distributed under some form o f work requirement; but in reality the capacity constraints o n EGS mean that a large proportion i s distributed as free food. 6.24 More generally, we found a real problem o f weakness o f data , and - except in some cases - a failure to measure the outputs or outcomes o f safety net programs. This i s in part the result o f the ‘relief-oriented nature o f efforts, which has focused o n accounting for inputs. Even where data i s available, i t i s almost impossible to marry information on costs with amounts actually transferred, and with the number o f beneficiaries, or how much transfer they received. 6.25 As far as we can estimate from the available data about 10 percent o f public safety net spending i s o n Food-for-Work; 20 percent on EGS, and 60 percent on free food distribution. lZ6 In addition there are a range o f smaller programs, including expenditures on the Emergency Food Security Administration; the School Feeding Program, cash-for-work pilots run by SCF(UK) and the EU, and a pilot asset generation program being supported by USAID, which together account for the remaining 10 percent. Expenditure on most i s small relative t o the large Gratuitous Relief and EGS transfers (which average about $250 million p.a. between them) - for example about $15 These data based o n very rough data provided by various sources, however they are not all consistent, and need to be refined following review with government and donors in Ethiopia. 133 million i s spent annually on the Food-for-Work program, and $3 million p.a. on school feeding. Cost-Effectiveness o f Current Program Spending 6.26 The table below shows the approximate cost per Birr o f transfer under a sample o f existing programs. Comparisons between them are o f course difficult, because the programs - especially the work-based ones - have very different overhead cost structures, and serve very different purposes. Furthermore, the costs and benefits are not consistently measured, due to the data problems mentioned above. Table 6.2: Some Estimates of Transfer Costs - Selected Programs Approximate Approximate Total Number of Transfer per Cost per Program Expenditure Beneficiaries Beneficiary Birr (in value to beneficiary) Food-for-Work \a $19.9 m. n/a $5.11 1.88 EGS \b n/a $ 11.22 n/a Gratuitous Relief \c n/a n/a $ 8.82 1.50 School Feeding \d $4.1 m. 258,000 n/a 2.50 Cash-for-Work \e $2.5 m. 20,341 $ 9.55 1.30 a/ FFW - Expenditure based on WFP country program table ($59.6 m. over 3years). No.of beneficiaries to be confirmed; benefitkosts based on distribution o f 134,824 mt. Over 3 years. [Food transfers valued at 2 birrkg. exchange rate of 8.6/US$.] b/ Based on WFP evaluation showing average transfer of 47.7 kg of food per beneficiary under EGS, and of 37.5 kg, under free food distribution; valued at 2 birr per kg. c/ Avg. benefit based on WFP evaluation that reports avg. receipt of 37.5 kg/person, valued at 2 birr.kg., at 8.5 birr/US$; cost-benefit based on reported costs of delivered food of $350/ mt, and benefit valued at 2 birr k g . (= approx. $235/mt.) d/ From WFP program table for 2003-06; $ 12.4 million over 3 years, to distribute commodities valued at $4.9 m. ($1.6 m. annually) internationalprocurement price; Tassew report implies 46 kg. per beneficiary. e/ based on SCF (UK) report on pilot for period Feb-May 2001; cost of Stg. 158,325 for 20,341 beneficiaries, of which stg. 121,425 distributed as wages; valued at US$l.6/Stg. 6.27 The average cost o f imported grain under safety net programs in Ethiopia (excluding program implementation and administration costs) i s made up approximately as follows: International Price $130 /mt. International Shipping : $50 Transport Djibouti-Regional center $65 Local distribution & transport: w Total cost: $285 134 When administrative costs are added, even free distribution can cost over $350 per ton, equivalent to about 3 Birr per kg. o f food; this compares with average local market prices o f about 1.5-2 birr per kg. for most foodgrains. 127 6.28 The costs o f delivering a dollar’s worth o f benefits would be substantially lower if cash were distributed. However, the problem i s that food i s just not available at any reasonable price in many o f the affected areas, especially during drought years. At the same time, however, the continued preponderance o f food distribution programs - o f whatever kmd - undermines the development o f inter-regional trade and price signals that are ultimately needed to ensure better-functioning markets. Clearly this i s a process that needs to be phased, but as progress i s made in developing functioning markets, the fastest-possible shift to cash-based transfers i s warranted. 6.29 The cost-effectiveness estimates for works-based programs are somewhat misleading, because we know that the current EGS program i s badly-underfunded. Complementary inputs, such as supervision, planning and design, materials and equipment are only financed at a very l o w level (or not at all). In some cases a donor or NGO provides some inputs - typically to about 20 percent o f program costs - but based on experience worldwide, we estimate that such complementary inputs typically cost one dollar for each dollar transferred as wages - so they should amount to about 50 percent o f program costs. 6.30 Furthermore, although table 6.2 shows cost per unit transferred, it does not really t e l l u s anything about the effectiveness o f these transfers. For a number o f reasons cited earlier, spending i s often ineffective: because food comes too late; or because the amount o f food distributed i s so diluted (due to insufficiency o f overall allocations, or the fact that it i s disbursed among many more beneficiaries than was intended), that each household receives too little to materially affect their welfare. In the case o f works programs, the stop-start nature o f programs prevents them having a sustained impact on incomes o f the poor; and the absence o f counterpart funds and integration with local capital plans means they often do not result in creation o f lasting, productive assets. C. WHAT COULD ETHIOPIA SPEND ON A SAFETY NETPROGRAM? POTENTIAL - NEEDS. PUBLIC EXPENDITURE 6.31 The amount Ethiopia could potentially spend o n a safety net for the poor i s obviously extremely high, because the number o f potential beneficiaries i s very large, both in terms o f those whose consumption i s at unacceptably l o w levels even in normal years, and o f the huge numbers susceptible to drought. 6.32 Clearly the solutions to this problem l i e outside o f the realm o f safety nets: in improved productivity o f labour - both in agriculture and elsewhere - in a more diversified production base, managing population pressure, and migration. But even with the best possible outcomes, there will remain unacceptably large numbers o f people who 127 See Harrison (2002) and SCF (2001); also note costs o f internationally procured food aid can be even higher than those indicated here; recent U S estimates at almost $450 /mt. delivered. ”* A study o f the impact o f food aid o n domestic prices concluded that in an average year prices were depressed by about 5 percent as the result o f each million quintals o f food aid inflows (Harrison, 2002). A major effect i s when large amounts o f food come too late, and overlap with the harvest. 135 will require direct transfers. The graph below shows projected number o f people living below the food-poverty line (that i s the level o f income required to sustain basic minimum food consumption) in Ethiopia over the next 20 years at both a 3 percent and 4 percent GDP growth rates. Figure 6.3: Projected Number of Persons Below the Food Poverty Line 35 1 01 I 2000 2005 2010 2015 2020 I+ 4% Growth + 3% Growth 1 Mission estimates based on population projections and income poverty-elasticity o f growth 6.33 To provide adequate food to all o f these people currently would require expenditure o f about $810 million p.a., representing 12 percent o f GDP, or about one- third o f all public spending. This i s clearly infeasible. Below we present projections o f what it might cost to provide a range o f more realistic levels o f safety-net coverage. 6.34 We have defined the options in terms o f the proportion o f basic food requirements that might need to be provided by a safety net. Table 6.3 below shows the broad unit costs, using different program instruments, associated with providing ‘light’ assistance (defined somewhat arbitrarily as 50 percent o f the minimum food requirement o f households for 4 months o f the year - this corresponds roughly to the level o f shortfall faced by many o f the moderately poor in Ethiopia, in many years) ; and with more substantial, ‘heavy’ support (providing 75 percent o f minimum food needs for 8 months o f the year - which corresponds roughly to the situation faced by the poorest in Ethiopia on a continuous basis, and by many additional households in bad drought years). 136 Table 6.3: Approximate Unit Costs of Large-Scale Transfer Programs (US$ millions per million persons served - constant 2002 US$ millions) Pure Food Public Works Distribution \a Program \b ‘Light’ Support $ 13.5 m. $16.8 m. [50 % offood requirement for 4 months o f theyear] ‘Heavy’ Transfer $40.5 m $50.2 m. [75 % of food requirement for 8 months of the year] Notes: a/ Food distribution assumes a full requirement o f 18 kg. per person per month (light support i s thus 36 kg. per person annually, and ‘heavy’ i s 108 kg.); assumes food cost o f $300/mt., and admin. costs of 25 percent; b/ works-based programs assume same requirements , and transfer required to purchase the defined level o f foodheneficiary; based on grain price o f 2 Biir kg. and program costs o f $1 per $1 transferred in wages. Note this i s cost per beneficiary, not works participant- wage rate and days worked could be adjusted in any combination to reached the desired transfer, depending on assumed number o f dependants per worker.’” 