Report No. 54952-ET ETHIOPIA Public Finance Review 2010 Draft August Poverty Reduction and Economic Management Document of the World Bank Africa Region CURRENCY EQUIVALENTS (Exchange Rate Effective) Currency unit = Birr US$1 = 13.65 ETB (End June, 2010) FISCAL YEAR (FY) July 8 – July 7 WEIGHTS AND MEASURES Metric System ABBREVIATIONS, ACRONYMS, and INITIALISMS AA – Addis Ababa APR – Annual Progress Report BOFED – Bureau of Finance and Economic Development BPR – Business Process Re-engineering DPT - Diphtheria, Pertussis and Tetanus FBG – Federal Block Grant FBG – Federal Block Grant FDH – Free Disposable Hull FDRE – Federal Democratic Republic Ethiopia FSF – Food Security Fund FSF – Food Security Fund FY – Fiscal Year GDP – Gross Domestic Product GER – Gross Enrolment Rates GOE – Government of Ethiopia GQUEP – General Education Quality Enhancement Programme HIPC – Heavily Indebted Poor Countries HP – Health Posts IMF – International Monetary Fund LIG – Local Investment Grant M&E – Monitoring and Evaluation MDRI – Multilateral Debt Relief Initiative MEFF – Macroeconomic Fiscal Framework MOFED – Ministry of Finance and Economic Development MOH – Ministry of Health MTEF – Medium-term Expenditure Framework NGOs – Non-Governmental Organizations OLSQ – Ordinary Least Square PASDEP – Plan for Accelerated and Sustained Development to End Poverty PBS – Protection of Basic Services PFM – Public Finance Management PFR – Public Finance Review PSCAP – Public Sector Capacity Building Programme PSNP – Productive Safety Net Program SDPRP – Sustainable Development for Poverty Reduction Programme SNG – Sub-national Governments SNNPR – Southern Nations Nationalities and Peoples Region SPG – Specific Purpose Grants TOT – Terms of Trade WCBS – Woreda and City Benchmarking Survey WDR – World Development Report WOFED – Woreda Office of Finance and Economic Development YoY – Year to Year TABLE OF CONTENTS EXECUTIVE SUMMARY....................................................................................................... VII A. Aggregate Fiscal Update .................................................................................................................................. vii B. Decentralized Service Delivery the Ethiopian Experience............................................................................. ix C. Conclusion........................................................................................................................................................ xiv 1. AGGREGATE FISCAL MANAGEMENT......................................................................... 1 1.1 Overview ......................................................................................................................................................... 1 1.2 Macroeconomic Context................................................................................................................................ 1 1.3 Medium Term Trend in Fiscal Aggregates.................................................................................................. 3 1.4 Revenue Mobilization under Inflationary Condition ................................................................................. 6 1.5 Aggregate Expenditure Management .......................................................................................................... 7 1.6 The FY 2010 Budget ...................................................................................................................................... 9 2. DECENTRALIZED SERVICE DELIVERY, THE INSTITUTIONAL CONTEXT ... 13 2.1 Introduction.................................................................................................................................................. 13 2.2 The Legal Framework ................................................................................................................................. 13 2.3 Administrative Structure ............................................................................................................................ 15 2.4 Functional Assignment ................................................................................................................................ 16 2.5 Expenditure Assignment ............................................................................................................................. 17 2.6 Revenue Assignment .................................................................................................................................... 18 2.7 Intergovernmental Fiscal Transfer ............................................................................................................ 19 Country Experience ................................................................................................................................................ 19 The Fiscal Gaps ...................................................................................................................................................... 21 The Nature and Types of Fiscal Transfers.............................................................................................................. 22 Federal Transfer to Regional States........................................................................................................................ 23 Regional Transfer of Block Grant to Woredas ....................................................................................................... 24 2.8 Interfaces between Institution Building and Devolution of Expenditure Assignment........................... 25 3. TREND IN REGIONAL SPENDING AND FINANCING FRAMEWORK ................. 27 3.1 Overview ....................................................................................................................................................... 27 3.2 Aggregate Fiscal Picture.............................................................................................................................. 27 3.3 Regional Revenue Mobilization .................................................................................................................. 30 3.4 The Pattern of Regional Expenditure ........................................................................................................ 32 3.5 Implication to Medium Term Regional Fiscal Health .............................................................................. 37 3.6 Major Achievements and Challenges of Decentralized Service Delivery................................................ 38 a) Local revenue base and entitlement ............................................................................................................... 40 b) Low local revenue and transfer for public investments ................................................................................. 40 c) Deteriorating in quality of services ................................................................................................................ 41 d) Weak decentralized capacity for local revenue mobilization......................................................................... 41 e) Local government expenditure responsibility ................................................................................................ 41 f) Borrowing ...................................................................................................................................................... 41 g) Community funding ....................................................................................................................................... 43 h) Transfer formula ............................................................................................................................................ 43 4. STATE OF PUBLIC FINANCE IN REGIONAL STATES: A CASE STUDY OF AMAHARA AND BENSHANGUL- GUMIZ REGIONS ....................................................... 44 4.1 Amhara Regional State................................................................................................................................ 44 i. Socioeconomic Profile ................................................................................................................................... 44 ii. The Institutional Context ............................................................................................................................... 44 iii. Regional Resource Envelop ........................................................................................................................... 45 iv. Regional Spending ......................................................................................................................................... 45 v. Local revenue Mobilization ........................................................................................................................... 47 4.2 BENSHANGUL-GUMIZ REGION .......................................................................................................... 49 i. Socioeconomic Profile ................................................................................................................................... 49 ii. The Institutional Context ............................................................................................................................... 49 iii. Regional Resource Envelop ........................................................................................................................... 50 iv. Regional Spending ......................................................................................................................................... 50 v. Local Revenue Effort..................................................................................................................................... 52 vi. Lesson from the Regional Case Studies......................................................................................................... 53 5. EXPLAINING VARIATION IN BASIC SERVICE DELIVERY ACROSS WOREDAS 55 5.1 Objective and Context ................................................................................................................................. 55 5.2 Measuring Public Sector Performance: Literature Review and Cross Country Experience................ 57 5.3 Bi-Variate Analysis of Variation in Woreda Service Delivery................................................................. 58 i. Woreda Specific Structural Factors ............................................................................................................... 58 ii. Public Spending and Service Delivery........................................................................................................... 62 a) Level of spending........................................................................................................................................... 62 b) Efficiency of spending on basic education..................................................................................................... 63 iii. Composition of Spending and Local Revenue Generation ............................................................................ 65 iv. Local revenue effort....................................................................................................................................... 66 v. Institutional Quality and Capability............................................................................................................... 67 5.4 RESULT from the Multi-Variate Regression............................................................................................ 70 a) Enrollment and Structural Factors ................................................................................................................. 71 b) Enrollment and Level of Spending ................................................................................................................ 72 c) Enrollment and Fiscal Policy ......................................................................................................................... 72 d) Enrollment and Institutional Factors.............................................................................................................. 73 5.5 Summary implications................................................................................................................................. 75 REFERENCES ............................................................................................................................ 76 Table of Figures Figure 1.1: Domestic Revenue, Expenditure and Fiscal Deficit ...................................................................4 Figure 1.2: Budget execution: Revenue, Expenditure and Fiscal Deficit .....................................................5 Figure 1.3: Evolution of Domestic Revenue: Nominal vs and Real .............................................................6 Figure 1.4: Evolution of Expenditure: Nominal vs. Real..............................................................................7 Figure 1.5: Functional Composition of Expenditure.....................................................................................7 Figure 1.6: Composition of Expenditure by Broad Function ........................................................................8 Figure 1.7: Share of Expenditure on Pro-Poor Sectors in Total Spending....................................................9 Figure 1.8: An Analysis of Revenue for the First half of FY10..................................................................10 Figure 1.9: General Government Expenditure in the first six months compared to Budget in FY 08, 09 and FY10..............................................................................................................................................................1 Figure 1.10: External Finance in the Budget: Commitment & Disbursement in the first half of the FY....12 Figure 2.1: Federal Structure of the Ethiopian Government ......................................................................15 Figure 2.2: Distribution of General Government Expenditure by Jurisdiction (FY 06-08 average, in %) ...1 Figure 2.3: Contribution of Administrations to the General Government Revenue (FY06-08, average in %) ..................................................................................................................................................................1 Figure 2.4: Fiscal Gap by Region in EFY 2001 ..........................................................................................22 Figure 2.5: Per capita Budget and Institutional Quality Indices .................................................................26 Figure 3.1: Share of Regional Financing Gap in General Government Deficit ..........................................28 Figure 3.2: Regional Revenue and Expenditure Performance ....................................................................28 Figure 3.3: Growth in Domestic Regional Revenue ...................................................................................29 Figure 3.4: Fiscal Transfer to Finance Decentralized Service Delivery......................................................30 Figure 3.5: Growth in Components of Regional Revenue ..........................................................................32 Figure 3.6: Growth in Regional Expenditure ..............................................................................................33 Figure 3.7: Share of Regional Expenditure in Key Sectors.........................................................................34 Figure 3.8: Share Regional Pro-poor Spending from the General Government ........................................36 Figure 3.9: Projections of the Resource Envelope and Fiscal Expenditure for General Government ........38 Figure 4.1: Fiscal Frame Work of Amhara Region in FY 09 in % of Total..................................................1 Figure 4.2: Pro-Poor Sectors Spending in Total Budget (in %): Amhara vs. All Regions .........................46 Figure 4.3: Per capita tax revenue in Amhara and country level ................................................................47 Figure 4.4: Woreda share in total regional spending in Benshangul Gumuz Region..................................50 Figure 5.1: Basic Educations and Health Coverage by the Degree of Urbanization...................................59 Figure 5.2: Comparison of Woreda GER (grade 1-8) by Remoteness........................................................60 Figure 5.3: Primary Education and health Service Coverage by Status of Food Security ..........................61 Figure 5.4: Primary School Enrollment and Per Capita Spending on Education........................................62 Figure 5.5: Health service coverage vs per capita health spending.............................................................63 Figure 5.6: Average GRE Under Different Budget Mix in %.....................................................................66 Figure 5.7: Average Primary GER Under Strong, Average and Weak Local Revenue Environment in %67 Figure 5.8: Average Primary GER v/s. PFM system ..................................................................................68 Figure 5.9: Average Primary Enrollment Ratio for Strong, Medium and Weak Woredas (in percent) ....68 Figure 5.10: Average Primary GER by the Degree of Devolution of Basic Services to Woredas .............70 Figure 5.11: Average Primary GRE by the Degree of Community Empowerment....................................70 List of Tables Table 1.1:Selected Aggregate Economic and Financial Indicators...............................................................3 Table 2.1: Functional Division between the Federal and Regional Governments ......................................17 Table 3.1: Regional Tax Base .....................................................................................................................30 Table 3.2: Regional Expenditure Share from the General Government in percent.....................................33 Box 4.1: Rural Land and Agriculture Income Tax in Amhara...................................................................48 Table 5.1: Result of FDH Analysis of Public Spending Efficiency on Education......................................64 Table 5.2: Results of the Regressions of Primary GER on Structural, Level of Spending, Policy, and Institutional Variables .................................................................................................................................74 List of Boxes Box 2.1: Legal Framework for Woreda Development Assignment..............................................................1 Box 2.2: Sub-National Budgeting Process ...................................................................................................1 EXECUTIVE SUMMARY 1. Ethiopia Public Finance Review (PFR) is an analytical input to the regular Government- Donor dialogue on public finance and aid effectiveness. It forms the basis for policy focused analyses and sustained dialogue that facilitate close partnership and enhanced mutual accountability between Government and the development partners. The PFR is also an important instrument for determining the level and quality of aid and it fulfills a due diligence requirement for operations like the Protection of Basic Services. 2. During the past few years there has been an evolution in the way Ethiopia PFRs are put together and in their contents. In terms of the process, the approach has been away from a Bank driven, big budget, multi-year review to a rolling process of information exchange/dialogue, analysis, and follow-up that is synchronized with client policy making process and budget calendar. In terms of contents, the approach has evolved from being a standardized review and comprehensive discussion of Bank selected topics to a short, timely and flexible document with a focus on contemporaneous topics of interest to the client. 3. The 2010 Public Finance Review (PFR) analyzes the effectiveness of decentralized service delivery in Ethiopia. While the Government of Ethiopia (GoE) has pursued decentralization for the last two decades, neither a comprehensive documentation of the functioning of the system nor a thorough evaluation of its effectiveness is available. This study attempts to fill this gap by linking policy and institutional variables related to decentralization with service delivery outcomes at the sub-national level. Taking advantage of the information generated through the Woreda and City Benchmarking Survey (WCBS), the study attempts to explain the cross woreda variation in service delivery through selected inputs including the level of expenditure and the quality and capability of local institutions. As part of the regular update, recent developments in aggregate macro/fiscal situation are included in the first part of the report. Summary of the major findings of the report are highlighted below. A. Aggregate Fiscal Update 4. In the aftermath of the recent global economic crisis, Ethiopia has made considerable progress towards restoring macroeconomic stability without compromising the pace of its economic growth. While maintaining the momentum for rapid and broad-based growth (real GDP growth has averaged 11 percent per annum in the past 6 years), year-on-year (YOY) inflation has gone down from 55.3 percent in FY08 to less than 3.0 percent in FY10. Though in recent months the inflation rate has risen again, it remains at single digit. At the same time, foreign exchange reserves have recovered from a low of 1.2 months of import cover in October 2008 to 2.1months by April 2010. Following a series of depreciation of the nominal exchange rate, the overvaluation problem has been alleviated. This has restored the prospect of improving the competitiveness of the economy. Exports have recovered from last year’s shock, growing at a faster pace than imports in FY10 (exports and imports are expected to grow at 38 percent and 12 percent respectively in FY10) based of IMF estimate, However, government reported growth of exports to reach to 38%. Nevertheless, narrowing of external imbalance remains a challenge reflecting the structural nature of the problem. 5. While the short-run adjustment process has relied largely on aggregate fiscal discipline, complementary monetary and exchange rate policies have been used to curb excessive demand and to restore external competitiveness. Between FY08 and FY09, fiscal deficit (including grants) to GDP ratio was lowered from 3 percent to 1 percent and domestic financing of the deficit was brought down to zero (from 2.7 percent of the GDP in FY08). The single digit inflation rate was achieved through a combination of tight fiscal management and slower growth of high powered money, though the reduction in global commodity prices played an important role. In the short-run, the burden of fiscal adjustment has relied mostly on expenditure control rather than on improving revenue effort. Between FY08 and FY09, aggregate spending to GDP ratio has dropped from 19 to16 percent. At the same time, revenue shrank from 12 to 11 percent of GDP. 6. Compared to the preceding years, the FY10 budget is less restrictive and based on realistic projections of revenue and spending. The realism of the FY10 budget is associated with the use of FY09 pre-actual data rather than the FY09 budget estimates. In FY10, the level of fiscal deficit (including grant) and domestic bank financing is expected to increase to 0.8 percent of GDP compared to the corresponding number of 0.2 percent in FY09. The plan also envisages greater reliance on domestic revenue than on grant but has limited space to react to adverse shocks. Thus, the development partners and GoE need to work together to sustain the recent progress in the predictability of aid and continue to protect the spending on basic services. This will require GoE to shield the share of total expenditure allocated to Federal Block Grant (FBG) from other competing expenditures. 7. At General Government level, the composition of spending continues to be pro- investment and pro-poor. In the past three years, spending on capital budget has grown by about 30 percent on average in nominal terms, twice the rate of growth of recurrent spending. Between FY03 and FY09, the share of pro-poor sectors spending in total spending increased from 52 to 64 percent. The increased emphasis on pro-poor sectors continued in FY10 as GoE generated more fiscal space from the decline in real spending on debt service and defense. Urban development, which used to compete with the traditional pro-poor sectors, got less allocation in FY10 in favor of natural resources and human development. 8. The medium-term economic prospect looks positive, with real GDP growth of around 7.5 percent, single digit inflation, and continued strong fiscal performance.1 Evidence on the performance of the FY10 budget so far has been encouraging with revenue surpassing budget, and spending remaining within target. Particularly, the strong revenue mobilization in the first half of FY10 has created greater space for federal spending. While revenue out-turn has been strong across the board, growth in foreign trade tax collection and the local currency equivalent of external grants and loans have been particularly robust. 9. Improved macroeconomic environment has helped decentralized service delivery effort. Sustained high inflation rate and exchange rate overvaluation in the past had lowered real revenue, and increased the cost of construction and provision of basic services. More importantly, in the past three years, inflation has undermined the real resources available for expanding and providing service delivery (see PFR 2009). The gain from improved macroeconomic environment on the fiscal side has already started to show results. According to the mid-term review (first six months of FY10) there has been an increase in domestic revenue 1 The 7.5% projection is that of the IMF. According GoE, the estimate for FY 10 and the projection for FY 11 is 10 percent. and grant from foreign trade taxes and exchange rate revaluation of budgeted external grant (nearly Birr 700 million from PBS alone in nominal terms). B. Decentralized Service Delivery: the Ethiopian Experience The Institutional Context 10. The Ethiopian decentralization initiative is wide in its scope involving legal, regulatory, administrative, civil service, and public finance management reforms. The Federal system of administration is at the heart of the Ethiopian constitution. Swift administrative restructuring, resumption of regional councils assuming constitutionally granted power, and quick staffing of newly established regional offices and devolution of expenditure assignment help sets up the system on a firm ground. The civil service and public finance reforms initiated around the mid- 90s, have gradually rolled out to regions and woredas with the objective of building an effective and efficient public service system. 11. Strong legal and administrative reform has contributed to the stability and predictability of the federal fiscal transfer system. Nevertheless most local administrations lack the capacity of developing strategic plan. The Ethiopian Federal system started to come into effect following the overthrow of the Leninist-Marxist regime of the Derg in 1991. The legal basis and the institutional framework for decentralization at the Federal level is the Federal Constitution. This framework is further elaborated in the Regional Constitution which launched the second wave of decentralization from Regional states (second-tier) to Woreda level (third-tier). In the absence of long-term local socio-economic strategy, unavailability of formalized planning methodology, and lack of clarity regarding responsibility of planning process forced sub-national administrations to go through incremental annual planning. Generally, sub-national governments’ development initiative is based on sectoral and Federal government directives and policy targets. 