Report No. ACS4421 Republic of Uganda Service Delivery with More Districts in Uganda Fiscal Challenges and Opportunities for Reforms Public Expenditure Review June 2013 AFTP5 AFRICA Document of the World Bank Standard Disclaimer: This volume is a product of the staff of the International Bank for Reconstruction and Development/ The World Bank. The findings, interpretations, and conclusions expressed in this paper do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Copyright Statement: The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development/ The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750- 4470, http://www.copyright.com/. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA, fax 202-522-2422, e-mail pubrights@worldbank.org. CURRENCY AND EQUIVALENT UNITS (as of May 30, 2013) Currency Unit: Ugandan Shilling (U Sh) US$ = U Sh 2,597 Vice President : Makhtar Diop Country Director : Philippe Dongier Sector Director : Marcelo Giugale Sector Manager : Albert Zeufack Task Leader : Anton Dobronogov iii ABBREVIATIONS AND ACRONYMS CAO Chief Administrative Officer CAS Commission on Administrative Structure CBMS Community-Based Management Systems DSC District Service Commission DTB District Tender Board DUCAR District Urban and Community Access Roads DWSDCG District Water and Sanitation Development Conditional Grant FDS fiscal decentralization strategy GDP gross domestic product HHSIP Health Sector Strategy and Investment Plan HSSP I Health Sector Strategic Plan LC local council LC1 village council LC2 parish council LC3 sub-county council LC4 county council LG local government LGFC Local Governance Finance Commission LST local services tax MAAIF Ministry of Agriculture Animal Industry and Fisheries MCBG Municipal Capacity-Building Grant MDB Municipal Demarcation Board MDG Municipal Development Grant MDGs Millennium Development Goals MoES Ministry of Education and Sports MoFPED Ministry of Finance Planning and Economic Development MoH Ministry of Health MoLG Ministry of Local Government MoPS Ministry of Public Service MoWE Ministry of Water and Environment MP Member of Parliament MTEF Medium-Term Expenditure Framework NAADS National Agricultural Advisor Services NDP National Development Plan NGO non-governmental organization NMS National Medical Stores NPA National Planning Authority NRA National Resistance Army iv NRM National Resistance Movement NSDS National Service Delivery Survey NWSC National Water and Sewerage Corporation OPM Office of the Prime Minister OSR own-source revenue PAF Poverty Action Fund PETS Public Expenditure Tracking Surveys PFM public financial management PHC Primary Health Care PMA Plan to Modernize Agriculture PRSP Poverty Reduction Strategy Paper PSC Public Service Commission RC Resistance Council RDC Resident District Commissioner RG regional government RGB Regional Governments Bill RING Resiliency in Northern Ghana SWAp Sector-Wide Approach TC town clerk UAAU Urban Authorities Association of Uganda UBOS Uganda Bureau of Statistics ULGA Uganda Local Governments Association UNPS Uganda National Panel Survey UNRA Uganda National Roads Authority URF Uganda Road Fund USAID US Agency for International Development v ACKNOWLEDGEMENTS This report was prepared jointly with the Government of Uganda and the team is most grateful to its government counterparts for this fruitful collaboration. A Steering Committee chaired by the Permanent Secretary—Secretary of Treasury Keith Muhakinizi (MoFPED) guided the team. Portions of the report are based on comments regarding the study and policy proposals prepared by the Ministry of Local Government (MoLG), which were requested in December 2012 by Hon. Adolf Mwesige, Minister of Local Government, and drafted in discussions with Patrick Mutabwire, Acting PS, MoLG. The OPM team, led by Albert Byamugisha, offered useful advice and support throughout the preparation of the report. The team would also like to thank Lawrence Banyoya, (Secretary, Local Governance Finance Commission - LGFC) for his useful guidance, as well as Kenneth Mugambe, Abdul Muwanika, and Winnie Mukisa for their cooperation. . The report was prepared by a team including Anton Dobronogov (Task Team Leader, AFTP5), Roland White, Martin Onyach-Olaa, Victor Ocaya (AFTU1), Sukhdeep Brar, Innocent Mulindwa (AFTEE), Peter Okwero (AFTHE), Samuel Mutono (TWIAF), Rasit Pertev (AFTA3), Liv Bjornestad (AFTP5), Melville McMillan, Paul Smoke, and Thomas Gole (consultants). Peer reviewers were Serdar Yilmaz (AFTP4) and Monali Chowdhurie-Aziz (WBIOG). Helpful guidance and comments were received from Philippe Dongier, Moustapha Ndiaye, Sajjad Shah, Albert Zeufack, Jacques Morisset, Shanta Devarajan, Anand Rajaram, Chiara Bronchi, Barbara Magezi, Stuti Khemani, Dan Mwanje, Willy Karagura, Jakob Fredensborg Rasmussen, Audrey Sacks, Dominic Haazen, Waly Wane, and Obert Pimhidzai. Lydie Ahodehou and Clare Busyngie provided administrative support to the team. vi Table of Contents I. Introduction........................................................................................................................................... 1 II. The Evolution of Local Governance, Decentralization, and Service Delivery in Uganda ................... 3 A. The Rise of Decentralization, Sub-national Governance, and Local Service Delivery .................... 3 B. The Current State of Decentralization .............................................................................................. 7 C. The Drivers behind the Post-Decentralization Evolution of the Local Government System ......... 10 III. Expenditures and Revenues of the Local Governments in the Context of National Public Finance 13 A. Government from a National Perspective ....................................................................................... 13 B. Local Governments ......................................................................................................................... 22 C. District Core Services: Fiscal Impacts of Changing Transfers and Shifting Responsibilities ........ 30 IV. Fiscal impacts of the increase in the number of districts ................................................................ 37 A. Ugandan Local Government in an International Context ............................................................... 38 B. Assessing the Fiscal Impacts of Redistricting in Uganda ............................................................... 41 1. What Happens When a New District Is Established? ................................................................. 42 2. Indicators of the Costs of New Districts: UBOS and BOOST data ............................................ 42 3. Indicators of the Costs of New Districts: Old and New Districts ................................................... 49 4. Expenditures Associated with the Creation of New Districts ........................................................ 54 5. Economies of Scale vs. Similarity of Preferences ......................................................................... 59 V. Value-for-money in service delivery: determinants and challenges ................................................... 62 A. Unequal Spending, Unequal Outcomes .......................................................................................... 62 B. Assessing Value-for-Money in District Spending .......................................................................... 66 C. Investigating the Determinants of Value-For-Money ..................................................................... 72 VI. Policy implications.......................................................................................................................... 76 A. Institutional proposals..................................................................................................................... 77 B. Fiscal proposals............................................................................................................................... 83 References........................................................................................................................................... 87 Annex 1. District Core Services: A Sectoral Analysis....................................................................... 95 Annex 2. Estimation of Fiscal Impacts of New District Creation..................................................... 116 Annex 3. Determinants of Value-for-Money Econometric Exercise................................................ 118 Annex 4. Value-for-Money Challenges faced by Uganda’s Local Governments: Lessons from the second LG PEFA Evaluation ............................................................................................................ 122 Annex 5. Assessment of a Proposal for the Creation of Regional Governments in Uganda ............ 125 Annex 6. Assessment of a Proposal for Enhancing Own-source Revenues of the Districts......... 129 vii Figures Figure 1: Current Structure of Local Governance in Uganda ....................................................................... 3 Figure 2. Real Gross Government Expenditure (2011/12 Prices)............................................................... 14 Figure 3. Real Government Expenditure Per Capita (2011/12 Prices) ....................................................... 15 Figure 4. Real Per Capita GDP and Government Expenditure (2011/12 Prices) ....................................... 15 Figure 5. Government Expenditures as a Portion of GDP .......................................................................... 16 Figure 6. Central Government Expenditures as a Portion of GDP ............................................................. 16 Figure 7. Central Government Total, Domestic, and Grant Revenues as a Portion of GDP ...................... 18 Figure 8. Central Government Domestic and Grant Revenues as a Portion of Total Revenue .................. 18 Figure 9. Central Government Grant Revenue as a Portion of Central Government Total Revenue ......... 19 Figure 10. Government Real Expenditures Per Capita (2011/12 Prices).................................................... 20 Figure 11. Central Government Recurrent and Development Expenditures .............................................. 21 Figure 12. Grants to Local Government as a Percentage of Central Government...................................... 21 Figure 13. Real Per Capita Funds of All Local Governments (2011/12 Prices)......................................... 23 Figure 14. Comparison of Local Government Real Per Capita Recurrent Funds: Expenditures from Statistical Abstracts and Budgets from the BOOST Database (2011/12 Prices) ........................................ 23 Figure 15. Real Per Capita Local Government Recurrent Expenditures by Major Components (2011/12 Prices) ......................................................................................................................................................... 24 Figure 16. Real Per Capita Local Government Development Budgets (2011/12 Prices) ........................... 25 Figure 17. Real Per Capita Budgets: Development, Recurrent, and Total (2011/12 Prices) ...................... 25 Figure 18. Real Per Capita District Budgets (2011/12 Prices).................................................................... 27 Figure 19. Real Per Capita District Recurrent and Development Budgets (2011/12 Prices)...................... 27 Figure 20. Real Per Capita Budgets of Thirteen Municipal Councils (2011/12 Prices) ............................. 29 Figure 21. Real Per Capita Recurrent and Development Budgets of Thirteen Municipal Councils (2011/12 Prices) ......................................................................................................................................................... 30 Figure 22. Real Per Capita Budgets in Five Core Areas (2011/12 Prices) ................................................. 31 Figure 23. Real Per Capita Budgets of Five Core and Other Expenditure Categories (2011/12 Prices) .... 32 Figure 24. Per Capita District General Public Administration Recurrent Expenditures (2011/12 prices) and Number of Districts.............................................................................................................................. 43 Figure 25. District General Public Administration Recurrent Expenditure as a Share of Total Recurrent Expenditures and Number of Districts........................................................................................................ 44 Figure 26. District General Public Administration Recurrent Expenditure: Per Capita (2011/12 prices) and Share of Recurrent Expenditure.................................................................................................................. 45 Figure 27. District Per Capita Total and Scaled-up General Public Administration Recurrent Expenditures (2011/12 prices) and Number of Districts................................................................................................... 45 Figure 28. District Per Capita Public Administration Accounts (2011/12 prices)...................................... 47 Figure 29. District Public Administration Budget Shares........................................................................... 47 Figure 30. Alternative Measures of Public Administration Costs: Per Capita............................................ 48 Figure 31. Alternative Measures of Public Administration Costs: Budget Shares ..................................... 49 Figure 32. Public Sector Management Budget Share of Old and New Districts ........................................ 50 Figure 33. Per Capita Public Sector Management Budgets in Old and New Districts ............................... 51 Figure 34. District Per Capita Total Budgets: New, Unchanged, and Mother............................................ 53 Figure 35. District Per Capita Total Budgets: New, Unchanged and Mother............................................. 53 viii Figure 36: Variation in Total Spending and Spending in the Health and Education Sectors (2011).......... 63 Figure 37: Variation in the Ratio of Pupils to Teachers............................................................................. 63 Figure 38a: Variation in Education Outcomes – Net Intake Rate............................................................... 63 Figure 39a: Variation in Health Outcomes –Pentavalent Vaccine 3rd Dose Coverage Rate....................... 64 Figure 40a: Variation in Education Outcomes............................................................................................ 65 Tables Table 1. Net Fiscal Impact on Districts of Recent Developments in the Finance of Five Core Areas ....... 33 Table 2. Local Authorities in Uganda: Number and Average Populations................................................. 38 Table 3. Approximate Number of Municipal Governments in Selected Developed Countries and Average Population, 1990s........................................................................................................................................ 39 Table 4 Number of Municipal Governments and Average Population in Selected Developing Countries, about 2002................................................................................................................................................... 40 Table 5. Indicators of Additional Costs Associated with New Districts..................................................... 54 Table 6. Relative Magnitudes of Additional Expenditures due to Adding Districts beyond 56: Three Scenarios ..................................................................................................................................................... 57 Table 7. District Numbers and Average Populations, Pre- and Post-Redistricting..................................... 60 Boxes Box 1. The Current Structure of Sub-National Governments in Uganda ..................................................... 5 Box 2. Decentralization in Uganda: Historical Background......................................................................... 6 Box 3. A Comparative Perspective on the Administrative Costs of Local Governments........................... 56 Box 4. A Process for the Creation of New Local Governments: Lessons from International Experience and Implications for Uganda....................................................................................................................... 78 Box 5. Thinking about the Optimal Size of a District................................................................................. 80 Box 6. Polycentric Governance .................................................................................................................. 81 Box 7. Principles for the Allocation of Responsibilities between Different FLevels of Government........ 82 Box 8. The Municipal Development Grant................................................................................................. 85 ix EXECUTIVE SUMMARY What are the fiscal and institutional effects of the rapid increase in the number of district governments in Uganda? How do these effects relate to recent trends in public expenditure and to the evolution of the country’s overall decentralization framework? Given these circumstances, how could service delivery by local governments be improved? Ugandan decentralization efforts of the 1990s represented an unusually authentic and powerful local government reform, compared to similar efforts pursued in many other low-income countries. However, over time the changing interests of the central agencies, dissatisfaction with service outcomes, and the overall dynamics of the country’s governance resulted in the adoption of a number of re-centralizing policies. On the fiscal side, over 80 percent of the local governments’ funding comes from conditional transfers from the central government, while own-sourced revenues account for less than 10 percent of their budgets. On the institutional side, the system remains more decentralized, with local councils being elected and exercising a certain degree of authority over the local civil services. However, the councils’ power to recruit and control top local government staff has been reduced. In order to increase the voice of local citizens, to better target local service delivery, and in response to the growing average population of districts (which increased by about 50 percent over a 15-year period), the central government chose to dramatically increase the number of districts in Uganda. From 45 in 1997 to 56 in 2001, their number had reached 112 by 2010. New district creation (also known as district proliferation) may produce its officially intended effect in some areas. However, a number of analysts believe that, over time, district proliferation is mainly driven by political forces and is designed to help the ruling party to win increasingly tough electoral battles in a more competitive electoral climate. Some also believe that increased public administration expenditures among district governments cause a decline in their spending on social services. The objective of this report is to take stock of the fiscal and institutional arrangements for service delivery by local governments in the context of district proliferation and in view of recent trends in national public finance, as well as to identify policy options that could facilitate improved service delivery. The report finds that, while district proliferation has not had any major effect on public finances so far, it may have serious adverse effects in the future if the institutional structures and funding mechanisms of district governments are not adjusted to the new realities. Its effect on public expenditure has been rather small since an increasing number of positions in district governments remain vacant, and because a recent bout of inflation has eroded their wage bills. Furthermore, at present the average population of a district is roughly equivalent to that of similar jurisdictions in other countries. The decline in local government spending is primarily caused by across-the-board shifts in the spending priorities of the central government—away from social sectors and towards infrastructure—as well as by a reduction in government revenues as a share of GDP, leading to a decline in transfers to district governments, particularly when it comes to those intended to finance education and health expenditures. However, district proliferation has increased the costs of local public administration more than actual expenditures on this item. Its fiscal effects are likely to increase over time if vacancies are filled and salaries are indexed, or it may have an adverse effect on value-for-money in service delivery if they are not, particularly because many vacant jobs in district governments are skills-intense. In either case, the creation of new districts may, in the end, amplify the problems it was intended to address. The report concludes with a number of institutional and fiscal proposals designed to reduce this risk and to improve value-for-money in service delivery more generally. These proposals are outlined in the table below. If Uganda manages to use district proliferation as an opportunity to implement these changes, the resulting fiscal savings and improvements in value-for-money would make it much easier to cover the costs of that process. x Institutional proposals Fiscal proposals Establish a systematic technical process for the Enhance own-sourced revenues of the district creation of new districts governments Reform districts’ civil services to adjust them to Simplify the conditional transfers system and new realities and budget constraints improve its horizontal equity Deconcentrate selected departments of the line Increase the share of unconditional transfers in the ministries to the regions and/or facilitate the districts’ funding, and design these transfers in a development of inter-district cooperative way that enhances district governments’ arrangements in the sectors where they are needed accountability for their performance Expenditures and revenues of the local governments in the context of national public finance Government expenditure as a portion of GDP has declined from about 21 percent to roughly 19 percent from 2001/02 to 2011/12, even though real per capita total government expenditures have increased by approximately 10 percent during the same period. The government’s spending priority has shifted to fixed capital. At best, real per capita government final consumption expenditure has remained constant. Fixed capital formation has increased about 50 percent in real per capita terms. Relative to GDP, government final consumption has decreased from 16 percent to about 11 percent, while government fixed capital formation increased somewhat (from about 5 to 6 percent of GDP). The central government maintained—if not increased—its relative position in the economy, despite declining revenues. In 2019/10, central government expenditures represented roughly the same percentage of GDP as in 2001/02 (about 15 percent, with variations during the period), followed by an increase during the two subsequent years. Making up about 12 percent of GDP, domestic-sourced revenues stagnated at a low level compared to those of other Sub-Saharan African countries. Grants to the central government declined from about 7–8 percent of GDP to about 3–4 percent. Central government spending priorities shifted towards development expenditures. Meanwhile, central development expenditures increased relative to GDP, while recurrent expenditures were, at best, constant. Real per capita spending by the central government increased (about 45 percent over 2001/02 levels). At the same time, real per capita development expenditures approximately doubled and development expenditures increased from about 25 percent to over 40 percent of central spending. Real per capita recurrent expenditures rose but only modestly. Local governments became relatively smaller. Uganda’s local governments rely on transfers from the central government for over 90 percent of their revenue; about 90 percent of these transfers are conditional. Central government transfers to local governments decreased as a share of GDP from about 5 to about 3.5 percent. The total real per capita budgets of the districts have trended downward since 2003/04, with an overall decline of about 10 percent. Further aggravating the revenue problem, district own-source revenue also appears to have declined over this period with the abolition of the graduated tax. Local governments’ spending on social services has been declining, while non-education and non- healthcare funding shows a modest increasing trend. Education budgets have been especially hard hit. Education represents almost 50 percent of local governments’ budgets. The education sector experienced a reduction of nearly 20 percent in real per capita transfers between 2001/02 and 2011/12. Health funding, which represents roughly 13 percent of all transfers, is also being reduced. xi Various recent changes in the central government’s spending policies and priorities imposed a net fiscal burden on district budgets. A review of funding and expenditures in the five core service delivery sectors of the districts (education, healthcare, roads, water supply, and agriculture) accounting for about three- quarters of district budgets suggests that, on aggregate, this burden amounts to approximately 10 percent of district budgets in 2011/12. These dynamics are consistent with and reflect a shift in the central government’s spending priorities, from social sectors toward infrastructure. This shift might be of a cyclical nature, reflecting Uganda’s inherent difficulty in finding a balance between spending on infrastructure and on social services. Uganda has very high fertility (with a total fertility rate of more than six children per woman) and a very young population (with a median age of about 15 years). Given the scarcity of its fiscal resources, the government finds it difficult to invest in both infrastructure and social services in amounts that would be sufficient to allow the majority of its young people to find jobs when they enter the labor market. An increase in one of these categories of investment leads to gaps in the other. In the 1990s and early 2000s, the government focused its efforts on social services until it became apparent that the country’s infrastructure required urgent attention. Sooner or later this recent shift in priorities might be reversed, following the same logic that led to its occurrence. However, judging by past experience, this reversal may not occur for a decade or more. Fiscal and institutional implications of the growth in the number of districts The impact of the creation of new districts on public expenditures has been low to date. Total expenditures per person and public administration costs per person are somewhat greater on average in new districts as opposed to old (unchanged and “mother�, which are those that have been reduced in size due to the creation of a new district or districts). Total government expenditures per person averaged 9.3 percent higher in the new districts. As for public administration costs, two definitions are examined: the broader definition generates a 3.5 percent premium in total cost per person in the new districts due to greater public administration sector expenditures, whereas the narrow definition shows a 1.26 percent cost premium due to added conditional funding specific to public administration. These costs translate to a 1.2 percent increase and a 0.47 percent increase, respectively, in the total expenditures of local government as a result of increasing the number of local governments from 56 to 112. This impact is apt to increase in the future, however, making district proliferation a potential “time bomb.� This is because the share of vacant positions in district governments (currently about 40 percent) has been increasing along with the number of districts, and also because the real wage bill of district governments has been eroded by a recent outburst of inflation. Longer-term fiscal costs could be even higher and increase over time given the risk that value-for-money in public services could decline. The district governments’ most critical staffing issues relate to difficulties in hiring and retaining highly skilled staff. If such problems persist, value-for-money is likely to decline, and some public services might simply be suspended. In addition, an increase in the number of districts increases the average costs related to operations requiring the use of expensive equipment (as in the case of road maintenance, for example). The effects of district proliferation on service delivery are also likely to be amplified by tightening of the central government’s budget constraints. This is because local governments’ budgets are largely financed through transfers from the central government, and fierce contests among various government agencies for increasingly scarce resources increase the risk that such financing will not be provided at appropriate levels. xii Moreover, if the problem of unfilled vacancies is not addressed, it may further undermine the overall decentralization framework. If value-for-money in service delivery declines as a result of this problem, the central government might be inclined to take over the functions that district governments would be increasingly unable to perform. Value-for-money, however, is more likely to improve if district governments are offered opportunities and incentives to improve their capacity to deliver public services. Preferences among a given district’s population do not differ as much as they do from one district to another; all things being equal, district governments are in a better position than the central government both to collect information about these preferences and to satisfy them. Value-for-money in education and health expenditures of the local governments Given the tightening of overall budget constraints, the decline of allocations for social services, and the increase in public administration costs, it is increasingly important to improve value-for-money in service delivery by local governments, particularly in the education and health sectors. Uganda is marked by inequality across districts in terms of public spending per capita. For example, in 2010 public spending on health differed by a factor of 11.5 across districts, and public spending on education differed by a factor of 6.3. There is also variation in the level of services provided in various sectors. For example, in the education sector, the number of students per primary school teacher varies from as low as 32 to as much as 100 across districts, and the share of deliveries in government health facilities varies from 4percent to 94 percent across districts. Increasing value-for-money in education and health expenditures in the districts that are lagging behind would help to achieve better outcomes without raising costs. Some districts are able to achieve better outcomes while operating under tighter budget constraints. Bringing every district up to the outcome-to- spending ratio of the districts that perform very highly in both health and education sectors could save about 0.8 percent of GDP. This exceeds annual recurrent public administration expenditures of all districts, old and new. The analysis undertaken in this report shows that, overall, value-for-money in social sectors can be increased by shifting funding from districts that are already well-served to districts that are relatively under-served. Additional spending appears to have a greater impact where current levels of spending are low; consequently, shifting resources to those areas will have a larger effect on education and health outcomes. Districts with lower levels of funding also tend to have poorer outcomes, and so are in greater need of additional funding. Accountability and the quality of local institutions are other key factors. For example, districts that received an unqualified audit in the Auditor General’s Annual Report had consistently higher levels of value-for-money. Stronger local government institutions improve the efficiency of spending in other areas, enabling limited resources to achieve better outcomes. Accountability in this sense can reflect both top-down and bottom-up mechanisms. Top-down auditing mechanisms—such as those employed by the Auditor General and the Inspectorate-General—can hold districts to account for both substantive and procedural weaknesses. Bottom-up mechanisms, particularly making district governments more accountable to citizens, can also align local government decision-making and the needs of citizens in order to better achieve sector outcomes. xiii Policy implications A number of broad short- to medium-term measures need to be implemented in order to: (i) bring greater stability to districts and to the intergovernmental system; (ii) enhance the system’s institutional design to strike the right balance between capturing economies of scale and better targeting local service delivery in a cost-effective way; and (iii) redesign funding of district governments in a way that helps to improve service delivery outcomes under given budget constraints. Institutional proposals The most immediate priority is to bring some order, stability, and technical rationality to the process of new district creation. As long as districting remains a “moving target,� other reforms will be difficult to implement. In this regard, the recent Cabinet decision to declare a moratorium on the creation of new districts is an important first step. However, in addition to this, it will be necessary to establish a systematic process governing the creation of districts (and other local governments) and the delineation of geographical boundaries. Reforming the institutional structure of Uganda’s civil services at the district level should be the next step, since they are disconnected from the human and financial resources that are currently available and often from the districts’ needs as well. The basic organizational structures of district governments—such as administrative configurations, staffing numbers, and conditions of service for staff positions—need to be reviewed and recalibrated in line with the service delivery mandates of the local governments, their increased numbers, and the budget constraints of the consolidated government. This reform could help to reduce a trade-off between filling all positions in district governments and increasing expenditures in local public administrations. Districts will need to be differentially resourced in line with the various needs they face. Other things being equal, smaller districts have smaller needs. For example, it may not be necessary for all districts to own and operate a grader for road maintenance. Where the road network of any given district makes this economical, a grader can be provided for; where this is not the case, the district can contract out on an occasional basis to meet such needs as they arise. Rationalizing districts’ civil services would go some way towards striking the right balance between capturing economies of scale and better targeting local service delivery. However, this problem cannot be addressed entirely at a district level, and other measures are needed in addition to district-level reforms. There appears to be a growing need to deconcentrate certain departments of central line ministries at the regional level to allow for more effective execution of their tasks. In fact, to some degree this is already happening: the Ministry of Education and Sports (MoES) has established four regional offices while MoWE has established eight technical support units to aid districts along with a number of Water Management Zones and Water Development Support Fund branches at the regional level. There is substantial scope for this sort of administrative and activity-based deconcentration when it comes to improving the capabilities of central line agencies to (i) execute their own functions more effectively; and (ii) support the districts in the execution of their currently mandated roles in areas such as primary health, education, roads, water, etc. more effectively by introducing a greater on-the-ground presence, shortening the “bureaucratic� and logistical distance between districts and ministries, and so on. There is also a possible need to establish inter-district structures or cooperative arrangements. In late 2012, a MoLG proposal to form Regional Governments as an intermediate layer between the districts and central government was circulated. Broadly speaking, it appears that this initiative would be unlikely to lead to service delivery improvements in the medium or long term while also being accompanied by a deterioration of service delivery in the short term. It also appears that support for this proposal may be xiv waning. Short of the establishment of Regional Governments or administrations, inter-district cooperative arrangements might be effective where cross-boundary planning and investment activities are required in service delivery areas for which district governments are responsible. An advantage of this “polycentric� approach is that the voluntary cooperation of district governments increases the bottom-up accountability of any institution they may decide to create. The Constitution of Uganda provides for this sort of collaboration but establishes very demanding procedural requirements in order for such arrangements to come into existence. As a point of departure for any arrangements that might emerge, it is suggested that districts and central line ministries retain their current powers and functional responsibilities. This does not mean that such powers and responsibilities should remain unchanged, since there may be some scope for their eventual review. If, however, any such re-allocation is undertaken, it should preserve the vertical integration of functional responsibilities. For example, if a newly constituted district has a relatively small roads network, hence an establishment size that does not allow it to effectively develop or maintain a major road having previously been characterized as a district road, it may be necessary to re-classify this road and place it entirely under the responsibility of the Uganda National Roads Authority (UNRA). However, it would be unwise to divide development and maintenance and operation responsibilities for this road between the district government and the UNRA, as this will weaken accountability, and hence is likely to reduce the quality of service delivery. Fiscal proposals In general terms, some of the main proposals to be found in the 2012 Local Governance Finance Commission (LGFC) report, “Review of Local Government Financing,� provide a good point of departure. It took on the task of developing an approach to addressing the fiscal dimensions of the local and intergovernmental system, although more work will be needed to make this approach operational. For example, with regard to own-source revenues, it is clear that substantial improvements can and should be made in this area. Before the abolition of the graduated tax several years ago, the district governments were able to collect substantially larger amounts of revenues compared to what they are collecting now, and there is no reason to believe that their capacity to do so cannot be restored. Enhancing own revenues of district governments is likely to have positive impact on both the political economy and the fiscal dimensions of local governance. It would strengthen the bottom-up accountability of district governments to their constituents. It would also help to alleviate, at least to some extent, the fiscal pressures experienced by local governments. The existence of such pressures at all levels of government could make it politically easier to legislate and implement an enhancement in district government revenues. However, a more in-depth study of particular revenue sources is required in order to develop a properly sequenced approach that is both viable and appropriately targeted. It is also important to consider any trade-offs that may exist between expanding own-source revenues of district governments and improving the central government’s capacity to collect its revenues. Value-for-money needs to be improved, in particular in the education and health sectors, and the system of inter-governmental fiscal transfers needs to be redesigned to serve this purpose. Unless the medium- term allocations for these sectors are revisited with a view to reversing this trend, increasing value-for- money in service delivery might be the only way to prevent further deterioration in the ability of local governments to deliver these services. Two broad directions for redesigning the system of fiscal transfers could be suggested. The horizontal equity in allocation of resources between local governments through conditional transfers needs to be improved. The current complex mixture of historical practices, need-based formulas, and ad hoc considerations determining amounts of such transfers to individual districts needs to be replaced by a simple, formula-based system that is based on transparent criteria. For example, in the case of education, xv conditional transfer to a district could be calculated as a standard amount per school-going child (adjusted, if necessary, for input-cost differentials between jurisdictions), multiplied by the number of school-going children. Such a system would be more equitable than the one currently in place and, as value-for-money analysis has shown, also more efficient. Reforming the system to allow for the allocation of conditional transfers would entail a redistribution of fiscal resources among districts, thus creating winners and losers. Managing the political economy risks and local budget shock implications of this can be handled by phasing in such a system gradually, to ensure that fiscal impacts of the reform are spread over time. It is also important to reverse the trend of declining unconditional transfers to districts, and to provide such transfers in a way that enhances district governments’ accountability for their performance. Such transfers provide the key means through which local governments can respond meaningfully to local needs. Increasing the share of unconditional grants in the transfers to local governments, with their amounts determined by district performance in relation to a number of governance and other efficiency indicators will create powerful incentives for improving top-down accountability mechanisms important for delivering value-for-money in local service delivery. Going forward, it will be important to ensure that funding for the existing Local Development Grant—which is targeted at district governments—is expanded, and that the assessment system for the allocation of this grant is strengthened. xvi I. Introduction 1.1 What are the fiscal and institutional effects of the rapid increase in the number of district governments in Uganda? How do these effects relate to recent trends in public expenditure and to the evolution of the country’s overall decentralization framework? And how, given these circumstances, could service delivery by district governments be improved? 