6.35 The total costs o f different safety net strategies will then depend on: ( i)the choices the government makes regarding: the numbers to be covered, ( ii)the degree o f support to be provided to them, and ( iii) the choice o f instrument. Below we present the costs o f 3 basic scenarios (details and assumptions are shown in the accompanying Box). Scenario 1: An “Aggressive” Safety Net Strategy: This strategy attempts to provide at least some support to the bottom 20-25 percent o f the population. I t corresponds to a situation o f persistent deep poverty (associated with 3 percent growth or less), and relatively bad luck with respect to exogenous shocks- foreseeing a major drought every 5 years; combined with Government taking a policy position that it i s willing to use transfer instruments to alleviate widespread suffering. The program details are shown in Annex 7, but broadly they involve providing a pure transfer to the bottom 10 percent (with the poorest 5 percent receiving a substantial amount, equivalent to 75 percent o f the minimum food requirement for 8 months o f the year); coupled with a sustained public works program that provides support for the second-poorest 10 percent o f the population in a normal year, rising to 15 percent in bad years. Such a program might cost something in the order o f US$ 375 million per annum currently, rising to over $500 million p.a. over the next decade. - Scenario 2: “Moderate” Safety Net Strategy which involves providing limited support to the very poorest (the bottom 5 percent o f the population); and a flexible Public Works Program to provide moderate levels o f support (equivalent to about half o f minimum household food requirements, for 6 months o f the year, o n average)) to between 5 and 10 percent o f the population in normal years, gearing up to provide more substantial support in bad years. Costs would rise from about $230 million annually now, to around $ 300 million by 2010. Note the assumption o f 1 Birr (or $) o f costs per Birr (or $) transferred represents a long-run cost estimate based o n international experience in low-income countries. While legitimate for long-run costing purposes, the short-run ratio w i l l almost certainly be lower in Ethiopia. To the extent that i t is, more person-days o f employment can be generated with the same budget. 137 Scenario 3: A Minimalist Scenario. This strategy involves a tightly restricted public safety net policy: essentially leaving poverty alleviation to growth and diversification, and providing only very limited humanitarian assistance to the poorest. I t assumes a declining number o f absolute poor. (associated with sustained growth o f over 4 percent p.a.); and infrequent droughts. The program would provide free food (or cash) to the poorest 2-3 percent o f the population, combined with a modest public works program, serving the next poorest 5 percent in most years, and no more than 10 percent o f the population in bad times.). Such a program could be delivered for approximately $135-150 million p.a. Table 6.4: Estimated Costs o f Safety Net Scenarios (Annual cost, in constant 2002 US$ millions) 2005 2010 2015 2020 ‘Aggressive’ Strategy 419.8 465.9 514.4 564.6 Moderate Intervention 255.0 283.1 312.5 343.0 Minimalist 152.1 163.1 173.6 190.6 6.36 While the costs o f the aggressive safety net strategy might seem excessive, i t must be remembered that the potential beneficiaries that we are talking about in the case o f Ethiopia are generally surviving on levels o f consumption that would place them among the very poorest almost anywhere else in the world, so considering some form o f transfer for them i s not inconceivable. Nonetheless, given the competing claims on public expenditure, it clearly makes sense to consider a more conservative approach. 6.37 The first strategy i s likely unaffordable; the $375 million p.a. would represent 18 percent o f total public expenditure, and given competing needs, in terms o f education, infrastructure, and other essential investments, it would seem unlikely to be a rational use o f public resources. 6.38 At the other end o f the spectrum, the ‘minimalist’ scenario would probably result in a level o f welfare loss, and human suffering, that i s unacceptable. Nonetheless it provides a bottom-line indication o f likely needs if things go well in terms o f raising consumption levels and diversifying away from dependence on rainfed foodgrains over the next two decades. 6.39 The middle scenario probably corresponds best to what i s likely to be required; and accords closely with the level o f resources currently devoted to transfers. However given the high opportunity cost o f public resources, combined with the ineffectiveness o f the current arrangements, there i s certainly room for improving the effectiveness, and probably reducing the incremental cost o f the program. 138 c 0 S n $ 3 r A iD0 E a 0 & ru 0 0 Y % 2 0 N d. 4 a 8 3 .r( iD iD d .. 69 x 6 z c B . c I 4 .. Y rc 0 ; i 0 .3 3 0 PI 6.40 Within this framework, Government and donors need to make a number o f basic choices: (1) forge a consensus on what the objectives of a safety net should be in Ethiopia (for example, whether to provide adequate minimum food consumption to most o f those at risk; to only cover those at risk o f death by starvation; or just to provide a true ‘safety net’ in times o f drought); (2) to be selective of which groups (or sub-groups) within the population should be provided with support (for example only those unable to look after themselves? (the elderly, disabled, labour-short female-headed households); whether to continue to focus on those living in selected geographic areas?) (3) to carefully manage the trade-off between spending on pure transfers and other poverty-reducing expenditures - by wherever possible using instruments that contribute to both objectives (such as employment o f the poor on public works that contribute to longer-term asset creation and income growth). (4) to maximize the welfare impact o f each birr spent on safety net transfers; (for example by timing transfers - either during the year or the lifecycle - so that they have the greatest impact, or using safety net spending to achieve multiple poverty-reducing objectives). 6.41 It i s our view that given the extreme scarcity o f public funds in Ethiopia, the level o f spending on pure transfers should be kept to a minimum, and most safety net spending directed towards works-based programs that can also finance investment in long-term assets. The fact that labour i s underutilized during the non-agricultural season i s an added reason for adopting an employment-based transfer scheme. Furthermore, Ethiopia i s fortunate in s t i l l having a high level o f investment in social capital in many areas; one manifestation o f this i s the fact that communities, or extended families, tend to look after many o f the elderly, the infirm, or the disabled, as a result o f which there are not many o f the very poor who are outside o f households which have able-bodied members that may be targeted by publics works programs. 6.42 Finally, we note however the extreme levels o f malnutrition compared to other countries, and believe that the safety net strategy should address the needs o f malnourished children below the age o f 30 months, given the long-term damage that can be avoided by selective nutritional interventions at this stage. D. WHAT SHOULD ETHIOPIA SPEND ON A SAFETY NETPROGRAM? 6.43 As a result we would recommend a program that focuses o n poor households with one or more members capable o f participating in public works programs, wherever they are in Ethiopia; and o n malnourished children. Such a program might consist o f : A public-works based transfer program (geared to reach the poorest 10-15 percent o f the population; but making use as far as possible o f funding already being spent o n rural infrastructure; run on a continuous basis, with a permanent institutional structure capable o f 140 operating the program, with adequate counterpart funding for supervision and capital inputs; run in a way that integrates the assets being constructed with local infrastructure plans; with employment focused in the dry season, and designed only to provide supplemental income (equal to about half o f household food requirements, but capable o f being scaled up quickly in bad years). 0 a nationwide nutrition program (geared particularly at children under 30 months old, but not completely restricted to them; that combines growth monitoring with counseling and incremental feeding). only very selected transfers for those unable to work (restricted to those who cannot be covered by the first two programs - that is, those who are not in households that contain able-bodied adults - single disabled, elderly or infirm; orphans and street children. Given the difficulty o f targeting such people, it i s recommended that a community-based or NGO programs be tested on a fairly limited pilot scale). 6.44 Estimated costs o f such a program might be as follows13o: Public Works: for 10 percent o f the population, about $110 million per annum. Nationwide Nutrition Program: About US$ 20 million equivalent per annum. 