12. The functional responsibility across jurisdiction depicts the principle of comparative and scale advantage in managing public resource. The Federal and Regional constitutions and subsequent proclamations clarify the functional, expenditure and revenue assignments at each level of government (Federal, Regional and Woreda level of governments). The regional functions that still remain with the Federal government include investment promotion and land management for large scale investment. Generally, functions international and macro in nature and services benefiting more than one region have been retained at Federal level. 13. The expenditure assignment to sub-national governments (SNGs) is consistent with the functional responsibility of the regions, while the revenue sharing arrangement favors the Federal government, resulting in huge vertical fiscal imbalance2. In FY08, about 57 percent of the public expenditure of the country was incurred by the Federal government while 80 percent of the revenue collection was made at the same level of government. On the other hand, about 43 percent of the public expenditure and only 20 percent of the revenue collection was made by the sub-national governments (SNGs). As a result of this, there is a large vertical fiscal imbalance which is filled through inter-governmental fiscal transfers. 2 One of the article of the constitution states that promoting equitable development across regions is the mandate on the Federal Government. 14. The intergovernmental fiscal transfer system has become more complex, with increasing number of instruments, raising the transaction cost. The Federal Government currently uses one block grant, and at least four Specific Purpose Grants (SPGs) to transfer resources to regions. The degree of expenditure discretion to regions vary across instruments ranging from full control in allocation in the case of FBG to partial control or delegated spending of Federally budgeted progrmmes (e.g., Local Investment Grant-LIG, Productive Safety Net Programme - PSNP). With narrowing of development gap across jurisdictions, the objective of the block grants across regions has evolved from being more concerned about equity to being more focused on efficiency. While not fully implemented, the potential advantages of the current formula over the earlier ones are that the former avoids using subjective variables and weighting, considers recurrent and capital expenditure separately, uses proxy for revenue raising capacity, includes major fairness criteria and minimizes unnecessary lobbying by cub-national governments. 15. While the implementation of supply side reforms has been on track, effort to build local capacity and the introduction of local governance package has not progressed as planned. Woredas particularly in remote areas are facing critical capacity gap and are unable to attract skilled professionals. At the same time, delay in implementation of ‘Woreda Governance Framework’ has limited the scope for community empowerment except the recent effort in budget transparency. Trend in Sub-National Revenue and Expenditure 16. Ensuring predictable spending at the regional level remains a top priority of the Government, as evident from the predictability of fiscal transfer to regions. In the past five years, the rate of growth of regional budget is close to that of the general government budget. Nevertheless, regional budget has been protected at times of expenditure contraction resulting in a stable 60:40 share between the Federal and Regional spending. 17. In spite of rapid growth of regional budget in nominal terms, the continued imbalance between wage and operating budget and between recurrent and capital budget poses challenges to reaching the vulnerable groups and addressing service quality. In FY09, Regions managed 62 percent and 28 percent of the general government recurrent and capital budget, respectively. With Federal delegated programmes financed through the SPGs, regionally executed budgetary resource will be close to 50 percent of total. With notable progress in service delivery, regions now need to focus more on quality improvement and targeting the vulnerable groups. So far, wages and salary absorb about three-fourth of the regional recurrent budget and more than ninety percent of the woreda budget. Compounded by limited resources for expanding and or maintaining existing facilities and low local revenue effort, regions have to revisit the current pattern of resource allocation. Moreover, the recent trend in expenditure centralization favoring the Federal along with growing competition for resources from non-pro-poor sectors at woreda level further exacerbates the problem. Between FY 09 and FY 10, the proportion of resources managed by regions declined from 44 to 42 percent. 18. Growth of regional revenue in the past has been sluggish and any improvement in the recent years is attributed to higher income tax collection from the growing cadre of regional civil servants. Addis Ababa (AA) stood unique among regions. All regions excluding AA collect only 10 percent of the overall revenue while AA alone collect about the same size of revenue as the rest of the SNGs. While Regions (excluding AA) rely more on tax, particularly income tax, AA depends on the less predictable non-tax revenue. In the past five years, the Regions’ tax revenue from business declined for four consecutive years until it turned positive in FY09. The performance of agriculture income tax has also been disappointing, declining by about 20 percent per annum between FY03-FY06. In spite of a modest nominal improvement since FY 07, the contribution of agricultural income tax to overall tax revenue remains less than 2 percent. 19. The period of decentralization has been associated with significant improvement in rapid and equitable coverage of basic services across the country. For example, Gross Enrolment Rates (GER) in Primary Schools (first cycle (1-4)) for girls increased 16 percent in FY91 to 85 percent in FY07. According to MoFED, access to clean water supply at national level increased from 19.1 percent in FY96 to 52.5 percent in FY07. During the same period, access to health service has increased from 33 percent to 89 percent. 20. Evidence from the regional case studies confirm that regional spending more than doubled between FY05 and FY09, albeit at different pace – reflecting the distributive effective of FBG to narrow the development gap across regions. In the past five years, aggregate regional spending increased from Birr 9 Billion to Birr 25 Billion. At the same time FBG to region, the major source of financing increased from Birr 7 to 16 Billion. Until now, the FBG distribution system has been driven by one of its objectives, equalizing development opportunities across regions, resulting in higher per capita allocation to lagging and war affected regions. As the development gap (in terms of public services) narrowing, the per capita FBG allocation and hence regional spending gap is narrowing gradually. For example, with per capita spending of 325 Birr , Benshngul spent three times higher than that of Amhara region five years ago and now the gap narrowed to only twice. Per capita spending gap across regions however is not expected to close in light of fixed administration cost which is common for all and the relatively higher cost in delivering services in distant and pastoral regions. 21. While the strategy followed in devolving expenditure assignment to woredas is similar across regions, the actual devolution varies reflecting uneven capacity at the woreda level. As of FY09, the most devolved region is Amhara with 70 percent of the regional budget devolved to Woredas, followed by Oromia and BG with Woreda budget share of 58 and 52 percent respectively. In terms of sectoral spending, the experience across woredas is uniform – with all of “basic services sectors” managed by woredas. Instead, variation exists along recurrent-capital line. In lagging regions and woredas of weak capacity, capital projects and major procurement are managed at regional level. 22. The regional case studies confirm increasing competition for resources from the non- pro- poor sectors as well as critical shortfall in capital and operating budget at woreda level. Between FY05 and FY09, pro-poor sector spending at Woreda level in Amhara dropped from 77 percent to 65 percent, in Oromia from 77 to 72 percent. In Benshangul, the proportion of pro-poor sector spending varies between 54 and 69 percent in past five years. Similarly, the proportion of capital in the woreda budget has been less than 10%, albeit a slight improvement has occurred following the introduction of Local Investment Grant (LIG) in recent years. 23. Budget and spending information at sub-national level is not comprehensive, with large and growing off-budget financing in recent years. Comparing estimated regional resource envelop and what is reported through the IBEX system in Amhara, off-budget resources have grown in magnitude. Between FY 05 and FY 09 the proportion off-budget to total regional resource increased from 10 to 29 percent. With the community and NGOs providing resources, the off-budget problem is more serious at woreda level. As per the woreda specific case study in Amhara, the proportion of off-budget financing ranges from 5 to 43 percent in selected woredas3. 24. Absence of a systematic approach to estimate and record community contribution undermines the efficiency as well as the transparency and accountability of local service delivery. Evidence from selected sample woredas confirms that the community is involved in the construction and equipping of primary and secondary schools, health posts, farmers training centers. As per the Amhara case study, community contribution could reach one-half of the woreda revenue collected in the form of tax. Given the significance of community contribution, recording the contribution and studying its link to the tax burden of the community is a priority. 25. The ability of Regions to finance expenditure from their own-revenue not only remains below 20 percent, but their revenue base continues to shrink. In per capita terms, there is wide variation in tax collection across regions– from Birr 61 in BG to 34 in Amhara region. The concern at sub-national level is that despite rapid economic growth, the tax base appears to have narrowed. In Amhara, the contribution of business profit tax to own revenue went down from 21 to 17 percent between FY05 and FY09. In BG, share of business profit tax in total own-region revenue went down from 11 to 5 percent and the contribution of agriculture income tax from 22 to 11 percent between FY05 and FY09. The only growing revenue in recent years has been payroll tax contributing for more than 50 percent of regional revenue. The regional case studies show that tax system on agriculture income, which is the main stay of growth and livelihood for most of the region’s economy, has been highly ineffective in raising resources. The experience from Amhara shows that the tax on agriculture income does not only reflect the growth of the sector. It also perhaps does not meet the fairness criteria, with a flat rate imposed across farmers irrespective of the fertility of the soil and the cycle of cropping per year. 3 See the Amhara regional report for detail findings on the sample woredas. Cross Woreda Analysis of Determinants of Variation Service Delivery 26. This report makes an attempt to explain the financial, policy and institutional determinants of service delivery in the education sector, based on the data generated through the Second Woreda and City Benchmarking Survey (WCBS). The report reviews the pattern of service delivery and institutional quality across the sample woredas and undertakes simple bi- variate regression analysis between access to basic education against potential determinants including, financial, policy, institutions, and structural factors4. Finally, using multiple regression technique, it makes an attempt to quantify the statistical significance of factors in determining woreda service delivery. The quantitative results are not straightforward to interpret given the uneven quality of and discrete nature of data in WCBS survey. The regression results show a strong and statistically significant relationship between enrollment and per capita expenditure, geographical location, composition of spending, and fiduciary quality. However, institutional capacity and empowerment indicators turn out to have positive relationship but statistically insignificant in part related to the problem in collecting, processing and ranking of institutional indicators in WCBS. 27. The rolling out of institutional reform focuses mostly on the woredas. Institutional quality and capability composite index of the WCBS shows notable progress across the 292 sample woredas with two-third of them scoring average to above average. However, progress in individual institutional quality scores showed wide variation across woredas. Overall, greater number of the jurisdictions have scored relatively well in public finance management, and in putting a place a practice of giving access to information to citizens. In areas like revenue generation and balanced allocation of public expenditure, majority of the jurisdiction, including the top performers in the compounded indicator score very low. 28. Bi-variate analysis shows a positive correlation between institutional indices and per capita expenditure by woredas. Thus, woredas who register a higher level of institutional quality also tend to spend more per capita. Improved institutional strength is expected to improve the efficiency of the allocation of resources. However, in Ethiopia, progress on institutional performance is limited. For example, there is a general lack of community participation in planning. Second, some of the fiscal pre-conditions for efficient service delivery, including local revenue generation, remain weak for most woredas. 29. According to the WCBS, the distribution of service delivery across urban and rural woredas is narrowing. Nevertheless, cross regional variation remains important. In particular woredas in pastoralist areas are lagging behind in expanding primary health and primary education services compared to woredas in the highlands. 30. On other hand, evidence shows that woredas which are food insecure and benefiting from SPGs have reasonably good service coverage. Among other factors, this can be explained by the possibility of positive externality from SPG programs. Specifically, these programs through food for work and co-financing of service facilities are contributing to the expansion of basic services. Further study might be needed to understand whether these improvements are happening through demand side interventions or are pure supply effect. 4 Attempt is made to analysis of the relative efficiency in public spending using Free Disposable Hull FDH method for single input (spending) and output (education service coverage) in one region. 31. Preliminary findings show larger variation in the relative efficiency in service delivery than access across woredas. The pilot case of SNNPR further indicates that if all woredas reach the level of efficiency of those woredas on the frontier, close to 15 percent of the current spending could be saved to deliver the ongoing level of service coverage or equally expand converge without additional resources. However, caution should be taken in interpreting the result recognizing the presence of other factors other than money as well as inherent methodological complications associated with frontier analysis. 32. Local revenue generation and improved intra-sectoral allocation matters in expanding the coverage and improving the quality of services. However, more than 90 percent of woredas suffer from poor local revenue mobilization and use most of their resources to pay wages. 33. Both the bi-variate and multi-variate analysis confirm that fiduciary quality is important to improve woreda delivered services. The progress in rolling out reforms to enhance local capacity building needs to be sustained for effective resource utilization. The level and quality of public service delivery is positively influenced by institutional capability. Simple scatter plot graph between primary enrolment and woreda capacity index shows positive relationship, i.e., woredas with higher capability tend to have high GER and vice versa. C. Conclusion 34. The next development challenge of Ethiopia is sustaining the recent pace of growth and service delivery among which requires continued effort to build human and infrastructure capital. For Ethiopia, progress on the MDGs beyond what has been achieved requires not only more resources but also a better functioning delivery system and capability. The next goal in service coverage will be reaching out disadvantaged areas and groups, and improving the quality and return on public service infrastructure. While progress generally is commendable across the MDGs, sector specific constraints and challenges remain to be addressed. Based on the assessment, the following are proposed as important policy areas for enhanced decentralized service delivery. 35. Decentralized service delivery benefits from stable macro and aggregate fiscal environment. Lowering inflation improves the real return on spending while prudent aggregate management improves the availability and predictability of resources for service delivery. Sustained growth and strengthened tax administration create more fiscal space but also incentive for regions to link spending and financing responsibility. 36. Clearly, both recurrent and capital are needed to ensure expanding and improve quality of services. Lesson from the WCBS shows better allocation can improve coverage and narrow gender and social gaps. Though it is beyond this paper, the quality and sustainability of service delivery is very much linked to the quality of public expenditure allocation. This in turn requires more resource (including SPG, e.g LIG, GQUIP), more institutional quality and capability to enhance community participation. In this regard, local revenue not only creates more fiscal space but also empowers local administrations which in turn could improve the allocation of resources and sustainability of service delivery. At local level, candidates are agriculture income tax and enhancing the tax administration system. 37. Progress and challenges in service delivery across regions could in part be fixed through better planning and budgeting including a) Regions and woredas to follow more realistic budgeting, b) Improve the comprehensiveness of the their plan by taking into account resources from all sources, c) Strengthen planning processes, notably among the sectors rural and natural resource development, d) Redress the challenge of local revenue effort. AGGREGATE FISCAL MANAGEMENT Overview 1. This chapter updates the fiscal performance of the general government. General government is composed of federal government, nine regional governments, two city administrations, more than seven hundred woreda administrations and nearly fifteen hundred kebele administrations (see Chapter 2 for a discussion on administrative structure and functional responsibilities of the four tiers of government). The first part of the chapter focuses on the fiscal performance between FY03 and FY09. The second part of the chapter examines the performance of the first-six months of the FY10 budget.5 2. Some of the key findings of this chapter are as follows: • The GoE has managed to successfully reduce its fiscal deficit by containing expenditure; • Performance of revenue been weak even though tax revenue has been rising with the composition of direct tax revenue to total tax revenue remaining fairly static; • Real revenue and expenditure has remained relatively static from FY03 to FY09. This implies declining per capita real revenue and expenditure; • Emphasis on capital expenditure directed at improvements in infrastructure; • The pro-poor sectors have seen their share of general government expenditure rise; and • There has been a shift in expenditure away from education towards agriculture and natural resource, health and roads. Macroeconomic Context 3. From a background of unprecedented macroeconomic imbalance, Ethiopia has made notable progress since the adoption of adjustment programme in February 2009. The three years to FY09, the performance of the Ethiopian economy was characterized by fast growth accompanied by accelerating inflation and deteriorating balance of payments situation. While unfavorable Terms of Trade (TOT) contributed to the problem, the macroeconomic adjustment programme adopted on February 2009 primary focused on aggregate demand management through active monetary, fiscal and exchange rate policies. The outcome so far confirms that the short term adjustment programme has been effective in lowering inflation, eliminating the overvaluation of the exchange rate, improving foreign exchange reserve. 4. In spite of the recent international economic crisis, real GDP continue to grow 10 percent in FY 09 and FY 10 and projected at continue at the same pace in FY11. The economic growth achieved in past five years was higher than anticipated in PASDP. On the other hand, the IMF was cautious in their projections at 7 and 7.8 percent for FY10 and FY11. 5 The data consists of actual numbers for FY03-FY04, pre-actual numbers (yet to be verified) for FY05-FY09 and budget numbers for FY10. 5. The annual average inflation rate after the unprecedented 25 and 36 percents growth in FY 08 and FY 09 has gone down to 2.8% in FY 10. Inflation is expected to continue to moderate in FY11 with the reduction in global food and fuel inflation. However, Ethiopia suffers from inertia with respect to non-food inflation. Buoyant growth in reserve money in the past has contributed to preserving relatively resilient non-food inflation numbers. The growth in reserve money prior to adjustment was driven by excessive monetization of public debt. 6. While public borrowing (including Public Enterprises) has been managed under the adjustment programme, credit rationing is present in the prevalent in Ethiopia due to excess bank reserves and credit ceilings imposed by the National Bank of Ethiopia. The negative real interest rate further stifles private savings and creates a disincentive for commercial banks to lend to the private sector. 7. The falling inflation rate and the managed nominal depreciation of the Birr has led to a depreciation of the real effective exchange rate from October 2008 to June 2009. This development will have a positive impact of the country’s current account with the added benefit of improving Ethiopia’s reserves. The IMF expects foreign exchange reserves to continue improving and exceed two months of import cover by the end of FY 10. 8. The GoE projects exports to grow by 26.7 percent and imports by 21.5 percent in FY 10 resulting a current account deficit of 5.2 percent. The value of exports increased by about 23 percent during the first 8 months of FY10 compared to the same period in FY09. The following products contributed to the rise in exports: Pulses, flower, and oilseeds increased by about 49, 29 and 19 percent, respectively while coffee increased marginally by approximately 4 percent. The value of imports declined by 3 percent over the same period. Table 0.1:Selected Aggregate Economic and Financial Indicators Ethiopia: Selected Economic and Financial Indicators, 2004/05-2012/13 2004/05 2005/06 2006/07 2007/08 2008/09 2009/10 2010/11 Est pre act pre act pre act pre act est. proj GDP at constant prices (at factor cost) growth, % IMF 12.6 11.5 11.8 11.2 9.9 7.0 7.7 GDP at constant prices (at factor cost) growth, % (GOV) 12.6 11.5 11.8 11.2 9.9 10.1 … Growth of Per Capita GDP, % (GOV) 10.0 8.9 9.2 8.6 7.3 7.5 5.1 Consumer prices (period average)growth, % 6.8 12.3 15.8 25.3 36.4 5.6 8.4 Non-food 5.1 8.0 14.8 12.5 23.7 17.8 … Gross domestic investment % of GDP 23.8 25.2 25.8 22.5 22.4 24.8 … Government investment 15.2 17.3 18.8 16.2 16.5 16.9 … Private investment 8.6 7.9 6.9 6.3 5.9 7.9 … Fiscal balance, including grants in percent of GDP -4.4 -3.9 -3.1 -2.9 -0.9 -2.0 -3.4 Domestic financing (including residual) percent of GDP -1.2 1.4 -1.6 -1.1 0.0 0.0 0.0 External Sector Export, growth % 41.1 18.1 18.7 23.1 -1.0 16.4 11.1 Current account balance, excluding official transfers % of GDP -12.4 -14.9 -10.6 -10.5 -10.0 -13.0 -14.4 Gross reserves in months of imports of GNFS of the following year 3.4 2.2 1.9 1.2 1.8 2.0 2.2 Source: The IMF 9. Prospect for FY 11 is positive, with projected double digit GDP growth and single digit inflation. While exchange rate adjustment restored the competitiveness of the economy, the country will continue to remain vulnerable from the structural imbalances in BoP and the internal saving investment gap. Medium Term Trend in Fiscal Aggregates 10. The fiscal deficit both including and excluding grants is lower in FY09 than it has ever been since FY03. The fiscal deficit to GDP excluding grants stood at 5.0 percent in FY09 while the fiscal deficit to GDP including grants stood at 1 percent in FY09. The GoE conscious decision to keep deficit low in face of high inflation, elimination of fuel subsidy, greater flow of grants, accounts for the adjustment process observed recently. The main instrument used by the GoE to reduce the deficit is moderating growth in expenditure and not increasing growth in revenue. Between FY08 and FY09 domestic revenue declined. Further, the Ethiopian government’s inability to increase revenue substantially has had an impact on its fiscal space. The policy of reducing the fiscal deficit has been achieved mainly by curtailing growth of expenditure. The decline in expenditure has been greater than the decline in revenue which has led to an improved budget balance. In particular, growth in recurrent expenditure has slowed markedly. 