1.2 Uganda is notable among developing countries for its degree of government decentralization. The decentralized system that emerged during the 1990s assigned responsibility for the delivery of significant services to local governments. Those included schooling and healthcare, as well as agricultural extension, roads, and water and sanitation, among others. Those services represent over 20 percent of government expenditures and about 4 percent of GDP. Districts, and municipal councils in the large urban areas, are the core units of Uganda’s local governments. After Uganda introduced a robust and high-profile decentralization policy in the early 1990s, it was for a time portrayed internationally as a case of considerable success with a type of public sector reform that has proved particularly challenging to adopt in many developing countries. 1.3 The positive image of Ugandan decentralization, however, has been changing. Over the past decade, the system of local government in Uganda, which substantially devolved service delivery responsibilities, has been compromised in a number of ways. The central government has greatly expanded the number of districts, especially during the past decade; from 45 in 1997 to 56 in 2001, Uganda counted a total of 112 districts by 2010. Although politically popular, the rapid growth in the number of districts (sometimes referred to as “proliferation� or “fragmentation�) has raised questions and concerns about the performance and viability of the local government system. Prominent among the concerns are possible adverse consequences for local service delivery (perhaps due to the loss of economies of scale and overly stretched administrative capacities), as well as higher costs to central government (possibly because of unnecessarily larger local public administration and decision-making costs, additional burdens on ministries and central agencies, and coordination problems). These concerns have generated suggestions to revise the political structure (for example, a realignment of local and central responsibilities, or the establishment of regional governments). The pressures and forces in play are multiple and diverse; the various actors have radically different views of the issues and what constitutes the most appropriate response to them. 1.4 Political economy realities need to be taken into account if productive and sustainable reform is to be undertaken. Indeed, it might be argued that Uganda has arrived at its present decentralization situation in part because the initial decentralization efforts did not adequately consider a number of important political economy constraints and other contextual factors that necessarily affected the ability of the envisioned decentralization to take root and prosper. 1.5 This report takes stock of the fiscal and institutional arrangements for service delivery by local governments in the context of district proliferation, as well as recent trends in national public 1 finance, and identifies policy options that could facilitate improved service delivery. 1 The report employs political economy analysis; fiscal analysis of recent trends in expenditures of revenues of both central and local governments and of the evolution of their functional responsibilities; and econometric analysis of the determinants of value-for-money in education and health across districts. Its principal sources of fiscal data include Statistical Abstracts published by Uganda’s Bureau of Statistics and the World Bank’s BOOST database. The value-for-money analysis relies on district league tables complied by MoES and MoH. 1.6 The report finds that, while district proliferation has not had any major impact on public finances so far, it may have serious adverse effects in the future if the institutional structures and funding mechanisms of district governments have not been properly adjusted to the new realities. The policy in question had a rather small effect on public expenditure in large part because an increasing number of positions in district governments remain vacant, and because their wage bills have been eroded by a recent bout of inflation. Further, the average population of a district is currently more or less on par with populations of similar jurisdictions in other countries. The decline in local government spending is primarily caused by across-the-board shifts in the spending priorities of the central government—away from social sectors and towards infrastructure—as well as by a reduction in government revenues as a share of GDP, leading to a decline in transfers to district governments, particularly those intended to finance education and health expenditures. This said, the fiscal effects of district proliferation are likely to increase over time if vacancies are filled and salaries are indexed. On the other hand, leaving these vacancies unfilled may have an adverse effect on value-for-money in service delivery, particularly because many vacant jobs in district governments are skills-intense. In either case, the creation of new districts may, in the end, amplify the very problems it was intended to address. The report concludes with a number of institutional and fiscal proposals designed to reduce this risk and to improve value-for-money in service delivery more generally. 1.7 The remainder of the report is organized as follows. Section 2 examines the evolution of governance, decentralization, and service delivery in Uganda over the last two decades, including the most recent trends and the current state of affairs. Section 3 analyzes expenditures and revenues of local governments in the context of national public finance. Section 4 assesses fiscal and political economy implications of district proliferation. Section 5 analyzes variation in and determinants of value-for-money in health and education sectors, which absorb more than half of district governments’ fiscal resources. Section 6 concludes with policy implications. 1 The report is not intended to be a comprehensive review of intergovernmental fiscal relationships in Uganda. For example, it does not review in any detail the central government’s rules and institutions for the allocation of resources across space and across sectors, nor does it attempt to determine appropriate public spending responsibilities among the central and local governments. 2 II. The Evolution of Local Governance, Decentralization, and Service Delivery in Uganda A. The Rise of Decentralization, Sub-national Governance, and Local Service Delivery 2.1 Uganda was one of the first countries in Africa to embark on a major decentralization in which significant administrative, political, and fiscal powers were devolved to local governments. The government announced the policy in 1992 and subsequently set out to develop a system that was defined in the Constitution (1995) and further elaborated in the Local Government Act (1997). The bulk of service delivery responsibilities were devolved to local governments, with the exception of urban water and sewerage and electricity, which are provided by national corporations. Line ministries were left primarily with the responsibility of setting national policies and standards, assuming inspection, supervision and monitoring roles, providing technical advice and support, and training local governments. The legal framework also provided for intergovernmental fiscal transfers (unconditional, conditional, and equalization) and several types of local own-source revenues. The elected local councils were empowered to develop and execute plans and budgets and to hire and fire staff. In short, decentralization in Uganda created several levels of local government with a fairly broad set of autonomous powers and functions typical of a devolved system. The current structure of sub-national governments in Uganda is presented in Figure 1 and Box 1. Figure 1: Current Structure of Local Governance in Uganda Source: Review of Local Government Set Up in Uganda, Ministry of Local Government 2012. 3 2.2 A range of subsequent major policy initiatives appears to promote and/or rely on decentralization as well. The Poverty Reduction Strategy (2000), Fiscal Decentralization Strategy (2002), Decentralization Policy Strategic Framework (2006), and Local Government Sector Investment Plan (2006) in various ways highlight the key role of local governments in public functions. The Fourth National Development Plan (NDP) 2011-15 also stresses the centrality of local governments to its implementation, although it raises questions about whether the current system is appropriately structured, sufficiently empowered, and adequately capacitated to play the expected roles. 2.3 In summary, Ugandan decentralization was unusually homegrown and initially substantial for a low-income, post-conflict country. A number of other factors about the emergence of local government in the 1990s, however, are important to keep in mind when analyzing possible further reforms. At first glance, it seems improbable that a least-developed African country emerging from a period of serious and debilitating conflict and with a history of contentious debate about the need for and nature of decentralization (Box 2) would look to local governments as a foundational element of the state. Potential opposition from those who might logically have opposed decentralization was relatively weak. It was muted by the dominance of the ruling party in the absence of a formal opposition, a broad-based desire to take positive steps to rebuild the country after an extended period of conflict, and the preoccupation of nascent national agencies with building their own policies and capacities. In this environment, some of the actors who stood to lose from decentralization might not have been aware of its significance for them and were engaged with other matters that were, at the time, more pressing. 4 Box 1. The Current Structure of Sub-National Governments in Uganda There are higher and lower levels of local government in Uganda. District and city local governments form the higher level. Kampala is the only city-level authority, so the 112 districts are the most common higher-level units. At the district level there are urban and rural local governments. As of 2010, 22 municipalities and 165 towns comprise those urban local governments. The municipalities and Kampala also have divisions within them. Rural local governments are the district councils and their 1,147 sub-county councils. The 162 counties are administrative units, not local governments. Other administrative units exist below the sub-county, town, and division council level. In the districts, the first level of units are called parishes and, in urban areas, wards. Below them are rural villages and urban cells or zones. Uganda’s subnational government activity functions through a unified system of elected authorities grounded at the district level, which has primary local planning and budgeting responsibility, and four lower government and administrative levels with more modest responsibilities. The district (local council 5 or LC5), sub-county (LC3), and village (LC1) councils are elected by universal adult suffrage, with county (LC4) and parish (LC2) levels elected indirectly or appointed ex-officio. District/city councils and sub-county councils (municipal divisions and town councils are urban equivalents) are corporate bodies and accounting levels. County councils (municipal councils and city divisions in urban areas), parishes (urban wards), and villages (urban cells) are administrative units. Two associations—the general Uganda Local Governments Association (ULGA) and urban-specific Urban Authorities Association of Uganda (UAAU)—represent local government interests in national policy discussions. At the national level, the Ministry of Local Government (MoLG) is charged with coordinating, advocating for, mentoring/supporting, and inspecting/monitoring local governments. The Ministry of Finance, Planning, and Economic Development (MoFPED) plays a key role in designing and executing intergovernmental transfers and financial oversight. The Ministry of Public Service (MoPS) and the National Planning Authority (NPA) have functions related to their specific mandates. The usual range of sectoral ministries (education, health, etc.) is also important for local government. They ensure compliance with policies and standards and play roles in inspection, monitoring, technical advice, support, supervision, and training in their areas of expertise. Several other bodies, e.g. for audit and inspection, also engage in local government oversight. Local Government Finance Commission (LGFC) advises the central government on intergovernmental fiscal relationships, and assist the LGs with improving efficiency of their spending and on enhancing their own-source revenues. The central government maintains a presence and exercises general oversight at the local level through the Resident District Commissioner (RDC). The RDC is appointed by the President and is responsible for coordinating the administration of government services in a district, advising the district chairperson on national matters, and carrying out other functions assigned by the President or prescribed by Parliament. Regional governments are allowed by the Constitution and an amendment to it, as well as by the Regional Governments Bill (RGB) of 2009, but they do not currently operate. The RGB allows two or more neighboring districts to cooperate in order to form a regional government to perform functions and services allowed under the Constitution. Each district has a council and administrative staff. The councils are elected and consist of about 12 persons. Districts require an administrative staff. A recommended complement of strategic positions for a district is 13 (10 for municipalities). In addition to a Chief and a Deputy Chief Administrative Officer, there are positions to cover personnel, finance, planning, education, health, production, engineering, community development, natural resources, and auditing. In addition, certain positions at the sub-county level may be needed. Due to financial constraints, the central government provides funding for (allows filling of) only about 62 percent of the positions recommended. The local governments have flexibility in filling positions and a large proportion of them go unfilled. This relatively standard complement of officials that represents a fixed cost for new districts has received much attention in the discussion of the growth of district numbers. 5 Box 2. Decentralization in Uganda: Historical Background During the early colonial period, Uganda’s government was organized as a top-down system of district administration led by appointed district commissioners; local governments were created later. Debates over a federal system built around traditional kingdoms were initially prominent and have arisen periodically since then. A federal system existed after Independence, from 1962 to 1966, but the kingdoms were abolished in 1967 under the first Milton Obote regime and absorbed into the district administration system. Over the years, formal local government institutions were alternatively strengthened or weakened. During the war that brought President Museveni and the National Resistance Movement (NRM) to power in 1986, the National Resistance Army (NRA) mobilized popular support by creating Resistance Councils (RCs) at the village level. After its victory, the NRM government built on this approach by creating a hierarchy of elected RCs in villages, parishes, sub-counties, and districts with authority to plan, make decisions, administer local justice, and provide services. The early NRM system formed the basis for the local government system (the LCs described above) that emerged in the 1990s. It seems likely that a combination of the success of the RC system for the NRA and the desire to build an effective and more equitable state with popular support were behind the decision to create empowered local governments. From the early stages of the NRM government, there was a strong emphasis on citizen engagement and political legitimacy. Decentralization was seen as critical for democratization, service delivery, and fighting poverty, and it was visibly promoted in these terms. Although the NRM government selectively consulted and relied on international development partners for advice and support, it is fair to say that the main impetus for decentralization came from within. The political importance of traditional kingdoms and their role in the rise to power of the NRM, however, did not go unnoticed. In 1993, they were permitted to reconstitute as parallel institutions with the understanding that they were to serve a cultural rather than a political function. Some of them have created the position of prime minister and have formed volunteer parliaments, so they have taken on some of the trappings of political entities; furthermore, by their very existence and nature, the kingdoms have a political meaning and import. They were, however, given no power to tax and receive only limited government funding, so they were clearly not intended to be major partners in formal governance and development efforts. Some analysts suggest that the allowance for regional governments is primarily a way to constrain the kingdoms by creating a formal government entity that would occupy some of their space. There are concerns in some circles, however, that regional governments as conceived in the RGB would mix administrative and service delivery obligations with cultural and traditional functions (where they are created in the territory of kingdoms); would increase operating costs by establishing regional administrations and assemblies; would result in changes in functions that require the revision of the intergovernmental fiscal transfer system; and would not ensure the universal application of the provision for their creation, which would create further challenges for acceptability and implementation. 6 2.4 On the other hand, even if the government’s commitment to decentralization was as genuine as it then appeared, it seems to have been based on an incomplete appreciation of its significant implications. The new local government system, on which high governance and functional expectations were placed, was very different from the RC system that had in part inspired formal decentralization. The considerable functions, resources, and levels of autonomy rapidly given to local governments under the initial reform program were, in hindsight, not commensurate with the accountability and especially the capacity levels of the local councils. Human capital in the public administration of a post-conflict, low-income country was too scarce to be spread across dozens of local governments. B. The Current State of Decentralization 2.5 Although the decentralization framework of Uganda is strong in constitutional, legal and policy terms, central government decisions and actions have not always followed the strategic orientation set out in the above-noted legal instruments and policy documents. In fact, there is growing concern about local governments and their performance in discharging their mandates. On the fiscal side, over 80 percent of the local governments’ funding comes from conditional transfers from the central government, while own-sourced revenues account for less than 10 percent of their budgets. On the institutional side, the system remains more decentralized, with local councils being elected 2 and exercising a certain degree of authority over the local civil services. However, the councils’ power to recruit and control top local government staff has been reduced and control over district procurement has been reassigned back to central civil servants. 2.6 Although local government budgets have continued to increase in absolute terms, the overall size of the public sector and the share of local government in total government spending are in nontrivial decline. Between FY 2001/02 and FY 2011/12, total government expenditure has declined from about 21 to less than 19 percent of GDP. The share of central government grants allotted to local governments in the national budget fell from FY 2003/04 to FY 2010/11, from more than 25 to just over 20 percent, a decline on the order of 20 percent (see Section 3 for more details). 2.7 There are reasons for concern about the composition of local government expenditure and service delivery performance. Although central government real per capita expenditures increased between FY 2001/02 and 2011/12, the comparable figure for local governments has remained relatively flat overall, but with a decline in urban areas. Interviews with government officials and opinion leaders conducted in the process of preparing this report indicate that service delivery coverage and quality is not commensurate with expectations. Social and economic outcomes have been improving in Uganda over the last two decades, but the perception is that they are not improving fast enough. While local governments can be held only partially responsible for this slow development, given the rapid population growth and high young-age dependency ratio, it is difficult to imagine them improving these outcomes much faster. 3 Furthermore, great disparities persist across districts, and the expected relationships among 2 Candidates from the ruling party receive support from the party’s headquarters in Kampala. 3 See the World Bank (2011) report on Demography and Economic Growth in Uganda. 7 improved inputs, outputs, and outcomes (e.g. primary education expenditures per student, primary completion rates, and literacy rates) are often not well correlated (see Section 5). 2.8 Local government own-source revenues have been in substantial decline. In the early years of decentralization, own-source revenues accounted for 8 to 10 percent of total local government revenues, whereas in recent years this figure has dipped to 5 percent or less. Available estimates indicate that local governments are collecting less than half of their potential revenue. 4 There is, of course, great variation (with urban governments typically more fiscally independent than rural governments). 2.9 Local governments have become increasingly dependent on conditional transfers. In accordance with the norms in a devolved system, unconditional transfers accounted for a significant proportion (nearly 35 percent) of total transfers in FY 1995/96. By FY 2010/11, the number had declined to only 10.8 percent, and it is estimated to be less than 5 percent in FY 2012/13. The number of conditional grants has more than doubled since 2000, and a dominant share of the remaining unconditional grants is used to pay for (underfunded) administrative core functions (see Section 3). 2.10 There has been a proliferation of new districts (and other levels of local government and administration) in the past decade. The number of districts increased from 56 in 2002 to 80 in 2008 and reached 112 in 2012. 5 There are indications that many new local governments in Uganda were created without adequate assessment. The political imperative, however, seems to be overwhelming, and some knowledgeable parties believe that this process will continue until there is a district for every parliamentary constituency. A cabinet meeting in early March 2013 decided to place a moratorium on the creation of districts and to withdraw from parliamentary consideration the pre-existing proposal for the creation of 25 new districts; this could be replaced with a proposal to create two new districts in Kakumiro and Kagadi. 2.11 The central government in recent years has adopted a number of centralizing policies that constrain local government powers. Most important among these has been imposing a reduction in the power to recruit and control top local government staff and the reassignment of control over district procurement from elected local councils back to central civil servants. 6 Specifically, local governments have lost the power to recruit Chief Administrative Officers, Deputy Chief Administrative Officers, and Municipal Town Clerks to the Public Service Commission (PSC), and the center now has more control over—or in some cases even takes direct responsibility for—district procurement. 2.12 Finally, a range of weaknesses in the design of the intergovernmental system have always been problematic and continue to create challenges for the public sector in general and for decentralization in particular. These include sometimes insufficient coordination among the central agencies that play a role in or influence local service delivery and development; the maintenance of central government staffing at pre-decentralization levels; weak linkages between the recurrent and 4 This is reported in LGFC (2012), although it is likely to be an underestimate, since the revenue base of local governments has not been well documented. 5 Smoke, Muhumuza, and Ssewankambo (2011) report that the number of subnational government and administration units increased from about 44,000 to more than 67,000 between 2004 and 2008. Current estimates indicate that there are more than 69,000 units. 6 For more details, see Muhumuza (2008). 8 capital budgets; and the lack of attention in the reform process to building downward accountability and civil society capacity to interact with local governments, among others. The combination of recent developments in the local government system along with initial weaknesses in the decentralization policy and the intergovernmental framework has generated pervasive and profound effects. 2.13 The actual and potential effects of these evolving changes to the local government system are considerable. Most are on balance negative, and many are mutually reinforcing. 2.14 The shrinking fiscal importance of local governments raises serious concerns about their ability to meet their extensive constitutional and legal functional mandates for delivering services. This problem is largely attributable to the steadily declining share of national public revenues transferred to local governments, particularly by a decline in conditional transfers for health and education, and is exacerbated by the weakness of and downward trend in local government own-source revenue generation (see Section 3 for further details). 2.15 There are major service delivery challenges in terms of efficiency, coverage, quality and equity, and these raise serious concerns for Uganda on several fronts. Public service provision reflects on the quality of governance and the success of the state, and it is ultimately critical for economic development and poverty reduction. Decentralization was in fact adopted in Uganda in great part to improve these very outcomes. 2.16 The composition of the intergovernmental transfer portfolio also raises major issues. The increase in conditional grants and the need to use unconditional grants on core administration reduces the fiscal space for local governments to target local priorities in the budgeting and service delivery process. This may frustrate attainment of the main efficiency benefit attributed to decentralization, weaken incentives for local revenue generation, and diminish the accountability relationship between elected local councils and their constituents. In addition, each new grant scheme comes with budget lines and regulatory processes, increasing complexity and transaction costs, further undermining local discretion and taxing limited local capacity. 2.17 The creation of many new districts generates a number of potential risks. 7 Additional districts may increase central government supervision burdens, may complicate the ability to capture economies of scale, may increase fixed costs for facilities and personnel, and may create logistical challenges for the disposition of staff and resources across the evolving jurisdictional landscape. As districts get smaller, available human resources are being spread more thinly across them. Many established positions in new districts remain unfilled, particularly those requiring greater skills; 8 this, in turn, creates capacity gaps and makes it tempting for central agencies to provide services that districts cannot and to recentralize functions. This blurs the lines of responsibility between central and local 7 There is a discussion of this topic in Reev Consult International (2012). A review of the MoLG proposal to the Parliamentary Committee on Public Service and Local Government regarding 25 new districts describes each in terms of geography, economic and demographic characteristics, and service delivery, but there is no analysis of fiscal capacity or of other factors that are important for district viability. 8 At the same time, even before so many additional districts were created, some districts did not have the volume of work in a particular sector to justify having certain full-time positions, such as a water or roads engineer. 9 governments, and it may distort performance incentives as well, since local governments know the central government will step in if they do not perform and develop the capacities needed to do so. 2.18 Finally, some of the centralizing institutional and procedural steps taken by the central government in recent years may not serve their stated purpose; furthermore, they send signals that reinforce mistrust of local governments. The recentralization of Chief Administrative Officers (CAOs) under the Public Service Commission, for example, was said to ensure recruitment of suitable CAOs, to protect hiring from local political interference, and to allow the transfer of CAOs to other districts without local re-interviewing. Critics of these changes believe that they have softened local accountability, confused lines of reporting and accountability between local leaders and the central government, and reduced control by councils over local government performance. 2.19 To sum up, not all features of the current local government system are likely to increase the ability of local governments to fulfill their obligations and generate the expected benefits of decentralization. Some of the reforms intended to improve on the early performance of the system may in fact have highly uncertain effects. C. The Drivers behind the Post-Decentralization Evolution of the Local Government System 9 2.20 How did the less-than-ideal situation outlined above emerge in a country that so strongly embraced decentralization early on? A number of factors played an important role. 2.21 Bureaucratic dynamics originating in the central government played an important role. As major national policies were articulated in the years after decentralization was adopted, influential ministries that had been developing their own capacities awakened to their potential roles in service delivery and local development, and they realized that decentralization had assigned many powers and resources for these purposes to local governments. As these ministries moved beyond developing their basic systems and capacities, the reforms they adopted increasingly began to come into conflict with decentralization, and new bureaucratic dynamics were set in motion. Public financial management (PFM) reformers in MoFPED, for example, began to pay attention to local budgeting, which had been under MoLG and the Decentralization Secretariat. In the early reform period, MoLG developed and instituted a new local financial management system. 10 Only a few years later, MoFPED decided the new system was inadequate and replaced it with one more consistent with the central government’s system. Around the same time, line ministries began to develop new service delivery approaches, primarily under the National Poverty Reduction Strategy Paper (PRSP) and Sector-wide Approaches (SWAps) promoted and financed by donors to support the Millennium Development Goals. These reforms had centralizing elements that clashed with decentralization policies and the early local systems and procedures developed by MoLG. In addition, they were in some cases financed by the same international development partners who were supporting decentralization, such that donor behavior reinforced some of the competitive tensions among government ministries. 9 This section is based in part on Smoke, Muhumuza, and Ssewankambo 2011. 10 Local Government Finance and Accounting Regulations 1998. 10 2.22 Empirical evidence, largely financed by donors, began to document problems with local expenditure management, raising questions about whether too much responsibility had been decentralized too quickly. The initial concerns arose from early Public Expenditure Tracking Surveys (PETS) conducted by the World Bank and further developed by additional research. 11 This information reinforced decisions by various government agencies to pursue recentralizing activities that were already on their agenda. 12 The convergence of centralizing interests resulted in multiple policies, systems, procedures, and programs that undermined local government powers defined in the legal framework. Steps to support the Poverty Action Fund (PAF), for example, started the process of making the transfer system more conditional in the late 1990s, creating the impetus for or reinforcing the above-noted PFM and SWAp reforms. 2.23 The 2002 Fiscal Decentralization Strategy (FDS), supported by development partners, introduced new budgeting and financial management processes. These reforms did lead to better information and monitoring, but they also limited the budgetary discretion of local governments, thereby stifling a core element of decentralization.13 The FDS provided for some local government flexibility, but it has been constrained in practice, and the stated intention of the FDS to further ease restrictions and increase discretion after local capacity improved was never implemented. 2.24 In addition to the bureaucratic dynamics surrounding intergovernmental relations, shifts in political dynamics also played a key role in reshaping decentralization. In particular, the emergence of multiparty elections in 2006 created incentives for incumbents to adopt policies to help them stay in power. Some believe that the abolition of GPT, the partial recentralization of control over senior staffing and procurement, and the creation of many new districts, among others, are largely or partially related to the NRM’s re-election strategy. New district creation, for example, is officially justified as a means to improve local citizen voice and to better target local service delivery. Although the policy may have that effect in some areas, a number of analysts see it as politically driven and expressly designed to help the NRM to win increasingly tough electoral battles in a more competitive electoral climate. 14 Creating a new district is highly visible and brings resources to a specific location, which benefits those in power. In addition, many of the original districts included multiple parliamentary constituencies, and MPs are thought to sometimes feel overshadowed by the district chairperson. This might be generating demands for breaking up these districts, perhaps to the point where districts reach a one-to-one correspondence with the constituencies. 2.25 Local political dynamics are also critical in how reform plays out, and central-local interactions can be powerful. For example, there are concerns about trust and accountability at the local level. There is some evidence to suggest that local governments do a better job of delivering services for which they have had full responsibility relative to services that involve overlapping central-local responsibilities, 15 but the increasing local government dependence on conditional transfers is likely to weaken local accountability. Moreover, it might be relatively easy for local elites to capture local resources and to use them for patronage in the relatively immature governance environment that prevails 11 See for example Ablo and Reinikka (1998), Jeppson (2001), and Reinikka and Svensson (2004). 12 See for example Jeppson (2002), Kasumba and Land (2003), and Wunsch and Ottemoeller (2004). 13 Details of the budgeting process are outlined in the 2002 Fiscal Decentralization Strategy. 14 There is a detailed discussion of the history of district creation in Green (2008). 15 Ahmad, Brosio, and Gonzalez (2006) elaborate on this point. 11 in many areas. 16 These local political dynamics and other factors discussed above complicate accountability and compromise local capacity, such that the local governance environment is far from ideal. Under such circumstances, citizens may be less willing to pay taxes to local governments that are not seen as credible service providers, further undermining the basic accountability required for effective local governments. 2.26 Finally, the issue of federalism seems to be resurfacing prominently yet again. The kingdoms have always favored a form of federalism that empowers them, and most have little interest in the regional government system discussed above. Some Kabakas (kings), especially the Kabaka of Buganda, the largest and most influential kingdom, have been vocally promoting a federal structure. 17 This trend concerns the government, not only because of the political threat, but also because of recent oil discoveries in territories associated with a number of kingdoms. How the situation with the kingdoms will play out in the coming years remains to be seen, but it is likely to figure prominently in future debates about intergovernmental structures and roles, particularly if districts continue to be weakened by further subdivision and if interest in regional governments continues to grow. 16 Francis and James (2003) provide a discussion of the dual-mode system of “technocratic� and “patronage� driven governance in Uganda, and Smoke (2008) discusses the politics of local revenue. 