0 Selected Gratuitous Transfers: about US$20 million p.a. 6.45 These are very rough estimates, and may be on the l o w side, but they give an indication o f the kind o f public spending that might be involved. Note the total amount i s less than i s currently being spent on transfers (about $265 million annually). The model being proposed i s now being pursued by the subsequent work o f the Coalition for Food Security, which i s proposing a safety net package comprised o f a mix o f flexible public works program which creates community assets, conditional transfers to households to protect human development assets, and a residual share o f the budget as unconditional transfers to the very needy who are unable to participate in workfare programs (the old, the disabled, or single women with many young children, etc.). Issues Government and Donors Need to Confront in Moving Forward 6.46 In this chapter we have only been able to sketch the outlines o f a solution, and what i t might cost. However we can identify a number o f key issues, or bottlenecks, that the government and donors will have to address in moving forward: 130 Based o n the following assumptions: For public works: coverage of about 6.5 million persons, providing the equivalent 50 percent o f minimum food requirements (18 kg. per person per month) for 4 months o f the year; at a food price o f 2 birrkg.; (implies about 1.4 million persons employed at an average of about 54 days per year each; assuming a ratio of 4.5 persons benefiting per worker; and a wage rate o f Birr 6 per day (US30.70); for nutrition program; assuming 25 percent o f under-30-month children (approximately 1 million in 2003) , and a cost o f $20 per beneficiary per annum; for pure transfers, assuming limited to pilot for bottom <1 percent o f population (say maximum 500,000 persons) and provides o n average 50 percent o f food needs, 12 months o f the year; at $300 per ton, and 25 percent administrative costs. 141 0 Capturing the Value o f Current Food Aid - Shifting from food to cash, and from emergency relief to a sustained system o f transfers, needs to be done in a way that minimizes the negative impact o f net resource inflows. Almost 90 percent o f current transfers to the poor are supported by commodity aid. O f this, we estimate perhaps two-thirds comes from sources which are unlikely to be available to Ethiopia except in the form o f emergency food relief aid 131- it i s important to manage this process so that Ethiopia does not forego major amounts o f net aid transfers (for example, by using resources from those donors who will only provide food aid to resource the nutrition and food-for- work programs). 0 Shifting to a continuous program, with associated functioning institutional arrangements. 0 The degree o f shift away from geographic targeting. In general the safety net system should move towards one which benefits the poor wherever they are located. However, the transition needs to be monitored, to ensure that particularly vulnerable areas are not suddenly left without. 0 H o w to maximize the use o f currently-budgeted expenditures (eg. on rural works and maintenance) to effect transfers. 0 The conditions under which a shift from food to cash distribution can take place, and the pace at which it should take place. Role o f local government, as opposed to a vertical program in operating a safety nets program. Given the impact o f inter-seasonal price swings o n the welfare o f the poor, it i s worth assessing the potential role o f price-smoothing mechanisms. 13' Assuming USAID, WFP, and about half o f bilateral current relief assistance i s not fungible. 142 7. WATER SUPPLY AND SANITATION A. INTRODUCTION 7.1 Sustainable improvements in water and sanitation conditions are essential to improve the quality o f l i f e o f the poor through: (i) positive impacts o n hygiene and maternal and child health, (ii) reduced drudgery and time spent o n collection o f water, especially for women, ( i i i ) reduced cost o f health services especially for water related diseases such as diarrhea - for both households and the government, (iv) increased productivity due to improved health, (v) improvements in education enrollment and attendance through better school sanitation, and reduced home duties for water collection or caring for siblings, especially for girls, and (vi) increased income from cattle that depend on water. In Ethiopia the poor themselves highlighted the linkages between access to water and sanitation and poverty in consultations (1999) and ranked water supply among the top five priorities in all woredas and, in urban sites, also ranked sanitation high. 7.2 According to the WHO, Ethiopia has the lowest level o f water supply coverage in Sub-saharan Africa at 24 percent, and i s second to last in sanitation coverage (15 percent)13*. Also according to WHO, water coverage increased by only 2 percent over the ten year period from 1990 to 2000. However, the Ministry o f Water Resources (MWR) and Regional Water Bureaus report higher coverage levels (30% for water supply and 17% for sanitation in 2000) and greater gains in recent years (6% increase in water supply coverage in the past ten years). The differences are likely due to different definitions o f access and coverage. All agree that the level o f service provided to those with access to improved water supply i s poor: on average less than the WHO minimum standard o f 20 liters per capita per day i s provided; service i s often intermittent in urban areas, and in rural areas between 20 and 30 percent o f systems are non-functional at any given time. The Government has put in place a good policy framework designed to provide efficient and sustainable service and i s committed to tackling the problem, but the challenge i s immense. 7.3 This chapter on water supply and sanitation (WSS) will focus on: analyzing WSS targets as set out in Ethiopian policy and strategy documents, the expressed wishes o f the population and in the Millennium Development Goals; 0 examining expenditure patterns in the sector to determine their consistency with policy objectives and priorities and their adequacy in terms o f meeting sector targets; 132 Source: World Health Organization and United Nations Children's Fund, Global Water Supply and Sanitation Assessment 2000 Report. 143 0 elaborating “development scenarios” for the sector for the next 20 years, tahng into consideration financial and other constraints on the Government’s ability to meet the targets; and 0 suggesting options for removing constraints and improving sector performance. 7.4 I t should be noted that much spending in the water sector i s not technically public expenditure. W h i l e the government pays the lion’s share o f capital costs, users contribute in the form o f connection fees for urban water supply, community contributions (cash and in-kind) for rural water supply, and direct purchase o f on-site sanitation. I n addition, user fees cover a large share o f the recurrent costs o f operating and maintaining urban water supply and sanitation systems. The Government’s Water Resources Policy should result in individuals covering an increasing share o f both capital and recurrent expenditures over time. Ethiopia’s Water Supply and Sanitation Objectives 7.5 The Water Resources Management Policy articulates policy goals, objectives and principles. The following excerpts from the policy sum up the primary policy goals: “As far as conditions permit it, every Ethiopian citizen shall have access to sufficient water, o f acceptable quality, to meet human needs. “the basic minimum requirements (basic human and livestock needs as well as the environment reserve) has the highest priority in any water allocation plan.” “Recognize that water i s a natural resource with economic value and ensure that fees are paid for services rendered.” “Ensure that pricing for urban water supply shall aim at full cost recovery.” “Ensure that the basic needs o f water for disadvantaged rural communities, who cannot afford to pay for the development o f water systems, shall be borne by the Government in so far as the communities are able and willing to cover the costs o f operations and maintenance.” 7.6 More specific targets for the sector are set out in the Water Supply and Sewerage Sub-sector Development Program (WSSDP). The main assumptions regarding targets for WSS are: Provision o f drinking water supply to practically all urban population by the end o f the program period (2016) - increasing coverage to 100 percent in Addis Ababa, Dire Dawa and all cities in the four large regions and doubling coverage in the smaller regions. Increasing rural water supply coverage from 23 percent to 71 percent by gradually improving the implementation capacity - 20 16 targets range from a high o f 90 percent for Dire Dawa and Tigray to a l o w o f 45 percent for Gambela. Increasing urban sewerage coverage by 3.5 percent annually in order to raise the urban sewerage coverage from the current 7 percent to 60 percent by the end o f the program period. 7.7 The spending and coverage targets for the first five years o f the WSSDP are reflected in the SDPRP (Ethiopia’s PRSP), which calls for increasing access to clean 144 water from 30% in 2000 to 39.4% by 2004105. Rural water supply i s to increase to 31.3% and urban water supply to 82.5% over a five year period. The SDPRP did not include coverage targets for sanitation. 