11. Broadly, trend in aggregate spending reflect available financing. Aggregate spending hit peaked at 24 percent of the GDP in FY 04. Since then it has been descending, reaching the lowest expenditure to GDP ratio of 16 percent in FY 09. Following the priority set under PASDEP, big drive on infrastructure expansion was supported by fast growth in capital budget (with an annual average growth of 31 percent in the past five years).6 This emphasis on infrastructure as a result growth of recurrent spending was contained below the growth of aggregate spending. Notably, expenditure on defense has been kept at Birr 3 Billion in nominal terms for a long time. 12. Ethiopia’s experiences weak revenue performance, as measured by general government revenue-to-GDP ratio. Although Ethiopia has undertaken several reforms to enhance tax collection, revenue-to-GDP has been in steady decline from FY03. The reforms undertaken include the imposition of valued-added tax, the expansion of branch offices to the regions, a crackdown on contraband trade and the introduction of taxpayers’ identity numbers. Despite all these measures, Figure 1.1 shows that this ratio fell 5 percentage points from FY 03 to FY09. The composition of tax revenue which accounted for 70 percent of total domestic revenue in FY03 rose to 72 percent by FY09. However, domestic revenue collection has not kept pace with rising GDP growth resulting in declining domestic revenue-to-GDP ratio from FY04. Figure 0.1: Domestic Revenue, Expenditure and Fiscal Deficit Source: MOFED 13. Predictability of aggregate fiscal balance (overall deficit including grant)in the past was poor showing volatile with the figures ranging from 41 percent to 95 percent between FY 04 and FY 09 (see Figure 1.2). There could be a number of reasons for this volatility. One is the inability of the GoE to provide a relatively accurate plan of the domestic revenue generation (particularly, non-tax revenue) through the Macroeconomic Fiscal Framework (MEFF). This seems to have translated into large variation between budgeted and actual fiscal deficit from year to year. The scepter of high inflation may provide some explanation for the volatility of budget execution with respect to the budget balance. In the face of high inflation the GoE consciously decided mid-year to revise its fiscal deficit target downward (as happened in FY09). Lastly, the difficulty in budgeting for grants may also have contributed to volatile budget execution with respect to the fiscal deficit. The volatility of deficit with grant has been greater as a significant share of anticipated flow of external grant failed to materialize. Recently with increasing programmatic support like the PBS the predictability of grant improved. Yet the performance of the less harmonized project support remains low. 6 In spite of fast growth in capital budget, a significant proportion of infrastructure investment is made outside of the budget through public enterprises (for more discussion see PFR 2009). Figure 0.2: Budget execution: Revenue, Expenditure and Fiscal Deficit Source: MOFED 14. Historically, budget execution at an aggregate level, generally strong averaging more than 90 percent. In part a reflection of weak planning, aggregate expenditure remains at about 85 percent of the budget in the past five years. In spite of unprecedentedly low performance in FY 09 (89 percent, lowest in five years), recurrent budget absorption average 92 percent in the past five years. As expected, capital budget utilization rate is lower than the recurrent averaging 85 percent in the past five years. Yet, from historical trend recent capital budget utilization is much better as domestic and programmatic aid increasingly becoming important in financing capital budget. 15. While tax revenue continues to perform well, averaging 94 percent of the budget in the past 5 years, non-tax revenue has been less predictable, averaging 86 percent of out-turn only. The problem of predictability has been more profound in grants; with an average out-turn of less than 70 percent in most of the years. Moreover, between FY06 and FY08, budget execution has increased from 80 percent to 95 percent. The narrowing of the gap between actual collection and budgeted revenue collection has led to an improvement in the realization of the government’s budgeted expenditure plans. However, this trend reversed in FY09 with a decline in the realization of revenue and grant from 89 percent in FY08. The performance of tax collection against budget in FY 09 was the lowest in six years (87 percent of the budget). The adverse implication of poor tax collection in FY 09 was moderated thanks to strong performance in non- tax revenue and grant with an outturn of 86 percent. Revenue Mobilization under Inflationary Condition 16. Nominal revenue growth steadily rose from FY03 to FY06 and accelerating in FY07 to FY08 after dipping slightly in FY07 due to decline in non-tax revenue. Non-tax revenue predominately comes from residual surplus from public enterprises, interest payment and dividends from public enterprises. In the past three years the contribution of residual surplus to non-tax revenue increased from 35 percent to 65 percent. In spite of slight decline in FY07, its magnitude growing much faster recently from 2.7 billion in FY 07 to 7.3 billion in FY 09. 16. Figure 1.3 depicts a similar pattern for growth in real revenue collection. However, from FY07 the gap between nominal and real revenue growth has widened with inflation growing at a rapid rate. This implies that while growth of nominal revenue collection looks impressive, the growth of real revenue collection has been extremely poor – barely enough to keep pace with the inflation rate. Figure 0.3: Evolution of Domestic Revenue: Nominal vs and Real Source: MOFED 17. Even though direct tax revenue has been rising, the composition of direct tax revenue in total tax revenue has remained fairly static at 35 percent over the period. In contrast, the growth in foreign trade taxes has been rising exponentially from FY05 to FY08. Figure 1.3 illustrates that this has contributed to a foreign trade taxes composing of 50 percent of tax revenue in FY08. Interestingly from FY05, openness as measured by the ratio of the sum of exports and imports to GDP declined. Recently, the share of foreign trade tax to total revenue increased from 44 percent to 49 percent (between FY 03 and FY 08) – this growth primary reflect increased import tax (sales and excises) and the introduction of sure tax at the end of FY 07. In the subsequent year (FY 09), its share drop again to all time low 40 percent, as the relative importance of income tax and domestic indirect tax increased. 18. The growth rate in domestic indirect taxes has been volatile. Its share of tax revenue has risen by 5 percentage points to 25 percent in FY09 from FY03. This is associated with imposition of a value-added tax. Figure 1.3 depicts domestic indirect taxes still contribute the least to tax revenue. Aggregate Expenditure Management 19. While the trend in nominal expenditure growth has been upward from FY03 to FY09, the trend in real expenditure growth has been downward since FY04. A high inflation rate in Ethiopia over this period has eroded the real growth in expenditure. The divergence between nominal and real expenditure growth has increased from FY05. Figure 1.4 shows that in FY09, the growth rate in real expenditure turned negative. This implies that the volume of public services provided by the GoE in FY09 actually falls in FY09. Figure 0.4: Evolution of Expenditure: Nominal vs. Real Source: MOFED 20. There has been a shift in emphasis from recurrent to capital expenditure from FY 03 to FY 09. The average growth rate in capital expenditure is 30 percent, twice that of recurrent expenditure growth. Figure 1.5 demonstrates that the GoE has increased the proportion of capital expenditure by general government to 53 percent from 35 percent while steadily decreasing proportion of recurrent expenditure from FY03. The increased expenditure on infrastructure is reflected in this shift in expenditure. However, reducing proportion of recurrent expenditure may adversely affect the delivery of services. 21. The shift in emphasis on expenditure to the economic and social sectors is highlighted in Figures 1.5 There has been a reallocation of expenditure from general and administrative sector to the economic and social sectors for both recurrent and capital expenditure. For recurrent expenditure, the proportion of expenditure in the social sector has steadily increased from 33 to 38 percent. Thus expenditure on the social sector is driving growth in recurrent expenditure. This reflects the fact that the GoE is placing more emphasis on health, education and social welfare. 22. Capital expenditure in the social sector increased from 29 to 32 percent. The average growth rate in social expenditure reached a high of 58 percent in FY07, composing 33 percent of capital expenditure. The share of capital expenditure in the economic sector remains relatively constant. This expenditure in economic and social sector is focused toward achieving the goals set out by PASDEP. Figure 0.5: Functional Composition of Expenditure Source: MOFED 23. With Ethiopia qualifying for HIPC and MDRI, the proportion of expenditure allocated to debt service (incl. amortization) has gradually declined by 6.5 percentage points to 1.1 percent in FY09 compared to FY04. This has allowed more resources to be devoted to the pro-poor reducing sectors as depicted in Figure 1.6. 24. The pro-poor sectors have seen their share of general government expenditure rise from 52 percent in FY03 to a peak of 64 percent in FY08 as illustrated in Figure 1.5. The rise in expenditure in the social sector reflects the increase in expenditure on the pro-poor reducing sectors. In addition, from FY05 to FY08, the growth rate in pro-poor expenditure has risen 10 percentage points to 35 percent. The pro-poor reducing sectors include health, education, rural roads and urban construction, and agriculture and natural resources. The emphasis on pro-poor reducing sectors by GoE is to increase the social capital of Ethiopia to make the society more productive by improving the quality of the human capital base of the country. 25. Considering expenditure in the pro-poor sectors for the FY09 budget, there has been a shift in expenditure away from education towards agriculture and natural resource, health and roads when compared to FY03. Across the pro-poor sectors, Figure 1.6 portrays the decrease in the proportion of education expenditure by 7 percentage points to 35 percent in FY09. There is a reallocation of expenditure towards agriculture, health and roads from FY03 to FY09. However, for this period, there has been increased overall expenditure on education in schools and universities. Figure 0.6: Composition of Expenditure by Broad Function Source: MOFED 26. Figure 1.7 indicates that the rise in pro-poor spending at the federal is replicated at the regional level. The share of pro-poor spending is increasing both at a federal and regional level. The federal government spends a greater share of its expenditure on pro-poor sectors when compared with regional governments. The federal government spent 38 percent of total expenditure on pro-poor sectors in FY09 while the regional governments only 26 percent. Figure 0.7: Share of Expenditure on Pro-Poor Sectors in Total Spending The FY 2010 Budget 27. Two important features differentiate the FY 10 budget: its relative lax aggregate fiscal stance compared to FY 09 and more realistic assumptions about revenue and spending plans than the earlier years. The FY 10 plan envisages greater reliance on revenue than grant, the performance last year implies limited space to react to adverse shock. Thus, donors need to sustain the recent progress in the predictability of aid while, protecting basic services spending may require protecting FBG from other competing expenditures. 28. Broad allocation of the FY 2010 General Government budget shows decline in defense and debt services spending in real terms while maintaining the share of pro-poor spending over 60 percent. However, the decline in real aggregate spending tends to affect all sectors including pro-poor spending. Ethiopia has one of the highest capital-to-recurrent expenditure ratio in Africa. Much of the growth in the recent years is accounted for pro-poor capital particularly in infrastructure investment. In FY 10 budget, , the emphasis on pro-poor sectors will sustain with notable increase in investment on roads, natural resource development (including water) and human development. In contrast, urban development (mainly of urban housing) that used to be one of the main focuses in the past two years now has less than one-sixth of the last year budget. 29. Decentralization seems to move in the reverse direction recently. Since last year, the share of the Federal government to total budget increased from 56 percent to 58 percent leaving the balance to regions. At the same time the share of Woreda budget decline from 22 percent to 20 percent. On the other hand, local revenue generation of regions surge significantly. In FY 09, regional revenue (excluding AA) grew overwhelmingly strong, with a performance rate of more than 60 percent above the budget. This temporarily narrows down the vertical fiscal imbalance from 85 to 80 percent. However, the sustainability of this performance seems in question as regional government’s plan to cover only 18 percent of their expenditure in FY 10 compared to 20 percent in FY 09. 30. One important feature of the FY 10 budget is the realism of the fiscal framework in which revenue forecast was moderate (about the same level of tax revenue as in FY 09). On the other hand, performance in the first half of the FY was impressive, and enabled the creation of a greater fiscal space for Federal spending. Strong tax revenue across its entire sources accounts for the more favorable revenue. While the overall tax revenue increased by about 46 percent in the first half of FY 10 over FY 09, revenue from direct tax, domestic indirect tax and import duty grew by 55, 41 and 37 percent respectively. The strong domestic revenue mobilization was crucial in this FY in light of sustained decline in disbursement of grant for the last two years. 31. Figure 1.8 shows that the share of tax revenue in total revenue has risen in FY10_H1 to 68 percent from 57 percent. This represents an increase of 46 percent. By contrast, the growth rate in grants have decreased by 19 percent resulting in the proportion of grants to total revenue falling by 7 percentage points to 14 percent. There has been a decline in the collection of charges and fees by 18 percent. In addition, residual surplus, capital charge and interest grew only by 6 percent. The latter category accounted for 73 percent of non-tax revenue in FY09_H1. Together this resulted in non-tax revenue rising marginally by 1 percent in FY10_H1. The creation of the unified revenue authority, the Ethiopian Revenue and Customs Authority, formed in 2008, is helping in boosting revenue. GoE hopes to implement measures to broaden the tax base, improve compliance and reduce evasion. Figure 0.8: An Analysis of Revenue for the First half of FY10 Source: MOFED 32. In the first half of FY10, General Government’s expenditure is slightly below the budget, but shows improved tendency. Thus, in the first six months of FY10, GoE spent 44 percent of the budget. While remaining within the plan, budget utilization improved over the same period FY09, when it was only 37 percent. Despite a modest increase in absorption of recurrent budget (increased from 42 percent to 44 percent), capital budget utilization rate improved timelessly from 32 percent of last year to 43 percent in FY10. Budget utilization at regional level was modest (about 37 percent) in the first half of FY10, and is slightly less than in the same period of last year. For most sectors (excl. economic sectors) recurrent was planned to increase by about 15 percent. In the first six months, regions utilized less than 40 percent of the budget (a decline from 45 Figure 0.9: General Government Expenditure in the first six months percent in FY09). Non-of the regions utilized their 50 compared to Budget in FY 08, 09 and FY10 percent of their respective recurrent in the first half. Compared to the first half of FY09 regional recurrent spending grew a mere 3 percent. On the other hand capital budget execution rate in the first half of FY continued to improve from 11 percent in FY 08 to 26 percent in FY 09, up to 34 percent in FY10. Yet compared to the Federal, regions lag behind in capital budget absorption. With FBG disbursement on track and improved local revenue generation, the low budget utilization (mainly accumulation of regional deposit) is not commendable for efficiency in cash management. 33. Share of spending on pro-poor sectors is higher in FY10 than FY09, particularly thank to the Road Sector. Comparing FY09 versus FY10 half year outturns, the Road Sector, both at federal and regional level, is the one with the highest growth rates. The share of pro-poor sectors from the national, Federal and Regional states spending account the lions share, more than 60 percent. 34. Aid, in absolute terms, has increased over the last few years, with the exception of FY 10 outturn. Despite the reduction in the budget in fact (-9.1 percent from FY 09), External Grants disbursement in the first 6 months of FY 10 has slowed down if compared to the three preceding years. The share of grants and loans has kept constant in the past four years representing 4 percent and 1 percent of GDP respectively. In spite of its importance, timely disbursement of grant such as PBS will be crucial not only for the health of the macro-economy but also ensure predictability of resource flow to basic services. Figure 0.10: External Finance in the Budget: Commitment & Disbursement in the first half of the FY Source: MOFED 35. To sum up, aggregate fiscal management is commendable, thanks to buoyant revenue performance. The positive sign of improved revenue mobilization, both as national and regional level, was badly needed for a more prudent macro management and better predictability of the financing flows. On the other hand, the timeliness of the grants disbursements is a matter of concern. General Government’s Expenditure: improved trends from previous fiscal years, but under performing compared to the budget. Particularly under spending at regional level on the one hand, and accumulation of deposit on the other hand, need due attention. As part of aid management, improvements of the quality and management of the data are required, as well as more frequent monitoring of the level and quality of aid and dialogue Donors/GoE need to be strengthen. DECENTRALIZED SERVICE DELIVERY, THE INSTITUTIONAL CONTEXT 1.1 Introduction 1. The objective of this chapter is to understand the institutional context of the decentralization process and its achievements and challenges in the Ethiopian context. The chapter is structured into eight sections. The first Section introduces some generic concepts on Federalism and fiscal decentralization. The second Section clarifies the legal framework on which the fiscal federalism is built on. In section three the administrative structure is defined. Section four and five describes the functional and expenditure assignments respectively. The sixth Section clarifies the revenue assignment. Section seven explores the different intergovernmental fiscal transfer systems existing between levels of governments. 2. Federalism involves the design of intergovernmental relation and decentralization of power. The rationales for federalism and devolution of power to intermediate and lower tiers of government vary from country to country. Some factors deter federalism, while others become its driving agenda. 3. In Ethiopia, federalism was introduced after decades of ethnic conflict and civil war, following the overthrowing of the military-Marxist regime in 1991.The transitional charter and then the Federal Democratic Republic of Ethiopia (FDRE) constitution in 1994 changed the political, fiscal and administrative structure of Ethiopia by forming autonomous Regional States. The constitution of FDRE in the preamble and elsewhere in the articles mention the significance of nations, nationalities and peoples in forming the federal government and the devolution of power to Regional states. 1.2 The Legal Framework 4. The 1995 Federal Constitution is the basic document that lays out the legal and institutional framework for decentralization in Ethiopia. Accordingly the Federal structure that evolved comprised of nine Regional States, viz. Tigray; Afar; Amhara; Oromia; Somali; Southern Nations, Nationalities, and Peoples Regional government (SNNPR); Benshangul- Gumuz; Gambella and Harari national Regional States and two autonomous administrative Cities, viz. Addis Ababa and Dire Dawa. It also outlined the respective spheres of authority and responsibilities of the Federal Government and Regional States. 5. The Federal and Regional constitutions as well as the subsequent proclamations delineate different expenditure and revenue assignments to the federal and sub-national level of governments. Thus, the principles of fiscal decentralization, which is emanated from federal and regional constitutions, devolve fiscal decision making power to lower tiers of governments, minimize vertical fiscal imbalances and provide complementary resources for effective and efficient delivery of public services. The government also attempted to introduce decentralization to reflect expenditure and revenue assignments and overseeing mechanism at district level. Thus, in Ethiopia, two waves of fiscal decentralization have been in place, one from federal government to regional states and the other reform regional states to Woredas with the view of revitalizing the Regional constitutional mandate for Woredas. 6. The prime legal basis for fiscal decentralization is the 1994 constitution, Article 97 & 52 which provides autonomy for revenue and expenditure responsibilities to Regional states respectively. Article 62 indicates the prevalence of budget subsidy and hence the approval of budget subsidy formula by the House of Federations. Article 94 states the need for providing loan and assistance to support regions and areas. The proclamation No. 33/1992 enacted through MOFED is the other legal document that describes the expenditure responsibilities of the sub- national governments (henceforth SNGs). 7. The fiscal decentralization strategy in FY 01 elucidates the landscape for general and specific grant to different tiers of government. It incorporates the principles governing fiscal decentralization, i.e., to devolve fiscal decision making power to lower tiers of governments, minimize fiscal gaps and provide complementary resource for effective and efficient delivery of services. Along this line, the specific objectives of fiscal decentralization in Ethiopia are also mentioned as follows. • To devolve fiscal decision making power to lower tiers of government. • To enable regional and Woreda governments/administrations provide standard services in accordance to their functional assignments • To narrow the horizontal fiscal gap and ensure horizontal equalization • To promote efficiency in the allocation of financial resources. • To maintain consistency between macroeconomic stability and fiscal decentralization. 8. In the year FY 01 the government’s decision to deepen the decentralization process to the Woreda level started by revising the Regional constitutions so as to define the responsibilities of the lower level of government in it. Box 0.1: Legal Framework for Woreda Development Assignment The revised constitution of Amhara Regional state clarifies the expenditure and revenue responsibilities of the Region in its articles 47, 49, and 58 as well as proclamation No. 59/2001. Articles 84, 86, and 91 as well as proclamation No. 59/2001, on the other hand defines the revenue and expenditure responsibilities of Weredas and also provide to them the right and power to exercise self administration, local development. The same power was provided to the major urban administrations that do not lie within the jurisdictional responsibilities of 1.3 Administrative Structure 9. The Government structure has four tiers. This was mainly a result of historical processes to provide autonomy, to shorten the tiers for administration, to reach the grassroots and upkeep the already setup physical structure. It is organized at the Federal, Regional (9), Woreda (671 Rural and 99 Urban Administration) and Kebele (15,000) levels of administration. Zones are by law coordinating and supervisory administrative structures with no executive or policy making authority except in Amhara Regional States and SNNPR. Zone officials are, however, delegated by Regions to support Woredas in their development activities. Furthermore, the constitution provides regional states to form its own lower tiers as it fits necessary. The following illustrates the organ gram for the government structure in Ethiopia. Figure 0.1: Federal Structure of the Ethiopian Government First Tier Second Tier Third Tier Fourth Tier Source: BOFED 10. Sub-national and local development is an important driver of a country’s progress towards achieving its development priorities, whether measured against the MDGs or another set of goals. Such development is not just development that happens locally; it is a process that leverages the comparative and competitive advantages of localities, mobilizes their specific physical, economic, social and political resources and institutions and is embedded in national development processes and frameworks, including existing national and sectoral development strategies and also the Medium-Term Expenditure Framework. 11. However, lack of long-term socio-economic strategy, unavailability of formalized planning methodology, and lack of clarity regarding responsibility of planning process forced the annual planning process of sub-national governments to be based on sectoral and Federal government directives and policy targets. This is more vivid when we look at the Woreda annual plans as shown in Box 2.2. Box 0.2: Sub-National Budgeting Process Local level planning prior to budget ceiling notification 9 Communities at sub-kebele and kebele level identify and prioritize new projects based on needs of the community 9 Kebele Council approves the draft Kebele need based plan 9 Wereda Development Committee with the assistance of the Wereda sector offices prepares the draft need based Wereda plan taking into account the inputs provided in the Kebeles’ plan 9 Wereda cabinet appraise the draft need based Wereda plan and the Wereda council approves it Local level Planning and budgeting after budget ceiling notification: 9 MOFED notifies the indicative Regional transfer to BOFEDs 9 BOFED notifies Wereda transfer to WOFEDs 9 WOFED makes budget call to the Wereda public bodies along with tentative budget Guidelines 9 Government development committee at Kebele level revise the need based plan and appraises the annual activities and projects as per the indicative budget 9 Kebele cabinet and the Kebele council approves the revised plans and budgets 9 Wereda sector offices and Wereda Development Committee reviews the kebele plans & budgets and prepares Wereda plan and budget 9 Wereda cabinet appraise the Wereda plan and budget 9 Wereda Council approves the Wereda Plan and Budget Source: BOFEDs 1.4 Functional Assignment • 12. The general guidelines for allocation of functions among levels of government as described in public-finance literature (Oates, 1972 and King, 1984) can be summarized into: • stabilization, which in most cases is central government responsibility; • distribution, which requires some degree of cooperation; and • Allocation function, which is not clear in most cases but would follow the principle of providing public services at a level of jurisdiction where the benefits accrue. 13. Although economic theory provides little guidance on expenditure assignment, examination of actual practices reveals that there is considerable variation in expenditure assignments among different countries. According to the Ethiopian Federal Constitution, matters of nation-wide concern and those which go beyond the jurisdiction of a single Regional States are left to the central Government while a great deal of responsibility for local economic development, including infrastructure and public services, is assigned to Regional States. The following table illustrates the general functions assigned to the Federal government and the Regional States. On the other hand, the Regional constitutions define the general functions of Local governments which are Woredas and Kebeles. Annex 1 gives detail functions of Woredas and Kebeles as per the revised constitutions of regional states. Table 0.1: Functional Division between the Federal and Regional Governments Federal Government has powers to: Regional Governments have powers to: • Print currency; • Establish a State administration that best • Administer national defense and advances self-Government security; • Execute economic, social and • Formulate foreign policy; development policies and plans of the • Regulate inter-state and foreign State; commerce; • To administer land and other natural • Formulate/execute national monetary resources in accordance with Federal policy; laws; • Establish uniform standards of • Collect and levy Regional governmental measurement; taxes; • Establish national standards; • To establish and administer a state • Determine nationalities police force, and to maintain public order and peace within the State. • Make Regional governmental laws; • Establish executive administrations; • Run Regional governmental courts; • Establish Regional governmental councils • To enact and enforce laws on the State civil service and their condition of work; • Keep law and order in the Regional government; • Enact a Regional governmental Constitution Source: FDRE constitution 14. Although the Federal and Regional States’ constitution defines the functional assignments of the different level of governments, in relation to its practicality, in some areas of responsibilities and in particular in developing Regions it is an evolving process. For instance, administrating of large scale farm investment has been delegated to the Federal government by the four developing Regions (Gambela, Afar, Somale and Benshangul). In the same manner, procurement of large scale works in most Woredas is delegated to either to the zones or Regions. 