17 For example, the Sunday Monitor of July 22, 2012 reported that the Buganda Kingdom is insisting on the adoption of a federal system, the return of lands claimed by the kingdom, and the payment of U Sh 27 billion the kingdom believes it is owed by the central government, among other demands. 12 III. Expenditures and Revenues of the Local Governments in the Context of National Public Finance 3.1 Local governments in Uganda rely almost entirely upon central government financing. Transfers from the central government provide well over 90 percent of local government revenues. Funds are transferred to the districts, the municipalities, and to Kampala. How these funds are shared with the lower governments and administrative units is specified in the Local Government Act. 3.2 This section examines the expenditures and revenues of the local governments in the context of national public finance, showing that the dynamics of the former to a large extent reflected changes in the latter. Government expenditure as a portion of GDP has declined from about 21 percent to about 19 percent between 2001/02 and 2011/12, even though real per capita total government expenditures have increased by roughly 20 percent during the same period. The government’s spending priority has shifted to fixed capital. The central government has maintained its relative position in the economy despite declining revenues; in 2011/12, central government expenditures made up approximately the same percentage of GDP as in 2001/02 (about 15 percent, with variation during the period). Its spending priority shifted towards infrastructure and away from social services. Local governments’ share in government spending became smaller. Local governments’ spending on social services (which account for about 60 percent of total spending) has been declining, while non-education and non-healthcare funding shows a modest increasing trend. On balance, various recent changes in the spending policies and priorities of the central government imposed a net fiscal burden on the district governments’ budgets. A review of funding and expenditures in the five core service delivery sectors of the districts (education, healthcare, roads, water supply, and agriculture) that account for about three- quarters of districts’ spending suggests that, on aggregate, this burden amounts to about 10 percent of district budgets in 2011/12. A. Government from a National Perspective 3.3 While real gross government expenditure had been rising in Uganda over the last decade before declining in 2012, it remained virtually flat in per capita terms. Figure 2 shows the levels of total government expenditure in Uganda and its two major components: outlays for final consumption and fixed capital from 2001/02 to 2011/12 (using 2011/12 prices). 18 Note that fixed capital expenditures have grown considerably faster than those for final consumption. As demonstrated in Figure 3, taking account of the rapid population growth in Uganda modifies the picture of government expenditure trends. Real per capita total expenditure has increased at a more modest pace. However, expenditures on final consumption have, at best, remained constant. Rapid growth in fixed capital formation (e.g., from about 55,000 to about 90,000 in 2011/12 shillings per person) account for the growth in real per capita expenditures of government. Most of that growth has taken place since 2004/05. 18 This analysis utilizes published Bureau of Statistics (Statistical Abstract) and Ministry of Finance (notably Background to the Budget) data. When considering aggregate or per capita values across years, it is necessary to adjust for price changes. The Statistical Abstract reports national accounts in constant 2002 prices. The values used here are scaled using the Statistical Abstract’s GDP price index to reflect prices in 2011/12. 13 3.4 Government expenditures as a share of GDP have been declining. Figure 4 provides an initial view from the GDP perspective. While real government expenditure per capita has grown, that growth has been modest relative to the growth in national per capita GDP. Real per capita government expenditures may have increased by about 10 percent since 2001/02, but real per capita GDP has increased roughly 40 percent. A consequence of the different rates of growth is that government now constitutes a smaller sector of the Ugandan economy (Figure 5). Since 2001/02, total government expenditures have declined from about 21 percent of GDP to about 19 percent. Government final consumption has experienced a larger reduction. Meanwhile, fixed capital formation seems to have maintained a relatively constant share, perhaps even increasing somewhat in the most recent years. Determining the right size of Uganda’s government is a matter that must be left to Ugandans, but the decline observed in its relative size is substantial. Figure 2. Real Gross Government Expenditure (2011/12 Prices) 10000 9000 Billions of 2011/12 Shillings 8000 7000 Total 6000 government 5000 Gov't final 4000 consumption 3000 2000 Gov't fixed capital 1000 0 Source: Government of Uganda. 14 Figure 3. Real Government Expenditure Per Capita (2011/12 Prices) 2011/12 Shillings Per Capita 300000 250000 200000 Total gov't pc 150000 Gov't final consumptio 100000 n pc Gov't fixed 50000 capital pc 0 Source: Government of Uganda. 3.5 While total government expenditure has declined as a portion of GDP, the expenditures of the central government seem to have remained a relatively constant share. There was a decline in 2005/06 followed by an upward trend (Figure 6). The blip in 2003/04 is the result of an extraordinarily large debt repayment that year. Large defense expenditures accompanied by unusual oil revenues (seemingly a one-time occurrence over this period) are the cause of the blip in 2010/11. Figure 4. Real Per Capita GDP and Government Expenditure (2011/12 Prices) 1600000 1400000 1200000 2011/12 Shillings Per 1000000 GDP pc 800000 600000 Govt 400000 (total) 200000 pc 0 Source: Government of Uganda 15 Figure 5. Government Expenditures as a Portion of GDP 0.25 Total gov't 0.2 Portion of GDP 0.15 Gov't final consumptio 0.1 n 0.05 Gov't fixed capital formation 0 Source: Government of Uganda. Figure 6. Central Government Expenditures as a Portion of GDP 0.3 Total 0.25 (recurrent & developmen 0.2 t) Recurrent Portion of 0.15 0.1 Developme 0.05 nt 0 Source: Government of Uganda. 16 3.6 The government’s spending priority has shifted to fixed capital. At best, real per capita government final consumption expenditure has remained constant. Fixed capital formation has increased about 50 percent in real per capita terms.Relative to GDP, government final consumption has decreased from 16 percent to about 11 percent, while government fixed capital formation increased somewhat (from about 5 to 6 percent of GDP). Central development expenditures increased relative to GDP while recurrent expenditures were, at best, constant. Real per capita spending by the central government increased (about 45 percent over 2001/02 levels). At the same time, real per capita development expenditures approximately doubled and development expenditures increased from about 25 percent to over 40 percent of central spending. Real per capita recurrent expenditures rose but modestly. 3.7 This shift in priority might be of a cyclical nature, reflecting Uganda’s inherent difficulty in finding a balance between spending on infrastructure and on social services. Uganda has very high fertility (with a total fertility rate of more than six children per woman) and a very young population (with a median age of about 15 years). Given the scarcity of its fiscal resources, the government finds it difficult to invest in both infrastructure and social services in amounts that would be sufficient for the majority of its young people to find jobs when they enter the labor market. An increase in one of these categories of investment leads to gaps in the other. In the 1990s and early 2000s, the government focused its efforts on social services, until it became apparent that infrastructure requires urgent attention. A subsequent increase in the investment in infrastructure explains the shift in spending priorities to favor fixed capital. Sooner or later this shift might be reversed, following the same logic that led to its occurrence. However, judging by past experience, such a reversal may not occur for a decade or more. 3.8 The difference between the central government’s expenditures and revenues reflects transfers to local governments that are accounted for separately. Expenditures and revenues tend to move together, so it is valuable to look at the revenue side as well as expenditures. The Ugandan central government’s revenues have ranged between 15 and 20 percent of GDP. This level exceeds the level of the central government’s expenditures, which have typically ranged between 13 and 14 percent of GDP. 3.9 The central government’s revenues as a share of GDP have trended downward. From 2003/04 to 2009/10, this share fell from 20 percent to 15 percent. This trend (Figure 7) corresponds to the almost steady decline in the contribution of grants, which fell from 7 to 8 percent of GDP to 3 to 4 percent. Domestic-sourced revenues have increased as a share of GDP, but that growth has only partially offset the reduction in the contribution of grants. Regardless, domestic revenue sources have grown from 60 percent to make up over 80 percent of central government revenue. The increasing importance of domestic revenues is demonstrated in Figure 8. Figure 9 shows that both budget support grants and project grants have diminished as a share of central revenue. However, the more significant budget support funding has had a larger absolute and relative decline. 17 Figure 7. Central Government Total, Domestic, and Grant Revenues as a Portion of GDP 0.25 0.2 Portion of GDP 0.15 0.1 0.05 0 Total Revenue Domestic sources Grants Source: Government of Uganda. Figure 8. Central Government Domestic and Grant Revenues as a Portion of Total Revenue 1 0.9 Portion of Total Revenue 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Domestic Grants Source: Government of Uganda. 18 Figure 9. Central Government Grant Revenue as a Portion of Central Government Total Revenue 0.45 0.4 Portion of Total Revenue 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 Grants Budget support Project support Source: Government of Uganda. 3.10 Despite the pressures on the central government budget caused by declining grants, the central government’s real per capita expenditure has increased rather significantly. It grew from about U Sh 140,000 to as much as U Sh 205,000 (2011/12 shillings) per person (or on the order of 45 percent) over six years (Figure 10). Both recurrent and development per capita expenditures have grown, but those for development have more than doubled to amounts on the order of U Sh 90,000, while recurrent outlays may have risen by 15,000 and seem to be leveling off at perhaps U Sh 120,000. Development expenditures have, since 2005/06, been the main driver of larger real per capita central government expenditures. Figure 11 shows development outlays having risen from about 25 percent to over 40 percent of the central government’s outlays. 19 Figure 10. Government Real Expenditures Per Capita (2011/12 Prices) 300000 2011/12 Shillings Per Capita 250000 200000 150000 100000 50000 0 Total (recurrent & development Recurrent Development Source: Government of Uganda. 3.11 Transfers to local governments are one of the central government’s major expenditures. Local governments receive most of their funding (over 90 percent) from central government transfers. 19 Transfers could be categorized into three broad categories: conditional grants, unconditional grants, and equalization grants. The first category is by far the largest in Uganda, representing about 90 percent of the total. Allocation of funds to specific local governments is determined by a complex mixture of historical practices, need-based formulas, and availability of funds. 20 3.12 Transfers to the local governments have recently exhibited downward trend. As reflected in Figure 12, transfers to local governments have declined as a share of central government revenues and expenditures in the last two years. The change is rather significant, from about 25 percent to something closer to 20 percent; that is, a reduction on the order of 20 percent. The decline of grants to local governments relative to GDP has been more gradual but has extended for a longer period and, relatively, is every bit as large. Possibly, this reflects a reaction on the part of the central government to reductions in foreign aid. 19 In MoLG (2012), it is stated that “Currently, there is no local government that contributes more than 2 percent of the total annual districts budgets from local generated revenue.� 20 For a more detailed discussion of this topic, see Steffensen and Ssewankambo 2011. 20 Figure 11. Central Government Recurrent and Development Expenditures (as Shares of Total Expenditure) 1 0.9 0.8 0.7 0.6 Portion of Total 0.5 0.4 0.3 0.2 0.1 0 Recurrent Development Source: Government of Uganda. Figure 12. Grants to Local Government as a Percentage of Central Government Revenue and Expenditure and of GDP 30 25 20 15 Percentage 10 5 0 % of central gov't revenue % of central gov't expenditure % of GDP Source: Government of Uganda. 21 B. Local Governments 3.13 While central government real per capita expenditures have increased, local real expenditure per person has remained relatively constant over most years but perhaps exhibits a slight downward trend overall. 21 Certainly, high inflation in 2011/12 eroded the real purchasing power of budgets. Figure 13 reports real per capita funds of all local governments from two sources, the World Bank’s BOOST database for Uganda and the LGFC. Data from the first source show the funds budgeted by local governments as transfers from the central government. The second source provides data on per capita funds actually released by MoFPED to the local governments, which, due to the need to meet various conditions, are typically somewhat less sizeable than the amounts budgeted. The two measures of funding track each other relatively well, with releases averaging at 95 percent of the budgeted amounts over the seven years with data in common. The BOOST data allows examination of both recurrent and development budgets for local governments in aggregate. Over the 2003/04–2011/12 period, the recurrent budgets averaged 78.5 percent of the total (and development budgets 21.5 percent). Figure 14 reports real per capita recurrent expenditures by local governments and also the real recurrent budgets from the BOOST database. 22 According to the BOOST data, real per capita recurrent budgets have trended downwards in recent years from a peak of U Sh 49,213 to U Sh 38,838 (2011/12 shillings, GDP deflated), representing a drop of about 21 percent. Recurrent expenditures are also reported in the Statistical Abstracts. The recurrent expenditures extend back an additional three years to 2000/01. The reason for the blip in 2005/06 is not obvious but may result from a change in accounting. Data for 2011/12 excludes Kampala City, which contributes (but only modestly, less than 1100 shillings per capita) towards the lower value that year. The two measures do not match up exactly but are similar. Over the common years (but omitting the unusually large expenditures in 2005/06), expenditures averaged U Sh 49,906, while the budgets averaged U Sh 44,945 shillings. To sum up, both suggest a downward trend during the most recent years. 21 A comparison of data from alternative sources of local data is useful. Data on local government own-source revenues are not reported in the public statistical sources. Similarly, development expenditures are not reported by the Bureau of Statistics, so information from that source is limited to recurrent expenditures. However, information on development functions and for total budgets is available in the World Bank’s BOOST database and from files provided by the LGFC. The LGFC data, however, do not cover quite as long a period of time and have additional limitations. Hence, more attention is focused on the BOOST data. Estimates covering all local governments appear in the LGFC (2012) and in Steffensen and Ssewankambo (2011). 22 This total is calculated by combining district and urban local governments for the period from 2000/01 to 2004/05. The Statistical Abstract reports these values in 2002 shillings. For ease of comparison, all values are converted to 2011/12 shillings using the GDP deflator for adjustment. 22 Figure 13. Real Per Capita Funds of All Local Governments (2011/12 Prices) 80000 70000 2111/12 Shillings Per Capita 60000 50000 40000 30000 20000 10000 0 Real PC Releases, LGFC Real PC Budgets, Boost Source: LGFC and BOOST database. Figure 14. Comparison of Local Government Real Per Capita Recurrent Funds: Expenditures from Statistical Abstracts and Budgets from the BOOST Database (2011/12 Prices) 70000 60000 50000 2011/12 Shillings Per Capita 40000 30000 20000 10000 0 Boost Budget Stat. Abst. Expend. Source: Uganda Bureau of Statistic (UBOS) and BOOST database. 23 3.14 It appears that education and health are (or certainly have been) a lower priority than other services. Figure 15 reports trends in three categories of local government expenditures as reported in the Statistical Abstract: education, health, and non-education and non-health (or other). The non-education and non-health category includes public administration, public order and safety, community services (water and other), and economic services (agriculture, roads, and other). The largest outlays are for education, but the real per capita expenditures have declined since 2005/06, falling from over U Sh 26,000 (2011/12 shillings) per capita before 2006/07 to about U Sh 20,560 in recent years. The health expenditures of local governments have been more stable but seem to be trending downwards as well. Other expenditures (i.e., those not related to education or health) trended upwards (from about U Sh 13,000 to over U Sh 23,000) until 2010/11. 23 Figure 15. Real Per Capita Local Government Recurrent Expenditures by Major Components (2011/12 Prices) 30000 25000 20000 2011/12 Shillings Per Capita 15000 10000 5000 0 Education Non-education & non-health Health Source: UBOS. 3.15 Recurrent expenditures dominate local government budgets but the share of development outlays is significant and has been increasing. The level and pattern of recent development funding of local governments is provided in Figure 16, which shows the real per capita amounts budgeted for development since 2003/04. For the reported period, development averaged about U Sh 12,300 (2011/12 shillings) per capita, or about 21.5 percent of budgets. Development funding has been rather uneven over the years, declining from U Sh 14,000 to just over U Sh 10,000, then increasing to over U Sh 15,000 but dropping again to about U Sh 12,300 per person in 2011/12 in real terms. Development budgets have increased since 2006/07 in real per capita terms, while recurrent budgets have declined. The combined 23 The GDP deflator has been used here to adjust for the inflation of local recurrent expenditures. An alternative is the price index for government final consumption. Using the latter makes minor changes in the amounts and does not change the patterns or trends observed. 24 total budget also seems to have trended downwards, especially if the 2011/12 figures are sustained. 24 The trends in development, recurrent, and total budgets are provided in Figure 17. Figure 16. Real Per Capita Local Government Development Budgets (2011/12 Prices) 18000 16000 14000 12000 2011/12 Shillings Per 10000 8000 6000 4000 2000 0 Source: BOOST database. Figure 17. Real Per Capita Budgets: Development, Recurrent, and Total (2011/12 Prices) 80000 70000 2011/12 Shillings Per Capita 60000 50000 40000 30000 20000 10000 0 Total Recurrent Development Source: BOOST database. 24 The total in 2003/04 exceeds the sum of the recurrent and development budgets because of (almost U Sh 8,000 2011/12 shillings of) donor funding that was not allocated between recurrent and development expenditures in the BOOST data. 25 3.16 The number of municipal authorities has been increasing along with the number of districts. In an approximate way, the division separates rural local governments from the (at least larger) urban local governments. 25 In 2003/04, the populations of municipal council governments (excluding that of Kampala) averaged about 54,600. In that year, district governments had an average population of about 426,000. Since then, the number of districts has expanded from 56 to 111 and the number of municipal councils has multiplied (as of 2010) from 14 to 23. Average populations declined to about 279,000 in districts and increased to 60,700 in municipalities. As discussed in Section 3, districts and municipalities have different arrangements for service delivery in some sectors. 3.17 The pattern of the districts’ total budgets over the nine years in question is somewhat uneven, but there is a downward trend in per capita real total funding. Over a period of eight years, the real per capita 26 budgets of the district governments have declined from U Sh 58,546 to U Sh 49,864, which represents a decline of 14.8 percent. Even in 2010/11, the real amount was U Sh 52,150, or 10.9 percent less than the 2003/04 level (Figure 18). 27 The recurrent and development funding of the district local governments moved in different ways, as indicated in Figure 19. Development budgets in the districts appear to have declined somewhat during the first half of the period and then to have recovered. The real per capita development funding was about 16 percent larger in 2010/11 than in 2003/04 and only five percent less than the base year in 2011/12. While the recurrent budgets experienced real increases until 2006/07, they have since declined with real per capita funding levels in 2010/11 and 2011/12 at 84.4 and 83.1 percent of that in 2003/04. The decline in the recurrent budgets dominated the gain in the development budget, so the real per capita total budget has in fact decreased, as noted above. 25 The Statistical Abstracts report recurrent expenditures of district and urban local governments. The districts there are differently defined than in the analysis above. In contrast to the districts above, the urban areas listed in the Statistical Abstracts include town councils as well as municipal councils. The urban population in the Statistical Abstract for mid-2011 is reported as 4,859,500 (and the rural population as 28,080,300 for a national population of 32,939,800 in that year). That total includes the populations of Kampala and of the municipal councils, which came to 2,994,600, and a total town council population of 1,864,900. The Statistical Abstracts also report recurrent expenditures for district and for “urban authorities.� Because the more comprehensive LGFC and BOOST databases distinguish financial information only for districts and for municipal councils (and not the town councils), the analysis in this report (unless otherwise noted) distinguishes only between districts and municipal councils as in the BOOST and LGFC sources. Examination of the more broadly defined urban and (rural) district local government finances would be informative but is beyond the scope of this study. 26 Population data is not available for all years. Using data available from the Statistical Abstract would require comparing 2003/04 with 2010/11 and 2011/12. That poses the possibility that the specific years are anomalies and not entirely comparable. To avoid that issue, populations are interpolated so as to obtain a sequence over the years and provide a better assessment of the trend. Additional population estimates from the LGFC necessitate interpolating populations for five years, from 2003 to 2007 (specifically those for the municipal councils). District local government populations are estimated by subtracting the estimated populations of the municipal councils from the national population. While the interpolation seems to pose no problems, the greatest confidence can be placed in the information for the years at each end (2003/04 and 2010/11 and 2011/12), as those are (or are closest to) being based on the official population statistics. 27 There is a complication with the accounting for donor funding in 2003/04 because it was not assigned to districts and municipal councils. To preserve as much data as possible, 2003/04 is still taken as the initial year. However, at the risk of understating budgets, donor funding is not included here. So estimated, the results for the district budgets from the BOOST data appear in Figure 18. If the unassigned donor funds in 2003/04 were known and included as they were part of the actual budgets, the reduction in real funding would be larger. 26 Figure 18. Real Per Capita District Budgets (2011/12 Prices) 70000 60000 2011/12 Shillings Per Capita 50000 40000 30000 20000 10000 0 Source: BOOST database. Figure 19. Real Per Capita District Recurrent and Development Budgets (2011/12 Prices) 60000 50000 2011/12 Shillings Per Capita 40000 30000 20000 10000 0 Recurrent Development Source: BOOST database. 27 3.18 District governments’ own-source revenues are very low. Own-source revenues are those that are obtained from revenue sources (taxes, charges, etc.) over which local governments have authority and are able to levy and use for their own purposes. District own-source revenues have traditionally not been significant since Uganda’s local governments have largely relied upon transfers from the central government. Estimates vary, but since the abolition of the graduated tax in 2003/04 it is clear that the own-source revenues of district governments have further declined from what already constituted a single-digit percentage of their budgets, and currently represent between 2 and 7 percent of these budgets. In addition, while the districts have come to rely even more on central transfers, the central government has made those transfers more conditional. In contrast, own-source revenues of the municipalities make up roughly 30 percent of their budgets (LGFC 2012). 3.19 The revenues available fund only 36 percent of the costs of meeting national standards for local government services, according to an assessment made by the LGFC. Extending information from an earlier Ministry of Local Government study estimating the cost to local governments of providing services to national standards, the LGFC estimated the (recurrent only) costs as of 2009/10 and compared those to the funds provided (LGFC 2012). The Commission’s figures indicate a 64 percent gap. 3.20 Municipal resources appear to be declining, although the expansion in the number of municipal councils complicates the measurement of changes in their funding. To estimate funding at the municipal level, the funding of the 13 councils that remained unmodified during the period is examined. The total per capita budgets of those councils since 2003/04 appear in Figure 20. Initially, the per capita budgets of the municipal councils are considerably greater than those of the districts. In 2003/04, these municipalities had average real per capita budgets of U Sh 93,970 (2011/12 shillings) while the districts averaged U Sh 58,546; that is, about 60 percent more than in the districts. The municipal resources have declined since 2003/04 to 83.5 percent of that level by 2010/11 and to 73.3 percent in 2011/12. The reductions are greater than those experienced by the districts over the same period. 28 Figure 20. Real Per Capita Budgets of Thirteen Municipal Councils (2011/12 Prices) 120000 2011/12 Shillings Per Capita 100000 80000 60000 40000 20000 0 Source: BOOST database. 3.21 The recurrent and development budgets of the municipalities have taken somewhat different paths. As Figure 21 demonstrates, despite considerable growth until 2006/07, real per capita recurrent budgets in 2010/11 were essentially at the same level as in 2003/04 and appear to have continued the downward trend as they declined by 11.7 percent (to U Sh 59,954) in 2011/12. Real per- person development funds in 2011/12 are at almost one-third (34.5 percent) of the 2003/04 levels in these municipal councils. Development budgets fell sharply in 2006/07 and experienced comparatively modest growth since then. 3.22 While the municipal councils seem to have fared better than the districts on the recurrent side, they have done less well with their development budgets. The declines in the overall budgets of these municipalities are somewhat greater than the reductions in the districts. 29 Figure 21. Real Per Capita Recurrent and Development Budgets of Thirteen Municipal Councils (2011/12 Prices) 100000 90000 80000 2011/12 Shillings Per Capita 70000 60000 50000 40000 30000 20000 10000 0 Recurrent Development Source: BOOST database. C. District Core Services: Fiscal Impacts of Changing Transfers and Shifting Responsibilities 3.23 Districts are responsible for delivery of a broad range of services. Five core service areas are selected for analysis in this report: primary education, primary healthcare, roads, water supply, and agricultural extension. These five are widely considered to be vital to residents’ wellbeing and, as a group, they dominate district budgets. Developments in those sectors (with some extensions, particularly in schooling) are examined. Those developments include changes in funding resulting from shifting priorities and changes in fiscal circumstances and also due to reallocations of responsibilities and the introduction of new services. A central concern is whether, after many changes, the districts are more or less able to provide local services. In an effort to address that issue, the examination concludes by making estimates of the net impacts by sector and of the overall capacity to deliver services. 3.24 This section examines the net fiscal impacts of changing transfers and shifting responsibilities on district budgets. It starts by providing an aggregate fiscal picture for the five core service areas and proceeds to look at fiscal impacts of changes in intergovernmental fiscal transfers and in the division of responsibilities between the central and local governments on each of them. Funding and the factors that affect funding are the primary focus. 28 Although there are limitations, funding levels over time indicate the levels of resources available and, assuming uniform management and conditions, those can be expected to correspond to outputs. In some cases, additional and alternative indicators of service delivery and performance are noted. The section is based in part on a detailed assessment of the fiscal capacities of district governments to provide services in each of the five areas named in Annex 1. 28 The World Bank BOOST database is the data source. That data is deflated by the GDP index. 30 3.25 The five core areas analyzed represent about 60 percent of an average district’s budget. Figure 22 shows the individual and aggregate real per capita district budgets in the five core areas. The dominance of primary education and, to a lesser degree, primary healthcare as expenditure categories is clear. Primary education averaged 59 percent of funding in these areas over this period and primary healthcare about 17 percent. Growth in the agriculture expenditures is probably the most obvious change in the expenditure pattern. Less obvious is the growth in road maintenance, which increased from about four to seven percent of these budgets. 29 Total per capita budgets in these five areas peaked in 2006/07 at U Sh 37,149 (2011/12 shillings) but have declined since then to U Sh 33,767 in 2009/20 and, if adjustment is made for the missing road maintenance data, to about U Sh 33,500 in the latest years. Primary schooling and primary healthcare account for most of that decline. Figure 22. Real Per Capita Budgets in Five Core Areas (2011/12 Prices) 40000 35000 2011/12 Shillings Per Capita 30000 25000 20000 15000 10000 5000 0 Agriculture Primary schooling (recurrent) Primary health care Water supply (rural and district urban) Road maintenance Source: BOOST database. 3.26 The other components of the budget include non-primary education, non-primary healthcare, social development, public sector management, accountability, and a residual. Three of those items account for about 94 percent of the other category shown here. Those are, in recent budgets, non-primary education (32.5 percent), non-primary healthcare (10 percent), and public sector management (51.5 percent). 30 Changes in the funding of the other category compared to the five core areas are better reflected in Figure 23. Real per capita funding of the five core areas rose, then fell, and in 2011/12 is about the same as in 2003/04. Other funding dropped but more or less to a lower plateau.31 Hence, while total real per capita funding of the districts has declined since 2003/04, funding in the other 29 Data on road maintenance expenditure is not available for 2010/11 and 2011/12 and not shown in the Figure. 30 Non-primary education and non-primary healthcare each represent about 25 percent of districts’ total education and healthcare budgets respectively. 31 The 2003/04 levels of expenditures on “other� and the “five core� categories need be viewed cautiously. The allocation of funds changed following that year and a smaller portion was assigned to the other category. Hence, “other� is unduly large (and the five areas unduly small) relative to post-2003/04 figures. 31 category has fared less well than that for the core five over the full period. However, the core five areas have seen funding diminish more since 2006/07. Figure 23. Real Per Capita Budgets of Five Core and Other Expenditure Categories (2011/12 Prices) 40000 2011/12 Shillings Per Capita 35000 Sum of five areas (with 30000 missing road 25000 maintenance estimated) 20000 15000 Other 10000 5000 0 Source: BOOST database and World Bank staff calculations. 3.27 Are district governments better or less able to fund local services than in the past? Estimates of the impacts based on the preceding analyses are presented in Table 1 by sector and by major program for the years examined (primarily for developments over the years 2003/04 to 2011/12). Those estimates are expressed in terms of 2011/12 shillings per person. In large part, this reflects a shift in the expenditure priorities of the central government, and most of it occurred in the education sector. 32 Table 1. Net Fiscal Impact on Districts of Recent Developments in the Finance of Five Core Areas Net Impact in 2011/12 Shillings Per Capita Program Area Program Explanation Sector Impact Impact Low High Low High est. est. est. est. Educat Primary ion -3070 Reduction in real per capita salary and non- salary recurrent grants from 2003/04 to 2011/12. -3875 -5925 Secondary Increased enrollment with universal secondary education but without corresponding increase in -805 -2855 funding. Health Primary Salaries: funding approximately constant. Non-Salary: changes associated with National Medical Stores have unclear net effect. 0 1000 0 1000 Development: Increased funding of about 1000 shillings. Roads Maintenance Increased district road maintenance grants. 800 800 Responsibiliti Transfer of responsibility for 9,000 km of es district roads to UNRA could imply an 800 1875 0 1075 additional district benefit of 0, 23, or 50 percent in road maintenance funding. Water DWSDCG Grant Approximately constant per capita real funding. 0 -70 DWSDCG is now funding only 75 percent of the number of new service connections as in 2002/03 to 04/05. If it were to fund the same ? 0? - 70 number, an additional U Sh 780 transfer per person would be required. Donor funding to NGOs appears to have met funding gap. Small Towns Smaller real transfers for operation and -70 -70 maintenance of small town water systems. Agricul ture NAADS and Large increase in real (NAADS-related) other funding; from 1,763 to 5,371. Expected district agriculture and sub-county contributions imply a 468 per 0 -468 0 -468 transfers capita extra cost at district level if met and zero if not met. TOTAL (net burden) -1070 -5663 Source: World Bank staff calculations. Note: DWSDCG stands for District Water and Sanitation Development Conditional Grant; NAADS stands for National Agricultural Advisor Services. 33 3.28 Real per capita recurrent funding for primary education has declined since 2003/04. That decline is estimated to represent about U Sh 3,070 (2011/12 shillings) per capita. Developments in secondary education have also had a substantial impact on the education sector, but the magnitude is less certain. Adoption of a policy of universal secondary education resulted in a 40 percent increase in secondary enrollments, but funding failed to match that growth. At the funding levels experienced in 2009/10 and 2010/11, the expansion of secondary enrollments was effectively an unfunded mandate, imposing a burden of U Sh 2,855 per capita on districts. Funding for secondary schooling increased substantially in 2011/12 (although not enough to compensate for enrollment growth). If that level of support was to continue, the net burden would be reduced to U Sh 805 per person. Thus, the net negative impact of developments in the education area is estimated to be between U Sh 3,875 and U Sh 5,925 per capita. These amounts are significant in that they represent between seven and 10 percent of the total transfers to the districts and between 17 and 25 percent of the recurrent transfers for primary and secondary schooling. 3.29 Various developments occurred in the health sector and the net effect there is nebulous. The National Medical Stores assumed responsibility for providing health supplies and medicines to districts. This resulted in a decrease in transfers and a decrease in costs. The net effect may be zero but the numbers to evaluate that fact are not yet available. Funding for staff has been relatively constant in real per capita terms but development grants increased sharply (perhaps by U Sh 1,000 per person) as of 2009/10. Some of that may have been catching up (grants declined earlier), but at least a portion of this increase should translate into improved district health facilities. That is, the districts may have experienced anywhere from no net improvement to perhaps a U Sh 1,000 per person improvement in health funding. 3.30 The impacts of changes in the road sector are even more difficult to assess. Support for roads in Uganda has definitely increased and districts benefited from that. Transfers for district road maintenance increased about U Sh 800 per person. In addition, approximately 9,000 kilometers of district roads were transferred from district responsibility to that of the UNRA. The impact of that transfer on district budgets is not obvious. It is possible that the central government simply substituted a responsibility transfer for a greater increase in conditional grants for realizing improvements in the national road network. If so, the net effect can be considered to be zero. Alternatively, if that possible substitution is ignored, the reduced responsibilities convert to an effective 23 percent increase in the support for maintenance of all (district and sub-county) roads in districts, or a 50 percent increase in support for those roads that are strictly the responsibility of the district government. So, the transfer of road responsibilities could be interpreted as providing a further benefit or additional support of U Sh 495 per capita, or of U Sh 1,075. Thus, the total increase in road funding could range from U Sh 800 to U Sh 875 per person. Funding for road rehabilitation and new development, however, remains a problem. 32 3.31 Funding of water supply and sanitation development in the rural areas comes almost entirely through the District Water and Sanitation Development Conditional Grant DWSDCG transfer and a small transfer to assist the operations and maintenance of water systems in small towns. The small town transfer has diminished to less than half in real per person terms, for a loss of U Sh 70 per 32 In the 2012/23 budget, the central government announced a program to provide districts with road equipment that would enable them to undertake additional road work themselves. Details of the development of that initiative and its fiscal implications are yet to be assessed. 