7.8 The WSSDP targets are more ambitious than the Millennium Development Goals, which call for halving the proportion o f people without sustainable access to safe drinlung water and halving the proportion o f people without access to improved sanitation by 2015. T o meet the MDGs, water coverage in Ethiopia would need to increase from the year 2000 reported level o f 24 percent133to 64 percent by 2015 - new service would need to be provided to an estimated 44 million people while maintaining service to those who already have it. Sanitation coverage would need to increase from 15 percent to 57 percent. B. ANALYSIS OF EXISTING PATTERN OF EXPENDITURE Trends 7.9 Decentralization: Decentralization reforms are changing the way in which urban and rural water supply and sanitation are financed. In recent years the federal government has made very little direct investment in WSS from i t s own budget resources. Urban water supply has been financed primarily from loans and grants channeled through the federal budget (from donors such as IDA, AfDB, JICA and the Netherlands) and from the regional budgets. With the latest decentralization reforms, responsibility for financing urban water supply will increasingly pass to municipalities and their utilities. Rural water supply, which has been financed primarily from regional budgets, the Ethiopia Social Rehabilitation and Development Fund (ESRDF), bi-lateral funding and NGOs worlung primarily at the local level, will be increasingly financed through the woreda budgets, with some capital contribution from communities (e.g., minimum o f 5 percent cash contribution). Support from NGOs and bi-laterals i s expected to continue in partnership with the woredas. However, anecdotal evidence suggests that in the early stages o f the latest decentralization reforms very little i s being spent by woredas on rural water supply and sanitation. This i s probably due to insufficiency o f block transfers to cover even recurrent expenses, and the absence o f trained water supply personnel in most woredas. 7.10 Prioritization of urban over rural water supply: Regions and the federal government, with donor assistance, have spent much more in recent years o n urban water ~~ supply than on rural water supply and ~ a n i t a t i o n 'despite the fact that coverage i s much lower in rural communities and that rural communities account for 85 percent o f the population. Water supply systems in most o f the largest regional cities have undergone major rehabilitation and/or expansion, including Adama, Mekele, hum, Bahir Dar, Gondar, Awassa, Gambella and Asosa. In addition, construction o f the water supply system for Harar i s now getting underway. The justification provided for prioritizing provision o f water supply in urban areas includes the poor level o f service provided in 133 Note again that the starting point used for estimating MDG requirements here i s lower than the starting point used by the WSSDP. The Ministry's calculations underlying the WSSDP rely o n coverage information reported by the regions, which differ from both the WHO and CSA reported coverage figures. 134 Though if estimated expenditures by NGOs and ERSDF are factored in, spending o n urban and rural water supply has been more roughly equal in recent years. 145 urban centers resulting f i o m the age and level o f deterioration o f existing systems, the related health risks in areas o f concentrated population, the high growth rate in cities, the need to develop urban centers in conjunction with decentralization and the fact that urban centers tend to be drivers for growth and development. The investment cost per capita i s higher for urban than for rural water supply, but so i s the ability o f users to pay for their upkeep and expansion once the initial investment has been made. Regions and the federal government have, however, indicated the need to shift focus to development o f water supply and sanitation in rural communities and smaller towns. 7.1 1 Futureflnancing o f urban water supply: The provision in the Water Resources Policy that states pricing for urban water supply shall aim at full cost recovery will change the way investments in urban water supply are financed. This provision i s being interpreted to mean that once an urban water supply system has been constructed with grant financing, all future investments are to be financed from internally generated funds, with tariffs set at levels sufficient to support system renewal and replacement as well as system expansion. Since replacement and expansion will likely require borrowing, there i s increasing emphasis on raising the level o f cost recovery and improving operational efficiency so that water utilities can establish themselves as creditworthy borrowers. The federal government has established the Water Resources Development Fund to channel loans to urban water supply systems in accordance with the policy. Regions are expected to continue to provide grant financing (including with support from donors) to establish urban water supply systems, but regional budgets for urban water supply and sanitation will increasingly be targeted to smaller towns and for recurrent expenditures for technical support and monitoring and evaluation. 7.12 Sustainability o f urban water supply systems: Tariffs in urban water supply systems have been insufficient to provide for adequate maintenance, renewal and replacement or expansion o f systems. As a result, many existing urban systems are providing a l o w level o f service - average consumption i s below the WHO minimum standard o f 20 liters per capita per day; rationing and unscheduled disruptions in service are common due to seasonal water shortages and equipment failure; customers wanting to connect to the system often must wait for months or even years until water, meters or pipes become available. At the same time water losses are high and operating efficiency levels l o w in most utilities. 7.13 Futureflnancing o f rural water supply - The Water Resources Policy calls for communities to cover operation and maintenance expenses through user fees. Woreda budgets will need to cover investment costs to improve coverage and recurrent costs for program implementation, monitoring and evaluation and to support communities. ESRDF i s to be phased out within the next two years and it i s expected IDA and other donors will support rural water supply through the Poverty Reduction Strategy Credit and through a multi-sector decentralized infrastructure fund 13’. However, the woredas are ill-prepared to take over this responsibility, and support will be needed in the form o f capacity building to help them develop their water supply and sanitation programs, develop in-house skills and partnerships with local NGOs and the private sector to support communities and develop improved systems for monitoring and evaluation. 135 Under discussion. 146 7.14 Improving sustainability in rural water supply: An estimated 20-30 percent of rural water systems in Ethiopia are non-functioning at any given time. Issues include lack o f access to and ability to pay for handpump spare parts and maintenance support and failure o f shallow wells in the dry season. Increasingly the Demand Responsive Approach, which requires demonstration by communities o f their willingness and ability to manage and properly maintain their rural water supply systems, i s being adopted as a means o f improving sustainability. 7.15 Improving sanitation and hygiene - The Water Resources Policy calls for sanitation planning and hygiene promotion to be carried out in conjunction with improvements to water supply. Government’s role has been primarily in sanitation promotion and hygiene education, but institutional arrangements are not well defined and water supply and sanitation have not always been well coordinated. 7.16 Capacity constraints - The sector suffers from lack o f financial resources, manpower, and construction capacity, which limit the ability to utilize available funds efficiently and effectively and to meet sectoral targets. With decentralization many staff in the public sector have been transferred and find themselves without the required skills or resources to do their jobs. Capacity building i s needed for staff in the public sector at all levels. In addition, decision makers at the woreda and municipal levels need to better understand the sector, especially the benefits that can be gained from improving water supply and sanitation coverage and what i s required to ensure sustainability. There i s an increasing recognition o f the need to involve new actors - local artisans, small scale private operators and NGOs as well as a broader range o f consultants, contractors and professional utility operators. However the enabling environment for private sector participation in the water sector i s weak, and effective demand limited by the ability to Pay* Composition of Water Supply Spending 7.17 An attempt has been made to summarize government spending in the sector. However, due to the serious problems encountered in obtaining and reconciling data from different sources, there are gaps in the data presented below and what i s presented should be taken as indicative rather than definitive. Overall, a reasonable estimate o f current annual government spending on water supply and sanitation, including external assistance, i s around 450 million Birr (US853 million). An estimated additional 100 million Birr per year (US$Z2 million) i s contributed by NGOs. Recurrent costs account for about 25 percent o f the total, with the balance going for capital expenditures for development o f urban and rural water supply, the majority o f public spending being for urban schemes. Only a negligible amount i s spent o n sanitation 7.18 Federal spending - Federal expenditures o n the water sector as a whole, including on water supply and sanitation, irrigation, basin management, master plan studies, etc., have fallen in recent years, in both real terms and as a percentage o f the total budget. Actual spending f e l l from 192 million Birr in EFY1992 (1999/2000), 7.7 percent o f the total federal budget, to 176 m i l l i o n Birr in EFY94 (2001/02), 5.5 percent o f the total federal budget. However, the federal allocation for water supply and sanitation was scheduled to increase from 85 million Birr in EFY 1994 to 225 million Birr in EFY95 (US$lO million to US$26 million). The increase i s primarily attributable to establishment o f the Water Resources Development Fund (WRDF), which will channel 147 loans for urban water supply and irrigation projects. The EFY 1994 budget for WRDF for both water supply and sanitation and irrigation projects i s as follows: Federal Budget 22.8 million Birr (US$ 2.7 million) Foreign Loans 99.8 million Birr (US$ 11.7 million) Foreign Grants 41.O million Birr (US$ 4.8 million) Total 163.6 million Birr (US$19.2 million) 7.19 Other external assistance included in the federal budget for the coming three years i s as follows 136: Table 7.1: External Financing I n Budget for Water Supply EFY 1995-97 EFY 1995 EFY 1996 EFY 1997 (Birr millions) (2002103) (2003104) (2004105) IDA Domestic Funds 0.7 4.2 4.0 Loans & Grants 36.1 JICA Domestic Funds 2.1 4.2 4.0 Loans & Grants 5.2 10.2 9.9 Netherlands Domestic Funds 2.6 Loans & Grants 18.5 Subtotals Domestic Funds 5.4 8.3 8.1 Loans & Grants 59.8 _. 