1.5 Expenditure Assignment 15. According to the Federal Constitution (Article 51), matters of nation-wide concern and those which go beyond the jurisdiction of a single Regional States are left to the central Government while a great deal of responsibility for local economic development, including infrastructure and public services, is assigned to Regional States. On the other hand, local governments (Woredas and Kebeles) are entry points for service delivery and their proximity to the people gives them a unique advantage to be responsive to community needs. Their functions vary in rural and urban areas. In both areas, they are major service providers like administrators and management of primary school, health posts, extension services, social shengos/tribunals, community policing. 16. Expenditure responsibilities related to sectors are also defined by level of governments. Sectoral expenditure responsibilities such as foreign affairs, defense, establishing national sectoral policies and standards, building and administrating trunk roads, administer and regulate air transport, administration and management of electricity, administration and management of telecommunication, Federal level police and courts, formulation and execution of national level capacity building programs , and food security are under the jurisdiction of the Federal (central) government. 17. The sector specific responsibilities of regional Figure 0.2: Distribution of General governments include regional water resource Government Expenditure by development and protection policy, human capital Jurisdiction (FY 06-08 average, in %) development, inter-woreda and intra-regional roads, Source: MOFED vocational and technical training, TTI, medium level colleges, Regional hospitals, and regional health service standard setting. On the other hand, the different State constitutions delineate sector specific functions such as primary schools (1st and 2nd cycles), coordinating primary health care preventive and curative activities (i.e., the functions of health posts and health centers), managing and maintaining rural roads, coordinating agricultural development activities, and making access to drinking water supply to the local level government. (for the detailed sector specific expenditure responsibilities refer Annex). 18. However, there are disparities among Regional states and specific local governments within a Region in terms of efficiency in the implementation of the expenditure responsibilities. The pastoral/boarder areas are relatively less performing in service delivery bestowed to them due to lack of manpower, institutional as well as structural challenges. 19. As it can been seen on Fig. 2.2 from the table about 58 percent of the expenditure in the specified years are incurred at the federal level while about 42 percent of the total expenditure was the responsibilities of the sub-national governments. If we further analyze the sectoral picture, education and health that contributes about 30 percent of the total expenditure, about 60 percent are incurred by Sub-national governments (for the detail of Regional and Woreda share refer the Annex). On the other hand, more that 80 percent of the expenditure on Road construction, transport and communication, and mines & energy are incurred by the Federal government. 1.6 Revenue Assignment 20. Public finance literatures McClure (1983), Musgrave (1983) and Bennett (1987) have provided some general guide-lines on revenue assignment to the different level of governments. Accordingly it is indicated that the Central Government should be assigned to taxes which are highly progressive, re-distributional, suitable for economic stabilization that needs economies of scale in tax administration, and tax bases that are distributed highly unequally across sub- national jurisdictions. The sub-national governments, on the other hand, are prescribed to rely on taxes that have relatively immobile bases like local real estate and user charges and fees 21. In line with this, the constitution contains reasonably specific provisions concerning the division of revenue raising powers between the Federal and Regional (State) Governments. It gives a broader summary of what is specified as to which taxes shall be levied and collected respectively by the Federal and Region Governments. Accordingly, the Federal Government revenue assignment includes tax types that Figure 0.3: Contribution of Administrations to the General Government Revenue (FY06- are highly progressive, have redistributive nature and 08, average in %) are important for economic stabilization. The Sub- Source: MOFED National’s tax responsibilities on the other hand include taxes that are more local in nature and levied on relatively immobile assets. 22. There are also revenue sources under the joint jurisdiction of the Federal and Regional States. Hence the Regional governments have two sources of revenue-own and joint – that would enable them to collect and use it. The Woredas, on the other hand, collecting taxes but do not have the authority to use it. The collection is counted to be part of the Regional transfer. Woreda has revenue assignments but the revenue woreda collect is offset by the regional government grant allocation. This condition is sometimes partially introduced in some federal countries having specific grant that requires matching fund from local own revenue. The following table depicts the detail revenue responsibilities by level of government. 23. Based on the revenue collection responsibilities by level of governments mentioned above, between the FY 06-08 sub-national governments was able to collect only 20.1 percent of the total revenue (Woreda level of government 11.8 percent and Regional level of government 8.3 percent), the rest 79.9 percent had been collected by the Federal government. The revenue collected by different tiers overtime has no marked improvement. Lower tiers collected a smaller proportion of the general revenue. Note that, the devolution of revenue assignment between regions and woredas is not complete. The current practice is that regions delegated woredas to collect revenue on behalf of them. 24. Looking the revenue by categories, 77.8 percent of the revenue is coming from tax revenue of which 37.2 percent of its part is import duties that is the sole responsibility of Federal government. The direct tax contributes 23.4 percent of the tax revenue of which Rural land use fee, agricultural income tax and urban land lease fee that are the responsibility of sub-national governments accounts only 2.3 percent of the tax revenue. (For the detail refer to Annex1). 1.7 Intergovernmental Fiscal Transfer Country Experience 25. The design of intergovernmental fiscal arrangements is one of the more complex areas of public finance, since it spans a number of policy and institution-building issues, requiring careful coordination and sequencing, and it is strongly influenced by historical, political, and social, as well as economic, factors. 26. There is no single “right” model in designing the intergovernmental fiscal arrangement, as countries adopt it based on their specific circumstances. However, there are some general lessons that can be drawn from the range of experiences. These include availing resources to sub- national governments as per the assignment of spending responsibilities; and the need for linking the basic minimum capacity of sub-national governments to the functional assignment. 27. For effective results and autonomy, sub-national governments either should have their own revenue or a predictable transfer mechanism. It is a fact, even in the most matured federations, that the division of expenditure and revenue responsibilities is never such that both orders of government are fully self- financing. If revenue responsibilities are relatively centralized and spending responsibilities are not, significant vertical imbalances will exist. Therefore, in most cases fiscal transfers from one order of government to another are used to deal with vertical imbalances. 28. It is obvious that a high level of vertical fiscal imbalance could lead to efficiency problems unless some measures are taken to correct the imbalance. Here comes the question of Inter- governmental transfer. However, if transfers are designed solely to close the vertical gap, there would be little incentive for the lower levels of government to raise own account revenues or restrict or manage expenditure efficiently 29. When compared with other federations, Ethiopia’s revenue collection is highly centralized as the Federal government collects most of the revenues in the country (due to the nature of the tax structure in the country about 80 percent of the tax revenue are collected from the Federal taxes) and shares them with the states through unconditional and conditional grants. Moreover based on expenditure assignments of each tier and due to the existence of vertical fiscal imbalances, an integrated system of transfer has been developed and adopted across tiers of governments. 30. The vertical imbalance in Germany, USA, Australia and India is very low compared to developing countries and could be attributed to revenue sharing and fiscal autonomy. The sub- national jurisdictions in Argentina, China, Canada and Brazil come closest to achieving fiscal balance. Ethiopia is second to South Africa and Bolivia for having high vertical fiscal imbalance (See annex 4). The Fiscal Gaps 31. The mismatch between the expenditure responsibility and the revenue capacity will create vertical fiscal imbalances. This will be aggravated in a situation where there are wide economic disparities among the states in a country. A high level of vertical fiscal imbalance, on the other hand, could lead to efficiency problems unless some measures are taken to correct the imbalance. 32. The basic rationale for a system of transfers is the existence of a fiscal gap at the local government level arising out of town-revenue and own-expenditure assignments. There are a number of methods to close the fiscal imbalances of sub-national governments, some of which also reduce imbalances between jurisdictions. In practice, we may distinguish between systems of revenue sharing and grants. 33. Like any Federal system, the aggregate revenue raising capacity and the aggregate expenditure obligation between levels of governments do not match in Ethiopia. Accordingly, the average Regional states’ revenue collection for the period FY 07-FY 09 is only 10.4 percent of the total national collection (excluding Addis Ababa as Addis Ababa is covering its expenditure from its own sources). On the other hand the Regional States’ expenditure share to the total national expenditure for the same period is about 46.8 percent. To make it clearer, for the same period, the total Regional states’ revenue collection would only able to cover 18.3 percent of their expenditure. Thus, more than 81 percent of the expenditure was covered from the Federal transfer (See Fig 2.4) 34. Moreover, the extent of incompatibility of aggregate expenditure requirement and revenue raising capacity among the same level of government is not the same. The introduction of inter-governmental fiscal transfer is meant to reduce these vertical and horizontal fiscal imbalances. The problem of ‘horizontal balance’ has to do with the fact that geographical areas usually differ with respect to resource capacity and needs. For instance, the tax base per capita often differs substantially between states and Woredas. The following table shows the fiscal imbalance of Regional States. Figure 0.4: Fiscal Gap by Region in EFY 2001 Source: MOFED 35. The imbalance is even worst when compared the expenditure responsibilities and the financing means at the Woreda level. As the Woredas do not have the use right of whatever small revenue they are collecting, the dependence on the Regional transfer mechanism is very high. The Nature and Types of Fiscal Transfers 36. At present, financial resources are transferred from federal government to regional governments in the form of both general purpose grant and specific purpose grant; and at the regional level, transfers are made to districts in the form of general purpose grant. The general- purpose grant is a system that encourages local decision making, as the various tiers of governments are the ones who decide on allocation. It is untied and provides discretion on the use of financial resources. The main justification for the central government to give unconditional grants to states and localities is that such grants can be used to equalize fiscal capacities of different states to ensure the provision of a minimum or reasonable level of public services. 37. The specific purpose grants (SPG), on the other hand, are given to fulfill the goals of national priorities which otherwise could not be attained by other tiers of government due to spillover effects and that encourage performance improvement. Federal Government provides specific-purpose grant for expenditure assignments that are under its mandate but executed by the different tiers of governments. At present, these include Public Sector Capacity Program (PSCAP), Food Security Program, productive safety net, Urban Local Government Development Program, Local Investment Grant, water and sanitation program etc. 38. Full-fledged data on SPG expenditure is not readily available, but the magnitude of financial resource through SPG has been increasing due to doubling of food security grants as well as the implementation of productivity safety net programs and PSCAP; and most are allocated at regional and local levels. SPG which amounted 3.7 Billion in 1998 increased to 16.9 in three year’s time. The challenges that apart from food security the behavior of allocation composition of specific purpose grant didn’t follow multi sectoral approach. 39. The SPG is a useful tool in promoting the performance base grant. This evidenced by the competition created on the Urban Local Government Development Program among the 19 beneficiary cities. Therefore, the SPG should be enhanced as an instrument to achieve national policies. 40. The transfer mechanism is backed up through overall and self-sustaining expenditure control arrangement. Following the regionalization process, efforts have been made to improve the expenditure management and control system. Federal Transfer to Regional States 41. The decision for inter-governmental fiscal transfer--the provision of untied block grant- and the introduction of formula for allocation dates back in 1994. The details of how the general purpose transfer/ grant have been divided between the nine Regions and Dire Dawa city Administration have varied over the last fifteen years. Until 2007, the ‘basic need approach’ formula was the basis for determining allocations. The formula took into account the size of population, expenditure need or level of development variables and fiscal effort as well as performance measurements in key sectors with weights assigned to the main indicators. However, it has been revised in phases to improve the methodology based on country experience and comments received from different stakeholders. The generic formula in all these cases was as follows: Where: Gi = Share of the Grant entitlement for Regional government i Popi = Index of population indicator Devi = Index of development level indicator Revi = Index of revenue raising effort indicator (0.6, 0.25, 0.15)= The figures represent the weights assigned for each variable which varies in different years. 42. The following table indicates the change in the General purpose grant formula through the period after its invention. As it can been seen from the table, every year (particularly FY 97- FY 06) the discussion by the Regional representatives at the House of Federation during the approval process was focusing on the weights of the variables. That is why the weight of the major variable (population) varies during these periods. 43. In 2007 the latest ‘per capita relativity based formula’, which is similar to the Australian Common Wealth Grant Commission system, has been adopted. It consists of an integrated Expenditure need and Fiscal capacity criteria. It has the objective of creating comparable financial capacity among States in order to provide standard public services. The advantages of this formula against the earlier once is that it avoids subjective nature of variables and weighting, considers recurrent and capital expenditure, provides the proxy to revenue raising capacity, is the major fairness criteria, is credible and minimize unnecessary lobbying. The improved Federal to Region transfer formula is expressed as follows: Gi =Pi (APCG +ADF1+ ADF2 +…+ADFi) Where: Gi= the budget grant to Regional States Pii= the population of each Regional State APCG= Average Per Capita Grant, and is simply the average per capita subsidy for all Regional States combined (total allocation divided by total population) ADFi=Assessed Difference Factor, and is for factor i, for example, the assessed expenditure need or revenue capacity of each regional government in per capita terms relative to the average for all Regional States, brought by calculating number of variables such as disabilities, sector standards etc 44. The formula is initially developed by MOFED but the responsibility of the technical preparation as well as the approval has been transferred to the House of Federation (HoF). MOFED every year allocates the general purpose grant to the nine Regional States and Dire Dawa City Administration using the approved formula directed from the HOF. The Macro Economic and Fiscal Framework (MEFF) is the base for determination of the distributive pool. The total resource available for the distribution includes domestic revenue, counterpart funds, external loan and external assistance. Regional states’ revenue collection is an additional resource available for the Regional expenditure (not offset as it has been the exercise at the early period of the allocation). Regional Transfer of Block Grant to Woredas 45. In practice, decentralization process at Woreda level got momentum at the early stage. However, in pastoral areas the decentralization process is at stand still. The legal framework is not practical 46. As the revenue collected from a woreda covers not more than about 20 percent to 25 percent of the recurrent expenditure, a woreda has not been able to cover the required expenditure from own sources of revenue (even this collection is not been entitled to be used). Therefore, since 2002, general purpose grant has been allocated to woredas from Regional Government on the basis of a simple and transparent formula approved by each Regional council. 47. The grant system has empowered Woredas to prioritize and decide financial resources for meeting their local development needs in line with the regional and national development policies. This fiscal dimension of the decentralization program (i.e. general purpose transfer) created an enabling environment for Woreda administrations to administer transferred budgets and determine the development goals without much interference of the higher administrative levels. In sum, the grant transfer approach enhanced the financial autonomy of Woredas and helped them to facilitate development efforts on their own. Woreda expenditure and revenue collection rights are protected by each Regional States’ constitution and supported by legal frameworks and approved guidelines. However, revenue use rights are not enacted. In general the formulas have to be improved. 48. Regional States are allowed to choose their own way of allocating the general purpose Grant to their Woredas as long as the formula they are using is objective and based on clear and predetermined criteria. The grant allocation formula used for Region- Woreda transfer in most Regions are the same type that the Federal Government was using until 2007 (the three simple parameter basic needs approach - population, expenditure need or development level variables and revenue effort combined with a unit cost calculation approach). However, the formulas need to improve. While the Bureau of Finance and Economic Development is responsible for the technical design of the formula, decision or approval is the mandate of the Regional Council. About 60-70 percent of each Region’s financial resource (Federal grant and Regional Revenue) is allocated to their Woredas using this approach. 1.8 Interfaces between Institution Building and Devolution of Expenditure Assignment 49. Administrative restructuring accompanied by a comprehensive institutional reform and capacity building program have been implemented in an integrated manner in the past decade. At the same time fiscal devolution with a big-push at the initial stage provide incentive to empower local administrations. Notable progress has been made in rolling out of public finance reform along with capacity building. 50. The institutional quality and capability composite index of the WCBS shows notable progress across the 292 sample woredas with two-third of them scoring average to above average.7 However, progress in individual institutional quality scores showed wide variation across woredas. Aggregate ranking of various institutional quality indicators put 11 jurisdictions as best performers including 4 jurisdictions from Oromia, 4 from Amhara, 2 from Tigray and 1 from SNNPR. 192 scored 70-84 percent and graded as middle performers and 82 of them scored below 70 percent and graded as low performers relative to each other. Overall, greater number of the jurisdictions have scored relatively well in public finance management, capacity development, and in putting a place a practice of giving access to information to citizens. In areas like revenue generation and balanced allocation of public expenditure, majority of the jurisdiction, including the top performers in the compounded indicator scored very low. The urban jurisdictions have scored slightly better than their rural counterparts in the quality of public expenditure, revenue generation, and capacity. They have similar scores in public finance management and the rural jurisdictions performed slightly better in participation and citizen consultation. 51. A scatter diagram of woreda per capita and selected institutional indices show positive relationship. Woredas with higher fiduciary quality and capability also mange relatively larger per capita budget. This indicates the fulfillment of one of the necessary conditions for successful decentralized service delivery. 7 Based on the results of the second WCBS, five sets of proxy indicators were used to rank the relative position of woredas in fiscal management, institutional quality and capability. These are i) quality of public financial management; ii) quality of allocation of expenditures; iii) revenue generation; iv) capacity; and v) participation. For definition of the variables and full discussion of the results refer to chapter 5. Figure 0.5: Per capita Budget and Institutional Quality Indices 52. The positive association of per capita woreda budget and institutional indices depicts effective implementation of administrative decentralization and fiscal devolution to woredas. One important factor for successful decentralized service delivery is complimentarily between institutional capability and expenditure assignment. The compatibility of institutional strength and expenditure assignment is expected to lower the risk of miss allocation of public resources. However, there are also areas where progress is limited. First, on the demand side of accountability, there is no strong evidence supporting effective community empowerment except in budget transparency. Second, some of the fiscal pre-conditions for efficient service delivery including local revenue generation and intra-sectoral composition of spending, most woredas remain weak. TREND IN REGIONAL SPENDING AND FINANCING FRAMEWORK 1.1 Overview 53. This chapter reviews sub-national finance at an aggregate regional level. First, it looks in to the overall fiscal stance of regions and their contribution to aggregate fiscal management at national level. Second, reviews regional financing framework with more emphasis on local revenue generations. Third, it assesses the level and Pattern of resource allocation and its adequacy in the context of the recent inflation. Fourth, based on the experience so far, it will draw preliminary conclusion in light of the health and sustainability of sub-national fiscal framework. Finally, the last section summarizes the progress made and the challenges encountered in providing services through the decentralized process in Ethiopia. 54. Major findings of sub-national finance include the following: • In the past five years, increasing imbalance between regional spending and their resource generation capacity contributes to aggregate fiscal strain at macro level; • Local revenue generation continue to be weak particularly after FY 07 financing no-more than 20 percent of their spending; • Increasing divergence between nominal and real expenditure due to nominal expenditure not keeping up with the inflation rate in the past three years; • Nominal domestic expenditure and revenue per capita has been rising steadily in the recent past while real expenditure and revenue per capita has been relatively stagnant; and • The inability of regions to generate sufficient revenue to meet there expenditure needs will continues to be an issue going forward. 1.2 Aggregate Fiscal Picture 55. Big push in regional spending and weak local tax effort contributed to winding aggregate fiscal deficit recently (See Fig 3.1). The growth in expenditure has exceeded the growth in revenue from FY03-FY09. Particularly after FY 06, slow growth in revenue compared to spending result in declining regional capacity in financing from their own source. Equally important, the relatively higher contribution of regional financing gap overall deficit reflect Governments commitment to protect resources going to basic services administered by regions even at times of fiscal restraint. Figure 0.1: Share of Regional Financing Gap in General Government Deficit Source: MOFED 56. As shown on figure 3.2, between FY06 and FY09, regions’ (including AA) local financing capacity dropped from 39 percent to 35 percent. Addis Ababa has a unique position among the regions in at least for two reasons. First AA generates at least half of the revenue of all regions together and this is clear from revenue vis-à-vis regional expenditure. At the peak of regional financing capability (FY06) regions with AA cover 40 percent of spending while excluding AA this ratio drops to 20 percent. Second, revenue of Addis is highly volatile compared to the rest of the regions. The fall in spending of regions between FY06 and 09 is purely attributed to Addis. Excluding Addis, spending coverage of regional revenue hardily changes from 20 percent in the past 6 years. Figure 0.2: Regional Revenue and Expenditure Performance Source: MOFED 57. The consequence of the recent inflation on budget particularly on regional revenue and spending cannot be overstated. At time when inflation reached pick in FY09, regional nominal revenue and expenditure is eroded such that real expenditure and revenue is only 43 percent of nominal expenditure and revenue. The food and fuel crisis together with the drought and poor transportation network has pushed prices upwards in the different regions by different magnitudes. 