34 capita. While DWSDCG has been relatively constant in real per capita terms, the number of persons added with that funding has declined to about 75 percent of those earlier in the decade. There appears to be a growing gap between population growth and the number of additional persons being added to water systems. At the moment, donor organizations appear to be playing an important and a growing role in compensating for that gap in order to mitigate the impact on districts’ finances and prevent the deterioration of services. 3.32 Funding for agriculture—essentially for extension services—has grown dramatically (three fold) as a result of a policy initiative to improve agricultural productivity. While rural communities have benefited from this effort, the impact on the district governments’ finances is small and mixed. Under these programs, district and sub-county governments are to contribute (initially five percent each and now six percent) to the funding. Thus, increased support also implied additional district resources. At the current level of transfer, that contribution amounts to U Sh 468 per capita. However, contributions at the district and sub-county levels are irregular and sometimes non-existent. This could mean that the additional financial impact is as small as zero. So, the overall added burden on districts is expected to be somewhere between zero and U Sh 468 per person. 3.33 Summing across these five areas, recent developments are estimated to have imposed some net negative financial burdens on district governments. Depending upon the assumptions made about the impacts of various programs, that burden could be a low as U Sh 1,070 per person or as high as U Sh 5,663. The negative impact is driven by developments in the schooling area, and especially by the inadequate funding to support the universal secondary schooling initiative. Funding for secondary education may be improving. However, the sustainability of that trend and the implications of secondary education on district finances need further examination. Even so, the erosion of funding for primary schooling still stands out. Differences across the functional areas indicate that, even if the overall net effect were small or nonexistent, the impacts on specific sectors may still be a concern. 3.34 The estimated negative net fiscal impacts are significant relative to district budgets. The U Sh 5,663 amount represents roughly 10 percent of district per capita budgets and U Sh 1,070 is almost two percent. 33 As percentages of conditional funding specific to those activities (including those for secondary schooling), the percentages are on the order of 15 and 3 percent, respectively. Overall, the negative impacts have been notable, especially on schooling. 34 3.35 This negative fiscal impact and the sectoral trends described in this section are in large part caused by the shift in spending priorities of the central government, from social services toward infrastructure. The central government remains the primary source of funds for local governments; as 33 These take U Sh 54,699 per capita as representative of district budgets. From 2008/09 to 2010/11, the average across districts was U Sh 55,800 (2011) and stable. The 2011/12 level fell to U Sh 51,337. The figure of U Sh 54,699 represents the average of those four years. 34 Some note the recentralization of CAOs and the central government’s payment of CAO salaries. That change and other similar changes on the administrative side were ignored in the discussion above because they fell outside the core areas and because they are relatively small. The CAO salary makes up 0.3 percent of the average district budget. Even if one includes the entire conditional funding specific to administrative tasks, the percentage runs at about 1.2 percent, and only increases to 2.5 percent after 2010/11. That increase is likely (at least in part) a product, perhaps related to start-up funding, of the number of new districts created that year. 35 such, any change in spending priorities of the former inevitably affects the latter. Social services represent a much larger share in the districts’ portfolio than does infrastructure. 36 IV. Fiscal impacts of the increase in the number of districts 4.1 This section considers the fiscal implications of the growth in the number of districts in Uganda over the last decade and a half. It draws in part on three recent studies of local government in Uganda. Each of these studies had rather different specific motivations and objectives. Therefore, the studies undertook a variety of analyses and made observations and recommendations on a wide range of issues. A common feature of the three is that the growth in district numbers receives attention, while concerns about that trend are highlighted. 35 The positions resulting from their assessments are expressed clearly in their summaries. Some examples illustrate the positions expressed. Steffensen and Sswankambo (2011) focus on the urban municipalities and the central-local transfer system. They note among other major challenges the “continued fragmentation� of the local government system and the resulting pressures placed on administrative capacity and the funding system. The report of the LGFC concentrates on the districts and examines 12 of them in detail (LGFC 2012). One of three problems in local government financing identified in this report is “…the rapid expansion in the number of districts compelling local governments to spend the bulk of their discretionary financing on administration.� They, too, point to increasing problems in filling administrative positions. The report to the Ministry of Local Government looks broadly at restructuring and reorganization and, in particular, at the option of regional government (MoLG 2012). Reference is made of the “burgeoning costs of public administration� resulting from the increasing number of local governments and of the fact that this “increase[es] drastically the cost of provision of decentralized services.� Understaffing is also noted; on average, about 40 percent of administrative positions in district governments remained unfilled in 2011 (over 90 percent in some districts). Obviously, growth in the number of districts is seen to impose large financial and human resource burdens that diminish their ability to deliver local services. However, all three studies lament the serious underfunding of local government that goes well beyond the cuts attributable to the growing number of governments. 4.2 The impact of the creation of new districts on public expenditures has been low so far but is likely to increase in the future, making district proliferation a potential “time bomb.� This is because the number of vacant positions in district governments has been increasing along with the number of districts, and also because the real wage bill of district governments has been eroded by a recent outburst of inflation. Longer-term fiscal costs could be even higher and increase over time because there is a risk that value-for-money in public services would decline if the understaffing of district governments persists. These problems, however, appear to be driven by the institutional design of district governments and their fiscal relationship with the central government rather than by the size of districts. The average population of a district is currently more or less on par with that of similar jurisdictions in other countries. Diseconomies of scale are hardly inherent to this size of a jurisdiction. 35 Although districts get the attention, a parallel increase in the number of urban authorities has occurred. 37 A. Ugandan Local Government in an International Context 4.3 For comparative purposes, it is best to consider Uganda’s local governments in terms of the populations under the jurisdiction of each. The number of local governments (since 2010) and the average population of each (as of mid-2012) by government level are reported in Table 2. Local governments are distinguished from administrative units. Kampala City is the largest, with about 1.7 million people. There are now 22 municipal councils with an average population per council of 62,527. Kampala and the municipal councils have divisions. The population per division is 344,660 in Kampala and 21,494 in the municipalities. The 111 districts (outside Kampala) encompass the towns and the rural areas falling under the sub-counties. The 165 towns have their own councils and an average population of 11,717 persons. The rural areas under the districts average 262,154 people per district and 25,370 per sub- county. There are many lower-level administrative units underlying the local governments. Out of a national population of 34.1 million, the average population per parish/ward (of which there are 7,771) is 4,392, and the average among the 66,739 village/cell/zone units is 511 persons. Table 2. Local Authorities in Uganda: Number and Average Populations Total Number, Type Population by Average 2010 Type, 2012 Population, 2012 Local Governments Kampala City 1 1,723,300 1,723,300 Divisions 5 344,660 Municipal Councils 22 1,375,600 62,527 Divisions 64 21,494 Districts Towns 165 1,933,400 11,717 District (rural) 111 29,099,100 262,154 Sub-County (rural) 1147 29,099,100 25,370 Administrative Units Parish/Ward 7771 34,131,400 4,392 Village/Cell/Ward 66739 34,131,400 511 Source: UBOS. 4.4 The population per local government differs considerably across both developed and developing countries. McMillan (1995) provided that information for a number of developed countries as of the early 1990s (see Table 3). Useful to note is that the numbers reported follow a period of substantial consolidation of local government in many of those countries, much of which occurred during the 1960s. The result was a reduction in the number of local governments and an increase in population per jurisdiction. At 1,300, the smallest is in France; the largest is in the United Kingdom, at 125,000. The average across the 19 countries is 22,700. Federal countries tend to have somewhat smaller populations per local government than do unitary countries (about half as large). Overall, of the 19 countries 38 considered, only two average more than 35,000, and only three come in under 5,000. 36 The pattern does not appear much different among developing countries. Table 4 reports information from Shah (2006) for nine developing and transition countries. Aside from India, where the numbers for district sizes in rural and urban areas are substantially different (3,278 versus 68,027), the range is again large, spanning from 4,331 in Kazakhstan to almost 240,000 per jurisdiction in South Africa. In this group, only three countries have relatively large averages. The median is Poland at almost 19,000 persons per local government (when one aggregates across rural and urban India). which is similar to the case of developing countries. Also, only South Africa has an average exceeding that of the United Kingdom. Although the sample for developing countries is small, in only three of the nine cases does the average population exceed 30,000; a pattern similar to that found in developed countries. Table 3. Approximate Number of Municipal Governments in Selected Developed Countries and Average Population, 1990s Number, Average Population, Country 1991 1994 Northern Europe Denmark 275 19,000 Finland 460 11,000 Netherlands 647 18,000 Norway 448 9,000 Sweden 286 30,000 West Germany 8,413 20,000 U.K. (England only) 401 125,000 Southern Europe Belgium 589 17,000 France 36,400 1,300 Italy 8,097 7,000 Portugal 275 34,000 Spain 8,022 47,000 Federal Countries Australia 838 19,000 Canada 4,993 5,600 United States 83,186 6,600 Austria 2,304 3,000 Switzerland 3,021 2,100 West Germany 8,413 20,000 Japan 3,255 37,000 Source: McMillan 1995. 36 A recent update that can be found in “Local Government Reform: In Brief,� which is available at the National Association of Local Authorities in Denmark website, www.kl.dk/English/Local-Government-Denmark/, indicates only a few changes. Due to reform in Denmark (discussed below), the country’s average population per district has increased to 55,000 and the average in the Netherlands has also increased to 34,900. On the other hand, decentralization in Spain has caused its average to drop to 5,300. 39 Table 4. Number of Municipal Governments and Average Population in Selected Developing Countries, about 2002 Country Number Average Population India (rural) 3,278 Kazakhstan 7,968 4,331 Indonesia 1,312 5,915 Argentina 2,154 14,972 Poland 2,478 18,881 Brazil 5,560 30,099 Chile 325 64,592 India (urban) 68,027 China 43,965 107,334 South Africa 282 238,839 Source: Derived from Shah (2006, p. 36-37) 4.5 Uganda is not an outlier in this context. Summing across local governments but excluding administrative units (i.e., including Kampala, municipalities, towns, districts, and sub-counties), there are 1,446 local governments. That implies an average population per local government of 23,604, a figure comparable to those of many countries noted above. With an average population of 25,370, the sub- counties in Uganda appear to be of a “typical� size. At the administrative level, on average the parishes or wards do not seem small. The small average number of people per village reflects the rural, agricultural and low-income features of the economy and region. 4.6 As in Uganda, there is a diverse array of local governments within countries. Hence, averages have limitations. It is possible to focus in a little more closely. Districts are the basic component of local government in Uganda and a center of attention for policy consideration. How do Uganda’s districts compare with similar units elsewhere? The districts are similar to upper-tier local governments in many countries. The alternatives are many but some illustrations provide a glimpse into the question. For example, the State of California in the United States has a population similar in size to Uganda’s. California’s population is 36.5 million and the state includes 57 counties and 478 sub-county municipal governments. Those numbers translate to averages of about 640,000 people per county and 76,000 per municipality. These averages are larger than those in Uganda, but California also has 3,800 other special purpose local governments, including school districts. In the less populous and more rural state of Iowa (population of three million), there are 99 counties and 947 municipalities, or 30,000 persons per county and 3,200 per municipality on average. In addition, Iowa has 386 school districts that are also independent, local governments and more than 500 other special districts. In the province of Alberta, Canada, with a population similar in size to Iowa’s, counties (or municipal districts, as they are officially designated) serve a relatively dispersed rural population. Their jurisdictions cover the rural areas only and exclude the towns and villages they surround, which have their own local governments. The average population per county in Alberta is about 6,700. Overlooking some specialized authorities usually serving 40 small populations, Alberta counts 58 school districts with an average population of 54,000 persons. Other than school boards, there are, unlike in the United States, few special district local governments. 37 4.7 Comparison to district-level governments in developing countries, especially in Africa, is complicated by differing circumstances. The commitment to and experience with decentralization varies widely among developing countries and often differs greatly from Uganda’s. Service delivery responsibilities differ from one country to the next, and political turmoil sometimes further complicates the picture. Ghana seems to offer a relatively comparable case to that of Uganda. It has some history of and commitment to decentralization and has adopted a similar structure. Ghana has 170 districts encompassing metropolitan, municipal, and rural (including towns and villages) areas. The number increased from 110 in 2004. Today, the country’s non-urban districts total 124. Approximately one-half of Ghana’s 24.5 million population is in the jurisdiction of the rural districts, so the average number of persons per district is about 104,000. The average in Mozambique is similar. Rwanda now has an average of about 360,000 persons per district, but it reduced the number of districts from 106 (with an average of about 100,000 persons) in 2006, largely as a means to increase ethnic diversity within each jurisdiction. In Tanzania, however, a country with a larger population and smaller number of districts, the average population per district is about 370,000. 4.8 Overall, international experience suggests that the increase in the number of districts in Uganda is at least partially justified by its demographic trends. Several observations lead to this conclusion. First, district-type governments are common in most countries. Second, although not given attention here, their responsibilities vary (for example, they may or may not include schooling), and sometimes considerably. Third, and likely most relevant for current concerns, the populations of viable district-level governments differ substantially across (and also within) countries. Uganda’s districts, with an average population of 262,000 and a range from 54,000 to over one million, are not at levels that would be considered small. Nor are the population sizes per local government unusual, relative to those observed in many other countries. Finally, if the number of Uganda’s districts had not increased, its population per district would definitely be large. In 1999, the average per district exceeded 500,000 persons and the increase to 56 districts reduced that figure to about 350,000. If the number of districts were still 56, the population per district would be about 560,000; even 80 would put it at about 390,000. Relative to average district numbers noted elsewhere, those numbers are quite high. B. Assessing the Fiscal Impacts of Redistricting in Uganda 4.9 This section assesses the fiscal impacts of redistricting in Uganda. Empirical analyses are undertaken from three perspectives and supplemented by a conceptual exercise using government norms. Because economies of scale loom large in the discussion of redistricting in Uganda, significant attention is given to that topic in the final portion of this section. 37 For more on U.S. local governments, see the U.S. Census of Government. LeSage and McMillan (2010) review general purpose local governments in Alberta. 41 1. What Happens When a New District Is Established? 4.10 The addition of a new district increases the direct fiscal demands on the central government. The dominant element is that an additional district administration needs to be put into place. A standard administrative complement is recommended. Depending upon the characteristics of the new district, one of three administrative staffing models is recommended to cover the office of the CAO and ten operational departments (finance, education, health works, etc.). The recommended staffing levels range from 147 to 218 positions. 38 Even if the actual staffing evolved from one complement of 218 to two of 147 as a result of division, a 35 percent increase in staff would result if both were fully staffed. Since the central government makes transfers specifically to cover some positions and, beyond that, local governments (especially districts) rely upon central funding, additional administrative staff add to the costs shouldered by the central government. Other costs exist as well—a new council, for example: the executive receive salaries and the councilors receive allowances. A new district requires a building and equipment. At least some of that cost of is met with special start-up grants. Also, a new district is supervised by a new Resident District Commissioner, who is a central government appointee. 4.11 Some of the additional costs are less direct. The central government ministries, departments, and agencies will have a larger number of districts with which to work. To the extent that additional training, monitoring, etc. is required, that adds new demands to those bodies. Less obvious indirect costs may emerge at the district level. There, shortages of qualified staff to fill the expanded number of positions may result in the appointment of under-qualified people, thus causing performance to suffer. 2. Indicators of the Costs of New Districts: Uganda Bureau of Statistics (UBOS) and BOOST data 4.12 The following analysis examines various sources in an effort to quantify the magnitude involved in the creation of new districts. The primary cost pressures on central government (and public administration, more generally) appear to be those associated with the direct outlays for funding the new positions in the district governments. Uganda Bureau of Statistics Data 4.13 The Uganda Bureau of Statistics reports expenditures of the districts for general public administration (as well as other expenditure categories) in its Statistical Abstracts. The reported information is for recurrent outlays (i.e., development expenditures are excluded); as such, it understates the full expenditures associated with each sector. However, recurrent outlays are the main component of public administration expenditure, although development expenditures specific to that category are largest when a district is newly established. 4.14 Public administration expenditures per person appear to move in parallel with the number of districts—that is, they increase as the number of districts increases—at least until 2010/11. The districts’ recurrent general public administration expenditures per person are displayed in Figure 24 in 2011/12 prices. Also presented is the number of districts, which increased from 45 in 2000/01 to 56 the following year, then to 69 in 2005/06, 77 in 2006/07, 80 in 2007/08 and finally to 112 in 2010/11. 39 In 38 See MoLG (2012, p. 25). 39 District numbers are as provided by the Local Government Finance Council. 42 2010/11, the number of districts increased sharply (from 80 to 112), but the administration costs turned down and dropped to levels approximating those in effect when the number of districts was half as large (56 or less). In 2010/11 and 2011/12, population growth and inflation took their toll, but total nominal expenditures also declined. 40 4.15 The administration budget share increases with the number of districts, but only until 2007/08. A somewhat similar pattern to that pictured in Figure 25 emerges when examining general public administration expenditures as a portion of districts’ budgets. That share is plotted in Figure 26, along with district numbers. After reaching 80 districts in 2007/08, the budget share allocated to public administration begins to decline, dropping steadily from over 0.25 to under 0.20. That trend continued even as the number of districts expanded to 112. 41 Despite that drop, recent budget shares are above those in vigor when the number of districts was half as large. Figure 24. Per Capita District General Public Administration Recurrent Expenditures (2011/12 prices) and Number of Districts 18000 120 16000 100 14000 2011/12 shillings per capita Number of districts 12000 80 10000 60 8000 6000 40 4000 20 2000 0 0 Per capita Number of districts Source: UBOS. 40 Another oddity exists. Administration costs increased in 2001/02 when the number of districts rose from 45 to 56. However, the per capita cost increased in 2004/05 (by 30 percent) the year before the number of districts increased to 69. 41 As with the per capita amounts, the budget share of administration rose substantially in 2004/05, the year prior to the beginning of the next growth phase in district numbers. 43 4.16 It is difficult to predict whether the recent trends will continue or how they may change. The patterns in district recurrent administration outlays, and even district total recurrent expenditures, are not closely associated with the number of districts. The information above presents a rather inconsistent pattern, so the data is examined somewhat further. Figure 27 plots the constant price per capita of recurrent public administration expenditures and that as a share of the recurrent outlays together. Per capita public administration expenditures shifted substantially higher from 2004/05 to 2009/10, compared to figures from 2001/02 to 2003/04—about U Sh 3,900 or about 30 percent more, on average. Public administration also rose as a share of recurrent expenditures, from 17.5 to 24 percent. However, as is clearly evident here, while per capita expenditures remained relatively constant in 2008/09 and 2009/10, the expenditure share was declining and that trend persisted as the real per capita expenditures fell sharply after 2009/10. The relationship between administration and total recurrent expenditures is more evident from Figure 44. To aid comparison, the per capita administration expenditures are shown at five times their actual amounts. The districts’ total real per capita outlays have been more stable than districts’ outlays for public administration. Both have declined sharply since 2009/10, especially the latter. Figure 25. District General Public Administration Recurrent Expenditure as a Share of Total Recurrent Expenditures and Number of Districts 0.3 120 0.25 100 Expenditure share Number of districts 0.2 80 0.15 60 0.1 40 0.05 20 0 0 Expenditure share Number of districts Source: UBOS. 44 Figure 26. District General Public Administration Recurrent Expenditure: Per Capita (2011/12 prices) and Share of Recurrent Expenditure 18000 0.5 16000 14000 0.4 2011/12 shillings per capita Expenditure share 12000 0.3 10000 8000 0.2 6000 4000 0.1 2000 0 0 Per capita Expenditure share Source: UBOS. Figure 27. District Per Capita Total and Scaled-up General Public Administration Recurrent Expenditures (2011/12 prices) and Number of Districts 90000 120 80000 2011/12 shillings per capita 100 70000 Number of districts 60000 80 50000 60 40000 30000 40 20000 20 10000 0 0 Total recurrent pc 5 x recurrent public administration pc Number of districts Source: UBOS. 45 BOOST Data 4.17 It is possible to use alternative data to broaden the perspective on the expenditures associated with the creation of new districts. The World Bank BOOST database for Uganda provides detailed information on public budgets, including those for district local governments. Information from the BOOST data is the focus of analysis in this section. The BOOST data provides information by sector (education, health, agriculture, etc.). One of those sectors is public sector management. 42 Unfortunately, the assignment by sector is presently available only to 2008/09 and begins in 2003/04. Although covering a relatively short period, that period includes years in which the number of districts was increased from 56 to 80. The Kampala District is omitted because of its urban character and its recent change of status, so the actual number of districts analyzed is 55 and 79. 4.18 This data suggests no relationship between public administration costs and the number of districts. The BOOST-based per capita public sector management district budgets for 2003/04 to 2008/09 are shown in Figure 28. The constant price amounts are quite consistent across those years. They average 12,270 over the six years, possibly with a somewhat downward trend. That time period includes two years during which there were 56 districts, one when they numbered 69, and three when there were 80.43 The public administration sector budgets increased in 2005/06 when the number of districts expanded from 56 to 69; it fell back, however, when the number increased to 80. The examination of budget shares attributed to public administration (Figure 29) reveals a pattern similar to that of per capita expenditures. The BOOST data show that public sector management represented between 19 and 21 percent of total budgets. 44 The share is not related to the number of districts. 42 Comparison of the BOOST and the UBOS public administration data indicate an anomaly. The UBOS per capita recurrent public administration expenditures that appeared in the previous section are also displayed in Figure 45. There is much more variation in those numbers than in the BOOST values over the 2003/04 to 2008/09 period. Also interesting is that the UBOS numbers equal or exceed the BOOST numbers in four of the six years for which both are available, although the UBOS data are recurrent outlays only while the BOOST reflect total amounts. Also note that the BOOST data used here are the budgeted central government funds, whereas the UBOS are reported as actual expenditures and, typically, for local governments, actual expenditures are slightly lower than the budgeted funding. There might be some difference in the allocation of funds to public administration in the two sources. 43 The number of districts in 2006/07 is 80 here, not the 77 reported by the LGFC in the previous section, because the BOOST database reports budgets for 80 districts during that year. 44 The UBOS data indicate a higher share of recurrent expenditures, almost one-quarter (at 0.23 for the common period). Although larger, since the UBOS are recurrent only, the data are not exactly comparable to the BOOST data. 46 Figure 28. District Per Capita Public Administration Accounts (2011/12 prices) 2011/12 shillings per capita 18000 140 16000 120 14000 Number of districts 100 12000 10000 80 8000 60 6000 40 4000 2000 20 0 0 BOOST public sector management UBOS recurrent general public administration Sources: BOOST database and UBOS. Figure 29. District Public Administration Budget Shares 0.3 140 0.25 120 Number of districts 0.2 100 80 Budget shares 0.15 60 0.1 40 0.05 20 0 0 BOOST public sector management UBOS public administration Sources: BOOST database and UBOS. 4.19 The conditional transfers must be insufficient to meet public administration expenditures or the BOOST and UBOS categories recognize a much broader range of public administration expenditures (or both). The BOOST database includes a detailed listing of conditional funding from the central government to local governments. That data extends from 2003/04 to 2011/12. All the conditional transfers specifically granted in support of district administration costs are aggregated to create a BOOST public administration conditional transfer category. Included in that category are transfers for CAO/Town Clerk (TC) salaries, District Service Commission (DSC) Chairs’ salaries, DSC operating costs, DSC non- wage operating costs, District Tender Board (DTB)/DSC/PAC/Land Boards/etc., ex-gratia for lower local governments (LGs), salaries and gratuities for LG-elected political leaders, women/youth/disability councils’ wages and operations, and IFMS recurrent costs. The per capita amounts of those funds averaged U Sh 763 (2011 shillings) per person up to 2009/10 and then approximately doubled for the next 47 two years due to the addition of new grants. The magnitude of these conditional funds specifically for district government administration is small and represents a small portion of overall public administration expenditures. The relative significance of the funds to meet those expenditures is indicated in Figure 30. The per capita amounts of the conditional transfers for administration are traced by the line at the bottom of the figure. Per capita public administration funding under the BOOST public sector management and the UBOS public administration categories are represented by the two lines at the top of the figure. They are about U Sh 12,000 per capita for the years when the public administration conditional transfers are about U Sh 760 per capita. Figure 30. Alternative Measures of Public Administration Costs: Per Capita Amounts (2011/12 prices) 18000 UBOS recurrent public administration 16000 14000 BOOST PSMgt sector 12000 account 2011/12 shillings per capita 10000 BOOST public 8000 administration & unconditional 6000 transfers BOOST unconditional 4000 transfers 2000 0 BOOST public administration conditional transfers Sources: BOOST database and UBOS. 4.20 The funding attributed to public administration by BOOST and UBOS probably includes at least some (possibly significant) administrative expenditures related to services such as health, education, etc. that are likely to be relatively uniform per person served rather than a fixed cost associated with any particular district. It has been argued that the conditional transfers are inadequate to cover districts’ administrative expenses and the only way for the districts to meet administration costs is to use their limited unconditional grants and other discretionary funds for that purpose. Some perspective on that possibility is achieved by looking at the available unconditional funding. Unconditional funds are calculated as the sum of unconditional grants, the graduated tax compensation grants to the districts and their local level local governments, and the equalization grants to the districts. As shown in Figure 30, those funds have averaged about U Sh 8,000 per capita. If one combines the available unconditional funds with the conditional transfers for administrative costs, the sum is less than the public administration costs of BOOST and UBOS. 4.21 The BOOST data provide no indication that districts’ administrative expenditures or conditional funding for administrative purposes changed with the increase in the number of 48 districts from 56 to 80. The consistency in administrative per capita budgets and administrative budgets as a share of total budgets is noted. The magnitude of the conditional funds for administrative purposes as a share of total funds is demonstrated in Figure 31. From 2003/04 to 2009/10, the administrative conditional transfers averaged 1.23 percent of districts’ budgets. Over those years, the number of districts increased from 56 to 80. The conditional transfers in both share and per capita terms did not change with the number of districts. 45 Figure 31. Alternative Measures of Public Administration Costs: Budget Shares 0.3 UBOS recurrent public 0.25 administration BOOST PSMgt sector Budget shares 0.2 account 0.15 BOOST public 0.1 administration & unconditional transfers 0.05 BOOST unconditional transfers 0 BOOST public administration conditional transfers Source: BOOST database and UBOS. 3. Indicators of the Costs of New Districts: Old and New Districts 4.22 If adding districts results in significant new expenditures, those expenditures should be evident in a comparison of old and new districts, and this section reports on such an investigation. The total budgets and the public administration budgets are examined for the new and old districts. When referring to old districts, that term encompasses those that have been unchanged throughout the study period and the “mother� district. It is useful also to distinguish between the unchanged and the mother districts in some instances. The new districts are those created out of existing (mother) districts. The data come from the BOOST database and extend from 2003/04 to 2008/09, the period for which the public sector management category is currently available. During that time, the number of districts increased from 56 to 80 although, again, with the exclusion of Kampala, the number of districts analyzed changed from 55 to 79. Hence, there are 24 new and 55 old districts. Of the old, 37 were unchanged and 18 are mother districts. 45 The increase in those grants in 2010/11 relates to an expansion in the number and type of grants and not to the increase in district numbers that year. The new grants were for salaries and gratuities for local elected leaders and ex-gratia funds for lower-level local governments. 49 4.23 Public administration expenditures are somewhat higher in new districts, at least during the early years. Figure 32 shows the shares for the new and the old districts. Because the shares for the unchanged and the mother districts (the old districts) are so similar, for greater clarity, only the shares for the two combined (the old districts) are shown in the figure. 46 The administration share for the old districts averages 0.195 and ranged from 0.187 to 20.9 during these years. The average for the new districts over the four years for which there were more than 56 districts is 0.219. However, the share was unusually large in 2005/06, the first year of that expansion when 13 of the 24 new additions were made. For the three years with 80 districts, the shares are consistent and the mean is 0.206. That is, the public administration budget share in the new districts was 1.6 percent larger, or administrative costs about 8 percent greater, in the new than in the old districts over the 2006/07—2008/09 period. 4.24 The new districts have higher per capita public administration budgets than do the old districts. Those are provided in Figure 33. The data for both the unchanged and mother districts are plotted, as well as for the new and old districts. The values are in current (not constant price) terms. The per capita public administration budget of the new districts was about 18 percent higher (about U Sh 1,300 shillings per capita at that time). 47 Note, too, that the mother districts have administration budgets consistently lower than those in the unchanged districts. That difference narrowed somewhat after the new districts were added, but mostly it was the result of a reduction in administration budgets in the unchanged districts. Figure 32. Public Sector Management Budget Share of Old and New Districts 0.30 0.25 0.20 Budget share 0.15 0.10 0.05 0.00 2003/04 2004/05 2005/06 2006/07 2007/08 2008/09 OLD (unchanged & mother) districts New districts Source: BOOST database. 46 The mean shares for unchanged is 0.20 and for the mother districts is 0.188. 47 2005/05 is unusual in that, although the administration share was high that year, the per capita administration cost was equal to that of the old districts, indicating below average outlays in the non-administrative areas. 50 Figure 33. Per Capita Public Sector Management Budgets in Old and New Districts 10000 9000 Current shillings 8000 7000 6000 5000 4000 3000 2000 1000 0 2003/04 2004/05 2005/06 2006/07 2007/08 2008/09 New districts Old (unchanged + mother) districts Source: BOOST database. 4.25 This new-old district analysis of the BOOST data suggests that some cost increases resulted from the expansion in the number of districts from 56 to 80. The per capita public administration expenditures in the new districts appear to be somewhat larger than in the other districts in both absolute and relative terms. Total per capita budgets are also somewhat higher in the new districts but by considerably more than the extra administration costs. The higher public sector management budgets accounted for 38 percent of the higher total budgets. 48 These differences may be due in part to initial start- up costs and funding, although the total budget data do not reveal convergence towards the levels of the old districts during the four years. An examination of new districts unaffected by the 2010 round of additions—when those data become available—would provide more information on whether the difference is persistent. 49 Looking at the differing patterns of the real per capita total budgets across the new, mother, and unchanged districts indicates that the unchanged districts have, with declining levels of resources, fared less well than both the new and mother districts from 2003/04 to 2009/10. For whatever reason, per capita resources have shifted towards both new and mother districts. 4.26 This analysis of administration budgets indicates that public administration costs did increase somewhat with the creation of new districts. The per capita administration budgets in those 24 new districts were about 18 percent greater than in the old districts during the three years following their establishment. There seems to have been little or no impact on the administration costs of the mother districts, and their costs correspond closely to those of the districts that were unchanged. 4.27 The new districts have total budgets exceeding those of both the mother districts and the unchanged districts. The total per capita budgets for the new, unchanged, and mother districts appear in 48 Also, looking back at the public administration costs funded directly by central government conditional transfers—often cited as the cost culprit—those funds represent only 38 percent of the higher public administration budgets and 12 percent of the larger total budgets. 49 Perhaps part of the cost difference is due to characteristics of the new districts (not observed here) that result in greater per capita funding than is true of old districts. 