10.2 - 9.9 Total 65.2 18.5 18.0 7.20 Regional Expenditure - The following table summarizes WSS spending in the four largest regions for the past several years, including funds received from foreign grants channeled directly through the regions. I t should be noted that changes to the system o f fiscal transfers can be expected to reduce direct spending by Regions o n water supply and sanitation as responsibilities shift to the woredas. Table 7.2: Regional Government Spending on Water Supply - Four Largest Regions Birr millions Amhara Oromiya SNNP Tigray EFY 1993 Capital: Actual Urban 11.1 73.9 3.9 4.8 (2000/01) Rural 3.8 12.0 9.5 7.8 Recurrent 113 . - 44.2 -1.9 - 2.3 Total 26.2 130.1 15.3 14.9 EFY 1994 Capital: Actual Urban 13.5 53.1 4.2 2.5 (200 1/2002) Rural 5.4 36.4 12.0 12.3 Recurrent 5.7 - 48.0 - 2.5 - 2.6 Total 24.6 137.5 18.7 17.5 EFY 1995 Capital: Budget Urban 49.2 32.2 5.0 2.4 (2002/03) Rural 0 30.5 9.1 1.o Recurrent -3.6 46.7 - 2.5 - 4.2 Total 52.8 109.4 16.6 7.6 136 Note that these are committed amounts and do not include estimates for new projects n o w under preparation, such as IDA’SWater and Sanitation project. 148 7.21 Data was not collected on actual and budgeted spending in the smaller regions for the past three years, but during EFY89-92 spending in the four large regions accounted for roughly 80 percent o f the total spent by all regions o n water supply and sanitation, excluding Addis Ababa. Therefore, a reasonable estimate o f the annual spending on water supply and sanitation by regions other than Addis Ababa i s 250 m i l l i o n Birr per year (US29 million). Capital expenditures in Addis Ababa have averaged about 60 million Birr per year (US$7 m i l l i ~ n ) ' ~ 'bringing , the total spending to over 300 million Birr per year (US35 million). 7.22 Ethiopian Social Rehabilitation Development Fund (ESRDF) has been one o f the primary sources o f financing for rural water supply and sanitation and has also financed a small amount o f urban sanitation. Rural water supply and sanitation has accounted for just over 30 percent o f the total spent under ESRDF over the five year period from 1996 to 2001 and urban sanitation for 0.8 percent. Over the past six years, EFY 1989 to 1994, ESRDF has budgeted close to 300 million Birr for rural water supply and sanitation, an average o f about 50 million Birr per year (USS.9 million). 7.23 Woreda Expenditure - Regional transfers to woredas are a recent phenomenon. Few data are available and they do not allow any breakdown o f where these expenditures go or what they are used for. Anecdotal information suggests that transfers to woredas to date have been barely sufficient to cover recurrent costs and that woreda spending on development o f new rural water supply and sanitation schemes has been negligible. 7.24 Municipal Expenditure - The city states o f Addis Ababa and Dire Dawa are responsible for provision o f water supply and sanitation and provide the service through their autonomous water authorities. Addis Ababa City Administration has financed investments o f roughly 200 million Birr (US23 million) over the past five years for the Addis Ababa Water and Sewerage Authority (AAWSA). In the future other municipalities will assume responsibility for provision o f water supply and sanitation service, with the service to be provided through autonomous water utilities in accordance with the Water Resources Policy. Establishment o f municipalities i s starting with the largest cities. Evidence suggests most o f the new municipalities will not generate sufficient own-source revenues in the near future to be in the position to finance significant investments in water supply and sanitation, though this could be the case in the long run. 7.25 Water Utilities - The Regional Water Bureaus indicate that tariffs o f most urban water utilities are now set to cover at least recurrent operating costs (salaries, power, chemicals, supplies), but not always at a level sufficient to provide for adequate maintenance, renewal and replacement o f existing assets. The largest urban water utilities (e.g., Addis Ababa Water and Sewerage Authority, Dire Dawa, Adama, Awassa, Mekele) sometimes generate operating surpluses, but so far not enough to finance significant capital investment. Addis Ababa has implemented the first in a series o f tariff increases designed to provide a sufficient level o f cost recovery to fund renewal and replacement and borrowing for future expansion from internally generated revenues. 13' Addis Ababa Water and Sewerage Authority has covered recurrent expenditures and a small share o f capital expenditures from its own revenues, an average o f about 60 million Birr per year (US$7 million). 149 Other urban utilities are also expected to phase in full cost recovery tariffs in accordance with the Water Resources Policy, including the 25 towns rehabilitated under the IDA- funded WSDR Project, Harar and 13 towns in Oromia. 138 7.26 Non-Governmental Organizations (NGOs): A number o f NGOs operate, especially in rural water supply and sanitation. I t i s estimated NGOs spend roughly 100 m i l l i o n Birr per year (US$12 million) on rural water supply and sanitation in the four largest regions. Data was not available for the smaller regions. 7.27 Sanitation: The Ministry o f Water Resources and Regional Water Bureaus have not been responsible for sanitation and hygiene promotion and, therefore, have neither tracked nor projected spending for sanitation, other than for urban sewerage and to satisfy requirements o f particular donors and bi-laterals. In any event, spending on sanitation has been limited- there has not been much to track. Observations on effectiveness, efficiency, adequacy and appropriateness 7.28 Effectiveness o f spending o n water supply and sanitation i s usually indicated by the increase in coverage corresponding to spending, and both quantitative and qualitative measures o f the sustainability o f systems installed. From the data available, it i s not possible to correlate spending in the sector to a change in coverage levels due to the lack o f an effective monitoring and evaluation system. However, the fact that coverage increased by only 2 to 6 percent from 1990 to 2000 indicates that there may be a problem with the effectiveness o f spending. There i s strong evidence that sustainability i s an issue for both rural and urban systems, with an estimated 20 to 30 percent o f rural systems non-functioning at any given time, and urban water supply tariffs insufficient to provide for adequate maintenance, much less renewal and replacement or expansion o f facilities. 7.29 Efficiency o f spending on water supply and sanitation i s usually indicated by the cost per capita o f investment, cost o f ongoing operating and maintenance, and operating nterms o f investment cost efficiency and financial performance ratios o f water utilities. I per capita, Ethiopia appears to be relatively efficient, though there i s always room for improvement. Government i s exploring ways to bring down the per capita cost o f investment by providing incentives in the form o f access to financing for selection o f the most cost-effective and affordable technology. 7.30 For urban water suppy, the Master Plan consultants estimate investments will cost roughly $80 per capita, while the WSSDP estimates the cost will be $71 per capita .139 These figures are consistent with estimates made for other urban systems in the Africa region. An independent study carried out by a local engineering firm in Ethiopia, Tropics Engineering Consultants, estimates lower costs are possible in most cases, depending on the type o f technology required and the size o f the town, as shown in the table below: 13* Only SNNP region provided data o n annual revenues from urban water utilities, showing an increase o f 60 percent from EFY1993 to 1994 and 40 percent from EFY1994 to 1995. 13’ Cost per capita includes capital cost and consulting fees for design and feasibility studies. 150 Table 7.3: Estimated Cost Per Capita For Different Types Of Urban Water Supply Systems Population Type Of System 1,000 2,000 5,000 10,000 20,000 50,000 -- -- -- -- -- Hand dug well and handpump $25 __ __ __ -- __ Borehole and handpump Borehole and motorized pump $35 $60 $35-60 $35-60 $35-60 $35-60 __ River intake -- -- $95 $80 $70 $60 Impoundment -- -- $130 $100 $85 $70 7.3 1 Data from recently designed, constructed and/or rehabilitated urban water supply systems in Ethiopia confirm these lower estimates. Table 7.4: Sample Costs of Town Water Supply Systems Design population Per capita investment cost Adama (Nazaret) 260,000 $48 WSDR Project =30,000 average $30 average 25 towns (rangingfrom 15,000 to 100,000) (ranging from $70 to $10) Oromiya Region 45,000 average $40 average 12 towns (ranging from 20,000 to 175,000) (ranging from $70 to $20) 7.32 The cost per capita o f urban water supply systems i s largely dependent o n the technology required and the size o f the town, but investment cost can be brought down and sustainability improved if consumers are involved in making choices about system design based on their willingness and ability to pay for different levels o f service. The Ministry o f Water Resources and regional bureaus are shifting from a technical, engineering-driven approach to project design and are incorporating cost effective design principles, including participatory methods and focus on financial and economic feasibility, into manuals for project design. 