58. Clearly, nominal and real domestic revenue growth appears to track each another with the growth rate in real revenue consistently below growth in nominal revenue. This implies that the inflation rate over this period has been faster than the growth in nominal receipts. The difference between nominal and real growth in receipts is approximately 25 percentage points in FY08. 59. For the FY09, the growth rate in nominal revenue excluding Addis Ababa is greater than when Addis Ababa is included. The growth rates are 61 and 50 percent, respectively. The opposite holds true when the growth rate in real receipts is considered. This implies that the inflation rate in the regions excluding Addis Ababa is greater on average than when Addis Ababa is included for the FY09. The recent increase in local revenue comes from income tax from the civil service. The new health and rural development approach to basic services required the recruitment of tens thousands of development agents and health extension workers. In addition growth of the service sector (the fastest growing economy in recent years) also contributes to increase in local revenue generation. 60. From FY03, total expenditure growth was at a high of 49 percent in FY08 before moderating to 28 percent in FY09. Figure 3.3, reveals that real expenditure growth also follows the same pattern, reaching a low of 1 percent in FY09 over the same period. Excluding, Addis Ababa, real expenditure growth for FY09 is -16 percent while nominal expenditure growth is 21 percent. The reduction in the growth rate in real expenditure when Addis Ababa is excluded from the analysis for the year FY09 indicates that the volume of expenditure, on average, has declined in the other regions. Figure 0.3: Growth in Domestic Regional Revenue Source: MOFED 61. With slow progress in local revenue generation, the inter-governmental transfer in the form of block grant and specific purpose grant fills the regions financing gap. As shown in chapter 2, block grant transfer plays an important role financing about 80 percent of their budgetary spending. In the past five years, Federal block grant to regions increased from 7 billion to 16 billion. In the past five years the size and number of SPGs to regions increased substantially. Key nature of these programmes is that they are budgeted at Federal level but executed at regional level for pre determined development interventions. 62. External assistance has been instrumental in financing decentralized service delivery. Among aid instruments include Protection of Basic Services (PBS), Productive Safety Net Programme (PSNP), General Education Quality Enhancement Programme (GQUEP), and Local Investment Grant (LIG). These programmes created additional fiscal space to the traditional project grant and loan available to regions. Among these programmes PBS is unique as it complements Federal block grant allowing full discretion to regions. Figure 0.4: Fiscal Transfer to Finance Decentralized Service Delivery Source: MOFED 1.3 Regional Revenue Mobilization 63. The share of regional tax revenue and regional non-tax revenue has been declining from FY06 to FY08 and picks up marginally in FY 09. The contribution of regional government revenue collection has depreciated for these years. Given the fact that general government revenue growth has been rising over this period, while regional revenue growth declined, the federal government has been the main driver of general government revenue growth over this period. 64. Addis Ababa is unique in its fiscal framework from regions. This ratio is around 20 percent when Addis Ababa is included and 10 percent when Addis Ababa is excluded as illustrated in Table 3.1. This implies that Addis Ababa contributes 50 percent of total domestic regional revenue. Excluding Addis Ababa, these figures are 11 and 8 percent, respectively. Thus on average, regional non-tax revenue is more significant vis-à-vis regional tax revenue when including Addis Ababa. The reverse is true when Addis Ababa is excluded. This implies that non-tax revenue collected by Addis Ababa contributes 70 percent of regional non-tax revenue. The other regions of the country clearly are unable to generate close to the non-tax revenue collected in Addis Ababa in terms of residual surplus, capital charge, interest payments, state dividend, and charges and fees. 65. Looking at the structure, tax revenue, on average, generates 70 percent of domestic revenue in Ethiopia between FY03-FY09. Moreover, Table 3.1shows that direct taxes account for 80 percent of tax revenue. For FY09, direct tax contribution to total domestic revenue has declined by 5 percentage points while non-tax revenue contribution has increased by 4 percentage points. When Addis Ababa is excluded from the analysis, the share of tax revenue to total domestic revenue falls to 80 percent from 86 percent from FY08 to FY09. Table 0.1: Regional Tax Base Regions Tax Share from the General The Composition of Regional Revenue by Government by Major Category (in % of GG Major Category (in % of Regional Revenue) Revenue) 2003 2004 2005 2006 2007 2008 2009 2003 2004 2005 2006 2007 2008 2009 Direct taxes 59.5 58.1 59.9 51.8 54.6 58.8 53.6 Tax revenue 17.7 19.1 21.7 20.2 18.6 17.2 19.9 Domestic indirec taxes 11.7 8.7 15.3 13.1 19.2 15.8 16.5 Non tax revenue 16.7 33.8 27.8 28.7 25.9 23.3 22.1 Non tax revenue 28.8 33.2 24.8 35.1 26.2 25.4 29.9 Source: MOFED 66. The share of regional average income and profit tax to total general government income and profit tax is 40 percent for the period FY03-FY09 and 25 percent (excluding Addis Ababa). Figure 3.8 portrays regional business profit tax as a percentage of total business profit tax declining from FY05 while personal income tax rises from FY05. The picture is similar when Addis Ababa is excluded. This reflects the increased number of civil servants recently and the effect of the FY 06 downward adjustment of tax on non-incorporated business. 67. Figure 3.11 depicts regional tax revenue rising by 22 percent annually from FY03 to FY08 when we exclude Addis Ababa and non-tax revenue increases by 22 percent as opposed to 26 percent and 27 percent, respectively, when Addis Ababa is included. The increase in non-tax revenue is brought on by rising collection from Sale of Goods and Services. Most of this increase is due to inflation. In nominal terms, collection from Sale of Goods and Services increases by 20 percent while in real terms it decreases by 1 percent for the same period. 68. From FY03 to FY09, regional tax revenue rising by 22% annually. Income and profit tax has been driving domestic revenue as portrayed in Figure 3.4 when Addis Ababa is excluded from the analysis. Non-tax revenue and indirect tax receipts have also been buoyant over this period but volatile. Business profit tax declined sharply in FY08 while business profit tax excluding Addis Ababa experienced moderately positive growth rate for the same year. Given that most of the businesses are located in Addis Ababa, falling business tax receipts in Addis Ababa dragged down overall business tax receipts. This could be a reflection of the challenge of small business operating in an inflationary environment. 69. Though not as fast as the growth of the sectoral value added receipt from agricultural income tax has been rising from FY06 at an increasing rate. However, on average, agricultural income tax contributes only 2 percent of total domestic tax receipts from FY06-FY09. Yet, the growth rates of 80 percent in FY09 for agricultural income tax amount only to a marginal increase in tax collection. Figure 0.5: Growth in Components of Regional Revenue Source: MOFED 70. Over the years FY03-FY09, Figure 3.5 displays the proportion of business profit tax to total direct tax receipts has been declining. This is true including and excluding Addis Ababa from analysis. This reflects the relatively faster growth of personal income tax collection from civil service workers than the rest of the tax category. 71. Sale/excise taxes have been driving growth in indirect tax receipts. Figure 3.4 reveals sale/excise taxes contribute to 46.6 percent of indirect tax receipts in FY 09. The average growth rate in sale/excise taxes is 33 percent from FY03-FY09. Excluding Addis Ababa, sale/excise taxes grew at 29 percent annually. However, the fastest growing segment of indirect tax receipts is service sales taxes. They grew at an annual rate of 54 and 40 percent, including and excluding Addis Ababa, respectively. 72. Addis Ababa contributes to 50 percent of regional indirect tax revenues over the period FY03 to FY09. When Addis Ababa is included and excluded, regional stamp sales and duty revenue contributes the largest share of total stamp sales and duty revenue. Including Addis Ababa, on average it is 56 percent for FY03-FY09 while excluding Addis Ababa this number is 17 percent. Thus when we include Addis Ababa, most of the revenue from stamp sales and duty is accrued from the regions. 1.4 The Pattern of Regional Expenditure 73. The share of total regional expenditure to general government expenditure has remained relatively constant at 40 percent from FY03 to FY09. This indicates that although the GoE has embarked on a program of decentralization of the budget, the allocation of expenditure to regional government has remained fairly static over from FY03 to FY09. On the other hand GoE was effective in protecting Federal block grant while slowing Federal capital at time of short-fall in financing. 74. Although the proportion of recurrent expenditure for general government has been falling, the share of regional recurrent expenditure to general government recurrent expenditure has been rising from FY03 to FY09. This indicates that recurrent expenditure for regional government is taking a greater share of recurrent expenditure in general government. Figure 3.5 illustrates the share of average regional government recurrent expenditure to total government expenditure for FY09 is 62 percent. 75. Recurrent expenditure by the regions outweighs capital expenditure. However, for FY 09, the proportion of recurrent expenditure to regional total expenditure declines to 67 percent from 73 percent in FY06. Excluding Addis Ababa, recurrent expenditure accounts for 75 percent of total expenditure for FY09. Moreover, in FY09 the proportion of capital expenditure rose from FY05 but its share of total expenditure remained less than it was in FY03. It appears that Addis Ababa gets disproportionately more capital expenditure than the average for the other regions. 76. Regional capital expenditure as a share of capital expenditure in the general government budget has been declining over the period till FY07 as illustrated in Figure 3.5. However, due to increased capital expenditure in the regions over the last two years, the proportion of capital expenditure has increased to 28 percent of total capital expenditure. Table 0.2: Regional Expenditure Share from the General Government in percent 2003 2004 2005 2006 2007 2008 2009 Recurrent 45.2 43.1 48.0 54.1 56.6 61.6 62.1 Capital 30.0 28.1 22.1 21.3 19.9 24.3 27.8 Source: MOFED 77. Growth rates in recurrent expenditure and capital expenditure track one another for the period FY03-FY09 as illustrated by Figure 3.6. This implies that the one does not appear to be a substitute for the other. However, from FY07, the growth rate in capital expenditure has exceeded that of recurrent expenditure. The implication is that the GoE has been placing increasing emphasis on capital expenditure in order to increase the productive capacity of the economy. 78. Excluding Addis Ababa, there is increased variability in capital expenditure by the regional governments. For example, in FY05, the negative growth rate in capital expenditure indicates that the regional governments, excluding Addis Ababa, on average, invested less for regions. Whereas the growth rate in capital expenditure has declined in FY08, when including Addis Ababa, the growth rate in capital expenditure rises for the same year and this is consistent with the vast urban infrastructure investment occurring in Addis. Figure 0.6: Growth in Regional Expenditure Source: MOFED 79. For the FY09 budget, the largest share of the regional expenditure is allocated to social expenditure, followed by general services and finally economic services. Social services account for 40 percent of total expenditure while general services accounts for 37 percent and economic services accounts for 23 percent. The allocation of expenditure across these broad categories in FY 09 is similar to FY 03. 80. The proportion of recurrent social service expenditure has remained relatively static at 50 percent over the course of FY03 to FY09. Most of the variation in recurrent expenditure is a reallocation between economic services and general services. The proportion of general services rises marginally from 29 to 33 percent in FY09 vis-à-vis FY06. A similar pattern for subcategories of recurrent expenditure occurs when Addis Ababa is excluded from the analysis. 81. Considering capital expenditure, general services accounts for the largest share at 50 percent for FY03 and FY06. This share declines to 33 percent in FY09 as the share of social services rises to 17 percent from 7 percent in FY 06. The importance attached by regional government to capital expenditure on social services in the form of schools and hospitals reflects the attitude of central; governments drive to improve the social infrastructure of Ethiopia. Excluding Addis Ababa, a similar pattern follows except that general services in FY03 and FY06 accounts for 40 percent of total capital expenditure. Pro-Poor Spending 82. Regional expenditure for the pro-poor sectors as a share of total expenditure for this sector has declined to 40 percent in FY09 from 53 percent in FY03. This is reflected for all the components of pro-poor sectors with the share of regional health expenditure declining by 23 percentage points over the same period in Figure 3.7. Even though the proportion of general government expenditure in health has risen over the period, most of this increased expenditure is at a federal level and has not filtered to a regional level. The share of pro-poor average spending to total spending by regional governments stands at 70 percent in FY03 and marginally decreases in FY09. This pattern is replicated when Addis Ababa is excluded from the analysis. Figure 0.7: Share of Regional Expenditure in Key Sectors Source: MOFED 83. Approximately 50 percent of the regional pro-poor expenditure is on education while 60 percent of the pro-poor expenditure is on education when excluding Addis Ababa over the years FY03-FY09. Figure 3.8 also demonstrates that recurrent is largest for education. This implies that the salaries of educational staff and other operating expenses have absorbed most of the recurrent regional pro-poor expenditure. Capital expenditure on schooling infrastructure is approximately 10 percent of total capital pro-poor regional expenditure. Given that education accounts for the largest share of pro-poor regional expenditure, the recurrent expenditure on education is creates a substantial drain on the coffers of regional governments. 84. There has been a shift in the proportion of education expenditure conducted by regional governments from FY03 to FY09. This is reflected (in Figure 3.8) by a decline of the proportion of recurrent expenditure on education allocated by regional governments. This indicates that there may be a reversal in the process of decentralization when it comes to spending on education from regional to federal government. The same trend occurs for regional agricultural expenditure. This is true when Addis Ababa is excluded from the analysis as well. 85. Both construction and agriculture account for 20 percent of the pro-poor expenditure when Addis Ababa is included, while construction expenditure share falls to 5 percent and agriculture rises to 25 percent when Addis Ababa is excluded. This implies that most of the public expenditure on construction occurs in Addis Ababa while most of public expenditure on agriculture occurs in the other regions. The limited expenditure on construction in the regions (excluding Addis Ababa) implies that more resources need to be directed towards these regional governments to spend on regional road construction projects. Most of the road expenditure is on highways and not on the arterial roads linking the regions to the main highways. 86. Health accounts for the least share of pro-poor spending at 10 percent. Moreover, the proportion of expenditure on health appears is relatively static over when comparing FY03, FY06 and FY09. This hold true when Addis Ababa is excluded except that the proportion of health expenditure rises to 15 percent. Regional expenditure on health remains inadequate in the regions. 87. Even though health accounts for the smallest share of pro-poor spending, increased decentralization in the allocation of expenditure for health has occurred between FY03 and FY09. While the magnitude of health budget has been low compared to other sector budget, the recent strategy to increase basic health service converge on one hand and increased availability of resource from various external sources (PBS component 2, global fund, HIV fund, etc) I on the other resulted in unprecedented increase in health budget at all administration levels. Figure 0.8: Share Regional Pro-poor Spending from the General Government Source: MOFED 88. Figure 3.8 demonstrates that from FY03 to FY09, the share of regional recurrent expenditure in agriculture and natural resource and health sectors to total recurrent expenditure for these sectors has increased by 8 and 4 percent, respectively. The share of recurrent expenditure to total recurrent expenditure for roads and education declined over the same period. 89. Figure 3.8 confirms that the share of pro-poor capital expenditure to total capital expenditure falls from 80 to 75 percent when Addis is excluded and 83 to 78 percent when Addis is included. This could be attributed to the increasing competition for resource from urban development and housing programme. 90. Construction accounts for 60 percent of total capital expenditure among the pro-poor sectors for the years FY03, FY06 and FY09. The proportion of capital expenditure on agriculture is relatively volatile with the figure falling from 20 percent in FY06 to 15 percent in FY09 while the proportion of capital expenditure on health and education rises marginally when comparing these years. 91. Excluding Addis Ababa, the proportion of capital expenditure on construction and agriculture are relatively equal at 30 percent for FY 03 and FY 06 and 25 percent in FY09. In addition, the share of capital expenditure on health rises to 20 percent when Addis Ababa is excluded. 92. The shift in sectoral capital is not uniform. The share of regional capital expenditure in health to total capital expenditure for this sector has declined by 50 percent. Furthermore, regional capital expenditure on agriculture and natural resource as a share of total capital expenditure in this sector registered the greatest decline compared to regional capital expenditure amongst the other pro-poor sectors. There is a decline of 67 percent in agriculture and natural resource expenditure when Addis Ababa is included and excluded. This in part reflects the complimentarity between food security and PSNP with the regular agriculture capital. 1.5 Implication to Medium Term Regional Fiscal Health 93. The rising share of regional deficit to total deficit is placing increasing strain on the national budget. This is as a consequence of growth in expenditure outpacing revenue growth from FY03-FY09. Hence regional governments continue to be dependent on Federal grants and transfers to provide basic services to their respective constituencies. 94. While the share of regional revenue to general government revenue declines, the share of regional expenditure to general government expenditure is rising. As the GoE goes ahead with decentralization of the fescues, the regional governments continue to be unable to maintain fiscal rectitude. This situation is likely to continue into the medium term. The dependence of regional governments on central government hand-outs will mitigate their ability to control the direction of fiscal expenditure according to their respective needs. 95. Although both revenue per capita and expenditure per capita has increased approximately three fold from FY03 to FY09. Since the base is much higher for per capita expenditure in FY03 vis-à-vis revenue per capita, per capita deficit has risen approximately three fold. 96. The inability of regions to generate sufficient revenue to meet there expenditure needs will continue to be an issue going forward. However, going forward, if there is acceleration in domestic revenue generation similar to that which occurred between FY07-FY09, the regional fiscus balance will improve as long as expenditure is kept in check. 97. Regional fiscal health is improving steadily with improvement in non-tax, direct tax and indirect tax revenue in FY09 from FY03. Furthermore, service sales taxes are contributing to rising indirect taxes while withholding income tax on imports contributed to rising direct taxes. 98. Non- tax revenue is contributing increasing amount to regional revenue while tax revenue contributes less to overall regional revenue. It is a positive development that non-tax revenue is rising but regional governments will need increase its efforts in improving collection of tax revenue. 99. The share of pro-poor regional expenditure, recurrent expenditure and capital expenditure has remained fairly stable from FY03. Whether the status quo remains depends not on the planned expenditure changes in PASDEP II or the MEFF, but the ability of the regional governments to implement the fiscal plans. 100. The systemic inconsistency between plans (in the MEFF) and budget for the federal government imply that predicting the fiscal position of regional (as well as general) government is an arduous task. (See Figure 3.9 below) Figure 0.9: Projections of the Resource Envelope and Fiscal Expenditure for General Government D R T E MEFF MEFF MEFF MEFF MEFF MEFF MEFF MEFF BUDGET BUDGET ACTUAL ACTUAL Source: MOFED 1.6 Major Achievements and Challenges of Decentralized Service Delivery 101. Fiscal Federalism in Ethiopia has started 15 years ago. There are positive impacts that have been achieved related to service delivery, local level empowerment, accountability and other fruits of decentralization. However there are also a lot of short-comings and challenges that should be removed for better results. This section will list the major achievements and challenges faced by this young fiscal decentralization system. Progress 102. Decentralization has significant contribution for improving the delivery of services at sub national levels. Prior to decentralization, there was no fiscal transfer to Woredas and the meager resources were remaining at the zonal level. As a result of decentralized arrangement and other capacity building measures, basic services, under the mandate of the Woreda government, such as education, health, water supply, rural roads and agricultural extension and natural development and protection activities are performed in a better way at local levels. Distribution of public sector responsibilities by expenditure shows on how far financing and service delivery actions improved following the implementation of decentralized system of arrangement. Accordingly, from FY 06- FY 08, 62.5 percent of health, 55.4 percent of education, 93.2 percent of urban development expenditure is committed at the sub-national level. At the same time 31.6 percent, 42.7 percent and 14 percent is committed at woreda level respectively. 103. The Gross Enrolment Rates (GER) in Primary Schools (first cycle (1-4)) has increased significantly. GER was 22 percent for boys and 16 percent for girls in FY 91 while it is 97.2 percent for male and 85 percent for female in FY 07. However, there are wide Regional variations for women’s enrolment rate in emerging Regional States and pastoralist areas which is still quit low. The national gender gap at primary level was 20.2 percent in FY 08 and 12.9 percent in FY 07 none the less this gap is still a cause for concern 104. According to the latest (2006/07) Annual Progress Report (APR) on the second year of PASDEP implementation (MOFED, 2007), access to clean water supply at national level increased from 19.1 percent in FY 96 to 52.5 percent in FY 07 (i.e., 82.0 percent for urban and 46.4 percent for rural). However the target has been to reach coverage of 85 percent by the end of PASDEP period (2010). 105. Access to health services has reached 89 percent in FY 07 from 33 percent in FY 97; Over 10,500 new nurses, technicians, and front line health workers have been trained; over 1,900 new health posts and centers have been built; and the proportion of the population living within less than 10 km away from a health post has increased from 51 percent to 64 percent. The Health Extension Program particularly made impressive progress in reaching the rural area. By FY 07, the number of health extension workers reached 17, 653 (about 60 percent of the targeted 30, 000), while progress in the construction of health posts (HPs) has been moderate. 106. In general, decentralization has provided autonomy for decision making, improved access of services through construction of more facilities despite quality shortcomings. These were possible due to the shift of financial resources to the local level, devolution of administrative power, strategic actions for meeting MDGs, mobilization of people, and shift to local financing approach as well as approaches for outreach effort & package introduction. Therefore, improved service delivery is attributed to the aforementioned factors. Challenges a) Local revenue base and entitlement 107. The system of providing a highly elaborated fiscal transfer mechanism (Federal Government Dominance of Inter-governmental Expenditure) in order to promote development and fiscal equalization is creating perverse and conflicting interest for state governments, and in more aggressive way, for Woreda Administrations as they do not have incentive to depend on their own resource. This resulted that the local governments to be more accountable to the higher level of government than the tax payers’ citizens. Local governments (Woredas) in Ethiopia, like many developing countries, are assigned to certain tax bases. But own revenue including external funds is offset in the allocation process. Hence, significant revenue autonomy and some tax effort incentives are critical, in particular at the local (woreda) level, to encourage downward accountability and increase the efficiency of local government operation. Such conditions, however, would not be effective without changing the behavior of financial allocation and utilization. 