51 Figure 34. The amounts are in current shillings and the data are available for an additional year, the fourth and final year in which the total number of districts was 80. For the years 2006/07 to 2008/09 (prior to some divergence from the trend among unchanged districts) the additional amount is about nine percent. Some increase is to be expected, because certain start-up funds are provided to new districts. Another factor may be the somewhat different characteristics of new districts, which might warrant a higher level of transfers per person from the central government. Note that, with the reduction in size of mother districts, total per capita budgets went from being somewhat smaller to slightly greater than those of the unchanged districts. It is possible that this upward shift is related to their reduced sizes, however that too would require further analysis. 4.28 The picture is similar if one looks at the total budgets in real terms. The per capita total budgets for the new, unchanged, and mother districts are plotted in Figure 35 in 2011/12 prices. The relative levels remain unchanged, but the absolute levels have rather different patterns that were not obvious before. The real per capita budgets of the new districts are the highest but, across all 80 districts, real funding has been relatively constant or declining. The budgets of the mother districts, while somewhat up and down, have been rather steady over the seven years. The unchanged districts, however, have seen their real per capita funding decline almost steadily.50 In 2009/10 their real per-person budgets were 20 percent smaller than in 2003/04. To the extent that there has been a cost involved in adding districts, perhaps it is the unchanged districts that have in fact borne that cost. 50 An examination of the funding allocated to the various budget sectors of new, mother, and unchanged districts might suggest reasons for differences in the levels of funding. 52 Figure 34. District Per Capita Total Budgets: New, Unchanged, and Mother Districts (current prices) 60000 90 80 50000 70 Number of districts 40000 60 Current shillings 50 30000 40 20000 30 20 10000 10 0 0 2003/04 2004/05 2005/06 2006/07 2007/08 2008/09 2009/10 New districts Unchanged districts Mother districts Number of districts Source: BOOST database. Figure 35. District Per Capita Total Budgets: New, Unchanged and Mother Districts (2011/12 prices) 70000 60000 50000 2011/12 shillings per capita 40000 30000 20000 10000 0 2003/04 2004/05 2005/06 2006/07 2007/08 2008/09 2009/10 New districts Unchanged districts Mother districts Source: BOOST database. 53 4. Expenditures Associated with the Creation of New Districts 4.29 What are the fiscal implications of creating additional districts, according to indicators? The comparison of new and old districts provides the only consistent and reliable indication of costs associated with each category; as such, the following analysis is based on results of such a comparison. The analysis relates to the years 2006/07–2008/09, the years during which Uganda counted 80 districts and data exists in the Public Sector Management category. Estimates of Fiscal Impacts at the District Level 4.30 The evidence indicates that new districts have somewhat higher public administration and total expenditures than do old districts. Because nominal values change with inflation and policy decisions, expressing the differences in percentage terms is more informative since these can be expected to remain representative and consistent over time. Such percentages appear in Table 5. In particular, the new districts’ expenditure premiums reported there are expressed as a percentage of the expenditures of unchanged districts, mother districts, and old (combined unchanged and mother) districts. For the sake of brevity, attention is focused on the old districts as being representative of non-new districts, while noting some differences between mother and unchanged districts. The values are the averages observed over the 2006/07–2008/09 period. Table 5. Indicators of Additional Costs Associated with New Districts PSMgt Total PSMgt budget budget budget PSMgt as a PSMgt premium as a premium, premium, percentage premium as a District Type percentage of percentage percentage of total percentage of public sector of total of total budget total premium management budget budget budget New 0 0 20.63 Unchanged 10.5 3.25 19.55 16.7 31.3 Mother 6.9 3.99 18.06 22.1 58.8 Old (Mother & Unchanged) 9.3 3.51 19.04 18.4 38.3 Note: PSMgt represents public sector management. 4.31 The cost of local government is expected to be somewhat greater in new districts given the complement of district officers and staff that are added to the government structure. Indeed, public administration budgets per capita were found to be greater in new districts, which is expected to positively impact total per capita budgets. Total budgets per person in the new districts were observed to be 9.3 percent above those in the old districts on average. However, a difference of that magnitude is unexpected. Public sector management budgets represent about one-fifth of total budgets and the public sector management budgets in new districts are about one-fifth greater than in other districts (Table 7). A fifth of a fifth amounts to about four percent, so an increase on that order should result from the extra 54 administrative costs. When the observed new district public sector premium is calculated as a percentage of the old district total budgets, it is on the expected order of magnitude at 3.5 percent. 51 4.32 The indicators of higher costs associated with new districts must be viewed with some caution. The total budget increases in the new districts exceed the increases due to public sector management budgets. The larger public sector management budgets account for only 38.3 percent of the total budget increase. Why should other budgets increase when the number of school children, schools, health units, roads, water systems, and farmers to whom extension services are provided do not change? If all else remains the same, non-public administration funding per person is not expected to increase (unless some additional funding and, presumably, improvement in services were to occur). In addition, there is even reason to be cautious of the full increase in the public sector management cost component itself. There will be a new complement of district officers and staff, the costs of which will be new, but the administration costs of schools, hospitals, existing lower-level governments and administrative units, etc. should not change. If this is indeed the case, the only additional costs associated with the creation of a new district should be those of the new district office. As previously noted, the conditional transfers funding district administration were found to represent on average 1.26 percent of total budgets. In sum, these observations yield three potential indicators of the additional costs that may be associated with new districts—9.3 percent, 3.5 percent, and 1.26 percent of total budgets—with an expectation that the lower estimates are the better measures, provided only district administration changes. 52 4.33 Administrative costs in Uganda, broadly defined, appear generally comparable to those recorded elsewhere. Box 3 elaborates on the international experiences in this regard. 51 It is possible that the figure of 3.51 percent overstates the long-term budget difference. New districts require start- up costs and funding, which should diminish over time; moreover, experience should generate greater efficiency. There may be some evidence to this effect. In 2006/07, 40 percent of the total per-person cost difference could be attributed to large administration budgets, but that figure had dropped to 32 percent by 2008/09, suggesting that the administration factor may be of diminishing significance. This implies a (slightly reduced) public sector management premium of 3.33 percent in 2008/09. 52 There is an alternative way to assess the potential impact of expanding the number of districts that would involve comparing those that change with those that do not. This approach implies a comparison of the new and mother districts, combined with the unchanged districts. This method has the advantage of taking into account the fact that the mother districts may be (adversely) impacted by their reduction in size as a result of carving off a new district. If evaluated in this way, total per capita budgets are calculated to be about 6 percent greater in the changed than in unchanged districts during the 2006/07–2008/09 period. Public sector administration costs are larger by about 1 percent of total budgets. A relevant comparison is also between the pre- and post-change periods; that is, when there were 56 districts and when there were 80. During that transition, however, total per capita budgets grew significantly in the mother (and also the new) districts in categories other than administration, while those in the unchanged districts declined; that is, the two converged (at least temporarily). The causes of those changes are not known but, for reasons noted, they are expected to be unrelated to the increase in districts. Hence, a comparison over the longer period seems questionable. Because the 6 and 1 percent differences noted for the three years between 2006/07 and 2008/09 are smaller than those taken from the new-old analysis, those of the new-old study are used with the knowledge that they present conservatively large estimates of the possible cost of adding districts. 55 Box 3. A Comparative Perspective on the Administrative Costs of Local Governments How do administrative expenditures in Uganda’s districts compare with those of local governments in other countries? Such a comparison is not easy to determine. Accounting systems vary from one country to the next and few provide a category comparable to the public sector administration item in the BOOST database. Also, many countries, especially in the developing world, do not report data to comparable international sources. However, the International Monetary Fund’s (IMF) Government Finance Statistics does report a general public services category for local governments in a number of countries. McMillan (2008) provides a summary for developing countries. While the general services category includes the legislative, executive, and financial branches, it also includes otherwise uncategorized services, debt repayment, and any transfers to other governments. Although there may be limits to comparability, across 20 developed countries, general public services average about 16 percent of local government expenditures. The range is wide, from about 4 percent in Denmark and the United Kingdom to 36 percent in France. Estimates for developing countries are difficult to find. The IMF source (again using general services) provides the following: 14.9 percent in Bolivia, 16.4 in Mongolia, and 33 percent in South Africa. Among transition countries, it reports 7.4 percent in Slovenia and 15.2 percent in Hungary. Also, McMillan (2007) notes 6.3 percent for India (in the category of “other� expenditures), 4.0 percent for legislative purposes in Brazil (but with another 16.0 percent for administration and planning), and 13.1 percent in Latvia. Available statistics not reported in these sources included 5.2 percent allocated to legislative activities in Argentina but 33.5 for general management. Clearly, administrative activities can make up a wide range of local government budgets. In this context, at about 19 percent Uganda’s public sector management category is not an outlier. The Estimated Direct Fiscal Impacts in a Broader Context 4.34 Taking budgets as stable in the old districts but somewhat greater in the new, the impact on the additional expenditures of all districts is a fraction of the impact on the new districts. Although estimated expenditures are larger in new districts, the impact on the total of district budgets (i.e., the cost of local government at the district level) is smaller because the new districts make up only a portion of the total number of districts and they contain only a fraction of the total population of the districts. In the case of the increase from 56 to 80 districts, 24 were added and those new districts included about 22 percent of the total population under the districts’ jurisdiction. For example, if total budgets were 9.3 percent greater per person in the new districts, and if those districts account for 22 percent of the districts’ total population, the impact on the total district budgets is just over two percent. Similar calculations can be made for the 3.51 and 1.26 percent increases in the new districts and for the expansion of districts from 80 to 112. 53 Those additional expenditures, or the cost premium due to adding districts, can also be compared to the expenditures of local government, the total government expenditures in Uganda, and to GDP. The relative magnitudes of the additional expenditures measured as percentages of those economic indicators are reported in Table 6. 53 It is assumed that the same cost premiums continue in the 2010 expansion. In that case, the new districts held about 16 percent of the total district population. 56 Table 6. Relative Magnitudes of Additional Expenditures due to Adding Districts beyond 56: Three Scenarios Additional Costs as a Percentage of Selected Economic Measures 2006/07 2007/08 2008/09 2009/10 2010/11 2011/12 Number of Districts 80 80 80 80 112 112 Scenario A. If extra cost is 9.3% of new districts’ total budget % of districts’ total budgets 2.01 2.01 2.01 2.01 3.41 3.41 % of local governments’ total budgets 1.91 1.91 1.91 1.92 3.09 3.21 % of total government 0.40 0.37 0.33 0.33 0.57 0.51 % of GDP 0.09 0.08 0.08 0.07 0.12 0.11 Scenario B. If extra cost is 3.51% of new districts’ total budgets % of districts’ total budgets 0.77 0.77 0.77 0.77 1.32 1.32 % of local governments’ total budgets 0.73 0.73 0.73 0.73 1.19 1.24 % of total government 0.15 0.14 0.13 0.13 0.22 0.12 % of GDP 0.03 0.03 0.03 0.03 0.05 0.04 Scenario C. If extra cost is 1.26% of new districts’ total budgets % of districts’ total budgets 0.28 0.28 0.28 0.28 0.48 0.48 % of local governments’ total budgets 0.26 0.26 0.26 0.26 0.43 0.45 % of total government 0.08 0.05 0.05 0.05 0.08 0.07 % of GDP 0.01 0.01 0.01 0.01 0.02 0.02 Source: World Bank staff calculations. 4.35 Clearly, the largest relative effects result in the case where the 9.3 percent increase in total expenditures in new districts is assumed to be the product of redistricting. The resulting budget increase is just over two percent of districts’ total budgets for the 56 to 80 change and about 3.4 percent when the numbers change from 80 to 112 (Table 8, Scenario A). As a percentage of total local government budgets, the percentages are only slightly smaller (about 1.9 and 3.1 percent, respectively) because district budgets dominate the local sector. Municipal council budgets accounted for about five percent of local government budgets when there were 80 districts (and prior to the increase in municipal councils that accompanied the 2010 redistricting). As a percentage of total government, the additional expenditures are smaller—0.4 percent in 2006/07—and, across the years during which the number of districts is constant, it declines due to the growth in central government relative to local government. The additional expenditures represent 0.09 percent of GDP in 2006/07 and similarly decline (e.g., to 0.07 percent in 2009/10), reflecting the faster growth of GDP than of local government. 4.36 The higher budgets in the public sector management category observed when the number of districts was increased from 56 to 80 (3.51 percent of the new districts’ total budgets) are a smaller portion of the four comparison measures. For example (Table 8, Scenario B), the budget premium accounts for less than 0.8 percent of total local government budgets when there are 80 districts and for about 1.3 percent when this number is expanded to 112. As a percentage of total government budgets, it was as low as 0.125 percent in 2008/09 and reached a high of 0.218 percent in 2010/11, just after the 57 expansion to 112 districts. The percentage of GDP those costs represent was about 0.03 percent when the country counted 80 districts and 0.04 percent when this number grew to 112. 4.37 The actual conditional funding specifically allotted for district administration amounts to almost 1.26 percent of the new districts’ total budgets. That is, again, even smaller in this relative context. As a percentage of the total of all district budgets, it was almost 0.28 percent when there were 80 districts and 0.48 when that number reached 112 (Table 5, Scenario C). In turn, that is about 0.05 percent of all government budgets and about 0.07 percent when the country counted 80 and 112 districts, respectively. As a percentage of GDP, the additional expenditure is 0.01 to 0.015 percent. 4.38 All the comparisons outlined here indicate that the relative magnitudes of the additional expenditures associated with the expansion of district numbers are at least modest and typically small. In particular, what can be considered the most representative and meaningful estimates—that is, those specifically attributable to public sector management and, more narrowly, funding for district administrations—those estimates are small. At most (i.e., those estimated for the case of a move from 56 to 112 districts), the additional costs add about 1.3 percent to the total cost of district local government, 0.2 percent to the cost of government, and represent 0.04 percent of GDP. If the extra administrative expenditures amount to the extra transfers for district administration, the additional “burden� is reduced to almost one-third of the percentages just mentioned. Clearly, these costs appear to be small relative to the non-quantified costs argued to burden the country—often excessively, it is argued—due the expansion of district numbers. 4.39 Even if relatively small, additional expenditures are not to be dismissed. If they contribute nothing, the forgone public services or private goods required to finance them are a real cost. Might those additional administrative costs generate real benefits? Given their magnitudes, a modest or even small improvement in the efficiency of the delivery of district services and/or an improvement in citizen satisfaction would offset the cost premium of new districts. Some might see any added cost as threatening the sustainability of district governments. Such small impacts hardly seem to pose such a threat. Even so, the level of local government costs and small changes to those are difficult to assess in terms of sustainability. There can be little doubt that in both Denmark, where local government expenditures amount to one-third of GDP, and in Australia, where that figure is 2.4 percent, local governments have proven solidly sustainable. Sustainability depends upon citizen satisfaction with the services realized given the cost involved in their provision. In this case, a discussion of the merits of an increase in the number of districts should be informed by a careful assessment of the benefits and costs involved. The estimates above indicate that the costs are less than is commonly suggested. 4.40 That the additional costs associated with adding districts could be relatively low is generally consistent with estimates from Ugandan authorities. As mentioned above, it was determined that additional administrative costs added between 1.26 and 3.51 percent to total costs in the new districts and between 0.28 and 1.3 percent to the total cost of district local governments altogether. Analyses by Ugandan authorities suggest that these estimates derived from budget data are reasonable. The Local Government Finance Commission draft report (LGFC 2012) presented information on the estimated cost of delivering services in an average district. Administration was one of the eight categories of services identified. As defined for that purpose, administration represented approximately 10.5 percent of total costs. Those estimates also appear in the local government set-up review (MoLG 2012). The Ministry of Local Government report additionally presents the three models of standard administrative staff 58 establishments recommended for local governments. This information is used to examine the consequences on the administrative and total costs of district services of a district’s division into two entities. The results of this analysis are presented in Annex 2. They are consistent with results obtained from the observed budgets of new and old districts, also suggesting that redistricting may involve relatively few additional costs. 5. Economies of Scale vs. Similarity of Preferences 4.41 Adding new districts makes districts more uniform in size, and, for some period of time, smaller on average. The populations of Uganda’s districts and the changes that have occurred as a result of increasing district numbers are summarized in Table 7. It is usually the more populous districts that are divided. Consider what happened when the number of districts increased from 56 to 80. Prior to redistricting, the mother districts had an average population of 614,000 persons. With the creation of 24 new districts, that population was reduced to 392,000, and the 24 new districts had populations averaging 195,000. The districts that did not divide had an average population of 376,000 in 2006. A similar pattern is observed in the 2010 expansion. In that case, there were 24 mother districts and 32 new districts. Again, the mother districts were relatively large, containing on average 523,000 persons, in contrast to an average 289,000 in the then 55 unchanged districts. With redistricting, the mother districts’ average population dropped to 294,000 (essentially the same as the unchanged district average) and the new districts had an average of 180,000 people (which is about 60 percent of the average of the other districts). One should also note that, with the Ugandan population growing at 3 percent per year, the population of an average district has been doubling every 20 to 25 years, a trend that will continue until the country’s fertility rate substantially declines. 4.42 Reducing the average size of a district implies potential costs, due to loss of economies of scale in public administration, as well as potential benefits, due to greater similarity in preferences among district government constituents. The activities of a local government may have some fixed costs, meaning that in smaller districts per capita costs of public administration are bound to be greater than in larger districts. Furthermore, the number of people possessing higher-level skills required for managing service delivery in low-income countries is typically limited, leading to another source of economies of scale in local public administration. At the same time, as the populations of these districts become smaller, the preferences of their constituents become more homogeneous, and, other things being equal, district governments may find it easier both to collect information about these preferences and to satisfy them. The benefits related to achieving greater similarity in preferences among smaller populations depend on the accountability of district governments vis-à-vis their constituents, as well as on their ability to allocate resources across various services for which they are responsible. That is, such benefits are larger if district governments have stronger incentives to collect information about these preferences and to satisfy them, as well as sufficient flexibility in allocating the resources needed to do so. 59 Table 7. District Numbers and Average Populations, Pre- and Post-Redistricting 2004 2006 2009 2010 Total number of districts 55 79 79 111 without Kampala Mother districts average population 614,059 391,837 513,071 294,446 number 18 18 24 24 New districts average population 194,680 179,634 number 24 32 Unchanged districts average population 352,377 376,469 289,305 298,976 number 37 37 55 55 4.43 Diseconomies of scale in local public administration appear to be a dominant effect of district proliferation. For a wide range of services and circumstances – but not always - economies of scale are exhausted for local governments at quite modest populations. A number of studies 54 find that, for the most part, local government services are horizontally integrated. That is, they are delivered efficiently by small units—for example, schools or health units—that can easily be replicated. As a result, costs per unit or per person served need not differ significantly depending on whether such facilities are numerous or scarce. Thus, local governments of a broad range of sizes can coexist and remain competitive with one another. The average population of a district remains sufficiently large, compared to similar jurisdictions in other countries (see Section 4.1), suggesting that diseconomies of scale for most (though not all 55) specific public services provided by the local governments are unlikely to exist. On the other hand, about 40 percent of jobs in district governments remain vacant, and jobs requiring greater skills are represented disproportionately among the unfilled vacancies (LGFC 2012), suggesting that district governments’ demand for skilled labor may exceed available supply at the prevailing wage structure. At the same time, some of the jobs are becoming redundant at a district level; for example, hiring a roads engineer makes little sense for a new district that contains only 50 kilometers of roads. 4.44 The benefits of greater similarity in preferences due to a district’s smaller size are limited. This is because district governments receive the bulk of their funding from the central government through conditional transfers (see Section 3), which leaves them with a very limited degree of freedom in the allocation of resources. Further, the district governments’ accountability to their constituents is limited (see Section 2). 54 See, for example, Hirsch (1970), Bergstrom and Goodman (1973), Borcherding and Deacon (1972), McMillan, Wilson and Arthur (1981), Deller and Rudnicki (1992), McGreer and McMillan (1993), De Borger and Kerstens (1996). 55 Of course, there are limits to the services of that kind. Vertically integrated services such as electricity, water, and sewerage have large fixed costs, usually in production units (power plants, water and sewerage treatment, etc.), and the volume of product and number of people served affect the unit cost. Such services typically realize greater efficiencies the larger the population served. 60 4.45 To ensure that changing jurisdictional boundaries will have fewer negative than positive effects, redistricting should always be based on a sound assessment of the pros and cons involved. Defining criteria and undertaking careful assessments is probably all the more valuable when, as in Uganda, districts are becoming progressively smaller and further divisions are under consideration. What are the objectives being pursued, and are those objectives (i.e., criteria) widely accepted? Only with generally accepted criteria and analyses based on those criteria can the question of how far these divisions should go be answered. 61 V. Value-for-money in service delivery: determinants and challenges 5.1 The purpose of this section is to understand what causes differences in outcomes in different areas of the country. More specifically, Section 5 looks at where value-for-money comes from in district government performance. With the overall budget constraints tightening, allocations for social services declining, and costs of public administration rising, it becomes increasingly important to improve value- for-money in service delivery by local governments, particularly in the education and health sectors. 5.2 Over the past decade, cross-district comparisons of performance have increasingly featured in the Government of Uganda’s statistical analysis of sector performance. For example, both the Ministry of Health and the Ministry of Education and Sports have begun issuing yearly District League Tables that assess the relative performance of district governments with respect to a range of different criteria. The analysis in this section combines those sources of data with statistics from the National Service Delivery Survey, the Uganda National Panel Survey, the Uganda Bureau of Statistics, the United Nations Development Program and the BOOST database. 5.3 Specifically, this section aims to answer three questions: How do spending and sector performance vary across districts in Uganda? How do district governments differ in their value-for-money? What factors contribute to high value-for-money? 5.4 The analysis shows that there is large variation across districts in spending, performance, and value-for-money. Bringing every district up to the outcome-to-spending ratio of the districts that perform very highly in both the health and education sectors could save about 0.8 percent of GDP. Overall value-for-money in social sectors can be increased by shifting funding from districts that are already well-served to districts that are relatively under-served. Accountability and the quality of local institutions are other key factors to consider. For example, districts that received an unqualified audit in the Auditor General’s Annual Report had consistently higher levels of value-for-money. A. Unequal Spending, Unequal Outcomes 5.5 There is inequality across districts in Uganda in terms of public spending per capita. Figure 36 shows the variation in spending between districts with the highest levels of spending (towards the left) and districts with lowest levels (on the right). Health and education spending feature more variation than overall spending. 56 There is also variation in the level of service provided in various sectors: for example, in the education sector the number of students per primary school teacher varies from as low as 32 to as high as 100 across districts (Figure 37). 56 In Figure 36, the scales are made equivalent by depicting each type of spending as a percentage of the average level of spending. 62 Figure 36: Variation in Total Spending Figure 37: Variation in the Ratio of Pupils to and Spending in the Health and Teachers Education Sectors (2011) 500 12 400 10 Frequency % of Mean 8 300 6 200 4 100 2 0 0 Best served Least served Districts Total Spending Per Capita Health Spending Per Capita Pupil-Teacher Ratio Education Spending Per Capita Source: BOOST database. Source: MoES. 5.6 Variation in public spending and levels of provided services seems to pass through to performance. There is wide variation across districts in education outcomes, such as in the Net Intake Rate, the share of school-age children enrolled in school (Figure 38a), and in measures of performance on the PLE (Figure 38b). There is also large variation in health outcomes among districts. Figure 39a shows variation in latrine coverage across districts, from 2 percent to 97 percent. Figure 39b shows variation in the share of deliveries in government health facilities, from 4 percent to 94 percent. Figure 38a: Variation in Education Outcomes Figure 38b: Variation in Education – Net Intake Rate Outcomes – PLE Performance Measure 25 25 20 20 Frequency Frequency 15 15 10 10 5 5 0 0 0.1-0.2 0.2-0.3 0.3-0.4 0.4-0.5 0.5-0.6 0.6-0.7 0.7-0.8 0.8-0.9 1.1-1.2 1.2-1.3 1.3-1.4 0.9-1 1-1.1 Net Enrolment Rate MoES PLE Performance Measure Source: MoES. Source: MoES. 63 Figure 39a: Variation in Health Outcomes – Figure 39b: Variation in Health Outcomes – Pentavalent Vaccine 3rd Dose Coverage Deliveries in Government and Nonprofit Rate Facilities 35 20 30 Frequency 25 15 Frequency 20 15 10 10 5 5 0 0 10-20 30-40 50-60 70-80 110-120 130-140 150-160 170-180 190-200 210-220 230-240 90-100 10-15 20-25 30-35 40-45 50-55 60-65 70-75 80-85 90-95 0-5 100- Pentavalent Vaccine 3rd Dose Coverage Deliveries in Government and Nonprofit Rate (%) Facilities Source: MoH. Source: MoH. 5.7 The Ministry of Education and Sports (MoES) and the Ministry of Health (MoH) compile League Tables that aggregate a range of sector performance measures to construct relative rankings. The MoES District League Table includes the Net Intake Rate, the Primary Completion Rate and an index of PLE Performance. 57 In 2011, the MoH League Table included vaccine coverage rates, the fraction of deliveries in government and private not-for-profit facilities, OPD per capita, the rate of HIV testing of children born to HIV positive women, latrine coverage, and measures of staffing and reporting completeness and timeliness. 58 These different aspects of sector outputs and outcomes are scaled and aggregated to create an overall ranking for each district. 59 5.8 Districts vary significantly in their performance in the District League tables. This variation (Figure 40) reflects variation in each of the underlying subcomponents. Because they act as a useful aggregate of a range of different factors and outcomes, the analysis for the rest of this section will focus on these District League Table scores to show how education and health outcomes are linked to variation in spending, social conditions and governance. 57 The analysis in this section focuses on Primary Schooling, as public spending at the district level has the clearest link to education outcomes at the Primary level. 58 One difficulty with comparisons of the MoH League Tables over time is that the criteria used vary from year to year. While it is important to update outcome measures to best assess the true quality of outcomes, consistency in measures used is also important to allow for inference over time. 59 This analysis scales the MoES League Table in a slightly different fashion to the MoES Sector Performance Report. Details are included in Annex 1. 64 Figure 40a: Variation in Education Figure 40b: Variation in Health Outcomes Outcomes 25 30 20 25 Frequency Frequency 20 15 15 10 10 5 5 0 0 10-15 20-25 30-35 40-45 50-55 60-65 70-75 80-85 0-5 MOES District League Score MOH District League Score Source: MoES District League Score. Source: MoH District League Score. 5.9 Education and health outcomes in part depend on a range of social characteristics, in particular income levels. Higher income is associated with higher education and health scores (Figure 41). Figure 41a: Higher Income is Associated Figure 41b: Higher Income is Associated with with Better Education Outcomes Better Health Outcomes 0.7 0.7 0.6 0.6 UNDP GDP Index UNDP GDP Index 0.5 0.5 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 0 0 10 30 50 70 30 40 50 60 70 80 MoES District League Score MoH District League Score Source: MoES and UNDP. Source: MoH and UNDP. 5.10 Public spending per capita is the factor most closely associated with outcomes in the health and education sectors. Higher levels of education spending are associated with better education outcomes (Figure 42a). Similarly, higher health spending is associated with higher league table scores (Figure 42b). There are strong reasons to believe that, all else being equal, increased resources should improve outcomes. This hypothesis is backed up by the formal econometric analysis in Annex 3. 65 Figure 42a: Higher Levels of Spending are Figure 42b: Higher Levels of Spending are Associated with Better Education Associated with Better Health Outcomes Outcomes 50 25 Budgeted Spending Per Budgeted Spending Per 40 20 30 15 Capita Capita 20 10 10 5 0 0 10 30 50 70 30 50 70 90 MoES District League Score MoH District League Score Source: MoES and BOOST database. Source: MoH and BOOST database. 5.11 Spending appears to explain part of the variation in sector outcomes; still, large variation remains unexplained. The unexplained variation in outcomes suggests differentials in the ability of districts to convert public spending into sector outcomes. Local managerial effectiveness, structural differences across districts and differences in the marginal impact of an additional unit of spending are all leading candidates as explanations for significant variation in observed value-for-money. B. Assessing Value-for-Money in District Spending Creating a measure of value-for-money 5.12 The first means of measuring value-for-money is to look at the ratio of spending to outcome measures. For both the education and health sectors, there is a wide range in the outcome-to-spending ratio in the District League Table results. One way of demonstrating this is to consider the outcome-to- spending by region, as shown in Figure 43. 5.13 In the outcome-to-spending ratio, the central and western regions perform on average better than the eastern and northern regions. The difference is more noticeable in the health sector than in the education sector. This variation across regions is echoed by variations across districts within a region, suggesting a potentially significant role for district government efficiency. 66 Figure 43a: Ratio of Outcome to Spending for Figure 43b: Ratio of Outcome to Education by Region Spending for Health by Region 3 14 Average Outcome-to- Average Outcome-to- 2.5 12 Spending Ratio Spending Ratio 10 2 8 1.5 6 1 4 0.5 2 0 0 Central East North West Central East North West Source: MoES and BOOST database. Source: MoH and BOOST database. 5.14 A second efficiency measure helps to take social conditions into account. Using the predicted relationship between social conditions and outcomes described in 5.1, it is possible to estimate the expected level of outcomes in a district if it is performing with the average level of efficiency for Uganda. The ratio of actual outcomes and these predicted outcomes is then used as a measure of value-for-money, where a score above one implies higher-than-average efficiency, and less than one implies lower-than- average performance (see Annex 3 for further details). Overall, this implied efficiency measure is correlated with the outcome-to-spending ratio in the District League tables, more strongly in the education sector (0.63) than in the health sector (0.23). Identifying high- and low-efficiency districts in Education 5.15 Most of the best performing districts in the Education sector have relatively high income and moderate levels of spending per capita. These districts appear to be well-managed, as they perform relatively highly in transforming spending into positive education outcomes. They have high outcome-to- spending ratios, and high implied efficiency measures. Table 8 summarizes the characteristics and value- for-money measures for these high-performing districts. 5.16 However, the highest value-for-money group of districts also contains districts that have low spending per capita and low income, and yet perform relatively well. These districts, which include Kotido and Moroto, have lower outcome-to-spending ratios. This suggests they do not perform as well as the other districts in the absolute sense. However, they can still be considered relatively high value-for- money, because they achieve the above predictions given their levels of income and social conditions. This results in a high implied efficiency measure. 67 Table 8: High Performing Districts in Education (2010 data) Units of outcome per U Implied Efficiency Public expenditure per GDP Sh 1,000spent per capita Measure (high is good) capita (U Sh thousand) Index Moroto 4.955 1.864 5.878 0.282 Kaberamaido 3.136 1.585 19.679 0.338 Kotido 3.483 1.457 7.574 0.286 Kiruhura 4.107 1.436 15.715 0.5 Kaabong 3.185 1.278 6.685 0.286 Mbarara 3.233 1.246 20.331 0.5 Mukono 3.316 1.237 17.688 0.504 Kumi 2.773 1.22 20.183 0.418 Wakiso 5.534 1.168 10.73 0.594 Mbale 2.796 1.113 19.215 0.456 Mayuge 2.749 1.1 14.733 0.398 Kamuli 3.082 1.098 14.401 0.417 Soroti 2.978 1.097 15.652 0.443 Masindi 3.169 1.087 14.344 0.484 Kampala 7.009 1.077 8.233 0.637 Average 3.700 1.271 14.069 0.436 Average district 2.493 1.007 19.439 0.430 5.17 The group of relatively inefficient districts typically features high spending and moderate- to-poor outcomes. Each of these districts is below average in their outcome-to-spending ratio (Table 9). Most have moderate levels of income and other social factors, and so are predicted to perform more effectively than they do. This results in a low implied efficiency measure. 5.18 There are, however, some districts that feature a combination of low spending and poor efficiency. This group includes Adjumani, Bundibugyo, and Kiboga. Adjumani, for example, has amongst the lowest levels of spending per capita, and therefore would be predicted to have low performance. However, even taking those factors into account, Adjumani is considered relatively inefficient in this analysis. This suggests that poor value-for-money is not always associated with spending that is too high, but can also be associated with the poor management of relatively scarce resources. 