7.33 Comprehensive data o n operating and financial performance o f urban water utilities i s not available, but the limited data available indicate that most water utilities have not been operating efficiently - water losses are high 140, customer billing often lags by two or more months and i s not based on actual meter readings, collection ratios are low and utilities have been over-staffed with under-qualified personnel. In addition, financial performance has not been good. Until recently many utilities were receiving subsidies for operations, but tariffs have been increased in most towns in recent years to cover at least recurrent operating costs. Decentralization i s expected to bring to bear increased pressure for improved financial and operating efficiency and improved service levels by utilities. Municipalities responsible for service provision are likely to require more accountability from the autonomous water boards. In addition, lenders such as the Water Resources Development Fund will require preparation o f business plans, including plans for improved management, better efficiency, financial performance, financial management, performance monitoring, and related capacity building as a prerequisite for lending. I4O Available data suggest unaccounted for water i s as high as 50 percent in many utilities. 151 7.34 For rural water supply the best source o f data o n average cost per capita o f different types o f schemes comes from a study carried out by Tropics Engineering Consultants. The findings o f the study are presented below: Table 7.5: Estimated Unit costs o f Rural Water Supplies Construction Total Type of scheme cost per capita cost per capita incl. technical assistance Hand dug well or spring $15 $25 Hand dug well whandpump $20 $30 Borehole with handpump $25 $35 Borehole w/electric or diesel pump $25-50 $35-60 7.35 These figures are consistent with data obtained from the RWSEP project in Amhara region (supported by Government o f Finland), which has budgeted between $15 and $30 per capita, including technical assistance and program running costs, for the next phase o f its project. ESRDF has apparently been able to achieve lower per capita construction costs, especially for hand dug wells and springs (regional averages range from $4 to S l l ) , but n o data i s provided on the total cost per capita including technical assistance and program running costs.'41 7.36 Ethiopia's rural water supply investment cost per capita i s in line with other countries in Sub-Saharan Africa, but regional experience shows that greater efficiency i s possible if the volume o f business increases and procurement packages are designed to permit economies o f scale, while at the same time encouraging competition (e.g., batching in lots o f around ten boreholes in a particular area to reduce driller mobilization and supervision costs). 7.37 Efficiency in utilization of funds available to the sector for urban and rural water supply and sanitation should also be examined. Data o n utilization o f Government budget from two o f the larger regions shows that actual capital expenditures have been consistently below planned levels as shown below: Table 7.6: Actual vs. Planned Water Supply Spending (Current ETB millions) Region Planned Capital Actual Utilization Budget Expenditures Rate /a (urban & rural) Amhara EFY 1993 24.4 14.9 61% EFY 1994 32.3 12.5 39% Tigray EFY 1993 9.4 7.8 83% a/ Note: as ?I EFY 1 4 other sectors, some reporting on donor-funded projects. 6d2fapparent .. under-spen2di;g - i s probably . 40° due incomplete 7.38 ESRDF reports uneven performance across the regions, where some regions utilized their initial allocations quickly and others were unable to do so within the first 14' Source: ESRDF review report, February 2003. 152 five years. The IDA-funded WSDR Project was extended twice so that all works and consultancies could be completed; actual disbursements lagged behind schedule throughout the project. 142 Under utilization has also been noted as an issue in AfDB and EC funded projects. Possible reasons for under-utilization may include unrealistic plans and resource estimates, failure o f donors or the Treasury to deliver funding according to plan, flow o f funds problems, absorption capacity and cumbersome rules for accessing funds. 7.39 Adequacy o f funding - As noted above, over the ten year period from 1990 to 2000 water supply coverage reportedly increased by only 2 to 6 percent 143, which indicates that spending in the sector has only managed to keep pace with population growth and deterioration o f systems. Clearly more will have to be done if targets are to be met, especially for rural water supply and sanitation, where coverage i s the worst and government spending has been low. Increases in block grants to the woredas will be needed so that budget can be allocated for both capital and recurrent costs to support implementation o f woreda water supply and sanitation programs. Projected funding requirements associated with targets are discussed in detail in Section C. 7.40 Appropriateness o f funding - Regions and the federal government, with donor assistance, have spent more in recent years o n urban water supply than on rural water supply and sanitation. Government officials at all levels have indicated they recognize the need to shift focus to development o f water supply and sanitation in rural communities and smaller towns. This will require increasing emphasis o n capacity building to ensure sustainability. For rural water supply this would include capacity building at the woreda level to develop Woreda Water and Sanitation Programs, capacity building at the community level for management and maintenance o f systems, and building partnerships with the private sector to provide maintenance support, technical assistance and spare parts. For town water supply, capacity building will be needed for the local governments to take up their new responsibilities for planning, governance and monitoring and evaluation, and the autonomous Water Boards require strengthening in the areas o f management and operation to improve coverage and the levels o f service provided, while at the same time improving cost recovery and efficiency. Consistent with i t s policy o f full cost recovery for urban water supply, Government grant funding to established water utilities in the larger cities may be limited to capacity building for improved management and cost recovery and support with planning and preparation o f business plans to support financing for further development through loans. c. PROJECTIONS OF EXPENDITURE REQUIREMENTS 7.41 The Water Sector Development Program and the Master Plan provide estimated investment requirements for urban and rural water supply and sanitation in 5-year blocks, with associated coverage targets. For urban and rural water supply, estimates are given for investment required to meet coverage targets outlined in the PRSP for the first five 142 Complicated implementation and procurement arrangements have been cited as the key constraints to timely utilization o f available funds under the project. Lessons leamed include the need to decentralize and streamline implementation and procurement arrangements by adopting a more programmatic approach. 143 World Health Organization and United Nations Children’s Fund, Global Water Supply and Sanitation Assessment 2000 Report shows coverage increased by only 2 percent from 1990 to 2000, but M W R and the Regions report an overall increaseof 6 percent over the same time period. 153 years and beyond. The targeted coverage levels are ambitious, and the starting point i s higher than that reported by the WHO and used as the basis for the MDGs. The WSDP projections are presented here as the unconstrained or ambitious scenario. The cost o f reaching the MDG coverage targets i s then presented as the moderate scenario - although in the view o f the team even these targets are very ambitious, compared to past rates o f expansion and likely implementation capacity. A low scenario i s not provided for urban water supply in the Master Plan. What i s presented here i s a extrapolation o f the current low spending levels and a “best guess” o f what this would mean in terms o f coverage. Note that in all cases the water supply cost projections are drawn from calculations done as part o f other exercises. They were not prepared from first principles by the PER team, and do not reflect the same degree o f detail and rigor that underlie the expenditure projections in the other sectors. As such they should be regarded as rough order-of-magnitude estimates only. 7.42 For sanitation the data provided in both the WSDP and the Master Plan i s much less comprehensive. The M P presents only one set o f spending scenarios for rural sanitation and on-site urban sanitation, which in any event will be the responsibility o f households, not the Government. Projections are also provided for urban sewerage. N o projections are provided for spending on sanitation promotion and hygiene education, which would likely be included in health sector projections. Ambitious ‘ MDGs Plus’ Projection Scenario’44 7.43 The WSDP and Master Plan estimates investment needs to bring urban and rural water supply coverage up to 100 percent by 2025, assuming no financing, institutional or manpower constraints. Note that under this ambitious scenario, Ethiopia would exceed the MDG targets o f halving the proportion o f the population without sustainable access to safe drinking water supply by 2015. The targets in this scenario correspond to those underlying the water supply strategy laid out in the SDPRP. In the view o f the PER team these targets are probably excessively ambitious, given implementation capacity, financing requirements, and historical rates o f expansion. Technically, there i s a big difference from the other scenarios: note they involve much higher level-of-coverage targets; in part because the initial conditions assumed in the Master Plan projections - in terms o f current coverage - are much higher than those revealed by the WHO and WMS surveys (about 42% coverage in 2000, as opposed to 24% in the WHO data). 