108. Autonomy to raise and utilize own revenue at local level can be improved by combining different mechanisms, i.e. organizing the mandates of local governments for revenue raising and utilization, conducting revenue base assessment to broaden the base and introducing performance mechanisms against matching funds- in doing so it is necessary to put in place accounting system to measure local revenue mobilization, compare the actual revenue collected against the agreed plan- provide matching fund as performance incentive. 109. Autonomy can be embedded in the transfer formula. It is necessary to provide autonomy of local governments to plan, raise and utilize their own revenues or partial off it instead of appropriating it in a balanced way together with the entitlement of the transfer formula. It would be necessary to build the capacity for assessing the revenue base at all level of the government tiers. The higher the revenue raised at a particular locality, the more the matching fund it would get. The pool of specific purpose grant requires to be increased overtime and the proportion could be used as matching fund to additional local revenue generated. b) /RZ ORFDO UHYHQXH DQG WUDQVIHU IRU SXEOLF LQYHVWPHQWV 110. Although the needs of Woredas have been supported directly and indirectly through resources transferred from the States and Federal Governments, financial resource for capital expenditure is low in most Woredas and is highly affected by horizontal imbalances. This is so because some of the Woredas and at the same time other sub -national governments have also to cope with fulfilling (in most cases short of ) their wage and recurrent requirements and the growing future capacity needs from the same source than allocating fund to capital expenditure. The push behind this fact is the dominance of general purpose grant (although Local Investment Grant (LIG) is exercised at a pilot level) which focuses on equalization than sector wide performance based grant. The relationship between the general purpose grant, local government structure, and salary against recurrent expenditure is not well studied. 111. In order to resolve vertical imbalances between federal government and regional governments as well as regional government and local government, it is necessary to analyze the proportion of different types of grants (specific and general purpose grants), the revenue and a series of expenditure data of the different tiers which at present is not fully available for local governments. c) 'HWHULRUDWLQJ LQ TXDOLW\ RI VHUYLFHV 112. Achievements in improving access to most of the services have not been accompanied by adequate improvement in quality. There is a need for improving the quality of services at the decentralized level. This requires designing to a more flexible and location specific approach of service delivery, increase in non-salary recurrent expenses, and strengthening the M&E system including effective monitoring of performance of local governments. Apart from mechanisms for overseeing local government functioning, monitoring of the expenditure and revenue performance of local government could be improved by improving the system in place as well as commissioning independent hired monitoring body. d) :HDN GHFHQWUDOL]HG FDSDFLW\ IRU ORFDO UHYHQXH PRELOL]DWLRQ 113. Many Regional States in Ethiopia lack adequate institutional and skilled personnel to collect revenue and also delivery services. The problem is aggravated in less developed and border area States as well as at the local (Woreda) government level. This situation has meant that the Federal Government will continue to dominate expenditure and revenue management, particularly in the revenue collection regime. It is, therefore, necessary to initiate capacity building programs and policies to enable Regional States to collect more revenue and mobilize resources for local economic development and expenditure assignment. e) /RFDO JRYHUQPHQW H[SHQGLWXUH UHVSRQVLELOLW\ 114. Expanded responsibilities and mandates for sub-national governments, in particular at the local (Woreda) level, for services that have been previously implemented by zones or Regions require that they should be endowed with discretion and capacity to procure services. But in most cases this capacity is not available at the sub-national level and is severe at the Woreda level. In order to attain MDGs, some activities (like procurement) related to expenditure responsibilities should be delegated temporally to Regions and Zones and also local governments are required to put matching targets. The functional assignment of woreda has to state the ways delegation could be provided to regional state on temporary basis for specific functions or urgent matters beyond the capacity of the woreda. Duplication of assignments could lead to dependency, and avoiding responsibility. f) %RUURZLQJ 115. In many federal states, sub-national governments are allowed to borrow from a variety of sources, including international and foreign sources, commercial banks and other financial institutions, mainly for capital projects. However, in the Ethiopian case, borrowing is limited to internal sources only. Borrowing by Regional governments is made against future grants. As per the law, Regional States are allowed to finance some of their activities through borrowing from the central government (article 35 of Proclamation No. 7/1992). Although the action of the government is needed, based on the rationalizing of macroeconomic stability considerations, borrowing should not be a means for overheating national borrowing and drastic backtracking. g) &RPPXQLW\ IXQGLQJ 116. There is low level of counterpart matching fund or community (sub-district) level revenue for carrying out projects at facility level. The fund available at Woreda level couldn’t sufficiently allow exercising decision making on priorities by the community and constituencies. The priority setting procedures at the Woreda level has to change and should incorporate pro- poor community development based priority setting procedures. As a result of shortage of fund, sections of the population at the grass roots could be pressurized to contribute funds. There is an attempt to improve the service delivery functions of Kebeles through institutional structure and training but requires a lot of capacity building efforts for a fruitful results. The recent BPR has not changed the pattern of structure in relation to preference shift to sector based recurrent allocations than community priorities. The pool for specific purpose grant has to be increased and constitutionalized. h) 7UDQVIHU IRUPXOD 117. The Federal transfer formula becomes more theoretical and conceptual. The transparence among the decision making people has decreased. The Region-Woreda transfer is still not improved. The institutional memory on mechanism of fiscal transfer and the clarity of financial pool determination is another debatable issue. 118. The basics of the transfer formula were replaced twice throughout the implementation periods. The different proposals gave provision to refine the formula every three years as explored from different countries experience. However, transfer formula were adjusted and changed at least every two years. The reasons behind it is associated with the quality of data, absence of sustainable independent body, lack of experienced and knowledgeable staff, lack of understanding respective bodies on the role of fiscal transfer improvement every year. 119. Initially, MOFED prepared the transfer formula methodology for approval by the house of federation. Later the responsibility is transferred to the house of federation and approved by the same. By preparing by its own or consultants. Internationally, there are three models of authorities which could be responsible for recommending the distribution of grant between sub- national governments, i.e. the separate authority model, the joint authority model, the finance or planning agency model. The separate authority model is independent and could take the work permanently and continuously. The joint authority model is useful for cross check of separated allocations but lacks system for the overall equalization process. STATE OF PUBLIC FINANCE IN REGIONAL STATES: A CASE STUDY OF AMAHARA AND BENSHANGUL- GUMIZ REGIONS 1.1 Amhara Regional State i. Socioeconomic Profile 120. With a population of 17.6 million in FY08, Amhara is the second most populated region in Ethiopia. Agriculture is the main stay of its economy, contributing 61 percent of the regional gross domestic product (RGDP) followed by services and industry sectors, with a respective share of 23 and 15 percent respectively. 121. Poverty is pervasive in the Amhara region and higher than the national average. According to MoFED FY10 report, in FY05 about 41.8 percent of the population in the Amhara region were below the nationally defined poverty line (42.9 percent for rural and 31.1 percent for the urban) compared to 38.7 percent for the entire country.. The food poverty situation in the region is critical, with nearly 3 million people chronically food deficient and dependent on food aid annually. About one-third of the population can meet their food requirement only for six months and rely on food aid for the rest of the year (BoFED, 2006). Besides, in years of natural disasters as drought, nearly one-half of the population in the region fails to earn a weekly income of Birr 100 or more. This shows that the weak resilience capacity of the Amhara region. ii. The Institutional Context 122. Amhara region is comprised of 10 zones and 150 woredas. There are three tiers of governments-- regional, woreda and kebele-- with their respective legislative, executive and judiciary organs. The region also has zonal administration serving as a bridge between the region and woreda. Since FY05, the region has been allocating block grants to woredas through a transparent and objective criterion. The woreda block grants are separately allocated for recurrent and capital budgets using different formula. In both the formulas the five pro-poor sectors are taken into account. Since the last three years, the following ‘four’ principles have used to guide block grant allocation: (i) the expenditure assignments; (ii) revenue collection ability; (iii) additionality of overall resources; and (iv)fairness measure. The information used for estimating block grant allocation is collected for all the woredas. 123. The fiscal planning process in Amhara starts at the sectoral level in consultation with woreda and other stakeholders. The draft report is consolidated by BoFED and submitted to the Regional Cabinet. The Regional Cabinet after discussing the budget with the various stakeholders submits it to the Regional Council for approval. The same process is undertaken at the woreda level. What makes the Woreda budget preparation different from Region is that the latter incorporates Kebele level plan and conduct community level consultations. 124. The analysis of the budget planning process in Amhara led to three important findings. First, the national and regional plans seem to be well aligned in road, health and education sectors (excluding community built facilities) due to the presence of sector wide planning, where the latter helps to gauge service quality standards. Second, the plan itself is not very realistic, since it is not constrained by available resources and is rather a wish list of projects. Such practice can be improved by notifying of the size of the FBG before the Region prepares its budget. Third, there is a need for integrating the planning process including resources made available through Special Purpose Grants (SPGs), channel 2 and 3, and contribution by the community. iii. Regional Resource Envelop Figure 0.1: Fiscal Frame Work of Amhara Region in FY 09 in % of Total 125. The Region’s aggregate resource envelop is comprised of federal block grant, regional own revenue, foreign loans and grants, road fund, water fund, Food Security and PSNP transfers from Federal Government, Federal Ministries support directly to the respective sectors in the region (for example, MoH’s health centre construction), off-budget resources like NGOs and community contribution. In FY09, the aggregate resource envelop for Amhara region amounted to 7 Billion Birr, where the largest share was from federal block grants (4.2 Billion Birr). The regional proclaimed budget is, Source: BOFED, ARS however, not comprehensive and do not include resource coming in the form of SPGs, road fund and other transfers. Moreover, channel three resources that come from NGOs into the region and community contribution are not fully recorded. iv. Regional Spending 126. In nominal terms, Amhara’s aggregate regional spending has doubled from about 2 billion Birr in FY06 to about 4.7 billion Birr in FY09, showing an annual average growth rate of 33.7 percent. In per capita terms, spending has increased from Birr 105 to Birr 235. The share of woreda expenditure in total regional expenditure has increased consistently and reached 74 percent in FY09. In the past five years, Amhara region has spent on average nearly 62 percent of its budget on pro-poor sectors, which I slower than the 70 percent average for all the regions (see figure 4.1). The Education sector stands out as the number one priority with a share of 35 percent in the total spending. The second area of priority is agriculture and natural resources with 16 percent share and health receives 9 percent share in the regional spending. The share for roads sector in the total budget remains insignificant at less than 5 percent. 127. The share of spending on the pro-poor sectors at the woreda level has declined consistently in Amhara region from 77 percent in FY05 to 64 percent in FY09. The largest share is accounted by education sector, which has consistently declined from 49 percent to 41 percent between FY06 and FY09. The share of agriculture has also declined throughout the period under consideration. The share of each of the pro-poor sectors in FY09 is below the level it was in FY05. With increasing competition from the general administration. Figure 0.2: Pro-Poor Sectors Spending in Total Budget (in %): Amhara vs. All Regions Source: BOFED, Amhara Regional State 128. In Amhara region, the share of capital spending in total regional spending has been remained at less than 20 percent. Most of the resources are being spent on financing recurrent expenses, primarily wage. Operating budget is also very much constrained below 30 percent at the regional level and less than 15% at the Woreda level. 129. At Woreda level, not only is the amount of resources spent on capital and operating costs declining, but Woredas are increasingly finding it difficult in the last two years to cover their salary costs. The main reasons for this includes; (i) the effect of the new population Census on the block grant allocation to Amhara; (ii) the rapid expansion of services particularly in agriculture, education and health; (iii) the increase in posts and/or processes as a result of the business process reengineering (BPR) carried out both at the regional and woreda levels; and (iv) the assignment of additional staff by the regional bureaus without concomitant increase in the budgetary resources. As a result of these, three of the four Woredas visited have doubts on their ability to cover their salary expenditures this fiscal year (FY10). Some have even borrowed from the regional administration to finance their last year’s recurrent expenditures. 130. Budget utilization in the Amhara region has been above national average, while aid predictability and utilization remains low From Birr 125 million project loan and grant included in FY05 to FY09 budget, less than half (53 million) has been disbursed. While project loan and assistance financed through budget remains low (at less than 5%), modest progress has been recorded in utilization. v. Local revenue Mobilization 131. The regional revenue in Amhara has grown from 424 million Birr in FY05 to 853 million Birr in FY10. On the average, 83.4 percent of this revenue has been collected by the Woredas and Bahir Dar town, while the remaining has been collected by the regional bureaus. In per capita terms, regional revenue increased from 17 to 34 Birr. Yet, with Birr 235 spending per capita in FY 09, the region managed to finance only 12 percent of its spending from its own source—which is one of the lowest rate in the country. Figure 0.3: Per capita tax revenue in Amhara and country level Source: BOFED, Amhara Regional State 132. The structure of revenue in the region is dominated by direct taxes. Its contribution to the total revenue collected rose from 50 percent in FY05 to 69 percent in FY09. The Woredas do not have the authority to set taxes in their own localities. Their mandate is to assess potential sources of revenue and then to recommend to the zone to consider the introduction or the revision of existing taxes. The decision to introduce new taxes and revise existing tax rates remains at the regional level. In spite of this, some Woredas do assess and forward proposal for changes and the regions do endorse such proposals as has been observed in Libo Kemkem and Dera woredas. 133. The tax collection at the regional level does not show any buoyancy despite rapid economic growth in recent years the revenue collected from agricultural activities and small businesses has not been growing as expected. There is no database for tax payers and the tax identification number is not known to the taxpayer. It is also difficult to ascertain that the tax is being paid as per the income generated. It is the tax payer that has the responsibility for presenting the financial statement on which taxes are assigned the implementation of the tax reform strengthening program has not been adequate. Although the Benishangul process for revenue collection has been decided in consultation between the federal and regional authorities, its capacity strengthening efforts does seem to be limited only to federal level. There is a need to strengthen implementation capacity (skills up-grading, ensuring protection and security, rewarding employees for better performance) at regional and woreda level similar to that of the federal level. Box 0.1: Rural Land and Agriculture Income Tax in Amhara In spite of the fact that the agriculture sector has grown rapidly over the last few years, the agriculture income tax rates remain very low and are based on land size and not production. In the word of one of the respondents, ‘the farmers were paying in real terms only 25 kgs of teff before and pretty much the same amount even after the revised tax’. It is reported that the tax rate per annum for a grind mill is only 90 Birr, which is considered very low given the income it generates. This also applies for mini buses that pay a tax of 170 Birr per year. The highest annual tax to be collected from a rich household in food secured area is 170 Birr per year while this is 140 Birr in the food insecure woredas. Table 0.1: Revised Agricultural Income Tax and Land use tax in Amhara region Size of land Food secure Woreda Food insecure Woredas Land use Agricultural Land Agricultural tax income tax use tax income tax <0.5 hectares 15 25 10 20 0.5-1 hectare 20 35 15 30 1-1.5 hectares 25 50 20 40 1.5-2 hectares 30 70 25 55 2-2.5 hectares 35 95 30 75 2.5-3 hectares 40 130 35 105 Source: ANRS, Proclamation number 161/2001. Rural land tax has also a fairness concerns. The farmers that produce three times a year through irrigation are paying as the same amount of tax as those producing once a year. In some Woredas, farmers are able to employ daily laborers paying 45 Birr a day. In these areas, there are suggestions from the woreda level officials to explore income related tax options for the relatively rich farmers There is no tax so far on incomes generated from vegetables and livestock and to large extent on chat. Further study is required on the cost benefit of collecting tax from the rural economy required. 1.2 BENSHANGUL-GUMIZ REGION i. Socioeconomic Profile 134. The Benishangul Gumuz Regional State is one of the 9 regional states of the Federal Democratic Republic of Ethiopia. The population of the region in FY09 is estimated to be 711,702. Administratively, the region is divided into three zones, 20 woredas and special woreda and 472 kebeles8. 135. Subsistence rain-fed agriculture is the main means of livelihood for more than 92 percent of the region’s population. A significant number of the farmers are not self-sufficient – filling this food gap from hunting and collection wild fruits. The indigenous community of the region use shifting cultivation and use hoe based farming. Only 44.5 percent of the farmers use animal power for traction. 136. The prevalence of poverty in Benshangul-Gumiz region is among the highest in the country. According to the FY05 Household Income and Consumption Expenditure Survey (HICES), the proportion of poor living below the poverty line was estimated to be 44.5 percent, compared to poverty headcount ration at national level at 38.7 percent. The region was second only to Tigray (48.5 percent) in having the highest proportion of poor people. Poverty in the rural areas is more prevalent (45.8 percent) than in the urban areas (34.5 percent). Childhood malnutrition in the region was among the highest in the country (46.8 percent in Benishangul Gumuz compared to 44.1 percent in the country) ii. The Institutional Context 137. Similar to other regions, BG has been using the three variable formula to distribute block grant to woredas. Since FY07, the old three variable formulas have been replaced with a two step process.9 First, the region ensures the regional “non-discretionary” spending, wages and salary of the civil service is fully funded. In the second step, the remaining balance is subject to distribution based a number of criteria including, Woreda expenditure assignments, their ability to collect revenue, and narrowing inter-woreda development gap. 138. Benishangul region has got regional strategic plan based on the nation plan SDPRP. Sectors such as health and education are also benefited from an elaborated sector development programmes. This practice has not trickle down to Woredas as a result the woreda planning is done incrementally and need based. The latter is caused by delay in the notification of woreda block grant. 8 Special Woredas and Woreda within a region who got special status due to ethnic factor. They may exercise greater autonomy compared to an ordinary woreda. 9 See annex for the evolution of block grant distributing formula used at Federal which is also practiced across regions. iii. Regional Resource Envelop 139. Though a comprehensive information was not available on the regional resource envelop, the regional strategic plan prepared in FY 05 indicates the following as sources of development financing ; the regions, own revenue, federal subsidy, loan and assistance, community and NGO contribution, and other funds such as the road fund. According to this plan, close to one-fourth of the financing was expected from NGOs and community contribution. On the other hand, actual flow from the on budget financings including the FBG and local revenue was reported with relatively lower local revenue out-turn compared to the FBG. On the other hand, the region benefited relatively larger project loan and grant than anticipated earlier. Table 0.2: Planned and Realized Resource between FY05 and FY09 In Benshangul Gumuz Region (in Mill Birr) Act in Plan Act percent Plan Reg. revenue 223 149 66.8 Fed subsidy 1727 1202 69.6 Loan and Ass. 102 110 107.8 Comm. Con 110 0 NGOs 626 0 Road Fund 0 33.2 ERTTP 0 11 Total 2815.8 1532 54.4 Note: Data on actual resource available from the community and NGOs not known. Source: BG, Finance and Economic Development Bureau. iv. Regional Spending 140. In nominal terms, aggregate regional spending in Benshangul-Gumuz has increased from Birr 192.7 in FY05 to Birr 335.1 million in FY09, showing an annual average growth rate of 15.3 percent. In per capita terms, spending has increased from Birr 324 to Birr 511 in five years. Out of the total spending, the share of capital spending has been low and declining -- around15 percent in FY09. The decline in the share of capital expenditure has been much higher at the woreda level. Between FY05 and FY09, the share of capital spending in total woreda spending has fallen from 16.7 percent to 3 percent. The fall in capital expenditure seems to be linked to the fall in FBG in FY09 (from Birr 290.3 mill in FY00 to Birr 271.5 mill in FY09). Devolution of expenditure assignment to woredas has continued to deepen, albeit at a modest pace. Between FY05 and FY09, the share of woreda spending out of total regional spending has increased from 45.6 percent to 51.5 percent. This increase has come from the modest increase in recurrent expenditure share. The woreda share in total regional capital spending has declined from 25.8 percent in FY05 to a low of 11.3 percent in FY09. Figure 0.4: Woreda share in total regional spending in Benshangul Gumuz Region Source: BG BoFED and MoFED 141. In terms of spending priorities, on average, nearly 60 percent of the resources have been spent on the five priority sectors (see Table 4.4). Education stands out as the number one priority in terms of spending, with its share averaging 28.1 percent in overall regional budget. The same has been true at woreda level. Regional level spending on education increased from 47.4 million birr in FY05 (24.6 percent) to 103.5 million birr in FY09 (30.9 percent). The amount of budget allocated to education, health and agriculture has consistently increased, while that of water and roads has fluctuated. The pro-poor sectors in general had accounted for 56.5 percent of total spending in FY05 and this had increased to 69 percent in FY06. But since then the share has steadily declined to 59.4 percent in FY09. Table 0.3: Sectoral composition of spending in Benshangul Gumuz Region (in percent) FY FY FY FY FY Sectors 05 06 07 08 09 Education 24.6 23.4 31.4 30.4 30.9 Health 10.8 10.6 10.6 10.1 8.3 Agriculture & natural resources 11.8 11.5 10.2 9.4 10.0 Water 2.5 5.3 5.6 3.0 1.8 Road 6.9 18.4 8.3 1.3 6.5 All poverty sectors 56.5 69.0 66.1 61.2 59.4 Source: BG BoFED 142. In per capita terms, spending on pro-poor sectors has increased from 183 Birr in FY05 to 294 Birr in FY09. The same trend can be observed in terms of per capita recurrent expenditure. Per capita capital spending has, however, declined since FY07. Per capita spending in Benshangul Gumuz has been below the national average, both in terms of aggregate spending as well as for capital spending. However recurrent spending on pro-poor sectors in per capita terms in Benshangul Gumuz is higher than the national average (see Figure4.5). This implies that regions are being progressively assigned to manage recurrent expenditure, while greater share of capital expenditure is being gradually implemented by the Federal government. Figure 0.5: Per capita Spending on Basic Figure 0.6: The Share of Operating Budget Services in Benshangul Region in Recurrent Budget in percent Source: BOFED, Benishangul Gumuz Regional State 143. The share of operational expenditure in the total recurrent expenditure has been declining both at the regional and woreda level, though there was a small reversal in this trend in FY09. While the share at regional bureau level was 47.4 percent, it was only 15.0 percent at the woreda level in FY09. v. Local Revenue Effort 144. Between FY05 and FY09, the Benshangul Gumuz region collected Birr 177 million and this represents 92.69 percent of the planned revenue. The per capita revenue collection in the region in FY09 was Birr 71 (which is twice that of Amhara). Tax revenue is the largest contributor to the total regional revenue. Tax revenue collected from FY05 –FY09 accounted for more than 70 percent of the total revenue. The share of non-tax revenue collected from FY05 - FY09 was less than 30 percent, but it was 46 percent in FY05. Significant share of the region’s tax revenue comes from salaries and wages. Between FY05 and FY09 the contribution of salary and wage to tax revenue increased from 59 to 75 percent. In contrast, agricultural income tax, rural land use fee and business profit tax accounted for 11 and 5 percent of tax revenue. Table 0.4: Contribution of Different Type of Taxes in Benshangul Gumuz All BG regions Tot Tot Type of Tax Tax Rev. Tax Rev. Salary and Wage 75 55 38 27 Agriculture and Land use Tax 11 6 Business Profit 5 4 18 13 vi. Lesson from the Regional Case Studies 145. While regions reported impressive progress in expanding basic services, the next challenge shall be improving the quality and reaching out the vulnerable groups. This again requires a more careful planning and allocation of resource to ensure efficiency in resource use. And the regional case studies underline that the choice of allocation between sectors and within sector between economic uses should be region specific. Official sources from the regions indicate notable progress in basic service coverage across a range of basic services including education and health. Progress in expanding primary education overwhelming however all regions suffer from poor quality of service as well as distributional concerns. For example Amhara region yet to narrow gender gap while BG to rise the enrollment of native population. At the same time, fast growth in primary enrollment created unprecedented demand for secondary schools which seems lagging behind. In expanding health service facilities including HP and HCs, all regions made notable progress however, low staffing and operating budget undermine service delivery. Above all, regions all lag in expanding clean water facilities compared to the medium term target. 