68 Table 9: Relatively Poorly Performing Districts in Education (2010 data) Units of outcome per U Sh Implied Efficiency Public expenditure per GDP 1,000 spent per capita Measure (high is good) capita (U Sh thousand) Index Adjumani 1.309 0.387 8.263 0.387 Yumbe 1.193 0.553 15.28 0.32 Bundibugyo 1.976 0.663 13.813 0.466 Dokolo 1.299 0.669 26.878 0.389 Amuru 1.31 0.706 24.07 0.349 Apac 1.494 0.721 21.744 0.385 Amolatar 1.391 0.731 28.067 0.389 Kiboga 1.964 0.733 15.035 0.443 Oyam 1.54 0.743 22.433 0.385 Rakai 1.697 0.752 23.998 0.471 Nebbi 1.765 0.767 19.259 0.404 Iganga 1.947 0.775 19.999 0.458 Mubende 2.404 0.828 14.494 0.465 Mpigi 1.786 0.86 27.943 0.448 Lira 1.857 0.885 21.714 0.389 Sironko 1.905 0.888 24.152 0.42 Kitgum 1.895 0.917 19.21 0.392 Pader 1.871 0.944 17.209 0.329 Average 1.700 0.751 20.198 0.405 Average district 2.493 1.007 19.439 0.430 Identifying high- and low-efficiency districts in the health sector 5.19 The high-performing districts in health are similar to those in education; typically, they feature moderate levels of spending per capita and achieve high results, but are also with a group that achieves relatively highly from low income levels. As Table 10 shows, the first set of districts has a relatively high outcome-to-spending ratio, and a high implied efficiency measure. It is worth noting that within this group, three districts in Table 10 also appear in Table 8: Kampala, Mbale, and Mbarara. 60 Districts of the second type, with high implied efficiency measures, include Oyam and Pader. 60 Several more of the districts in Table 1 could not be included in the analysis for Table 3 due to data issues, but may also be high-performing in the health sector. 69 Table 10: High-Performing Districts in Health (2010 data) Units of outcome per U Implied Efficiency Public expenditure per GDP Sh 1,000 spent per capita Measure (high is good) capita (U Sh thousand) Index Mityana 11.234 1.21 6.979 0.465 Pader 9.485 1.158 7.138 0.329 Kamwenge 17.666 1.137 3.872 0.423 Mbale 12.527 1.119 5.828 0.456 Kiboga 11.458 1.113 5.978 0.443 Lira 11.407 1.103 6.102 0.389 Oyam 9.03 1.103 7.652 0.385 Kayunga 10.355 1.085 6.664 0.426 Mbarara 18.461 1.078 3.792 0.5 Busia 11.193 1.077 6.138 0.449 Jinja 12.087 1.066 6.131 0.528 Namutumba 15.233 1.058 4.267 0.458 Kampala 36.895 1.046 2.466 0.637 Arua 12.248 1.032 5.152 0.377 Masaka 18.578 1.028 3.784 0.492 Isingiro 16.416 1.018 3.978 0.5 Ibanda 12.109 1.017 5.401 0.5 Average 14.285 1.082 5.429 0.455 Average district 10.554 1.000 7.413 0.425 5.20 The relatively low value-for-money districts in the health sector also fall into two groups: high-spending districts with relatively poor performance, and districts with moderate spending but poor overall outcomes. Districts in the first category (Table 11) have poor outcome-to-spending ratios, and underachieve relative to predictions. This category includes Bukwa and Kapchorwa. The second category includes districts that do not appear to have high levels of spending, but achieve outcomes below what might be expected. Examples from the second category include Adjumani, Kaabong, and Sironko. 70 Table 11: Relatively Poorly Performing Districts in Health (2010 data) Units of outcome per U Sh Implied Efficiency Public expenditure per GDP 1,000 spent per capita Measure (high is good) capita (U Sh thousand) Index Bukwa 3.242 0.802 17.057 0.417 Adjumani 6.756 0.805 7.327 0.387 Bulisa 4.927 0.821 11.001 0.484 Kaabong 6.586 0.87 7.228 0.286 Sironko 6.762 0.889 8.636 0.42 Moyo 6.286 0.911 9.004 0.334 Kapchorwa 4.56 0.921 13.62 0.417 Budaka 6.569 0.957 9.118 0.385 Average 5.711 0.872 10.374 0.391 Average district 10.554 1.000 7.413 0.425 5.21 What is the potential for savings if laggard districts can be made more efficient? Taking the performance of the best sectors as a benchmark, it is possible to estimate the potential savings from bringing all districts up to that standard.61 This is calculated by keeping current outcome levels constant, and predicting the level of spending required to achieve those outcomes if all districts had the outcome-to- spending ratio of the highest-performing districts. Achieving these savings would likely require improving not just governance but also social conditions and living standards more generally, but this exercise gives some sense of the size of potential gains from better value-for-money in district governments. 5.22 In the education sector, eliminating inefficiency could potentially save U Sh 197 billion. This represents a savings of 34 percent of the total district education budget for 2010, and a potential savings of 0.4 percent of GDP. The average outcome-to-spending ratio in the education sector of the highest performing districts is 48 percent higher than the average of all districts, reflecting a large potential for improvement. 5.23 In the health sector, eliminating inefficiency could save at least U Sh 49 billion. 62 This represents a savings of 34 percent of the 2010 health budget for the districts analyzed for 2010, and a potential savings of 0.1 percent of GDP. The average of the highest-performing districts in Health is 35 percent higher than the average of all districts. 5.24 Bringing every district up to the outcome-to-spending ratio of those that perform very highly in both sectors could save U Sh 323 billion, or about 0.8 percent of GDP. This figure is 61 Technical details for this exercise are included in Annex 1. 62 Some districts have been omitted due to data limitations, and so this potential savings figure does not consider all districts. 71 calculated using the districts in both Tables 11 and 13 (Kampala, Mbale, and Mbarara) as a benchmark. These districts are outliers in terms of performance, and particularly in the case of Kampala, are qualitatively different from the other districts in Uganda, but they represent the possibilities of achieving very high outcomes with each dollar spent. 5.25 This figure reflects only part of the potential gain from improved district-level efficiency. The calculations above hold government performance fixed, and improve all districts to the outcome-to- spending levels of the best performing districts. If, instead, all districts could improve both their performance and outcome-to-spending levels, there would be gains from more efficient spending and flow on benefits in terms of higher economic performance and improved standards of living. Those gains would likely be significantly larger. C. Investigating the Determinants of Value-For-Money 5.26 It is important to understand what causes value-for-money. A better understanding of the factors associated with high value-for-money enables governments to better pursue the aims of fiscal decentralization, in particular when seeking to improve service delivery. This question was investigated using an econometric analysis. Its results are discussed below (see Annex 3 for technical details). 5.27 The highest levels of value-for-money are found in districts with low levels of spending. This result reflects decreasing returns to funding; as funding for a particular district increases, each individual dollar has a smaller and smaller impact on outcomes. 63 One reason for this could be that early results in each sector are relatively inexpensive to achieve, but then higher levels of outcomes require increasing amounts of resources. Alternatively, the problems faced by relatively well-funded districts may not be as responsive to funding as problems faced by less well-served districts. This effect is significantly more pronounced in the health sector (Figure 44). 63 The analysis cannot rule out the possibility of increasing returns to spending at low levels. It is possible at very low levels of spending that additional funding has a very high impact in enabling districts to achieve sufficient scale to perform effectively. 72 Figure 44a: Decreasing returns to higher Figure 44b: Decreasing returns to higher spending in Education spending in Health 3 12 Average Outcome-to- Average Outcome-to- 2.5 10 Spending Ratio Spending Ratio 2 8 1.5 6 1 4 0.5 2 0 0 1 2 3 4 5 1 2 3 4 5 Spending by Quintile Spending by Quintile Source: MoES and WB staff calculations. Source: MoH and WB staff calculations. 5.28 Districts that received an unqualified audit in the Auditor General’s Annual Report had consistently higher levels of value-for-money. This suggests that accountability and control of corruption are important for achieving more efficient outcomes. Where public officials are more accountable for the use of their funds, funds appear better used and services more efficiently provided. 5.29 This result may reflect the importance of accountability, governance, and, more generally, the overall institutional environment. Data limitations prevent the analysis from directly tackling other questions of governance, but it is possible that the significance of accountability as measured by the Auditor General’s report comes not just from control of corruption, but also from good governance and management mechanisms more generally (Figure 45). Tsimpo et al. (2012) highlighted the prevalence and effect that concerns such as absenteeism raise in the health sector, and the importance of management practices in underpinning good performance. Figure 45a: Good Governance is Figure 45b: Good Governance is Associated Associated with More Efficient Education with More Efficient Health Provision Provision 3 20 Average Outcome-to- Average Outcome-to- 2.5 15 Spending Ratio Spending Ratio 2 1.5 10 1 5 0.5 0 0 Unqualified Qualified or Unqualified Qualified or Disclaimer Disclaimer Rating in Auditor-General's Report Rating in Auditor-General's Report Source: Auditor General’s reports. Source: Auditor General’s reports. 73 5.30 In Education, higher income per capita in a district makes it easier to achieve better outcomes. In the data, higher incomes are associated with higher value-for-money (Figure 46). A likely cause for this is that higher income better enables families to take advantage of government-funded opportunities and to purchase complementary goods, and so facilitates more efficient government (Tsimpo et al. 2012). An alternative explanation is that when citizens have higher incomes, they are more likely to pressure local governments for better services (see, for example, Afonso and Fernances 2008; De Borger et al. 1994). Figure 46: Higher Income is Associated with More Efficient Education Provision 4 Average Outcome-to- 3 Spending Ratio 2 1 0 1 2 3 4 5 Income by Quintile Source: National Service Delivery Survey. 5.31 In health, there is some limited evidence for small economies of scale effects. The analysis identifies the possibility of an effect whereby, as districts become larger, they can more efficiently deliver health services. However, the estimated coefficient is not very large, and Figure 47 suggests that the relationship is not straightforwardly linear. Moreover, as mentioned in Section 4, care should be taken not to rely too heavily on results about economies of scale in this context. 64 Figure 47: Larger Population is Associated with More Efficient Healthcare Provision 15 Average Outcome-to- Spending Ratio 10 5 0 1 2 3 4 5 Population by Quintile Source: National Service Delivery Survey. 64 Section 4 mentions the difficulties with identifying economies of scale using the usual techniques in the literature, as spending is determined by central government funding and patterns are likely the product of granting criteria. 74 5.32 There is little evidence to currently suggest a significant difference in value-for-money between old and new districts. This is consistent with Section 4, which suggests that there is not a significant difference between public administration expenditures in new and old districts. However, it is worth repeating the warning expressed in Section 4: although economies of scale effects currently appear small, and no significant increases in management expenses have been observed so far in the proliferation of districts, this does not imply that these effects will continue to be small should redistricting continue. In particular, the analysis of 5.2 and 5.3 considers district governments prior to the creation of new districts in 2011, so care must be taken when extrapolating these results. 5.33 To sum up, value-for-money in social sectors can be increased by shifting funding from districts that are already well-served to districts that are relatively under-served. Additional spending appears to have a greater impact where current levels of spending are low, and so shifting resources to those areas will have a larger effect on education and health outcomes. It should be noted that the benefits of shifting funding from well-served districts to under-served districts in terms of avoiding decreasing returns come in addition to the fact, observed earlier in this report, that districts with lower levels of funding also tend to have poorer outcomes, and so are more in need of additional funding. 5.34 Accountability is another key factor. Stronger local government institutions improve the efficiency of spending in other areas, enabling the spending of limited resources to achieve better outcomes. This finding echoes the suggestion made in LGFC (2012) that greater emphasis should be placed on improving auditing and reporting processes in lower-level governments. This finding also highlights the need for investigation at a finer level of detail, in which specific institutions at the local government level are the most important for value-for-money. Analysis of outcomes typically focuses on national institutions, however this result shows that variation in local institutions is also of great importance in pursuing effective policy outcomes. The Second LG Public Expenditure and Financial Accountability Evaluation provides useful information on such variation for a number of districts and municipal councils (see Annex 4.) 5.35 Accountability in this sense can involve both top-down and bottom-up mechanisms. Top- down auditing mechanisms, such as the Auditor General and the Inspectorate-General, can hold districts to account for both substantive and procedural weaknesses. Bottom-up mechanisms, particularly those making district governments more accountable to citizens, can also align local government decision- making and the needs of citizens, better achieving sector outcomes. This second channel might be strengthened, for example, by increasing the share of local sources of funding in the local governments’ budgets. Taxpayers are more likely to demand services from the governments to which they pay taxes. 5.36 One important way to improve both outcomes and effective local government spending is to continue to achieve economic growth. Policies that focus on economic growth can come from both local and central government. Economic growth will allow districts to achieve better outcomes, enable citizens to spend money on complementary goods to health and education, and encourage citizens to take better advantage of government. Higher income per capita in a district will also make it easier for the district to develop local sources of revenue. 75 VI. Policy implications 6.1 As indicated earlier in this report, in the early 1990s Uganda made a clear policy decision to deliver key social and municipal services and infrastructure on a devolved basis, by establishing a relatively autonomous local government system. Over the next 10 years, the legal architecture of institutional and fiscal arrangements that were established was supportive of this objective; the functions and powers of local governments were fairly clearly defined, they were given substantial authority over their civil services (including hiring and firing), they were assigned substantial own-revenue sources, and local government fiscal transfers—including those that were largely unconditional—increased over time. As a result, Uganda established a local government system that is among the most decentralized in Africa. 6.2 However, beginning in the mid-2000s and for a variety of reasons explored in Section 2, key elements of this framework began to crumble. The number of district governments grew rapidly as existing districts were sub-divided through an ad hoc process that appears to have paid little regard to technical considerations or capacity and resource realities. Districts lost appointment and dismissal powers of senior staff; substantial local revenue powers were scrapped; in real, per capita terms intergovernmental transfers declined, with the largest (proportional) cuts affecting grants that were unconditional (hence, which provided the greatest fiscal autonomy). In the case of Kampala, the local government authority was, for all intents and purposes, discarded and replaced with a central Ministry. 6.3 The quality of the local and intergovernmental framework has been increasingly marred by three serious problems. These are: a lack of coherence (i.e. an increasingly contradictory mix of decentralized and centralized features); a mismatch between the staffing structures of the district governments, their financing, and the availability of human resources (resulting, for example, in chronic vacancy rates at the senior staff level); and instability. Together, these features have combined to create a situation of local government dysfunction in which it is increasingly difficult to improve local service delivery. The system has exhibited increasing signs of strain and dissatisfaction with service delivery has been on the rise. Moreover, it appears that some services are not in fact being rendered. As a result, the government of Uganda appears to be increasingly concerned about the performance of the local government system and is seeking measures through which to address the problems that have emerged. 6.4 The remainder of this section outlines institutional and fiscal proposals based on the analysis presented in this study. A number of broad short- to medium-term measures need to be implemented in order to: (i) bring greater stability to the districts and intergovernmental system; (ii) enhance its institutional design to strike the right balance between capturing economies of scale and better targeting local service delivery in a cost-effective way; and (iii) redesign funding of the district governments in a way that helps to improve service delivery outcomes under given budget constraints. 76 6.5 These measures—institutional and fiscal—are outlined in Table 12 and elaborated on below. In all cases, further detailed institutional and fiscal analysis, often sector-specific, would be needed in order to realize the broad policy thrust that is proposed. Table 12. Summary of Policy Recommendations Institutional proposals Fiscal proposals Establish a systematic technical process for the Enhance own-sourced revenues of the district creation of new districts. governments. Reform districts’ civil services to adjust for new Simplify the conditional transfers system and realities and budget constraints. improve its horizontal equity. Deconcentrate selected departments of the line Increase the share of unconditional transfers in the ministries to the regions and/or facilitate the districts’ funding and design them in a way that development of inter-district cooperative enhances district governments’ accountability for arrangements in the sectors where they are needed. their performance. A. Institutional proposals 6.6 The most immediate priority is to bring some order, stability, and technical rationality to the process of new district creation. In this regard, the recent Cabinet decision to place a moratorium on the creation of new districts (see Section 2) is an important first step. However, in addition to this, it will be necessary to establish a systematic process to govern the creation of districts (and other local governments) and to delineate geographical boundaries. A proposal relating to this is suggested in Box 4, while Box 5 offers a conceptual framework for technical work on the optimal size of a district. 6.7 Reform of the institutional structure of districts’ civil services should be the next step. The organizational structures of Uganda’s district governments are disconnected from the human and financial resources currently available, and often from the districts’ needs as well. The basic organizational structures of district governments—such as administrative configurations, staffing numbers, and conditions of service for staff positions—need to be reviewed and recalibrated in line with the service delivery mandates of the local governments, their increased numbers, and the budget constraints of the consolidated government as discussed in Section 3. This reform can help to reduce a trade-off between filling all positions in district governments and increasing expenditures related to local public administration. Districts will need to be differentially resourced in line with the different needs they face. Other things being equal, smaller districts have smaller needs. For example, it may not be necessary for all districts to own and operate a grader for road maintenance. Where the road network of any given district makes this economical, a grader can be provided; where that is not the case, the district can contract out on an occasional basis to meet such needs as they arise. 77 Box 4. A Process for the Creation of New Local Governments: Lessons from International Experience and Implications for Uganda Lessons from international experience Haphazard processes of local government creation and boundary determination as witnessed in Uganda are not uncommon historically. Learning from these experiences, some countries have developed fairly successful approaches to ensure that the technical factors mentioned above are taken into account and play an important role in new local government creation, while also fostering an orderly process that does not create the sorts of instability and under-resourcing characteristic of the current situation in Uganda. In Denmark, a Commission on Administrative Structure (CAS) was established in 2002 to provide technical analysis of the existing LG system. After a long hearing process and consultations with local government, a major reform of the LG system was introduced in 2007, during which a number of local governments were consolidated (the total dropped from 271 to 98) and minimum and maximum sizes for local governments were established, based on service delivery cost, quality, and other technical factors. In South Africa, the Municipal Demarcation Board (MDB) has been established in law as an entity with responsibility for determining local government boundaries (hence also creating new or consolidating existing local governments) on the basis of a host of factors, such as financial viability, administrative capacity, topographical and physical features, and so on. In reviewing such experiences, three broad lessons emerge: a. In determining the number, size and boundaries of local governments, it is critical that technical factors such as those already stated in Uganda’s Constitution, as well as those used by the Danish CAS and South African MDB, are taken into consideration. While it is impossible to fully insulate LG determinations from political factors, it is key that a clear, robust, and institutionalized process be established, and that such a process ensure that the local governments emerge from this process properly resourced, in line with existing fiscal, human resource, and other capacity constraints. b. Actual numerical ranges or targets relating to population, geographic size, etc. may or may not be established. Based on prior analysis, Denmark’s CAS process led to a stipulation of minimum and maximum sizes, with an average population of 55,000 per local government. In South Africa, where the average LG population is about 239,000, size ranges are not stipulated; rather, the emphasis is on ensuring that appropriate criteria are applied on a case-by-case basis. What matters most is to ensure that efficiency and effectiveness factors play an important role in the determination of new local governments, that the districts created be viable, and that any quantitative targets or ranges be based on these factors. In Uganda today, because this sort of technical work has not yet been undertaken, there is no immediate way of determining the optimum size of local governments, and it would be hazardous to the future health of the local government system to impose quotas until this analysis is completed. c. Particularly in the Ugandan context, any process of new district creation needs to ensure sufficient stability of the local government system; specifically, it should be rolled out according to a timetable that is clearly stated and is sufficiently long to ensure that the local government system is not constantly disrupted by the ongoing creation of new entities. Implications for Uganda It follows from the above that the priority for Uganda is to institute a systematic and institutionalized mechanism for the process of new district creation to replace the current one, preferably through a Ministerial regulation promulgated in terms of the LG Act. Such a mechanism could have the following features: a. A Commission (or similar entity) established on either an occasional or a permanent basis. To avoid the risk of overly frequent new district creation, it might be best to establish it on a one-off basis, at least to begin with. A minimum period for the applicability of the Commission’s determinations should be specified, e.g. for the period of the censual cycle. b. The core function of the Commission should be to undertake any necessary technical and consultative work on the basis of which it should propose the optimum sizes, numbers, and boundaries of districts in Uganda, according to specified criteria. These criteria should include those already specified in the Constitution, as well as financial viability, administrative capacity, human resource capacity, existing and expected human settlement, infrastructure endowments and land use patterns, and environmental impacts. Until the Commission has completed its work and developed its proposals, the specification of district sizes should be avoided. 78 6.8 Rationalization of districts’ civil services would go some way towards striking the right balance between capturing economies of scale and better targeting local service delivery. However, this problem cannot be addressed entirely at a district level, and other measures are needed in addition to district-level reforms. 6.9 There appears to be a growing need to deconcentrate certain departments of central line ministries at the regional level to allow for more effective execution of their tasks. In fact, to some degree this is already happening; MoES has established four regional offices, and MoWE has set up eight Development Support Fund branches at the regional level. There is substantial scope for this sort of administrative and activity-based deconcentration to improve the capabilities of central line agencies to (i) execute their own functions more effectively; and (ii) support the districts in the execution of their currently mandated roles in areas such as primary health, education, roads, water etc. more effectively by introducing a greater on-the-ground presence, shortening the “bureaucratic� and logistical distance between districts and ministries, and so on. 6.10 There is also a possible need for some sort of inter-district structure or co-operative arrangement. In late 2012, MoLG circulated a proposal to form regional governments as an intermediate layer between the districts and central government. Broadly speaking, this change appears unlikely to lead to service delivery improvements in the medium- or long-term, but could cause a deterioration of service delivery in the short term (see Annex 5 for details). It also appears that support for this proposal may be waning. Short of the establishment of regional governments or administrations, inter-district cooperative arrangements might be effective where cross-boundary planning and investment activities are required in service delivery areas for which district governments are responsible. An advantage of this “polycentric� approach (see Box 6) is that voluntary cooperation of district governments increases the bottom-up accountability of any institutions they may decide to create. The Constitution of Uganda provides for this sort of initiative, but establishes very demanding procedural requirements for such arrangements to come into existence. 6.11 As a point of departure for any arrangements that might emerge, it is suggested that districts and central line ministries retain their current powers and functional responsibilities. This does not mean that such powers and responsibilities should not be changed at all; there may be some scope for their eventual review (see Box 7 for some basic principles; further work is needed in this area). If, however, any such re-allocation is undertaken, it should preserve the vertical integration of functional responsibilities. For example, if a newly constituted district has a relatively small roads network, hence an establishment size that does not allow it to effectively develop or maintain a major road which was previously characterized as a district road, it may be necessary to re-classify this road and place it entirely under the responsibility of the UNRA. However, it would be unwise to divide development and maintenance and operation responsibilities for this road between the district government and the UNRA, as this will weaken accountability, and is thus likely to reduce the quality of service delivery. 79 Box 5. Thinking about the Optimal Size of a District Many factors that vary with the country and the region in question can affect the optimal size of a local government jurisdiction. The key factors include: (i) the existing structure, responsibilities, and revenue sources of local governments, (ii) the service delivery conditions, including: a. geography and topography; b. homogeneity and types of service demands; c. availability of skilled local government employees; d. existing variability in service delivery across the country; and e. political strength of local leaders and bureaucrats. Gains from having either a larger or smaller jurisdiction size are far from certain. More careful analysis of objectives and realistic outcomes is appropriate before changes in either direction are made. The potential costs or savings must be carefully and realistically evaluated in terms of the actual setting in which these changes will take place. Several generalizations can be made. First, the extent of scale economies is lower than may be anticipated—bigger does not always imply lower costs. The traditional means of producing many public services using many small facilities near people (such as schools) and employing labor-intensive technologies does not lend itself to economies that extend to wide geographic areas. Some other services, of course, offer a greater potential for economies. Second, there are likely to be significant transition costs and time spent in moving from the existing government structure to a new one. Some of these costs will be related to additional expenditures and others will be in the form of poor service delivery and citizen dissatisfaction that will arise in operating a newly-created government. Third, making changes to the structure of local governments involves different actors with various individual goals and motivations. Decisions to undertake such changes should not be made under the assumption that the goals are altruistic and that the various actors will do whatever is necessary to make these changes succeed. The involved people—local and national bureaucrats, local and national politicians, and service deliverers—may in fact thwart or enable the new structure, depending on the design. They will, to a very real degree, determine whether it works or fails. Fourth, competition between governments increases the quality of service delivery and the effectiveness with which government delivers services, but competition is more likely to be generated by means other than increases in the number of local governments. Use of the private sector to deliver services should be carefully investigated, as it has the potential to offer cost savings because firms can deliver services to multiple jurisdictions at the same time. The private sector is already frequently employed in providing specialized components of public services in many countries, such as construction and engineering, and in many cases to deliver entire services. Finally, a number of other public sector options exist. Governments can work together in a variety of ways. For example, Monterrey, Mexico delivers sanitation services with a metropolitan-wide authority, Chile uses voluntary clustering of municipalities and Turkey has a greater metropolitan model. These experiences illustrate that countries can be successful using a variety of different structural models. This set includes a consolidated government, a cooperative arrangement between governments and a single service delivery district. As noted above, perfectly mapping every service may create considerable uncertainty and confusion on the part of citizens, but specialized systems for services with area-wide effects or for services with economies of scale should also be considered when decisions on restructuring government are being made. Source: Fox and Curley 2006. 80 Box 6. Polycentric Governance Polycentric governance is an alternative approach to either highly centralized government or the development of regional governments. The key idea is that each unit of government organizes and governs itself while fostering interdependent relationships with other government units. In a polycentric approach, district governments in Uganda could collaborate with other local governments, central government units, or other enterprises, such as the private sector or non-profit organizations, to pursue common interests. This approach is flexible and tailored to the needs of each community. For polycentricity to work, however, it requires the ability to develop mutually beneficial associations between participants, to manage conflict, and to resolve disputes. All participants must have political, administrative and fiscal authority of their own. Examples of polycentric governance include some forms of federal government, metropolitan government, and special purpose districts. The benefits of a polycentric approach arise primarily from the ability of governments to work effectively on the interests most connected to them. Local governments are better positioned to assess and act on the needs of the community, but they can also be assisted by pooling resources with other nearby governments and by the greater resources and coordinating abilities of national entities. Cooperation based on common interests can help foreground each of these strengths. This approach is most likely to be of use where traditionally defined jurisdictions are less relevant, such as in densely populated regions of bordering districts, in situations where local government interacts with community associations, and in the management of shared resources, such as watersheds. Experience in Africa and in developing countries on other continents shows that a polycentric approach enables local governments to more effectively coordinate and deliver public services than can central government agencies or non- governmental organizations. Case studies include: x In Rwanda, a polycentric approach was designed to promote self-governance and development and to empower local community groups. Under the Ubudehe program, trained facilitators convene volunteer community representatives, who meet to prioritize local problems and identify requirements to assess deficits. Disbursements from a central development fund are then combined with local resources and management to implement the plans developed. This project has changed public attitudes and beliefs about development, improved local government problem solving, increased accountability for resources, and strengthened national statistics for common development planning. x The Government of Ghana, working with the U.S. Agency for International Development (USAID), developed the Resiliency in Northern Ghana (RING) program to improve food security and nutrition in that region. Local and central government agricultural and nutrition specialists have teamed up to provide integrated extension services in rural communities. Over the next three years, this program will provide $28 million directly to selected local governments in the northern region to fund public investments that will build local capacity in the agriculture and health sectors. USAID funds will augment local government revenues and transfers from the government of Ghana. The program includes an additional $32 million in technical assistance to help local governments develop the capacity to effectively manage public service delivery. x In South Africa, the e’Thekwini Municipality worked with the Ministry of Water Affairs and Forestry to tackle spiraling water debt due to non-payment or irregular payment of consumer water bills in low-income areas. Local officials campaigned to explain problems to citizens, propose solutions to reduce water consumption, and identify water leaks. Social workers were trained to educate indigent consumers in their homes about options and programs. As a result, 70 percent of targeted customers pay their accounts regularly, water consumption has been reduced from an average of .63 kl per day to .55 kl, municipal debt has been reduced, and public funds to help those who are unable to pay for water have increased. x The Zambia Ministry of Health created the Public Service Capacity Building Program and the Public Service Reform Program to improve the efficiency and accessibility of public health services. Through these initiatives, the Ministry delegated responsibility and authority for public health management to district health boards. To meet its responsibilities, the Masaiti District Management Board built partnerships with a broad range of organizations working in the health value system, including the Planned Parenthood Association of Zambia, the National Food and Nutrition Commission, the University of Zambia, and the Zambia Flying Doctor Services. As a result of these collaborative efforts, the district was able to decrease maternal mortality from 660/100,000 to 330/100,000; increase infant survival from 54 percent to 84 percent; increase the TB cure rate from 63 percent to 74 percent; increase the family planning new patients rate from 46/1,000 to 130/1,000; and increase antenatal coverage from 50 percent to 84 percent. 