144 Note the scenarios in the water sector differ from those elsewhere in the PER, in that the ‘ambitious’ scenario, based o n the government’s Master Plan projections, already involves targets well in excess o f the MDG goals. Consequently the costing o f the M D G s i s presented as the ‘moderate’ scenario. 154 Table 7.7: Projected Investment Needs for Water Supply - ‘MDG Plus ’ Scenario 145 (Costs in constant 2002 ETB millions) Urban Rural Total Spending Coverage Spending Coverage Spending Coverage 2000 74% 23% 42% Up to 2005 2,311 88% 4,674 37% 6,985 48% 2006-2010 2,549 97% 6,085 52% 8,634 56% 2011-2015 1,224 98% 6,971 71% 8,195 78% Subtotal 6,084 17,730 23,814 US$700 US$2,IO0 US$2,800 2016-2020 2,900 99% 6,000 86% 8,900 89% 2021-2025 a 3,000 100% 6,000 100% 9,000 100% Total 11,984 30,730 42,714 US$I,400 US$3,600 US$&000 a/ The WSDP provides spending and coverage projections only through 2015; Master Plan projections “high scenario” are used for 2016 through 2025. Neither the WSDP nor the Master Plan provide estimates of spending and coverage after 20 15; the figures shown here assume spending at a level required to reach 100% coverage by 2025. 7.44 For urban sanitation, the Master Plan estimates that investments o f 14.8 billion Birr (US$1,741 million) will be needed by 2015 for on-site sanitation facilities (septic tanks and latrines) and an additional 690 million Birr (US$81 million) for desludging equipment to provide adequate sanitation under the high growth scenario. I t should be noted that much o f this investment should come from private households and the private sector, not the Government. However, Government resources would be needed to stimulate household demand through sanitation promotion and this would require additional budget for recurrent expenditure. 7.45 The Master Plan provides projections for expansion o f the Addis Ababa sewerage system at a cost o f 3 billion Birr, 1.2 billion by 20 15 and the remainder by 2025 (US$353 million). The targeted levels o f sanitation coverage for Addis are as follows: Table 7.8: Sanitation Coverage in Addis Ababa 2010 2020 Type o f service: Sewerage house connections 200,000 960,000 Pit latrines and septic tanks 2.5 million 2.7 million Total population with improved 2.7 million 3.7 million sanitation Coverage 79% 88% 7.46 The Master Plan also projects, in the ambitious scenario, construction o f sewerage facilities in the 36 towns that will have a population over 80,000 by 2025, at a cost o f 820 million Birr ( U S 9 6 million). Under this scenario 75 percent o f private households with water connections would be connected to a sewer and 80 percent o f non- domestic water users would dispose o f their wastewater through sewers. The Master Plan note that the construction cost o f sewerage i s roughly four times more expensive per m3o f water processed than water supply and that willingness to pay for sewerage i s very 145 Source: InterimMaster Plan; constant prices; all costs are in year 2002 price levels. 155 low, concluding that about 75 percent o f the investments in sewerage would need to come from grants. This i s a critical issue, since the provision o f piped water needs to be accompanied by a complementary ability to remove the sewage and waste water generated. 7.47 Rural sanitation: The Master Plan extrapolates the trend in autonomous growth o f rural sanitation facilities - 7.5 percent per year - to conclude that coverage would reach close to 24 percent by 2015, and 50 percent by 2025. The investment cost to private households to construct an estimated 9 million household latrines would be 7.2 billion Birr (2.9 billion by 2015 and an additional 4.3 million by 2025). N o projection i s provided for Government spending (recurrent) on rural sanitation, which would be needed for promotion o f sanitation and hygiene education. Moderate Projection Scenario - Spending required to meet MDG targets 7.48 Urban and rural water supply - The following table shows estimated spending to reach the MDG targets o f 64% coverage by 2015, and projects continued spending at the same level through 2025. Note that the starting point for coverage i s different for both urban and rural water supply. For these projections prepared for the mission, the starting point i s based on WHO reports, whereas the M W R has used figures provided by the regions, which are higher. (Note however that the Welfare Monitoring Survey data tend to support the lower estimates). As a result this scenario shows close to the same level o f spending to achieve lower coverage levels by 2015. Obviously, improved systems for monitoring and evaluation are needed in the sector so that the different coverage figures can be reconciled. Table 7.9: Projected Investment Needs for W a t e r Supply- Moderate ‘ MDG’ Scenario Period Urban Rural Total Spending Coverage Spending Coverage Spending Coverage 2000 146 77% 13% 24% Up to 2005 2,700 81% 4,200 28% 6,900 37% 2006-2010 2,700 85% 4,200 43% 6,900 50% 2011-2015 2.700 89% 4.200 57% 6.900 64% Total 7,100 12,600 19,700 US836 US$1,500 US$2,300 2016-2020 2,700 93% 4,200 72% 6,900 2021-2025 2.700 97% 4.200 87% 6.900 Total 12,500 21,000 33,500 US$1,471 US2.4 71 US$3,942 7.49 Urban sanitation: In the Master Plan “low growth” scenario, urban sewerage systems would be constructed in only ten towns (aside from Addis) at a cost o f 450 million Birr (US$53 million)- and then only at the end o f the planning period (after 20 15). The number o f connections would be small - only 6 percent o f the population would be connected to the sewer system. A lower projection i s not provided for expansion of the sewerage system in Addis. Neither i s a more moderate projection provided for household and private sector expenditures o n on-site sanitation facilities and desludging equipment. 14‘ Note that coverage levels in 2000 are different according to MDG tables than those reported by the Government in the PRSP and WSSDP. 156 7.50 For rural sanitation the Master Plan (MP) assumption that the trend in autonomous growth o f rural sanitation facilities will continued unabated seems unrealistic, especially if each latrine costs 800 Birr (US$94). I t seems more reasonable to assume slow growth in private latrine construction with little or n o support from the Government. T o construct only half as many latrines as projected in the ambitious scenario, spending by households would need to be 1.5 billion Bin by 2015 and an additional 2.1 billion Birr by 2025 (total o f US$424 million). Even to reach this level, Government will need to allocate recurrent funds for sanitation promotion, hygiene education and support for development o f the small scale private sector. Low scenario 7.51 The l o w scenario for urban water supply assumes federal and regional spending will continue at the EFY95 budgeted level throughout the projection period (half for rehabilitation and half for new systems in the first ten years and all for new systems thereafter), the result being that coverage would decrease in the first five years and then remain constant over the 25 year period, taking into consideration population growth. In addition, it assumes Addis Ababa will spend 100 million Birr per year (US$I2 million) to maintain its current level o f coverage - the average actual capital spending for the four year period from EFY89-92. For rural water supply, the l o w scenario according to the MP would require spending o f roughly 1.7 billion Birr up to 2015 and another 1.8 billion until 2025 (total o f US$412 million) to maintain and only slightly increase coverage. M u c h o f this would be for rehabilitation o f systems as they break down. Table 7.10: Projected Investment Needs for Water Supply - Low Scenario Water Supply Period Urban Rural Total Spending Coverage Spending Coverage Spending Coverage 2000 74% 14% 24% U p to 2005 850 71% 400 14% 1,250 24% 2006-2010 850 71% 600 14% 1,450 26% 2011-2015 - 850 71% 700 15% 1.550 27% Total 2,550 1,700 4,250 US$300 US$200 US$SOO 2016-2020 850 71% 800 1,650 29% 2021-2025 850 71% 1.ooo 1.850 31% Total 4,250 3,500 7,750 US$500 US$412 US$912 7.52 Urban sanitation: The l o w scenario for urban sanitation assumes n o spending on urban sanitation (consistent with recent years) under the assumption that i t i s highly unlikely sufficient grant financing will be found for the envisioned expansion o f the Addis Ababa sewerage system and construction o f even minimal sewerage systems in other cities. A l o w projection for household and private sector expenditures o n on-site sanitation facilities and desludging equipment i s not provided in the MP, which in any event would not be Government expenditure. Rural sanitation: The M P does not provide a l o w growth projection for private latrine construction by households in rural communities. 157 D. OBSERVATIONS ON POLICY IMPLICATIONS AND TRADE-OFFS, AND THE LINK WITH A WIDER DEVELOPMENT STRATEGY 7.53 Shwt in focus from large urban to smaller towns and rural communities: In recent years government has devoted more resources to improving urban water supply in the largest cities than to rural or small town water supply, the result being that coverage in rural communities and smaller towns i s much lower than in the larger urban centers. As noted above, there are justifiable reasons for having prioritized urban over rural in the past, but a shift in focus to rural and smaller towns, as the Government has indicated i s i t s intent, i s certainly warranted going forward. 