146. So far, the policy-making, planning and budgeting processes are not well connected at the sub-national level. This is seen from the mismatch between the resource required to fully implement strategic plans and what is available on an annual basis. Policy making and planning are not constrained by budget realities. The regional governments need to make a realistic forecast of their medium term resource envelops. Preparing regional level Medium –Term Expenditure Framework (MTEF) might assist in addressing this problem. Although the regional governments have made commendable achievement in expanding basic services during the last five year period (FY05-FY09), they are not be able to achieve most of the targets set in the regional five year strategic plan. 147. Important step towards better resource allocation requires a comprehensive knowledge of the resource available from various sources. All regions seem to have systemic problem in estimating resources coming outside of the budget including from NGOs and the community. As shown on figure 4.7, both off budget and community contribution becoming an important sources of financing Therefore, creating a system that can comprehensively record the community and NGOs contributions is among the priorities of regional and woreda administrations. Source: BOFED, Amhara Regional state 148. Local revenue mobilization remains a challenge for all regions. Urgent actions therefore needed including better documentation of the regional revenue base, review the existing tax policy in light of efficiency and equity, enhance institutional capability, and devising an incentive for woredas improve their revenue effort. EXPLAINING VARIATION IN BASIC SERVICE DELIVERY ACROSS WOREDAS Objective and Context 149. The objective of this chapter is to understand the factors that explain the wide variation in service delivery across Woredas (third-tier of government) in Ethiopia. This issue is particularly interesting since many observers allege that Ethiopia’s decentralization initiative is top-driven and too centralized. As far as we know this is the first ever attempt to link public finance and institutional reforms of sub-national governments to the level of basic service delivery in a systematic manner. While the initial idea was to look at a gamut of services, because of data constraints, the chapter is limited to exploring the determinants of basic education service delivery. 150. In principle, decentralization is believed to make governments more efficient, responsive and accountable leading to effective and efficient service delivery. Ethiopia’s district (woreda) level decentralization was first launched in 2003/04 and has registered impressive progress in expanding public services with understandable variation between regions and woredas..Is this variation a result of the disparity in resources available, or do policies and the mode of institutional arrangement influence the choice, coverage and efficiency of service delivery? It is this question that is at the heart of our analysis here. 151. In this chapter we analyze a number of hypotheses. To what extent geographical location affect access to basic services? Does the level of public resources deployed explain the variation in service delivery across Woredas? What about the composition of spending between recurrent and capital as well as between wage and operating expense? Does institutional quality including the quality of public financial management (PFM) system and capacity in woredas influence service delivery? By seeking answers to this question, this chapter attempts to identify the relative importance of factors in basic service delivery. The results can be used for cross- regional and cross-woreda learning—currently missing in public policy and discourse—as well as to help the federal government to come up with initiatives to scale-up good practices from high performing woredas to the rest of the country. 152. The analysis is based on the result of the second Woreda and City Benchmarking Survey (WCBS II). The survey was initiated primarily as a monitoring and evaluation tool for the Public Sector Capacity Building Program Support (PSCAP) Project. The main purpose of the survey was to establish a wider information base and benchmarks for monitoring changes in institutional capacity, financial autonomy, service delivery and accountability of woredas and cities over time. The survey consists of supply and demand modules to enable sustainable monitoring of the impact of development projects and programs 153. In the past five years, there have been two rounds of WCBS (in FY05 and FY08) conducted with varying sample size and scope. Thus, the results from the two rounds are not fully comparable. Therefore, the quantitative analysis in this chapter is limited to the second round WCBS only10. The second round covered 291 local jurisdictions (240 woredas and 51 city administration). Data sources for the survey are local governments and data extracted from records related to finance, staffing, revenues, budgets and expenditures. Different documents were also reviewed as source of secondary data. 154. While the WCBS provides wide coverage of institutional information to monitor the decentralization process, the database suffers from many limitations. It focuses on collecting a vast amount of data on inputs, but not enough on outputs and outcomes. For example, on basic education service delivery, the only information available is the number of students enrolled in public schools. It does not have information on the total number of children by age and hence it is not possible to calculate enrollment rate by woredas. We therefore complemented the WCBS data with the population census data to generate enrollment rate. In absence of data on private schools, the enrollment data used here is inadequate11. In the health sector, WCBS has information on the number of health facilities at the woreda level, but not their utilization rate. In this case, standard service coverage per facility set by the Ministry of Health has been used to estimate health service coverage. 155. While WCBS has inadequate information on the outcome side, it has perhaps excessive information on the input side. We therefore had to choose a set of input indicators for our analysis based on their relevance and coverage. The study is thus based on the following four sets of variables: : Outcome variable: gross enrollment in government run primary schools and primary health service coverage. Fiscal policy variables: per capita spending at the aggregate and sectoral level, composition of spending (recurrent versus capital and salary versus non-salary recurrent), and local revenue effort. Variables measuring the quality of institutions: Quality of PFM system; local capacity, and community participation. Woreda specific control variables: urbanization, remoteness degree of food security/insecurity (see for the detail list of information see Annex 7). 156. The WCBS the survey collected data on 18 indicators on financial autonomy, institutional capacity, service delivery, participation and transparency. Fourteen of these indicators have been aggregated into 5 thematic areas which latter are used as proxies for fiscal policy and institutional quality indicators. All indicators have been given equal weight. Six jurisdictions were dropped from the sample for lack of full information, leaving us with a sample of 235 woredas and 50 urban municipalities. 157. The chapter is structured in four sections. The second section explores some of the recent empirical works on decentralization and measuring the performance of public service delivery. The third section assesses variation in education and health service delivery across woredas in light of the level of spending, the quality institutions and capability (measured by an 10 The first round WBMs one was conducted in 2005 covering 80 local governments (48 woredas and 32 city administration) 11 For our analysis of public service delivery the public school enrolment is sufficient and consistent with input data. index) and other structural (geographical) factors. As part of the bi-variate analysis, a single input-output efficiency analysis for primary education service in one region ( SNNPR) has been attempted and the result discussed. This includes constructing an efficiency frontier and generating of input and output efficiency scores for individual woredas (detailed methodology and results are presented in annex 7). Section four explores the relative importance of various factors (money, policy, institutions and structural) on the level of service delivery using econometric techniques. Measuring Public Sector Performance: Literature Review and Cross Country Experience 158. According to the World Development Report (WDR) of 2004, putting people at the center of service provision is key to reach the marginalized and the poor through public service delivery. To this effect, decentralization of public service delivery is considered as a necessary step to redress the limitations of a centralized system. According to Ahmed and Devarajan (2005), public spending alone has limited impact on outcomes. They argue that a number of factors could result in misalignment between functional responsibility and expenditure assignment including the detachment of funds and the frontline service providers, power imbalance between the central and local governments in deciding resource transfer, lack of capacity at sub-national level to managing public resource and incomplete decentralization process. Above all, Devarajan and Ahmed argued for strong accountability between the various actors in service delivery chain including the people, the policy makers and the service providers. 159. Recent empirical works to quantify and measure the relative importance of factors in public service delivery provide important technical and methodological impetus to assessing the effectiveness of decentralized service delivery system. These works concentrate more on cross country scenarios than across jurisdiction within a county and mostly focus on health and education services. 160. Gupta and Verhoeven (2001), for example, estimate cross-country public spending efficiency in education and health for 37 African countries for the 1984-1995 periods. Using Free Deposable Hull (FDH) method for a single input and single output, they found that African countries are less efficient in providing education and health services compared to their peers in Asia and the West. 161. On the role of finance, using parametric approach, Herrera and Pang (2003) construct efficiency frontier for 76 countries. Their main conclusion is that there is no relationship between level of expenditure and education and health outputs. On the other hand, Haque, Persaran and Sharma (1999), using the data of Evans and Tandon (2000), found positive relationship between efficiency of service delivery and the level of spending. 162. While spending matters, its effectiveness is contingent on other institutional factors. A study by Alfonso, Schuknecht and Tanzi (2003) shows that variables such as the quality of administrative functions are important factors in explaining service delivery across countries leading to the conclusion that countries with small public sector exhibit better performance. 163. The work of Herrera and Pang (2003) further sheds light on the role of structural (e.g., geographical location) and institutional quality in explaining cross country variation in service delivery. According to their study, variation in efficiency scores is explained largely by the degree of urbanization and the quality of bureaucracy and not so much by the level of spending. The rationale for greater efficiency in urban areas rests on the assumption that clustering of agents makes it cheaper to provide services in urban than in rural areas. 164. These studies however suffer from some limitations. First, cross country analysis assumes homogeneity in the production technology of the services. This entails the use of a small number of factors of production and the quality of the inputs is the same across countries. Another limitation is the “Belassa Samuelson effect,” which entails that price levels in wealthier states tend to be higher than the poorer countries resulting in positive association between the level of per capita spending and the level of development. In such cases, Gupta and Verhoeven (2001) suggest splitting the sample based on the level of development. Other challenges include unequal disaggregation of input and output data and the use of monetary and non-monetary factors of production as explanatory variables. 165. Important lessons can be drawn from the existing literature. The framework provided by Ahmed and Devarajan underlines the significance of institutional factors in influencing public service delivery. The studies find that the effectiveness of a decentralized public service delivery system depends on better resource allocation, institutional capability, and appropriate incentive to all agents in the service delivery chain. At the same time, the approach, i.e., measurement and methodology used in cross country cases can be applied to our cross woreda analysis with some modifications. Bi-Variate Analysis of Variation in Woreda Service Delivery vii. Woreda Specific Structural Factors 166. Ethiopia has made impressive progress in expanding basic service delivery in the past decade. According to official source, net enrollment in first and second cycle primary education has increased 3 fold between FY95 and FY08 from 24 percent to 78 percent12. Similarly, access to basic health service coverage has increased substantially. Based on the Social Development Indicators of the World Bank, child immunization of Measles has increased from a mere 12 percent in FY85 to 74 percent in FY08. At the same time, DPT immunization grew from 6 to 81 percent.13 The same source also shows that resource going to these sectors has doubled in the past 12 years. As noted in chapter 2, decentralization has been instrumental in managing public resource for the expansion and delivery of basic services. According to PFR 2009, between FY03 and FY08, the share of national education and health sector budget being managed by regions have been delivering primary education and health services are the two main functions of the Woredas. According to the second WBMS, Ethiopia made notable progress in expanding primary education across the entire country. As of FY07, Gross Enrollment Ratio for first and second cycle primary (Grade 1-8) in the WCBS sample woredas averaged 66 percent14. The distribution of primary education is more equitable with 50 percent of the woredas having 12 Social Development Indicators, the Wold Bank, 2009. 13 See annex for detail social indicators. 14 Primary school enrolment from WCBS is less than what is reported at national level recently. This could be a result of lag, the sample composition of the WCBS were about 70 of the woredas are rural woredas, and the exclusion of private schools. reached a GRE rate of 70 percent. Yet, there are woredas that are lagging behind– with less than 30 percent GRE (e.g. Awara and Filtu). Figure 0.1: Basic Educations and Health Coverage by the Degree of Urbanization Source: Based on 2nd WBMS 167. The evidence shows that the urban–rural gap in access to education is narrowing. GER in urban woredas with urban population of 90 percent and above average 66 percent while the ratio for rural woredas (with more than 50 percent rural population) stood at 62 percent. The graph below, and a simple correlation between GRE and remoteness confirm the narrowing gap albeit decline in GER as we move away from asphalt road. Nevertheless, there remain two extremes of woredas – those whose enrollment rate exceeding 100 percent and those who lag behind in expanding basic education. Among the obvious ones are pastoralist areas with an average of 30 percent or less GER. Figure 0.2: Comparison of Woreda GER (grade 1-8) by Remoteness (Remoteness measured by distance of a Woreda from asphalt road in KM) Source: Based on 2nd WBMS 168. While there is wide variation across individual woredas in access to primary health service, difference between the average coverage of urban and rural woredas was found modest - 86 and 84 percent respectively15. However, the relatively narrow gap between rural and urban woredas in part is a reflection of the exclusion of private service providers and referral hospitals. The exclusion of Hospitals is on the ground that they provide referral service across woredas and their resource is managed by regional offices. 169. As per the WCBS, private health service providers were reported in only 34 woredas out of the 292 woredas covered. The regional distribution of private health service providers is skewed – more than 70 percent are found in Addis Ababa, Oromia and Amhara of which Oromia alone accounts for more than one- third of the operators. Most of these woredas are located in urban setting. In these woredas private service coverage average 28 percent with wide variation ranging from less than 10 percent to 90 percent. Thus, it is not surprising to find that the correlation between health service coverage and urbanization while positive is less than 0.5. 170. The status of food security of a woreda came out as an important explaining factor of variation in average GER across woredas. As shown on table below, average GRE is higher in food secured areas with an average rate of 70 percent followed by food insecure but getting supplementary fund from Specific Purpose Grants (SPGs) most importantly Food Security Fund (FSF) and Productive Safety Net Programme (PSNP)16. Status of food security matters more in the rural areas with food secured woredas showing stark contrast with food in secured ones with 15 Here health service converge show potential converge of existing facilities. Mode has been used instead of mean as a better representative average. 16 The FSF and PSNP programmes transfer close to 3 percent of the GDP each year for targeted vulnerable woredas. The resource primarily used for building household asset through direct transfer and to construct rural infrastructure through food for work and community participated programmes. l a respective average GER ratio of 80 percent and 49 percent. Rural woredas receiving transfer through SPGs are much better from non receiving with average GRE of 56 percent. 171. The food security factor explaining variation in health service coverage across woredas turn out to be stronger in health than in education. As shown on table below, food insured woredas and receiving SPG transfer achieved higher health service coverage than the rest. This in turn could be attributed to positive externality of SPGs via demand and supply of health service. Improving nutritional gap (either through direct transfer of food or income to households) believed has a strong potential of reducing household vulnerability to health related shocks. At the same time, the contribution of SPGs to rural infrastructure expansion including health facilities (through food for work and as matching grant to community contribution) compliment sector specific resource and effort. Figure 0.3: Primary Education and health Service Coverage by Status of Food Security Source: Based on the 2nd WBMS 172. The expansion of Health Posts (HPs) recently believed substantially improved access to health service in rural areas. However, coverage here only shows s potential capacity of facilities than utilization or the quality and delivery of service. More information on service inputs such as availability of health workers is needed to assess the effectiveness of the system. At the same time, private provision of the service is still limited to only 10 percent of the woredas and concentrate in urban areas and around the capital. Thus, subsequent studies need to address broader and deeper issues of service delivery including the comparative advantage of public vs private service delivery. Moreover, the observed wide gap across woredas in health service coverage along the status of food security line needs further research including the potential effect of SPGs on basic service delivery. viii. Public Spending and Service Delivery a) /HYHO RI VSHQGLQJ 173. The WBCS data show a positive correlation between per capita spending in education and primary GER at the woreda level (see figure 5.4). This is not unexpected given the missionary zeal with which GoE has mobilized its population to send kids to school and at the same time hire a large number of teachers to bring down the student-teacher ratio. The below graph also shows the importance of non-monetary factors in explaining access to basic education. There are a number of woredas that are spending more than 100 birr per capita and yet their enrollment rate is often lower than woredas that are spending less than 40 birr per capita. In fact the below graph seems to imply diminishing return to GER sets in once the per capita spending reaches 80 birr. Figure 0.4: Primary School Enrollment and Per Capita Spending on Education Source: Based on 2nd WCBMS 174. Closer look at education spending across woredas reveal huge variation in per capita spending than the variation in service coverage measured in GER in public schools.. While average spending per capita on education stood at Birr 45, about 25 percent of woredas on the average spent Birr 26 only (about 50 percent of the average and just one quarter of high spenders.17 On the other hand, per capita spending tend to be higher in urban than rural and remote woredas. Thus, there is a strong and positive relationship between per capita spending and urbanization (0.51). This seems unexpected because clustering of agents is expected to lower unit cost in urban areas than rural (See Herrera and Pang, 2003). This could be attributed to the relative better staffing of urban schools than rural ones (in the number, qualification, and experience of teachers) resulting in larger wage bill in urban than in rural areas (both in terms of 17 Per capita spending at national level in FY 07 was Birr 115 while that of regions was Birr 58. In this particular FY, tertiary and primary education had equal weight in sharing the education sector budget. salary per-teacher as well as in the aggregate wage bill). In contrast, there is a high possibility that distant woredas suffer from under staffing as well as employing less experienced teachers. This question is beyond the scope of this PFR and deserves further study. 175. As shown on figure below, the relationship between access to basic health service and level of per capita spending on health at woreda level is non-linier and three groups of woredas could be identified based on average and marginal return of spending. The first group of Woredas is those spending less than the national average (Birr 13 per person). For these woredas, additional resource leads to increase in access. With an additional resource of Birr 10 per person, Woreda service coverage increase by 23 percent. While average return remain positive, marginal return increases for woredas spending up to Birr 10 per capita and decline between Birr 10 to13. The second groups of woredas are those who spend between Birr 13 and 20 per person. This group of woredas experience negative average and marginal return to spending. The third categories of woredas are those with the coverage rate in excess of 100 percent and a unit cost of more than 20 Birr per person. These are woredas expected to provide service including to neighboring woredas. In this group, both average and marginal return seems positive. Further in-depth study is required to understand this phenomena specifically what explains variation in average and marginal return across woredas. Figure 0.5: Health service coverage vs per capita health spending Source: Based on 2nd WCBMS b) (IILFLHQF\ RI VSHQGLQJ RQ EDVLF HGXFDWLRQ 176. Preliminary result from public spending efficiency on education across woredas show wider gap in the cost of delivering service than access across woredas. Attempt was made to assess the efficiency of education service delivery based on the WCBS data for one region, Southern Nations and Nationalities Region (SNNPR). The choice of the region is random. Using FDH approach, input and output efficiency scores have been generated (see result on table 5.1 below). Detail of the methodology, assumptions and the efficiency frontier constructed for the region is provided in annex7. 