81 Box 7. Principles for the Allocation of Responsibilities between Different Levels of Government The question of appropriate allocation of responsibilities and powers among the various levels of government can be analyzed in five key steps. Step One: Determine the Economic Nature of the Public Service Where the provision of a good or a service is non-rival (one person’s consumption does not affect the supply available for others), non-excludable (it is difficult or impossible to prevent specific individuals from consuming), or both, there is role for public sector provision. Goods that fall into these categories include basic education and literacy, health care that prevents communicable diseases, law enforcement, natural resource management and some physical infrastructure. The extent to which goods are non-rival or non-excludable across districts should serrve as a guide for when there is a role for heightened coordination or central government involvement. Step Two: Analyze the Value Chain of the Public Service Deciding how to best organize and govern public service delivery requires a detailed and realistic understanding of the relevant value system, including local physical conditions and support activities, such as finance and planning work. For example, are services most efficiently delivered on a small or large scale? How does the delivery of one service affect the delivery of other services? What are the opportunities for private sector or non- profit involvement? Where does expertise currently reside? Which set of actors has the strongest links to and interest in value creation? Step Three: Separate Production and Provision Activities In this context, production specifically refers to activities involved in transforming inputs into outputs, for example, construction of a road. Provision, on the other hand, refers to activities associated with ownership, design, planning, finance and distribution. It can be the process of determining the need for the road, acquiring property rights and finance, and maintaining the road once it has been built. The actor that produces the service need not be the actor that provides the service. Being aware of the potential separation of these activities and allocating them to groups best able to perform them is a key step in institutional design. It is possible for different governments to jointly contract for services, to jointly produce or provide, or jointly produce and provide. Step Four: Develop Design Criteria for Governing Public Service Delivery Design criteria provide guidelines for creating governing arrangements that prevent failures, improve coordination, remove uncertainties and reduce the risks and costs associated with public service delivery. These criteria provide guidance as to where the relevant opportunities and risks lie, and which actors are best suited to deal with them. General design criteria include predictability, efficiency, efficacy, distributional equity, accountability, checks and balances, independence, due process, and flexibility. Step Five: Assess Alternatives Against Design Criteria Once analysis of the underlying good, value chain, and production/provision distinction are understood, alternative governance arrangements should be compared, identifying design gaps and recognizing trade-offs. Constructing a comparison in the form of a “scorecard� helps designers clarify and communicate important values, preferences, and beliefs, and provides a basis for reasoned discussion and further creative design. 82 B. Fiscal proposals 6.12 In general terms, some of the main proposals found in the 2012 LGFC report “Review of Local Government Financing� provide a good point of departure. The report began to develop an approach to addressing the fiscal dimensions of the local and intergovernmental system, although more work will be needed to make this approach operational. 6.13 For example, with regard to own-source revenues, it is clear that substantial improvements can and should be made in this area. Before the abolition of the graduated tax several years ago, the district governments were able to collect substantially larger amounts of revenues compared to what they are collecting now, and there is no reason to believe that their capacity to do so cannot be restored. Enhancing own revenues of district governments is likely to have a positive impact on both the political economy and fiscal dimensions of local governance. It would strengthen the bottom-up accountability of district governments vis-à-vis their constituents. It would also help to alleviate, at least to some extent, the fiscal pressures experienced by local governments. The existence of such pressures at all levels of government could make it politically easier to legislate and implement the enhancement in district government revenues. A preliminary assessment of the LGFC proposal is presented in Annex 6. However, a more in-depth study of particular revenue sources is required in order to develop a properly sequenced approach, which is both viable and appropriately targeted. It is also important to look at any trade-offs that may exist between expanding own-source revenues of district governments and improving the central government’s capacity to collect its revenues. 6.14 Value-for-money needs to be improved, in particular in the education and health sectors, and the system of inter-governmental fiscal transfers needs to be redesigned to serve this purpose. Unless the medium-term allocations for these sectors are revisited with a view to reversing this trend, increasing value-for-money in service delivery might be the only way to prevent further deterioration in the ability of local governments to deliver these services. Two broad directions for redesigning the system of fiscal transfers could be suggested. 6.15 Horizontal equity in the allocation of resources between local governments through conditional transfers needs to be improved. The current complex mixture of historical practices, need- based formulas, and ad hoc considerations determining the amounts of such transfers to individual districts needs to be replaced by a simple, formula-based system based on transparent criteria. For example, in the case of education, conditional transfer to a district could be calculated as a standard amount per school-going child (adjusted, if necessary, for input-cost differentials between jurisdictions) multiplied by a number of school-going children. Such a system would be more equitable than the one currently in place, and, as analysis in Section 5 has shown, also more efficient. Reforming the system for allocation of conditional transfers would entail a redistribution of fiscal resources among districts, thus creating winners and losers. Managing the political economy risks and local budget shock implications of this can be handled through phasing in such a system, gradually ensuring that the fiscal impacts of the reform are spread over time. 6.16 It is also important to reverse the trend of declining unconditional transfers to districts and to provide such transfers in a way that enhances district governments’ accountability for their performance. Such transfers provide the key resource through which local governments can respond 83 meaningfully to local needs. Increasing the share of unconditional grants in the transfers to local governments, with their amounts determined by district performance in a number of governance and other efficiency indicators, will create powerful incentives for improving top-down accountability mechanisms important for delivering value-for-money in local service delivery. Going forward, it will be important to ensure that funding for the existing Local Development Grant, which targets district governments, is expanded and that the assessment system for the allocation of this grant is strengthened. The Municipal Development Grant (MDG) could serve as a model for these changes (see Box 8). 84 Box 8. The Municipal Development Grant The Municipal Development Grant (MDG) aims to enhance the capacity of targeted municipalities to improve urban service delivery in the context of a rapidly urbanizing society in which significant service delivery responsibilities have been decentralized. It will provide substantial additional funds to the targeted municipalities for investment in urban infrastructure, designed in such a way as to leverage and incentivize improved institutional and delivery performance of these bodies. It will include the Municipal Capacity Building Grant (MCBG), which will provide municipalities with the resources necessary to access the capacity-building inputs needed to achieve the performance that the MDG will incentivize. Municipalities will have significant discretion in selecting priority projects, from a menu of typical investments that are currently not supported through earmarked transfers and that are key to developing and managing the built environment, namely: (i) roads and associated infrastructure; (ii) liquid and solid waste management; (iii) water and sewerage; (iv) local economic infrastructure (e.g. markets); and (v) urban transport (e.g. bus terminals). In order to achieve improved performance on the part of targeted municipalities, MDG funds will be allocated annually to municipalities on the basis of a transparent, equitable formula, combined with a performance score determined through an annual assessment. The indicative allocation of the MDG to the participating municipalities will be based the formula taking into account municipal (i) administrative land area (15 percent weight); (ii) projected population (45 percent weight); and (iii) poverty head count (40 percent weight). The actual allocation to each municipality will, in reality, be determined by the actual weighted performance scores for four areas (see below), so as to provide the necessary incentives for institutional improvement as well as progress in refining urban infrastructure services. The MDG assessment criteria will focus on areas of critical relevance to improved municipal performance, namely: a. Improved linkage between Municipal Physical Development Plan, Five Year Development Plan, and Budgeting; b. Increased municipal own-source revenue (OSR) performance; c. Improved procurement performance; d. Improved financial management performance; e. Improved budget execution performance; f. Improved accountability and transparency; g. Improved environmental and social management; h. Improved delivery of urban infrastructure (quantity and quality). In order to maximize the objectivity and robustness of the annual assessment, it will be conducted by an external private firm contracted to perform the task. The assessment will cover four areas: i. A Minimum Conditions assessment, which will determine whether the Municipality has met a number of key minimum conditions to ensure that it is capable of handling at least a proportion of the MDG amount and to provide basic comfort in respect of fiduciary and other (e.g. social and environmental management) risks. In order to receive any MDG funding, a municipality will need to comply with all the minimum conditions, and such compliance (alone) will allow it to receive around 21.4 percent of the total grant amount. ii. An Institutional Performance assessment that will cover areas a. through g. above. Assuming that a municipality has met all the minimum conditions, it will receive an additional allocation of up to 43 percent of the grant amount in direct proportion to its performance score. iii. An Infrastructure Delivery assessment that will measure the performance of the Municipality in the actual delivery of urban infrastructure in both quantitative and qualitative terms (i.e., area h. above). This will account for up to 28.2 percent of the grant amount. As with the institutional performance assessment, individual Municipal allocations will be determined in direct relation to the score they achieve in this area. iv. A Capacity-Building Plan assessment that will focus on whether the municipality has a capacity-building plan in place and assess whether the previous MCBG expenditures have been within allowed parameters. If so, the municipality will receive its annual capacity-building grant, amounting to roughly 7.4 percent of the total grant. 85 86 References Afonso, Antonio and Sónia Fernandes. 2008. “Assessing and explaining the relative efficiency of local government.� Journal of Socio-Economics 37 (5): 1946-1979. Afrobarometer. 2009. The Quality of Democracy and Governance in Africa: A Compendium of Public Opinion Findings from 19 African Countries, 2008. Afrobarometer Public Opinion Research Survey Round 4. http://www.afrobarometer.org/. Ahmad, Ehtisham, Giorgio Brosio, and Maria Gonzalez. 2006. Uganda: Managing More Effective Decentralization. Working Paper 06/279. Washington, DC: International Monetary Fund. Apter, David E. 1997. The Political Kingdom in Uganda: A Study of Bureaucratic Nationalism. London: Frank Cass Publishers. Asiimwe, Delius and Nakanyike Musisi. 2007. Decentralization and Transformation of Governance in Uganda. Kampala: Fountain Publishers. Azfar, Omar, Jeffrey Livingston, and Patrick Meagher. 2001. Decentralization in Uganda. Working Paper. College Park, MD: Institutional Reform and the Informal Sector (IRIS) Center, University of Maryland. Bahl, Roy. 2011. “Financing Metropolitan Areas,� Chapter 9 in Local Government Finance: The Challenges of the 21st Century (Second Global Report on Decentralization and Local Democracy). Edward Elgar for United Cities and Local Governments. Benin, Samuel et al. 2007. Assessing the Impact of the National Agricultural Advisory Services in the Uganda Rural Livelihoods. IFPRI Discussion Paper 0724. Bergstrom, T. C. and R. P. Goodman. 1973. “Private Demands for Public Goods,� American Economic Review June 1973: 280-296. Bish, Robert L. 1990. Local Government in British Columbia (2nd edition). Vancouver: Union of British Columbia Municipalities. Boadway, Robin and Anwar Shah. 2009. Fiscal Federalism: Principles and Practice of Multiorder Governance. Cambridge University Press. Borchording, T. E. and R. T. Deacon. 1972. “The Demand for the Services of Non-Federal Governments.� American Economic Review (December): 891-901. Bratic, Vjekoslav. 2008. “Local Self-Government in Central and Eastern Europe – A Strong and Independent Local-Level Management Tool or Just a Paper Tiger?� Financial Theory and Practice, Vol. 32 (2): 139-157. British Columbia, Ministry of Community Services. 2006. A Primer on Regional Districts in British Columbia. www.crd.bc.ca/about/documents/primer_rd_bc.pdf. Byrne, Rita M., ed. 1990. Uganda: A Country Study. Washington, DC: GPO for the Library of Congress. 87 Chicoine, David L., Steven C. Deller and Norman Walzer. 1989. “The Size Efficiency of Rural Governments: The Case of Low-Volume Rural Roads.� Publius: The Journal of Federalism. Vol. 19 (Winter): 127-138. Danish Ministry of Health, Municipalities and Regions. www.im.dk/English/Municipalities-regions.aspx. Danish National Association of Local Authorities. 2005. “Local Government Reform: In Brief.� www.kl.dk/English/Local-Government-Denmark/. De Borger, B., K. Kerstens, W. Moesen, J. Vanneste. 1994. “Explaining Differences in Productive Efficiency: An Application to Belgian Municipalities.� Public Choice 80 (3/4): 339–358. De Borger, Bruno and Kristiaan Kerstens. 1996. “Cost Efficiency of Belgian Local Governments: A Comparative Analysis of FDH, DEA, and Econometric Approaches.� Regional Science and Urban Economics Vol. 25: 145-170. DeConinck, J. 2004. “The State, Civil Society, and Development Policy in Uganda.� In K. Brock et al. (eds), Unpacking Policy: Poverty Reduction in Uganda and Nigeria. Kampala: Fountain Publishers. Deller, Steven C. and Edward Rudnicki. 1992. “Managerial Efficiency in Local Government: Implications on Jurisdictional Consolidation.� Public Choice Vol 74: 221-231. DENIVA. 2006. Civil Society in Uganda at the Crossroads? Kampala: DENIVA. Eaton, Kent, Kai Kaiser, and Paul Smoke. 2011. The Political Economy of Decentralization: Implications for Aid Effectiveness. Washington, DC: The World Bank. Economic Policy Research Centre (EPRC). 2011. An Insight into the Performance of the NAADS Programme, Makerere University. Surazska, W. (ed.). 1997. Special Issue: From Regional Administration to Regional Government. Environment and Planning C: Government and Policy Vol 15 (4). Francis, Paul and Robert James. 2003. “Balancing Rural Poverty Reduction and Citizen Participation: The Contradictions of Uganda’s Decentralization Program.� World Development 31(2): 325–337. Fox, W. F. and T. Curley. 2006. Will Consolidation Improve Sub-National Governments? Policy Research Working Paper no. 3913. Washington, DC: The World Bank. Government of Uganda. 2012a. Budget Speech 2012/13: “Priorities for Renewed Growth and Development.� Kampala: Ministry of Finance, Planning and Economic Development. Government of Uganda. 2012b. Background to the Budget 2012/13. Kampala: Ministry of Finance, Planning and Economic Development. Government of Uganda. 1994. Decentralization in Uganda: The Policy and Its Implications. Kampala: Ministry of Local Government, Decentralization Secretariat. Government of Uganda. 2000. Poverty Reduction Strategy Paper. Kampala: Ministry of Finance, Planning and Economic Development. 88 Government of Uganda. 2001. Fiscal Decentralization in Uganda: The Way Forward. Kampala: Fiscal Decentralization Working Group. Government of Uganda. 2002. Fiscal Decentralization in Uganda: Strategy Paper. Kampala: Fiscal Decentralization Working Group. Green, Elliot. 2008. District Creation and Decentralization in Uganda. Working Paper No. 24. London: Crisis States Research Center, London School of Economics. Hedger, Edward, Tim Wiliamson, Tom Muzoora, and Justina Stroh. 2010. Sector Budget Support in Practice: Case Study, Education in Uganda. Overseas Development Institute. UK and Mokoro. Hepburn, Neil, Edward LeSage, and Melville McMillan. 2004. Shared Service Arrangements: Determinants of Success. A Study of Economic Development and Recreation and Culture Shared Service Arrangements among Municipalities of the Alberta Capital Region. Information Bulletin 78, Western Centre for Economic Research, University of Alberta, June 2004. Hirsch, Werner Z. 1970. The Economics of State and Local Government. New York: McGraw-Hill Book Co. ITAD, Performance Evaluation of National Agricultural Advisory Services, June 2008. Jeppsson, Anders. 2001. “Financial Priorities under Decentralization in Uganda.� Health Policy and Planning 16(2): 187–192. Jeppsson, Anders. 2002. “SWAp Dynamics in a Decentralized Context: Experiences from Uganda.� Social Science & Medicine 55(11): 2053–2060. Kanyeihamba, G. W. 1975. Constitutional Law and Government in Uganda. Nairobi: East African Publishing House. Kasumba, George and Anthony Land. 2003. “Sector-Wide Approaches and Decentralization: Strategies Pulling in Opposite Directions? A Case Study from Uganda.� Maastricht: European Centre for Development Policy Management. Kauzya, John-Mary. 2007. Political Decentralization in Africa: The Experiences of Uganda, Rwanda and South Africa. New York, NY: UN Department for Economic and Social Affairs, Division for Public Administration and Development Management. LeSage, Edward and Melville McMillan. 2008. Alberta: Municipal System Overview. Edmonton: University of Alberta, Western Centre for Economic Research, Information Bulletin 110. March 2008. Updated January 2010 as Information Bulletin 135. http://www.business.ualberta.ca/Centres/WCER/Publications/InformationBulletins/ Local Government Finance Commission. 2012. Review of Local Government Financing: Financing Management and Accountability for Decentralized Service Delivery. Kampala: Local Government Finance Commission. Local Government Finance Commission. 2012. Review of Local Government Financing: Initial Draft Report. McGreer, Edward and Melville McMillan. 1993. "Public Output Demands from Alternative Congestion Functions." Journal of Urban Economics Vol. 33 (January): 95-114. 89 McMillan, Melville and Wing Chan. 2006. “Comparing University Efficiency Using Stochastic and Non- stochastic Methods: The Case of Canadian Universities.� Education Economics Vol. 14(1): 1-30. McMillan, Melville L. 2008. “A Local Perspective on Fiscal Federalism.� In Macro Federalism and Local Finance, Anwar Shah (ed.), Washington, DC: The World Bank. McMillan, Melville L. 1976. “Criteria for Jurisdictional Design: Issues in Defining the Scope and Structure of River Basin Authorities and Other Public Decision-Making Bodies.� Journal of Environmental Economics and Management Vol. 3: 46-68. McMillan, Melville L. 1989. “On Measuring Congestion of Local Public Goods.� Journal of Urban Economics Vol. 26 (September): 131-137. McMillan, Melville L. “A Local Perspective on Fiscal Federalism: Practices, Experiences, and Lessons from Developed Countries.� Public Finance Working Paper. World Bank (processed 1995). McMillan, Melville L. 2007. “Intergovernmental Transfers and Rural Local Governments.� Chapter 18 in Intergovernmental Fiscal Transfers: Principles and Practice, Robin Boadway and Anwar Shah (eds.). Washington, DC: The World Bank. McMillan, Melville L. 2012. “Patterns and Trends in Uganda Local Government Expenditures with Special Attention to the Districts: Service Delivery Implications.� World Bank Uganda Public Expenditure Review on Decentralization and Service Delivery. McMillan, Melville L. 1997. “Taxation and Expenditure Patterns in Major City-Regions: An International Perspective and Lessons for Canada." In P. Hobson and F. St. Hilaire (eds.), Urban Governance and Finance: A Question of Who Does What. Montreal: Institute for Research on Public Policy. McMillan, Melville L., W. R. Wilson and L. M. Arthur. 1981. “The Publicness of Local Public Goods: Evidence from Ontario Municipalities.� Canadian Journal of Economics Vol. 14 (November): 596-608. Mentor Consult Ltd and Dege Consult. 2008. Africa Local Council Oversight and Social Accountability (ALCOSA). “Studying Local Council Oversight in Uganda.� Draft Synthesis Report. MoES (Ministry of Education and Sport). 2010. “Uganda Education Statistical Abstract.� Ministry of Education and Sport, Kampala. MoES (Ministry of Education and Sport). 2011. “Uganda Education Statistical Abstract� Ministry of Education and Sport, Kampala. MoH (Ministry of Health). 2010, “Annual Health Sector Performance Report.� Ministry of Health, Kampala. MoH (Ministry of Health). 2011. “Annual Health Sector Performance Report.� Ministry of Health, Kampala. MoH (Ministry of Health). 2012. “Annual Health Sector Performance Report.� Ministry of Health, Kampala. Muhumuza, William. 2010. “State-Civil Society Partnership in Poverty Reduction in Uganda.� Eastern Africa Social Science Research Review 26(1): 1–21. 90 Muhumuza, William. 2003. “The 1998 Local Government Elections and Democratization in Uganda.� In Sabiti-Makara et al. (eds) Voting for Democracy in Uganda: Issues in Recent Elections. Kampala: LDC Publishers, pp. 226–254. Muhumuza, William. 2006. Participation, Empowerment and Accountability in Local Governments in Uganda. Sponsored by Rockefeller/Makerere I@mak Research Foundation Grant. Kampala: Makerere University. Muhumuza, William. 2008. “Pitfalls of Decentralization Reforms in Transitional Societies: The Case of Uganda.� Africa Development 33(4): 59–81. National Association of Counties (United States), www.naco.org/Pages/default.aspx. Oates, Wallace E. 1975. Fiscal Federalism. Harcourt Brace Jovanovich, Inc. (reprinted by Edward Elgar Publishing, 2011). Office of the Auditor General. 2009. “Annual Report of the Auditor General for the Year Ended 30th June 2009.� Office of the Auditor General, Kampala. Office of the Auditor General. 2010. “Annual Report of the Auditor General for the Year Ended 30th June 2010.� Office of the Auditor General, Kampala. Office of the Auditor General. 2011. “Annual Report of the Auditor General for the Year Ended 30th June 2011.� Office of the Auditor General, Kampala. Okidi, John and Madina Guloba. 2006. Decentralization and Development: Emerging Issues from Uganda’s Experience. Kampala: Economic Policy Research Center, Makerere University. Onyach-Olaa, Martin. 2003. “The Challenges of Implementing Decentralization: Recent Experiences in Uganda.� Public Administration and Development 23(1): 105–113. Reev Consult International. 2012. Review of the Local Government Set-Up in Uganda. Prepared for the Ministry of Local Government of Uganda. Reinikka, Ritva and Jakob Svensson. 2004. “Local Capture: Evidence from a Central Government Transfer Program in Uganda.� Quarterly Journal of Economics 119(2): 679–705. Reinikka, Ritva and Jakob Svensson. 2010. “Working for God? Evidence from a Change in Financing of Nonprofit Health Care Providers in Uganda.� Journal of the European Economic Association 8(6): 1159– 1178. Republic of Uganda. 1995. Constitution of the Republic of Uganda. Kampala: UPPC. Republic of Uganda. 2010. National Development Plan. Kampala. Saito, Fumihiko. 2004. Decentralization and Development Partnership: Lessons from Uganda. Tokyo: Springer Publishers. Sejjaaka, Samuel. 2004. “A Political and Economic History of Uganda.� In Frederick Bird and Stewart Herman (eds.) International Business and the Challenges of Poverty in Developing Areas. London: Palgrave-Macmillan, pp. 98–110. 91 Shah, Anwar. 2006. “A Comparative Institutional Framework for Responsive, Responsible, and Accountable Local Governance.� Chapter 1 in Anwar Shah (ed.) Local Governance in Industrialized Countries. Washington, DC: The World Bank, pp 1-40. Shah, Anwar. 2006. “A New Vision of Local Governance and the Evolving Roles of Local Governments.� Chapter 1 in Anwar Shah (ed.) Local Governance in Developing Countries. Washington, DC: The World Bank, pp 1-46. Slack, Enid. 2006. “Fiscal Aspects of Alternative Methods of Governing Large Metropolitan Areas.� Chapter 6 in Richard M. Bird and Francois Vaillancourt (eds.) Perspectives on Fiscal Federalism. Washington, DC: World Bank Institute, pp. 101-122. Smoke, Paul. 2012. “Political Economy of Institutions for Service Delivery.� For World Bank Uganda Public Expenditure Review on Decentralization and Service Delivery, September 2012. Smoke, Paul, William Muhumuza, and Emmanuel Ssewankambo. 2011. Uganda Decentralization Assessment. Washington, DC: US Agency for International Development. Smoke, Paul. 2008. “Local Revenues under Fiscal Decentralization in Developing Countries: Linking Policy Reform, Governance and Capacity.� In Gregory Ingram and Yu-Hung Hong (eds.) Fiscal Decentralization and Land Policies. Cambridge, MA: Lincoln Institute Press, pp. 38–69. Steffensen, Jesper and Emmanuel Ssewankambo. 2011. Diagnostic of Intergovernmental Fiscal System – Urban Authorities in Uganda. DEGE Consult APs and Mentor Consult Ltd. June 2011. Steffensen, Jesper, Emmanuel Ssewankambo, and Per Tidemand. 2004. A Comparative Analysis of Decentralization in Kenya, Tanzania and Uganda. Final Country Study. Tsimpo, C., D. Lukwago and I. Mulindwa. 2012. “Key Constraints to Child Education in Uganda.� Tsimpo, C., P. Okwero, D. Lukwago, S. K. Bakeera. 2012. “Health Facility Survey: Assessment of the Management of Resources for Health in Uganda.� Tumushabe, Godber, Elijah Dickens Mushemeza, Lillian Muyomba Tamale, Daniel Lukwago, and Eugene Ssemakula. 2010. “Monitoring and Assessing the Performance of Local Government Councils in Uganda.� ACODE Policy Research Series No. 31. Kampala: Advocates Coalition for Development and Environment. UBOS (Uganda Bureau of Statistics). 2008. “Statistical Abstract.� Uganda Bureau of Statistics, Kampala. ———2008. “Uganda National Service Delivery Survey.� Uganda Bureau of Statistics, Kampala. ———2009. “Statistical Abstract.� Uganda Bureau of Statistics, Kampala. ———2010. “Statistical Abstract.� Uganda Bureau of Statistics, Kampala. ———2010. “Uganda National Household Survey 2009/2010: Report on the socio-economic module.� Uganda Bureau of Statistics, Kampala. 92 ———2011. “Statistical Abstract.� Uganda Bureau of Statistics, Kampala. ———2012. “Statistical Abstract.� Uganda Bureau of Statistics, Kampala. Uganda Ministry of Works and Transport. 2011. Works and Transport Sector: Performance Report 2010/11, October 2011. Uganda National Roads Authority. 2011. UNRA’s Performance Report for FY 2010/11, September 2011. Uganda Road Fund. 2011. Annual Report 2009/10, October 2011. Uganda, Ministry of Agriculture, Animal Industry and Fisheries. 2010. Agricultural Sector Development Strategy and Investment Plan: 2010.11-2014/15, July 2010. Uganda, Ministry of Health, Annual Health Sector Performance Report, various fiscal years. ———Statistical Abstract, 2010. Uganda, Ministry of Local Government. 2012. Review of the Local Government Set-Up in Uganda: Draft Report, Reev Consult International, June 2012. Uganda. 2010. National Development Plan (2010/11-2014/15), April 2010. Uganda, Office of the Auditor General, Annual Reports, various years. Uganda. 2011. Ministry of Water and Environment, Water and Environment Sector Performance Report, September 2011. UNCDF. 2003. Local Government Initiative: Pro-poor Infrastructure and Service Delivery in Rural Sub- Saharan Africa. New York: UN Capital Development Fund. UNCHS. 2002. Local Democracy and Decentralization in East and Southern Africa. Nairobi: United Nations Center for Human Settlements. UNDP. 2007. Uganda Human Development Report. ———2003. Decentralization and Local Governance: Towards a “Home Grown� Solution in Uganda. Pretoria: United Nations Development Program. USAID. 2004. “USAID Uganda Success Stories: Decentralization in Uganda.� http://www.usaid.gov/regions/afr/ss02/uganda5.html. Villadsen, Soren and Francis Lubanga. 1996. Democratic Decentralization in Uganda: A New Approach to Local Governance. Kampala: Fountain Publishers. Wallace, T. 2000. “Is the Role Played by Donors in Supporting Uganda’s NGO Sector Enabling it to Develop Effectively?� In Paul Collins (eds.) Applying Public Administration in Development: Guide Posts for the Future. New York: John Wiley and Sons Ltd., pp. 227–247. 93 World Bank. 2011. BOOST: Spending and Performance, Data and Results, User’s Manual. October 2011. World Bank. 2010. Uganda Public Expenditure Review: Strengthening the Impact of the Roads Budget. Report No.5567-UG. World Bank. 2011. Uganda: Agriculture for Inclusive Growth in Uganda. Draft Final Report, June 2011. World Bank. 2003. Decentralization Policies and Practices: Participants Manual. Washington, DC: The World Bank. Wunsch, James and Dan Ottemoeller. 2004. “Uganda: Multiple Levels of Local Governance.� In James Wunsch and Dele Olowu (eds.) Local Governance in Africa: The Challenges of Democratic Decentralization. Boulder, CO: Lynne Rienner Publishers, pp. 181–210. 94 Annex 1. District Core Services: A Sectoral Analysis 1. This annex takes a detailed look at each of the five core service delivery areas for which district governments are responsible. Education 2. The delivery of schooling is the responsibility of local governments. Education typically represents between 45 and 50 percent of local governments’ budgets. Local governments are responsible for schooling, which is entirely or almost entirely funded from conditional transfers from the central government. These grants cover teacher salaries, instructional materials, and facilities. Primary schooling dominates enrolments and education budgets. Primary students account for about 87 percent of school enrolments. A policy of universal secondary schooling has been in place since 2006. 3. Primary education is the largest, most common and most consistent schooling program; the largest share of conditional transfers is intended to finance the salaries of primary teachers. To assess the trends in service delivery in the districts, the funding of primary education is examined. In particular, attention is focused on the recurrent conditional transfers for primary teacher salaries and for primary instructional materials (and other non-salary inputs, such as co-curricular activities and school administration). Those two items account for about 78 percent of the total educational budget across the districts, with the first category being much larger than the second. This funding is examined in real per capita terms. 65 4. Over the past decade, the funding of primary schooling in the districts has not kept pace with the population/student growth and inflation. Despite a temporary improvement, real per capita funding for primary education was about 10 percent less during 2009-12 than it was during 2003-06. Funding for teacher salaries has declined, resulting in less competitive salaries and/or relatively fewer teachers. Non-salary support for primary schooling in the districts has, despite an increase in the capitation grant, declined dramatically. 5. During the period 2003–2012, the per capita salary funding for primary teachers did not increase but rather declined by about 6 percent. The districts’ budgets for primary teacher salaries are shown in Figure A1-1 in real per capita terms. While some fluctuation has occurred over the years, the real per capita amount during the past three years (2009/10 to 2011/12) is somewhat lower than it was between 2003/04 and 2005/06. Just how that funding translated to salaries per teacher in the districts is not available here, but this data suggests that, if teacher numbers are proportionate (or increasing) relative to the population, salaries are not improving. 6. Meanwhile, although the comparable data are imperfect, labor incomes elsewhere appear to have improved. From 2003/04 to 2011/12, real GDP per person increased by 32 percent as household 65 It would be logical to consider the number of teachers and the number of students as the relevant bases, but figures for those operating in the district (non-municipal) schools are not easily accessible. Rather, the per capita amounts are used, as the district local government populations have been determined and can be applied as a base in other assessments. For the population size not to be misleading, the number of students per capita must be stable over the period analyzed. That appears to be the case. The share of the population between the age of five and nine was 16.6 percent in 2002 and 16.8 percent in 2011. Also, nationally, the portion of the population enrolled in primary schools seems quite constant. On the other hand, education statistics indicate that the gross enrolment ratio for primary education is increasing. (The gross enrolment ratio for primary schools is the number of students enrolled relative to the primary school age group. Rates exceeding 100 percent indicate somewhat older children attending primary school.) This suggests that the per capita measures used here may understate the per-primary- student measure. These qualifications need to be kept in mind when examining the funding indicators. 95 consumption rose by 23 percent. Thus, since funding per capita for primary teacher salaries has remained relatively unchanged, the number of primary teachers relative to the population could only be maintained if teacher salaries were constant in real terms. However, if teacher salaries were constant in real terms, it is likely that they have declined relative to salaries in other employment sectors. Moreover, if that is the case, teaching as a profession could be perceived as less attractive and teaching less competitive as an occupation. Thus, even with constant real per capita budgeted transfers for teacher salaries, the ability to place qualified teachers in schools may have deteriorated since relatively fewer can be hired (if salaries are keeping pace with other employment) or because (if salaries are not keeping pace) teaching might hold less appeal than other occupations. The change in primary teacher salary transfers relative to GDP and relative to average household consumption is shown in Figure A1-2. Over nine years, the relative amount has decreased about 25 and 20 percent respectively, suggesting that the teacher services purchasing power of the salary transfer has diminished. Figure A1- 1. Real Per Capita Primary Salary Budgets of Districts (2011/12 Prices) 25000 2011/12 Shillings Per Capita 20000 15000 10000 5000 0 Source: BOOST database. 96 Figure A1- 2. Per Capita Primary Teacher Salary Transfer Relative to Per Capita GDP and Average Household Consumption (2002/03 = 1.0) 1.2 1 0.8 0.6 0.4 0.2 0 2003/04 2004/05 2005/06 2006/07 2007/08 2008/09 2009/10 2010/11 2011/12 GDP per capita Average household consumption Source: BOOST database and World Bank staff calculations. 7. The second, and much smaller, source of recurrent funding specific to primary schooling is a transfer intended mostly for instructional materials. The BOOST database designates this as a conditional transfer to primary education and it consists (entirely or mostly) of a per-student “capitation� grant that has remained at U Sh 6,000 for many years and was increased to U Sh 7,000 in 2011. As a result of inflation, the value of this source of funding has decreased dramatically. The trend in the per capita real value of that non-salary transfer to the districts is depicted in Figure A1-3. The per capita amount of U Sh 1,557 shillings in 2011/12 is 46 percent of the constant price level in 2003/04. Even the aggregate nominal amount budgeted for the districts did not exceed the 2003/04 amount until 2011/12. Figure A1- 3. Real Per Capita Non-Salary/Capitation Transfers for Primary Education (2011/12 Prices) 4000 3500 2011/12 Shillings Per Capita 3000 2500 2000 1500 1000 500 0 Source: BOOST database. 97 8. Total per capita funding for education in the districts has been declining. Transfers for teacher salaries and the general, largely “capitation� grant reflect recurrent funding for primary schooling. The plots of the real per capita total and the components appear in Figure A1-4 below. Total funding per capita in the districts has declined since 2003/04 (and especially since 2006/07). Even the drop since 2003/04, from U Sh 20,701 to U Sh 17,631 (2011/12 shillings) represents a decline of approximately 15 percent. About 60 percent of the reduction resulted from the lower non-salary funding, which fell from U Sh 3,398 to U Sh 1,557. The smaller relative decline in the per capita salary transfer may mask deterioration in the capacity to provide primary teachers. By way of comparison, consider the real per capita primary school funding in the 13 municipal councils that existed over the full time period. Across comparable years, per capita funding for teacher salaries in those councils was 17.5 percent greater than in the districts. However, the non-salary transfer was 11.6 percent less. The combined total for the municipal councils was 14.2 percent larger per capita than the total for the districts. Figure A1- 4. Real Per Capita Transfers for Primary School Teacher Salaries and for Non- Salary Recurrent and Total (2011/12 Prices) 30000 25000 2011/12 Shillings Per Capita 20000 15000 Total 10000 Salary 5000 General 0 Source: BOOST database. 9. Uganda has made progress in reducing the student-teacher ratio, but the national averages mask large district and regional variations. Figure A1-5 shows the average student-teacher ratios across the four regions and nationally. 66 The figure also shows the ratios for the districts in the region with the lowest and with the largest ratios (the minimum and maximum bars). There are significant differences from one region to the next. In the central region, the student-teacher ratio averaged 42.3, but the averages ranged between 57.7 and 68.2 in the other three regions. Even within the central region, the ratios across districts ranged from 27 (in Kampala) to 61. The greatest difference between the minimum and maximum figures is in the northern region, where the minimum is 44 and the maximum 105. Circumstances and conditions differ greatly across regions and districts, and major difficulties—including in providing schooling and other services—continue to exist in many. Still, these data indicate important 66 For primary schools, the Statistical Abstracts report a decline in the national average from 50 students per teacher to 49 between 2005 and 2010. Other sources indicate success over a longer period. 98 demands for services as well as major challenges in providing the resources required and in allocating sufficient resources to meet educational objectives. Figure A1- 5. Student-Teacher Ratios in Primary Schools by Region, 2010 120 100 80 60 40 20 0 minimum average maximum Source: UBOS. 10. The adoption of universal secondary education increased the demands on district schools without providing an adequate increase in funding. In 2006, Uganda adopted a policy of Universal Secondary Education. That change resulted in an expansion of the number of students in secondary education. As shown in Figure A1-6, secondary enrolment increased from about 27 per 1,000 children to over 38 per 1,000—an increase of over 40 percent. It is safe to assume that secondary enrolments increased relative to population size in the districts at roughly the same ratio as they did nationally 67 However, per capita real funding for secondary schooling in the districts has not increased correspondingly. In fact, real per capita transfers were the same or less per person from 2008/09 to 2010/11 than they were prior to the policy change. Although the 2011/12 funding package indicates a substantial improvement, it does not correspond to the enrollment increase (about 25 percent funding increase versus 40 percent rise in enrolment). Despite the increase in demand on schools due to expanded secondary education, the total funding to schools for primary and secondary purposes is less in real per capita terms in 2011/12 (even after the sharp increase in secondary funding that year) than it was in 2003/04, at U Sh 25,340 (2011/12 shillings) per person versus U Sh 23,836. 67 Information was not obtained (and may not be readily obtainable) about the public-private split in secondary education in the districts (i.e., as defined here, excluding the municipalities) but district numbers dominate the national data. Private schooling is more important in secondary education and the reliance on public secondary schools may be greater in the rural areas. 