7.54 Approach to financing improved water supply and sanitation: The larger cities, most o f which have a new or recently rehabilitated water supply systems, will be expected to pay for the upkeep and expansion o f those systems with internally generated funds. The Government plans to facilitate this process through the Water Resources 147 Development Fund with support from donors. Financial and economic analysis indicate that this will be feasible - that customers are willing and able to pay the required tariffs if service i s improved. What remains to be seen i s whether there i s the political “willingness to charge” higher tariffs. Some capacity building will be needed to help the newly formed autonomous water boards improve service levels, efficiency, financial performance, financial management and monitoring and evaluation systems so that they can prove themselves creditworthy entities. 7.55 Many smaller towns (most under 20,000 population) do not have a functioning water system. Grant financing will be required for the initial investment and capacity building to establish financially viable water utilities capable o f funding their own operations and maintenance, renewal and expansion going forward. Rapid growth i s expected in these towns and a functioning water supply system will help lay the groundwork for their contribution to the overall economic development o f the country. 7.56 Rural water supply: The greatest needs are in the rural areas, with 85 percent o f the population and only 14 percent water supply coverage. The documented high failure rate o f these systems points to the need to address the underlying causes, such as inadequate preparation o f communities to manage their own systems and lack o f spare parts and maintenance support. The Demand Responsive Approach should be followed for implementation o f all rural water supply systems to involve the communities in the planning stages so that they choose systems they are willing and able to maintain and to bring down the per capita investment cost. Woredas, who will be the ones responsible for financing and supporting rural water supply, will need assistance to develop their own water supply and sanitation programs and will need to work closely with NGOs to ensure that a consistent approach i s followed by all. 7.57 Urban sanitation - It i s unclear who i s expected to finance the expensive urban sewerage systems identified in the WSDP and Interim Master Plan. Cost recovery i s notoriously difficult as willingness to pay for piped sewerage and treatment i s usually l o w (even for the ongoing operation and maintenance costs). Given the l o w levels o f household income in Ethiopia, it i s unlikely customers will be able to afford the high 147 AfDB has already agreed to fund improvements to water supply and sanitation systems in Harar. 158 operating and maintenance costs o f extensive sewerage systems. Therefore, a multi- prong approach i s suggested: Planning for sanitation - including sewerage, where applicable, and drainage - should be carried out in conjunction with planning for improvements in water supply; this planning will normally be the responsibility o f municipalities. Sewerage systems should be developed to serve only the most densely populated sections and the business districts o f the large urban centers, where affordability i s usually higher and health issues dictate the need for piped sewerage. Urban customers with inside plumbing (currently a very small percentage o f water connections in most Ethiopian cities) should be required to be connected to the sewerage system where one exists or, in less densely populated areas, to have a septic tank (this i s the current policy in Ethiopia). Vacuum truck operators should be regulated to ensure proper disposal o f septage. In general, construction o f latrines, installation o f septic tanks and septage hauling would be carried out by the private sector and would be financed by households, not from the Government budget, but Government should support development o f the market for private sector artisans to construct latrines and for vacuum truck operators to haul septage. Government should also support sanitation and hygiene promotion, with the help o f NGOs. 7.58 For rural sanitation woredas will be expected to finance sanitation promotion and hygiene education carried out at the community level in conjunction with improvements to water supply systems, and can expect support from donors and NGOs. Support will also be needed to t r a h local artisans to construct latrines and to help them to get into business. Institutional arrangements will need to be addressed to ensure coordination o f the various governmental agencies involved in planning and promoting sanitation and hygiene. 7.59 Prioritization o f rural over urban: Despite the stated intention by the Government to withdraw over time from using Government resources to finance investments in urban water supply and to devote more funding to improvement o f rural water supply and sanitation, there i s a risk that the real politique may work against this shift in focus. Tariffs for urban water supply will have to increase, but political realities dictate that they must be phased up to full cost recovery levels, and corresponding programs will need to be carried out to increase understanding by customers and decision-makers o f the reasons for the increases. Tariff increases will need to be accompanied by improvements in service and efficiency and utilities will have to be more accountable to the public. At the same time, efforts will be needed t o raise awareness o f the benefits and need to prioritize rural water supply and sanitation among local government decision makers, who will need to budget resources for rural water supply and sanitation. The Ministry o f Water Resources and Regional Bureaus will have to play the role o f advocate for the policy objectives and achievement o f targets. 7.60 Capacity constraints in the sector need to be addressed as a matter o f priority. With decentralization this i s at the same time more critical and more difficult. There i s an increasing awareness that it will not be possible to employ sufficient staff in the public 159 sector and that, in fact, this i s not necessary or desirable. Other actors, including the private sector and NGOs should be brought in as partners in development - from local artisans, mechanics and suppliers to national and international consultants, contractors and operators. M u c h work i s needed to create an enabling environment and to attract private sector partners to the water sector in Ethiopia. 7.61 Monitoring and evaluation in a decentralized environment - As decentralization i s being introduced, it i s becoming increasingly difficult to monitor the level o f spending in the sector and to evaluate the efficiency and effectiveness o f spending. Data i s not collected and reported in a consistent manner from region to region and what does exist i s not passed up to the federal level. A concerted effort will be needed to implement effective monitoring and evaluation systems throughout the country in order to provide the data and feedback necessary to attract resources for sector development. 7.62 Water resources management - This chapter has focused on water supply and sanitation, but it should be noted that without accompanying measures for water resources management, water supply and sanitation improvements cannot be sustainable. In separate sector work, the World Bank and the Government will be reviewing the water sector in an integrated way, examining the linkages between water resources management, hydropower, irrigation and water supply and sanitation as they relate to poverty and economic development in the country. 7.63 Linkages with other social sector spending - Studies show strong linkages between water supply and sanitation and performance in other sectors, such as health and education. Studies consistently show that increasing access to clean water and to improved sanitation reduces the incidence o f disease and mortality. Table 7.10 i s from a review o f a large number o f studies that have sought to understand and estimate the impact o f improved water and sanitation facilities on water-related diseases (Esray, et a1 , 1991). In addition, when water points are installed, especially in rural communities, women and girls spend less time fetching water and are, therefore, able to engage in more productive activities and attend school. In many cultures, older girls often stop going to school when sanitation facilities are inadequate. However to measure these linkages requires tightly-controlled surveys to isolate the effects o f improved water-supply from other factors. Unfortunately, quantitative studies linking spending on water supply and sanitation to improvements in health and education have not been conducted in Ethiopia; although a very preliminary modeling exercise, based on the 2000 DHS data, suggests that full rural access to safe water could reduce under-5 mortality by 9.5 deaths per 1,000; with the benefits being greatest for households in the poorest consumption quintiles (13.1 per 1,000)'48 14' See Wang (2003) Environmental Determinants o f Child Mortality - Empirical Results fiom the 2000 Ethiopia DHS. 160 Table 7.11: Expected Reduction in morbidity/mortality from improved water and sanitation Disease All studies Rigorous Studies Number o f Median Number o f Median reduction studies reduction (%) studies (%) Ascariasis 11 28 4 29 Diarrhoeal disease 49 22 19 26 Morbidity Mortality 3 65 Dracunculiasis 7 76 2 78 Hookworm infection 9 4 1 4 Schistosomiasis 4 73 3 77 Trachoma 13 50 7 27 Child mortality 9 60 6 55 161