177. The SNNPR region is the third largest in terms of size and population and more- or -less is a representative region in level of development. Average enrollment across woredas in the region is about the level of the national average. However, per capita spending on education in SNNPR woreda on the average is Birr 41- slightly less than the national average of Birr 45. On the other hand, the pattern of serve delivery and the unit cost differ from the national pattern in a number of ways. First, per capita spending is lower in urban and high and in rural woredas. second, food secured spend more than in secured and, food in secured woredas and receiving SPG transfer are better off than the secured ones in average enrollment rate. 178. Using FDH approach for a single input and output, the following woredas prevailed as relatively efficient in basic education expansion falling on the production frontier: Arbaminch, Derashe, Loka, Hulberga, and Meskan. Compared to efficient woredas, the regional average input efficiency score is 0.866 showing an excess use of budgetary resource of 13.4 percent. In other words, SNNPR has a potential to save resource for providing the current level of service (63.4 percent GER). Table 0.1: Result of FDH Analysis of Public Spending Efficiency on Education In SNNPR Woredas Input Output Input Output Woreda Eff. Eff. Eff. Eff. Rank Rank Score Score Arbaminch 1 1 1.000 1.000 1 1 1.000 1.000 Derashe Special woreda Hulbareg woreda 1 1 1.000 1.000 Loka Abaya woreda 1 1 1.000 1.000 Meskan 1 1 1.000 1.000 Kindo Koyisha 2 19 0.993 0.821 DEMBA GOFA 3 15 0.989 0.852 Aletawondo woreda 3 16 0.989 0.840 Misrak Badawacho 4 24 0.985 0.761 Sheko Wereda 5 33 0.969 0.618 Merab Azernet woreda 6 13 0.966 0.889 KUCHA 7 12 0.948 0.905 Shei Bench Woreda 8 23 0.944 0.787 Shebedion 9 25 0.931 0.745 DAMOT PULASA 10 9 0.919 0.925 Alaba Special woreda 11 7 0.913 0.939 Borcha 12 29 0.893 0.728 KONTA 13 34 0.880 0.568 DAMOT SORE 14 5 0.852 0.962 Cheta Woreda 15 11 0.850 0.912 Endegang 16 30 0.846 0.698 Yirga Cheffe 17 6 0.843 0.948 Debub Bench 18 28 0.840 0.729 Kochore 19 32 0.835 0.646 Ezha 20 4 0.824 0.980 ISSERA 21 16 0.823 0.840 Deniboya 22 22 0.813 0.796 Sankura Woreda 23 21 0.809 0.811 Woliqite Municipal 24 20 0.801 0.818 Sodo 25 18 0.790 0.828 Kedida Gamela 26 8 0.787 0.933 MELOKOZA 27 27 0.771 0.740 Dilla town 28 26 0.714 0.744 HUMBO 29 14 0.699 0.859 Arbegone Woreda 30 2 0.694 0.990 Lanfuro Woreda 31 3 0.692 0.981 Yeki Woreda 32 31 0.677 0.651 Amaro 33 10 0.622 0.919 Average 0.866 0.846 Lowest 0.622 0.568 Source: Authors computation based on the 2nd WCBMS 179. Generally, the correlation between input and output efficiency scores is weak. Woredas like Arbegona and Lanfaro rank high based on output efficiency score reflecting the high enrollment achieved compared to others. However, these woredas stood among the least in input score entailing that high expansion is achieved at high cost. Thus, for these woredas to be on the frontier, they should cut currant spending by about a third without affecting output. 180. Another important observation is that, there is an inverse relationship between efficiency scores and remoteness. The more distant the woreda is the less efficiency score they achieve (though this might not always true for all woredas). Therefore caution need to be taken in interpreting the result as distant woredas tend to achieve lower education coverage at high transaction cost due to geographical location. As this analysis is based on single input, it does not take into account the implication of difference in structural nor institutional factors. Moreover, the robustness of the result needs to be verified with updated figures as well as alternative methodologies. ix. Composition of Spending and Service Delivery 181. While per capita spending is important in influencing the level of GER, the quality of expenditure allocation and local revenue generation are expected to have a bearing on the coverage of education services. Theoretically, better intra-sectoral allocation, i.e., the right balance between wages and operating expenses and between recurrent and capital spending, should help improve effectiveness of the education system. 182. Two indicators from WCBS are used to measure the quality of resource allocation at the woreda level. Indicator 2 measures the wage intensity of woreda budget with high wage bill in the total education sector budget is given a lower score than woredas with lower wage intensity (the indicator can take only 4 values: 0,1.2, and 3). Indicator 6 measures the share of capital to total budget. It assumes that if the salary portion of expenditure is very high, leaving limited resources for capital and operational expenditures, it constrains mobility of service providers as well as reduces the readiness of the woreda to meet future service demands. 183. As shown on figure 5.6, access to primary education services is positively associated with greater share of budget going towards salary and/or investment purposes. Average enrollment rate in woredas where the proportion of capital to total budget is the highest stood at 89 percent against 60 and 66 present in woredas with lower share of capital budget. This might not be surprising given that increased enrollment requires more school facilities which require more capital. The inverse relation in case of wage cost also reflects the need for more teachers as enrolment expands. But higher enrollment may have been achieved at the cost of declining quality—an issue that is beyond the scope of this report. 184. As per WCBS, most woredas suffer from high wage and low capital implying less fiscal space to accommodate new comers as well as undermine quality of service. The quality of expenditure allocation index is poor across the board with 78 percent of the jurisdictions graded as low performers as result of allocating almost all the budget for salaries leaving very little for operational and capital expenditures. Only 10 percent of the surveyed are graded as best performers and even among the top performers, only 45 percent of them are also graded best in the quality of their expenditure allocation. The urban jurisdictions perform much better in expenditure allocation than their rural counterparts at all performance levels. Figure 0.6: Average GRE Under Different Budget Mix in % Source: Based on the 2nd WCBMS. Note: The first variable measures salary in percent of recurrent budget, while the second capital to total woreda budget. Strong, average and weak corresponds with the rating of A, B, and C in the WBMS. x. Local Revenue Effort and Service Delivery 185. Theoretically, local revenue effort is expected to positively correlate with access to services. In WCBS, revenue generation capacity indicator is a composite indicator of 3 of variables. Indicator 3 measures the financial self-sufficiency level of a jurisdiction in term s of collection of taxes and fees. Indicator 4 depicts the trend towards improving own revenue and measures the deviation between budget and outturns for collection of taxes, fees and services of two consecutive years (2004 and 2005). Indicator 9 measures the strength of tax administration. 186. As shown in figure 5.7, the strength of local revenue effort is associated with greater access to basic services. As demonstrated by average GER across woredas, those with higher proportion of locally financed budget made more progress in local revenue generation and those with a stronger tax administration are associated with higher enrollment rate than the weaker woredas. Figure 0.7: Average Primary GER Under Strong, Average and Weak Local Revenue Environment in % Source: Based on the 2nd WCBMS. Note: Strong refers to woredas who scored A while weak refers to those ranked c in WBMS. The first index shows the proportion of woreda budget financed from its own source. The second index measures progress in local revenue generation in three consecutive years prior to the survey. The third index measures effort in broadening the tax base and institutional capacity to review the policy and administration of tax in their jurisdiction. 187. Though it is important for improving access and quality of services, local revenue generation remains a weakness for the woreda administration with 93 percent of the jurisdictions graded as low performers. Only 2 percent of the surveyed woredas are graded as best performers and even among the top performers, only 1 percent is also graded best in revenue generation gain the urban jurisdictions performed slightly better than their rural counterparts. xi. Institutional Quality and Capability 188. Though it is difficult to prejudge the relationship between the level of service coverage and the quality of PFM system, the evidence from the WCBS shows positive correlation. Thus, better planning and strong fiduciary system corresponds to higher primary school GER. 189. The variable measuring institutional quality and capacity is a combination of 3 indicators (WCBS indicator 1, 7 and 8). Indicator 1 measures variation between budgeted and actual expenditure. It shows how accurately woredas and municipalities forecast and use their resources. The budget and actual figures are summary of sectoral expenditures and not the total revenue. Indicator 7 measures the presence, transparency and inclusiveness of the strategic plan. If a strategic plan is used and budget is prepared based on this plan, the local jurisdiction is considered to have a high degree of transparency in reporting and as a result is considered as a top performer. Indicator 8 measures the efficiency and comprehensiveness of accounting and auditing procedures. A jurisdiction is graded as top performer if it uses double entry accounting; ICT assisted accounting, regularly reconciles its fiscal and bank records, has no account backlogs and has a system for internal and external auditing of the books. Figure 0.8: Average Primary GER v/s. PFM system Source: The 2nd WCBMS 190. As shown on figure 5.8, enrollment has a positive association with woreda strategic plan (indicator 2) and the quality of the fiduciary system in a woreda. The difference in average GRE across the three layers of woredas is however modest compared to local revenue effort. For example, the difference in the average GER across groups in weak and strong fiduciary system woredas is just 4 percent (63 percent and 67 percent respectively). This probably reflects the progress in rolling out of core reforms under the Expenditure Management and Control Program (EMCP) to most parts of the country. 191. Overall capacity: Capacity indicator is measured in terms of fiscal performance (as reported by WCBS indicator 5, 10 and 11) and human capability. Indicator 5 measures the realization of resource mobilization and spending obligation compared to plan at woreda level. Indicator 10 measures the appropriateness of staffing level by comparing the filled positions against the approved positions of a jurisdiction. The last indicator (indicator 11) measures the compliance with modern human resource approach by checking whether job descriptions and performance measures are issued to employees and if performance appraisals of employees are carried out regularly. 192. The level and quality of public service delivery is expected be positively influenced by institutional capability. As shown in graph below, woredas with better capability tend to have high GRE and vice versa. However, the gap in average GRE between strong and weak woredas is modest. Figure 0.9: Average Primary Enrollment Ratio for Strong, Medium and Weak Woredas (in percent) Source: Based on the 2nd WCBMS 193. Devolution and service delivery: Two indicators of the WCBS where devoted the extent of devolution of basic service delivery to woredas. The first looks in to all basic services while the latter assess agriculture services. In this report we refer to the first, the extent of a woreda providing basic services. Thus, higher or an “ A” ranking for a woreda mean greater devolution of basic service delivery to that particular woreda and lower rank reflect less devolution relative to others. As shown on figure 5.9, higher average enrollment corresponds to grater devolution and vice versa. 194. Community empowerment: The conventional wisdom is that greater community empowerment represents good governance with a strong potential to enhance efficiency in resource allocation and utilization. As per WBMS, empowerment indicator is a composite indicator of 2 of the Woreda and City Benchmarking Survey indicators (WCBS indicator 12 and 13). Indicator 12 measures the public access to basic information such as budgets, audit reports, strategic plans, tax assessment and other services provided by the jurisdiction. Indicator 13 measures the level of consultation with citizens and stakeholders in preparation of budget and plans. 195. The relationship between community empowerment and average GER is shown on graph 5.10. While budget transparency is associated with higher average GER, community participation in the woreda planning process show no clear pattern. Generally, effort to link between primary GER and our empowerment indicator seems inconclusive. The result could be a reflection of poor design, measurement of empowerment and ranking of woredas. Figure 0.10: Average Primary GER by the Figure 0.11: Average Primary GRE by the Degree of Devolution of Basic Services to Degree of Community Empowerment Woredas Source: based on the 2nd WCBMS Results from the Multi-Variate Regression 196. This section explains the relative importance of factors in contributing to the variation in service delivery across woredas. The scope of the analysis however is limited to identifying statistical regularities between the level of enrollment and potential factors across woredas rather than causality. 197. The analysis uses multiple regression analysis with primary GER from 1-8 as the dependent variable. In addition to spending, three set of variables are on the right side including; a) Woreda-specific structural factors including urbanization, remoteness, food security status and four regional dummies for the three large regions and Addis Ababa; b) Policy variables including student/teacher ratio, intra-sectoral allocation (measured by the proportion of capital to total budget and wage in percent of total recurrent), and local revenue effort; c) Institutional variables including index of fiduciary quality, community empowerment, and woreda capacity (human and budget absorption). 198. The regression equation is defined as follows. Ei represents primary school enrollment ratio for woreda “ i” (i= 1…292) the general model used to assess the significance of various factors in explaining variation in woreda service delivery is as follows 18: 18 For spending per capita as dependent, enrolment and health converge appears on the right side. Where: = level of per capita spending in woreda i; = Woreda “i” score in policy variable “j”; j = 1, 2 Woreda “i” score in institutional quality and capability variable “k”; k= 1, 2, ..5 = The location or food security status of of Woreda “i” ; l= 1, 2, ..5 199. The estimate the equation using Ordinary Least Square (OLSQ) technique and undertook robustness tests to correct for the presence of hetroscedasticity, multicollinearity, and specification errors. With the exception of per capita spending case, all variables are in their linear form. High variability of per capita spending for woredas with a relatively lower enrollment rate necessitated logarithmic transformation (see attachment 8 for the detail). 200. Step wise regression with a gradual introduction of a new set of independent variables has been conducted and the results of the seven estimated equations are reported in table 5.1. The first and the second regressions show the relationship between enrollment and the structural variables. The third regression introduces per capita spending as an additional explanatory variable. The fourth and fifth equations introduce policy variables. The six and seventh equations add institutional variables as additional explanatory variables. 201. Empirical evidences broadly converge in the notion that strong institutions positively impact on the efficiency and quality of service delivery. With respect to service coverage however, the directional relationship with institutional variables is ambiguous. Yet, it could be reasonable to assume that fiduciary quality could help save resources from efficient utilization of budgetary resource, while higher capacity help effective utilization of resource. At the same time, greater devolution of service delivery to woredas and better participatory environment expected to positively impact on access to services. Greater devolution is expected to be associated with greater budget utilization, and better opportunity to engage the community and hence mobilize more resource to priority sectors. The regional case studies have shown that the expansion of primary education facilities in most woredas is left for the community. a) (QUROOPHQW DQG 6WUXFWXUDO )DFWRUV The main findings of these are summarized as follows. 202. The first regression which includes urbanization and food insecurity variables as explanatory variables, shows that enrollment rate increases with urbanization, while decreases with food insecurity. The coefficient of urbanization and the constant term are significant at 5 percent level while the coefficient of food insecurity is insignificant. 203. The second regression which includes regional dummies in addition to urbanization and the status of food insecurity shows that regional difference is more important than urbanization and all the regional dummy coefficients turn out to be significant at 1 percent level19. All regional dummies are positive and significant. The relative size of the coefficients show that average primary school enrollment at time of the survey was the highest in Tigray 19 The regression turnout to have hetroscediasticity and multi-collinearity problems. followed by Amhara, SNNPR, and Oromia regions. Note than geographical factors explain only 24 percent of the variation in primary school enrollment across woredas. b) (QUROOPHQW DQG /HYHO RI 6SHQGLQJ 204. The regression results show a positive and statistically significant coefficient for per capita spending on education implying that availability of resource is one of the important factors in determining primary school enrollment at the Woreda level. While the country has made notable progress, Woredas are yet to fulfill the universal education target which require among other things additional resources. More interestingly, the regression has been re-run with the square of per capita spending as an additional variable and the coefficient turn out to be negative. This indicates diminishing return at higher level of per capita spending20. The fact that per capita spending has a negative relationship with service delivery at a higher level of spending is an indication of lower absorptive capacity of the local governments. The response of enrollment to per capita spending (measured by the size of its coefficient) increases with the introduction of institutional variables confirming complementarily of resource and institution. As can be observed in table 5.1, the coefficient of spending increases from 13.7 (with money and structural factors as in regression 2) to 15.1 (with additional policy and institutional factors shown on regression no. 7). c) (QUROOPHQW DQG )LVFDO 3ROLF\ 205. Apart from the level of resource available, the composition of expenditure contributes to improved service delivery in a more balanced fashion. As shown on regression 4 and 5, availability of teachers (measured by higher teacher- student ratio) and higher wage share in the budget are associated with higher enrollment. Particularly, the relationship between the proportion of wage in the budget and enrollment rate is found to be strong and statistically significant21 . Given that no less than 90 percent of Woreda education budget is spent on wages, the result may not be surprising. 206. Equally important, woredas with higher share of capital in their budget are also those who achieved higher enrollment and the result is significant (regression no. 4). While three-fourth of the sample woredas suffer from low capital, any effort to increase capital budget at woreda level is key in reaching universal access to basic education. To this effect, programmes like Local Investment Grant (LIG) can help ease the capital budget. At the same time, the coefficient of capital to budget ratio increases with more institutional variables entailing complementarily between these variables. Though fell short of significance, local revenue effort is also positively associated with enrollment. Currently, there is little variability across woredas as ninety percent of them generate limited revenue compared to their spending demand. 20 A regression run has been run with the sample breaking down in to quartiles (25 percent,50 percent, and 75 percent) and size of the coefficients decline as we go-up to woredas already reached higher level of enrolment. The quintile based regression also shows the importance of spending for lower quintile (Q1) than Woredas who achieved higher enrolment rate. The coefficient of spend under QI, QII, and QIII are 0.102, 0.083, and 0.035 respectively. 21 As noted earlier the variable in the regression measures the inverse of wage to recurrent (higher index value corresponds to lower wage to recurrent) as a result the coefficient for wage turn out to be negative and significant. d) (QUROOPHQW DQG ,QVWLWXWLRQDO )DFWRUV 207. Here the objective is to understand the relationship between enrollment and institutional quality and capability of woredas, specifically, fiduciary quality (the presence of accounting, auditing and reporting system) and woreda capacity (human capacity measuring the planning and executing strength of woredas). 208. As shown on regression no. 5, fiduciary capacity of a woreda is positively associated with higher enrollment and the relationship is statistically significant. However, inclusion of more institutional variables lead to lower coefficient for fiduciary quality in part reflecting the correlation between the rolling out of PFM reform and capacity building initiative under PSCAP programme. Table 0.2: Results of the Regressions of Primary GER on Structural, Level of Spending, Policy, and Institutional Variables Structural variables Budget Policy variables Institutional variable quality of quality of Dummy U Teacher public public quality of F value R Adj R urbaniz food dummy dummy dummy rban AA Log pceds student local finance financ public participati prob Constant ation security Oromia Amhara SNNP DD Tigay pen ratio revenue wage capital finance capacity on ’– ’– ’– ’– ’– ’– ’– 209. Interesting evidence shown in regression equations 6 and 7 is the positive association of the level of enrollment and woreda capacity and community participation indices. Since capacity here is measured in terms of the planning and budget capability of woredas, its impact on enrollment is indirect. Though community participation is found to be positively associated with enrollment, the coefficient remains insignificant. This could be attributed to data problem, poor definition and measurement of community participation variable. While, community empowerment to influence resource allocation is still limited, the community is seen to be active in funding basic services including primary education (see regional case studies). The upcoming WCBS is expected to improve the quality and availability of demand side accountability as well as the community’s role in supporting the supply of service. In alternative specification to capture the relation between enrollment and devolution of service delivery to woredas turns out to be positive and significant. Thus, community engagement including financing services is better facilitated in woredas who assume service provision responsibility. Summary implications 210. The WCBS provided a unique opportunity to look at factors influencing decentralized service delivery in Ethiopia. In spite of data limitations, the review provides important insights to the relative importance of finance and institutional factors in explaining variations in the level of access to primary education and primary health services at woreda level. 211. There is some evidence that supports the view that institutional reforms and capacity building efforts go side by side with devolution of spending to woredas. Woredas with strong institutional quality manage their fiscal resources more effectively. While the implementation of public finance reform seems to be progressing well, most woredas under perform in local revenue generation and in balancing recurrent and capital and wage and operating budget. 212. Strong relation between spending and education and health output underline the importance of additional financing for the woredas to meet the MDGs. However given the diminishing return to per capita spending at higher level of spending, any additional spending should be complemented by measures to increase absorptive capacity at the local level. 213. Local revenue generation and improved intra-sectoral allocation matters in expanding the coverage of basic services. However, more than 90 percent of woredas suffer from weak local revenue mobilization and devote most of their resources to paying wages. 214. Greater devolution of service delivery responsibility to woredas has been associated with higher service coverage. However, the indicator measuring community empowerment is found to be insignificant except for budget transparency variable. 215. Finally, fiduciary quality is important to improve woreda delivered services. The progress in rolling out reform supported by local capacity building need to sustained, 216. Generally the effectiveness of spending increases with better institutional quality as the coefficient progressively increases from single to multiple regressions. 217. As this is the first attempt to understand access to service delivery at the woreda level, there are more questions than answers. Therefore future studies need to get deeper in linking inputs and outcomes including the effect of demand side factors in decentralized service delivery. To this effect improving the quality and scope of data in the future WCBS is important. Moreover, primary data sources such as census and surveys can supplement the qualitative data coming from future WCBSs. References Afonso, A.;Schukencht,L.;and Tanzi,V.(2003). Public sector efficiency: An international comparison, Working paper No.242, European Central Bank Ahmed,J. and Devarajan, S.(2005). Decentralization and service delivery. World Bank policy research paper 3603, World Bank. Amhara National Regional State (ANRS), Regional constitution. ANRS, (2001), Proclamation No. 161/2001 ANRS, (2006). Regional; Development Report. ANRS, Various Regi8onal Offices . ARS, (2001), Proclamation No. 59, 2001. Central Statistical Authority (CSA), 2007, Ethiopian Population and Housing Census. Charnes, A.; Cooper, W.; Rhodes, E. (1978). Measuring the efficiency of decision making units, European Journal of Operational Research 2(6), 429-444 Evans, D.; Tandon, A.; Murray, C. JL.; and Lauer J.A (2000). The comparative efficiency of national health system in producing health: An analysis of 191 countries, World Health Organization GPE discussion paper series No. 29. Federal Democratic Republic of Ethiopia (FDRE), (2004). The Ethiopian Constitution. Gupta,S.and Verhoeven,M. (2001).The efficiency of government expenditure: Experience from Africa, Journal of Policy Mopdelling23,433-467 Haque, N.; Pesaram H; Sharma, S. (1999). Neglected heterogeneity and dynamics in cross- country saving regression, IMF working paper. Herrera, S.;Pang, G.(2005). Efficiency of public spending in developing countries: An efficiency frontier approach, World Bank policy research working paper 3645, World Bank. House of Federation (HoF), (May 2009), “The federal Budget Grant Distribution Formula to Regional States” International Monetary Fund (IMF), (2010). The FDRE 2019 Article IV consultation King, David N., 1984, Fiscal Tiers: The Economics of Multi-Level Government (London: Allen & Unwin). McLure, Jr., Charles E., 1983, “Assignment of Corporate Income Taxes in a Federal System,“ in Tax Assignment in Federal Countries, ed. By Charles E. McLure, Jr. (Canberra: Australian National University). Ministry of Finance and Economic Development (MoFED), , Proclamation No. 33/1992 MoFED (2009), Annual Progress Report (APR) MoFED, (2005), Plan for Accelerated and Sustainable Development Programme (PASDEP). Musgrave, Richard A., Peggy B. Musgrave, and Richard M. Bird, 1984, Public Finance in Theory and Practice (Toronto: McGraw Hill). Oats, Wallace E., 1972, Fiscal Federalism (New York : Harcourt Brace Jovanovich). The Constitution of the Federal Democratic Republic Of Ethiopia, 21, August, 1995 World Bank, (2004). World Development Report 2004. World Bank, (2009). Social Development Indicators. World Bank, AFTP2, (2008). Ethiopia: Public Finance Review 2008. World Bank, AFTP2, (2009). Ethiopia: Public Finance Review 2009, Report No. 50278-ET.