99 Figure A1- 6. Real Per Capita Transfers (Salary and Non-Salary) for Secondary Education and Number of Secondary Students per 1,000 Population 7000 45 Secondary Students Per 1,000 Population 6000 40 35 2011/12 Shillings Per Capita 5000 30 4000 25 3000 20 15 2000 10 1000 5 0 0 Per capita secondary transfers Secondary students/1000 population Source: BOOST database. Primary Healthcare 11. Healthcare programs and policy in Uganda have evolved considerably since the turn of the century, partly as a result of disappointments with the re-establishment of the healthcare system in the 1990s. Then, and still today, Uganda’s health sector relies heavily upon donor services and inputs. The lack of coordination among health agencies led to the adoption of a (more) sector-wide approach (SWAp) reflected in the first Health Sector Strategic Plan (HSSP I, 2001/02–2004/05). Improved coordination and a focus on improved access to primary care (realized in part by the abolition of user fees in public health facilities) were targets. HSSP II (2005/06–2009/10) and the Health Sector Strategy and Investment Plan (HHSIP, 2010/11–2014/15) followed, producing improvements. Still, the coordinated funding going to the health sector under the Medium Term Expenditure Framework (MTEF) incorporates less than half of donor support. 68 The record of the publicly managed allocation to the health sector since 2000 outlines some important characteristics. Donor funding has provided about 40 percent of the funds and the government about 60 percent. The government contribution has remained relatively stable at nine percent of the central government’s expenditure. The per capita US dollar funding gradually rose, from about US$6 to US$11 until 2009/10, but dropped to US$9.40 in 2010/11 with a sharp decline in donor support. 69 68 According to the Ministry of Health Statistical Abstract 2010, of national health spending, donors contribute 35.6 percent and the government 14.4 percent. Households fund 50 percent. 69 See Annual Health Sector Performance Report 2010/11, p. 26. 100 12. Health delivery is quite decentralized. Local governments receive about 52 percent of the health sector budget. There are national and regional referral hospitals, general hospitals, and three levels of health centers (IV, III and II). The national and regional referral hospitals are managed and funded directly by the Ministry of Health. Local governments are responsible for the general hospitals and the health centers. Central government grants fund 98 percent of the local governments’ health budgets and health represents about 15 percent of local government outlays. Approximately 80 percent of the local government funding for health goes directly to primary healthcare, and over 90 percent of that is for recurrent purposes, as opposed to development. 13. Primary healthcare dominates the local governments’ health budgets. That feature applies equally to the rural districts. A perspective on the funding and services to the districts is provided by an examination of the real per capita funding of primary healthcare at that level. Recurrent expenditures for wages and non-wage inputs are major and constitute the initial focus of attention. Transfers for primary care development are modest but have changed dramatically. Comparisons are made with municipal funding. 14. Funding to the districts for the wages and salaries of primary healthcare workers has remained quite stable since 2006/07. The level and trend in real per capita funding is presented in Figure 28 in 2011/12 shillings. The average amount from 2006/07 to 2011/12 is approximately U Sh 4,600. As for teachers, stable real per capita levels need not ensure maintenance of relative salaries and competitiveness in the job market. Figure A1-7 also shows the same transfers to the 13 municipal governments. The per capita amounts indicated there are somewhat greater, with an average of U Sh 6,700 shillings over the full period, but with considerably more year-to-year variation. The differences have narrowed. Over the most recent three years of data, the difference amounts to about U Sh 700. Figure A1- 7. Real Per Capita Transfers for PHC Salaries: Districts and Municipalities (2011/12 Prices) 12000 10000 2011.12 Shillings Per Capita 8000 6000 4000 2000 0 Districts 13 Municipal Councils Source: BOOST database. 15. Transfers to support non-wage recurrent health services demonstrate a declining trend but a change in practices is reflected in the most recent reductions. Since 2004/05, the patterns for the districts and the municipalities are more consistent than for wages, as the two lines parallel each other. The municipalities again experienced a premium that averaged 18 percent (Figure A1-8). The trend in 101 funding is downward. Non-wage transfers to the districts fell almost 30 percent between 2003/04 and 2008/09. The dramatic decline following that period was due to dissatisfaction with the management of health supplies and medicines by the districts and hospitals, which resulted in a shift of funding to the National Medical Stores (NMS), now charged with allocating and supervising the provision of these items to the districts and hospitals. Funds formally directed to PHC non-wage grants for the purchase of health supplies and medicines were henceforth to be directed to the NMS, so that there would be no reduction in the ability to supply these items or the related services at the district or hospital levels. 70 The transition was complicated by a reduction in donor assistance in 2009/10. Figure A1- 8. Real Per Capita Transfers for Non-Wage PHC: Districts and 13 Municipalities (2011/12 Prices) 2500 2000 2011/12 Shillings Per Capita 1500 1000 500 0 Districts 13 Municipal Councils Source: BOOST database. 16. In contrast to PHC recurrent and especially non-wage transfers, PHC development transfers have increased substantially since 2007/08. Averaging U Sh 400 (2011/12 shillings) per person from 2005/06 to 2007/08, they ranged between U Sh 1,465 and U Sh 1,850 after 2009/10 (as seen in Figure A1-9). No mention of the reason for this change has been noted. 70 See the 2009/10 Health Sector Performance Report for discussion. 102 Figure A1- 9. Real Per Capita PHC Development Transfers to Districts (2011/12 Prices) 2000 1800 1600 1400 2011/12 Shillings Per Capita 1200 1000 800 600 400 200 0 Source: BOOST database. 17. To the contrary, transfers for current expenditures are declining, raising questions about a possible imbalance between the development of and current expenditures in this sector. In the districts, prior to the change in funding for health supplies and medicines, wage and non-wage transfers had declined in real per capita terms by about 10 percent between 2004/05 and 2008/09. Despite the notable changes in non-wage and development PHC transfers, total funding for PHC to the districts has actually remained relatively stable during the period reviewed. In particular, recent reductions in the non- wage grants were approximately “offset� by increases in the development funding. Assuming that NMS allocations of medicines and health supplies at least maintain the per-person quantities under the previous PHC non-wage funding, real resources to the districts’ PHC sectors may have increased. The latest surge in development transfers does, however, imply a shift towards development expenditures. 18. The expansion of PHC units is improving access to care. Under HSSP I, the percentage of households living within five kilometers of a health facility rose from 49 percent to 72 percent. Although the number of government level II health centers increased by 50 percent—from 1,055 to 1,578 between 2004 and 2010—the percentage of those living within five kilometers has not changed. 71 In the context of distance, the gains in access were realized quickly and early. Even with this expansion, government centers account for just over 50 percent of level II facilities, but that is up from 46 percent. Though all types increased in number, the private for-profit facilities now make up a smaller share. One might pick and choose indicators, but the rates of child mortality and maternal mortality are indeed declining. Since 2005, immunization rates seem to have held steady compared to most other indicators. Funding remains a challenge. A large proportion of positions are vacant and significant sectors of the population are inadequately served. Such problems constrain the implementation of plans and the achievement of objectives. In regard to the districts, some interesting initiatives to improve resource utilization and service are to be noted, such as the transfer of allocation and supervision of health supplies and medicines to the NMS. Also, the Ministry has introduced comparative performance assessments of districts in its league tables. Although caution is advised in the implementation of their underlying methodology, in the interpretation of results, and in structuring policy conclusions on the basis of these assessments, 72 this effort may prove valuable. 71 Ministry of Health Statistical Abstract, 2010. 72 For discussion, see M. McMillan and W. Chan 2006. 103 District Roads 19. Roads in Uganda consist of national roads and district roads. The national roads come under the management of the Uganda National Road Authority (UNRA). As of July 2010, the country contained about 21,000 km of national roads, 73 approximately 16 percent of which are paved. District roads are more accurately classified as District, Urban, and Community Access Roads (DUCAR). While about 53,000 km of DUCAR roads exist, it is important to distinguish among the three classes subsumed in this category. There are 18,500 km of district roads; these provide access to the major trading areas within districts and are the responsibility of the district local governments. Community access roads—estimated at 30,000 km—link the smaller communities (e.g., villages) and provide access to the district roads; . these, too, fall under the districts’ scope, and specifically under the jurisdiction of the sub-county governments. Urban roads, of which Uganda counts 4,500 km, are those within the municipal and town councils. 20. It is the rural roads that are managed by the districts and their sub-counties that are of prime interest here—hat is, the district roads and the community access roads. Again, funding is employed as the indicator. Two elements are important in the assessment of road funding: the maintenance of existing roads and the construction of new roads (and major rehabilitation of old roads). Unfortunately, data on investment in new construction and on rehabilitation has been difficult to acquire for any length of time. Hence, the focus here is on funding for road maintenance. In particular, information from the BOOST data is reported on a per capita basis when it comes to conditional grants from the central government to the district governments for the maintenance of district and community access roads. 21. The road development budget represents a relatively small portion of district budgets. Although allocations for road construction and rehabilitation (development funding) are not separately identified in the BOOST database, some sense of their magnitude is possible. The per capita constant shilling total development budgets of the districts were reported earlier. That information is available up to 2011/2012. The average level of that funding over the nine years in question is U Sh 12,403 (2011/12 shillings) per person. Development budgets are used for many purposes, of which road maintenance and development is only one. The portion of the budget devoted to this specific activity could be estimated by subtracting road maintenance from the total budgets for works and transport. Road maintenance averages about 85 percent of that total budget (or 15 percent goes to road development or to non-road works or both). 74 If that difference represents road development, it would amount to about 2.7 percent of the total development budget in the most recent years. This is a small portion but is not inconsistent with figures from other sources. In its analysis of funding released to local governments for 2010/11 (the only year for which development funds are separately identified), the LGFC reports funds for road rehabilitation accounting for 6.8 percent of the districts’ development expenditures (but reports nothing for routine maintenance). This perspective suggests road development funding on the order of U Sh 310 to U Sh 830 (2011/12 shillings) per person in recent years. 22. Transfers for the maintenance of district roads have varied considerably but have also grown significantly in recent years. Figure A1-10 reports funding per person in constant 2011/12 shillings. While the level of funding declined by almost one quarter in real per capita terms from 2003/04 to 2006/07, it rose substantially over the following three years. These transfers reached U Sh 2,300 shillings per capita in 2009/10, a 53 percent increase over the 2003/04 level. Unfortunately, the latest levels are unknown because 2009/10 is the last year for which this series is currently available. 73 See the 2009/10 Annual Report of the Uganda Road Fund. 74 Note that the percentage going to road maintenance has increased from 83 percent pre-2007/08 to 88 percent post- 2006/07. 104 Regardless, this trend demonstrates a substantial improvement in the funding for district road maintenance after 2006/07 and, presumably, represents an improvement in the conditions of rural roads. Figure A1- 10. Real Per Capita Transfers to Districts for Road Maintenance (2011/12 Prices) 2500 2000 2011/12 Shillings Per Capita 1500 1000 500 0 2003/04 2004/05 2005/06 2006/07 2007/08 2008/09 2009/10 Source: BOOST database. 23. Funding of the roads sector appears to be generally in line with its needs. Although the per capita transfers are small, Uganda has many people per kilometer of road; for example, it averages about 1,526 people per kilometer of district road. However, costs are also substantial. Using mid-point estimates of the annual cost of routine maintenance of unpaved district and community access roads (US$650 and US$350 per kilometer, respectively), 75 the cost in Uganda shillings of maintaining those roads can be estimated at about U Sh 53 billion. In 2009/10, road maintenance transfers to the districts amounted to 50.9 billion. The funding provided appears almost sufficient to provide routine maintenance but left little room for periodic maintenance or road rehabilitation (costing US$5,000 and US$18,000 per kilometer, respectively). Also, if road development/refurbishing budgets were U Sh 800 shillings per person (which is at the upper end of the range noted in the previous paragraph), that would fund the refurbishing of 555 km, or about three percent of district (but not community access) roads. Thus, even at recently improved levels of road funding for the districts, it appears that the districts may have sufficient funding to carry out routine maintenance but have limited scope for major repairs, let alone expanding their road systems. However, between 2006/07 and 2008/09, real funding per kilometer of district road almost doubled, so the ability to maintain districts roads improved greatly as a result in a very short period of time. 24. The recent increase in real per capita road funding for the districts reflects a shift in Uganda’s public expenditure priorities towards infrastructure investment. In 2008/09, the Government of Uganda launched a decade-long plan for investment in DUCAR roads. 76 The consequences of that program appear to explain or at least contribute to the increased real funding in road maintenance budgets after 2006/07 (and, probably, contribute to the increase in development budgets in recent years). Those increases for the districts were noted above. The funding for the municipal councils also improved. From 2006/07 to 2009/10, per capita constant shilling road maintenance budgets in the 75 Estimates come from Table 1.1 of the Works and Transport Sector Performance Report 2010/2011. 76 The case for enhanced infrastructure was demonstrated in earlier World Bank studies. See Uganda Public Expenditure Review: Strengthening the Impact of the Roads Budget, The World Bank, 2010. Improving rural roads, largely to promote agricultural development, is one of the resulting initiatives. 105 municipal councils increased more than 3.5 times, from 2,366 to 8,669 shillings. Also, across all local governments, the total development budgets increased by one-third in real per capita terms after 2007/08, and a larger portion of local budgets are being devoted to development expenditures. 25. The change to the road transportation sector was not limited to road financing. A new Uganda National Roads Authority (UNRA) became operational in July 2008 and a Uganda Road Fund (URF) went into effect as of January 2010. The URF is to finance routine and periodic maintenance of all public roads in Uganda, relying primarily on funds from user charges such as fuel taxes. The UNRA is to maintain and develop the national (but not the DUCAR) road system. The Ministry of Works and Transport is assigned a policy, planning, and regulatory role. As of July 2009, a significant portion of the districts’ roads (approximately 900 km) was reassigned to the national road system and placed under the authority of UNRA. Estimates of the assignment of Uganda’s roads in 2008 and in 2010 are reported in Table A1-1. Reports of the length of district roads transferred to the UNRA, or at least the distance of roads serviced by the districts, differ (from 6,500 to 10,000 km), but the transfer was substantial. This shift reduced the quantity of roads for which the districts (or their sub-counties) are responsible. Thus, funding for road maintenance increased in real terms and the road distance that districts were to maintain decreased. While this combination of changes suggests a substantial gain for the districts, the net benefit may be more modest than it initially appears. Table A-1 1. Distribution of Uganda's Road Network (km) 2008 2010 UNRA 10,800 21,000 DUCAR Districts 27,500 18,500 CAR (sub-county) 30,000 30,000 Municipal Councils 4,300 4,500 Sources: The 2008 figures are from the UNAR website and the 2010 figures from the URF 2009/10 Annual Report. Both are as of July. 26. The changes in funding and assignment affected a significant (but not disproportionate) increase in real funding for the maintenance of district roads. Although real per capita transfers for road maintenance to the districts were essentially constant from 2008/09 to 2009/10, the funding per kilometer of district road rose by approximately 50 percent because of the reduced road distance for which districts were responsible. 77 Despite that sharp increase in funding, it is not possible to say that the districts realized a disproportionate windfall because road funding was improving throughout this period. By contrast, real per capita funding to the municipal councils increased 80 percent in 2009/10 with no (or only a marginal) increase in the length of municipal roads (see Figure A1-11). Had the 9,000 km (or thereabouts) of road remained a district responsibility, transfers to the districts may have increased by an additional amount to fund the higher level of maintenance required by the 2009/10 arrangements. 77 Note that if community access roads are also included in the calculation, the increase in funding per kilometer is a more modest 23 percent. 106 Figure A1- 11. Real Per Capita Transfers for Road Maintenance: Districts and 13 Municipalities (2011/12 Prices) 10000 9000 8000 2011/12 Shillings Per Capita 7000 6000 5000 4000 3000 2000 1000 0 2003/04 2004/05 2005/06 2006/07 2007/08 2008/09 2009/10 Districts 13 municipalities Source: BOOST database. 27. To summarize, a shift in emphasis toward infrastructure development supported significant real increases in the road maintenance budgets for DUCAR roads after 2006/07 (at least until 2009/10, which is the latest available data). After a period of declining resources, transfers to the districts for the maintenance of the roads for which the districts and their sub-counties are responsible approximately doubled in real terms between 2006/07 and 2008/09. That increase, plus the reassignment to the UNRA of approximately 9,000 km of district roads (about one-third of the roads for which districts were responsible, or one-fifth of those for which the districts and sub-counties were responsible) seems to have resulted in maintenance grants sufficient for the districts (including sub-counties) to provide routine maintenance to the roads currently under their charge. That represents a substantial improvement in funding and service capability. However, those funds still do not appear adequate to provide the major repair and rehabilitation necessary for long-term maintenance of the district road system. 78 Water Provision in Rural Areas 28. Water provision in the rural areas of Uganda falls under two categories: that provided through the district local governments and that provided through the small towns, with rather different arrangements for each. Water (and sanitation) services for the “large towns� (Kampala and 22 major urban centers) is provided by the National Water and Sewerage Corporation (NWSC). 79 The “small towns� are the urban centers outside the NWSC and include, in addition to other town and municipal councils, smaller urban communities under town boards and “rural growth centers� of 2,000 persons or 78 It can also be noted that dissatisfaction with the results of contracting road services in combination with external assistance in the form of road maintenance and construction equipment led the central government to initiate a program in 2012 to allocate equipment to districts, allowing them to undertake more road work themselves. 79 Service of the NWSC can extend beyond the jurisdictional boundaries of the 23 municipalities. 107 more. The areas outside the NWSC and the “small towns� come under the jurisdiction of the district local governments. 80, 81 29. The district local governments are responsible for the development and (indirectly) the operation and maintenance of water supplies in their rural areas. Through the Ministry of Water and Environment, the districts are allocated a District Water and Sanitation Development Conditional Grant (DWSDCG). This transfer represents by far the most substantial funding to the districts for water services, and it is the current focus of attention here. These funds are used by the districts to develop new water supplies by utilizing springs, shallow wells, boreholes, gravity flow systems, and rainwater harvesting and storage. Responsibility for the operation and maintenance of those systems is assigned to community/facility-based Water and Sanitation Committees (also called Water User Committees). This arrangement reflects the Community-Based Management Systems (CBMS) policy to promote community responsibility for the operation and maintenance (and, to some extent, development) of their water systems. Responsibilities include the collection of fees to cover operating and maintenance expenses (and possibly some contribution towards capital costs). 82 The Ministry also directly engages in developing some water supplies in rural areas, in part on an emergency basis. In 2010/11, 123 new boreholes were provided centrally. That compares with 2,863 new water supplies provided by the districts under DWSDCG. 83 30. The District Water and Sanitation Development Conditional Grants (DWSDCG) represent about 96 percent of transfers to the district governments for water services; as such, they are the prime determinant of the districts’ ability to extend and enhance water services to their citizens. The trend in per capita 2011/12 shilling funding of that source is depicted in Figure A1-12. The funding has increased and decreased but, until 2011/12 when prices increased more than usual, it remained relatively stable in that the budgeted amounts in 2003/04 and in 2010/11 were only marginally different— U Sh 2,270 versus U Sh 2,355 (2011/12 shillings) per capita. The U Sh 1,800 per-person level in 2011/12 is the notable deviation from that pattern. 31. More important for water service delivery is the capacity of the development funding to meet the need for new services. The Ministry reports the cost per newly serviced person experienced by the districts through their development expenditures. Those costs have been rising in real terms. It is informative to look at the number of additional persons in the rural areas provided with water services under the DWSDCG (see Figure 34). The number of additional people served by the districts’ new water developments shows a declining trend, particularly from 2002/03 to 2005/06. That trend need not indicate a potential problem if the need is being matched. However, the expansions do not appear to keep up with population growth (Figure A1-13 also shows the estimated increases in district populations). 84 80 Donors (development partners) play and important role in supporting water supply and sanitation, both through funding contributions to the Ministry of Water and Environment’s budget and through directly supporting “off- budget� programs in the communities. 81 The annual sector reports from the Ministry of Water and Environment provide substantial information on the water and related sectors. This section draws particularly from the 2011 Water and Environment Sector Performance Report. 82 The districts and the Ministry also support the self-supply of water. Funding for self-supply accounts for 8.4 percent. 83 The Ministry provides important support to the rural sector in the supply of water for production; that is, largely to support livestock operations and provide irrigation. 84 The population numbers include persons in the districts that would be served by the “small town� programs, so it overestimates the numbers to be served through the districts’ development grants. However, in 2010/11, 80,145 people were added through the small town services, compared to 559,136 in the districts and an estimated population change outside the municipal councils of approximately one million people. 108 Figure A1- 12. Real Per Capita Transfers to Districts for Rural Water (2011/12 Prices) 3000 2011/12 Shillings Per Capita 2500 2000 1500 1000 500 0 Source: BOOST database. Figure A1- 13. Additional Persons Supplied Water via District Development Grants and Estimated Population Increases 1200000 1000000 Number of People 800000 600000 400000 200000 0 Additional persons served Estimated population change Source: Ministry of Water and Environment. 32. It is possible that other sources compensate for any deficiency in the additions emerging from the DWSDCG transfers. The Ministry’s investments provided another 77,964 people in rural areas with service in 2010/11 and donors in off-budget activities (typically in partnerships with NGOs) extended water service to 410,885 in rural and urban areas. A total expansion of water services to more than one million additional people largely addresses the gap suggested here but implies a growing reliance on non-district and particularly donor sources. More analysis is required to clarify the situation and the trends. 109 33. The problem of keeping up with new services does not yet appear in the Ministry indicators. The percentage of persons in rural areas considered to have access to an improved water source has been stable at 65 percent from 2008/09 to 2010/11, although it is predicted to decline to 64 percent in 2011/12. 85 The percentage did increase from 61 percent in 2004/05 before stagnating. 86 Although accessibility seems to have stalled, the index of equity of service has improved. Since initially reported in 2007/08, the mean sub-county deviation from the national average in the number of persons per improved water point has been halved. The functionality of the rural systems has remained constant during the years examined here. Thus, water service in the rural areas does not appear to have deteriorated, although with rising costs, stable annual additions to the number of persons provided access, and a growing population, maintaining accessibility levels and reaching future objectives (e.g., 77 percent in 2014/15) will be difficult. 34. Cost data offer insight into issues with the expanding number of districts. The cost per additional person provided access is reported for both total cost and “hardware� cost. Hardware costs are the direct costs of building a facility (construction costs), while total costs incorporate all other costs, including the administrative components. From 2004/05 to 2010/11, the total costs per additional person served increased 60 percent but the hardware costs increased only 30 percent in real terms. A relationship between facility or hardware costs and total costs and the number of districts is suggested in Figure A1- 14. The number of districts increased from 56 to 80 between 2004/05 and 2006/07 and then to 111 in 2010/11. 87 The creation of new districts involves the creation and operation of new district water offices. Those new administration units involve some resources that will leave a smaller share of funding for facility construction, at least initially. Looking at the percentage represented by hardware/construction costs with respect to annual total cost, the share declined as the number of districts increased to 69 and then to 80. However, the percentage began to recover, and by 2009/10 it was back at the 2004/05 level (indeed, at 72 percent, it even exceeded the earlier 66 percent 88). The addition of another 31 districts in 2010/11 resulted in another downturn in the percentage of costs going to new facilities (or an increase in other/administrative costs). This pattern suggests that the costs associated with new districts may not be a permanent detriment but, rather that adjusting to and absorbing these changes only takes some time. Of course, the question is whether a similar pattern will emerge in response to the 2010 increase. 35. Water supply for the small towns is organized differently. The Ministry of Water and Environment is responsible for development of the systems, and the urban communities for their operation and maintenance. Development is guided through four regional Water and Sanitation Development Facility (WSDF) organizations (north, east, central and south-west) that emerged from the successful South-Western Towns Water and Sanitation Project. The WSDFs develop strategies, technologies, and standards appropriate for specific regions. Development costs per additional person served are higher than in the districts. Although they averaged US$40 in 2010/11 (or U Sh 93,000 2011/12 prices), they have ranged from US$40 to US$96 over the eight years ending in 2010/11. The projected or target cost is US$75 per person. Costs will vary dramatically depending upon the particular technologies utilized and the specific conditions at the location. In 2010/11, across developments, costs ranged from US$19 to US$72 per person. The US$40 average level in 2010/11 implies a total outlay of U Sh 7.5 billion and US$75 per person would require U Sh 14 billion. The Ministry budget for water and sanitation in small towns was about 28 billion. 85 See 2011 Sector Performance Report, page 2. Access is considered as being within one kilometer of an improved source. 86 Accessibility rose steadily from 21 percent in 1991 to reach 60 percent in 2002. 87 The Kampala District, in the NWSC, is omitted. 88 The water and sanitation guidelines to districts set the minimum share that has to be spent on the hardware at 75 percent. 110 Figure A1- 14. Construction Cost Percentage and Number of Districts 120 100 Cost percentage/Number of Districts 80 60 40 20 0 Hardward costs as a percentage of total costs Number of districts Source: UBOS and World Bank staff calculations. 36. The operation and maintenance of the small town systems is, as with districts, a local responsibility. The urban councils are responsible for the local systems and normally discharge that through a Water Supply and Sanitation Board or similar body. The local authorities, especially the 105 with piped water systems, typically contract with private operators to operate and maintain their systems. About 10 percent are operated by the public water authority itself. Smaller-scale operations often rely on contracts with private individuals (rather than firms) to serve as operators of facilities. The Ministry monitors and assists in operations and maintenance. In addition, five regional membership associations of WSSBs provide operations and maintenance services too costly for the individual members to provide. 89 37. Small grants to assist the operations and maintenance of the small town systems are channeled through the districts. Those transfers amounted to U Sh 1.67 billion in 2010/11. These conditional transfers have been declining in relative terms. As shown in Figure A1-15, the amounts of the per capita transfers in 2011/12 shillings have been declining (almost) consistently since 2003/04. The per- person amount in 2011/12 is about 40 percent of that in 2003/04. The per capita amounts shown in the figure seriously understate the true amounts available per person in the small towns. 90 However, the downward trend in funding to 40 percent of the 2003/04 level is still valid. In fact, due to relative population changes, it may be understated. 91 38. Data on the performance of the small town water systems is incomplete but, in certain instances, encouraging. Accessibility, measured by the portion of the urban population within 0.2 kilometers of an improved water source, has increased from 51 percent in 2005/06 to 65 percent in 2010/11. Based on an incomplete number of responses, 71 percent of the towns were considered to have actively functioning water and sewerage boards. However, the functionality of the respondents’ systems 89 The number of urban centers falling under the “small town� designation is unclear. For example, there is reference to 163 urban councils outside the NWSC and to 183 water suppliers to small towns and rural growth centers served by umbrella organizations. Similarly, the numbers over time are uncertain. 90 Lacking reliable numbers for the small town populations, the district populations (excluding people in municipal councils) are used. The small town population is, of course, much smaller than the district populations. 91 The urban population is growing faster than the population generally. This is due in part to migration but also to the expansion of the recognized urban councils. The small town populations covered by the urban water transfers to the districts is expected to grow at least as fast as the district populations, so the pattern depicted is still relevant. 111 (percentage of required hours water was supplied) is reported as 91 percent, up from 82 percent in 2006/07. 39. Although important, conditional funding specifically identified for sanitation is small compared to that designated for water services. For that reason, in part, sanitation is overlooked in this analysis. Sanitation goes hand-in-hand with water services. Improved sanitation focuses on access to latrines and hand washing. Access to both latrines and hand-washing facilities is improving in both rural and urban areas. Both are central to good health and have benefits for healthcare services. Figure A1- 15. Real Per Capita Transfers to Districts for Urban Water Services (2011/12 prices) 140 2011/12 Shillings Per Capita 120 100 80 60 40 20 0 2003/04 2004/05 2005/06 2006/07 2007/08 2008/09 2009/10 2010/11 2011/12 Source: BOOST database. 40. The funding of rural districts (and perhaps small town) water systems has remained relatively good in comparison to that of the Ministry of Water and Environment. The Water and Environment budget allocation has been approximately constant in nominal terms. The 2011/12 budget of U Sh 242 billion is almost the same as the 2004/05 budget of U Sh 236 billion. Those amounts imply a 45 percent reduction in real value. The Ministry’s share of the national budget declined from 7.5 percent to 2.6 percent over that period. Similarly, the donor contribution to on-budget programs declined from about two-thirds to less than one-third, thereby placing greater demands on the central government’s resources. The fact that the transfers to the districts for water development have been relatively stable not only in real terms but in real per capita terms indicates some success and suggests the priority assigned to that program. 41. Donors contribute substantially to water and sanitation outside the government’s budgeting process. Of the total sector (on- and off-budget) funding, about 27 percent was off-budget during 2010/11. With donor on-budget support amounting to about 30 percent of the publicly managed funds, donor support contributed about half of the total funding to the water and environment sector. A considerable portion of donor funding goes toward the extension and improvement of water and sanitation services. In terms of people newly served by improved water sources, it is estimated that 410,885 were added in 2010/11 thanks to donor programs, mostly in partnership with non-governmental organizations. With 637,100 added from district and Ministry developments, the donor share represents almost 40 percent of the total expansion. Substantial investments have also been made in sanitation facilities, much of which were slated for schools. Donor investments have been much reduced since the financial crisis in 2008, however. Umbrella networks contribute to coordination among donors and between the donors, local governments, and the Ministry. 112 Agriculture 42. Agriculture is a large and underperforming sector of Uganda’s economy. Most of the population (about 88 percent) lives in rural areas, three-quarters of households are engaged in agriculture in some way and two-thirds of households are directly involved in farming, but often at a subsistence level. Agriculture accounts for less than one-quarter of GDP. Although poverty rates are declining, productivity and incomes remain low. The growth of agricultural output lags behind that of the rest of the economy and even that of the agricultural-based population. Improvements in incomes, poverty reduction and economic growth depend substantially upon improved performance of the agricultural sector. Agriculture is recognized as a key sector92 and the place accorded to the Ministry of Agriculture, Animal Industry and Fisheries (MAAIF) in the national budget is scheduled to expand. Strengthening agricultural extension is part of the strategy 93 and figures prominently in the Ministry’s program. The delivery of much of agricultural extension, however, is the responsibility of the district and sub-district local governments. Initiatives expanding agricultural extension programs already preceded the latest plans and efforts. 43. Agriculture-related conditional transfers to the districts increased substantially since 2003/04. Figure A1-16 shows those increases in real (2011/12 shillings) per capita terms. From 2003/04 to 2008/09, those grants increased from U Sh 1,763 per person to U Sh 5,997 shillings per person. Grants per person continued to increase until 2010/11 at a slower pace before falling back somewhat, and over the last four years the funding averaged U Sh 5,371 shillings. This change represents a threefold increase and moved such transfers from 2.9 percent of districts’ budgets to 9.8 percent. Figure A1- 16. Real Per Capita Transfers for Agriculture (2011 Prices) 7000 2011/12 Shillings Per Capita 6000 5000 4000 3000 2000 1000 0 Source: BOOST database. 92 See, for example, National Development Plan, 2010; Agricultural Sector Development and Investment Plan: 2010/11-2014/15; and World Bank (June 2011). 93 As already noted, road improvements are also an important component of the program to strengthen the agricultural sector. 113 44. The dramatic increase in agricultural transfers to districts stemmed from the growth of a new program—National Agricultural Advisor Services (NAADS)—that began in 2001/02. NAADS was introduced in 2001 as a joint long-term initiative by the government of Uganda and its development partners to improve and expand agricultural extension services. The primary aim was to make extension services more responsive to the agriculture community by making them demand-driven, farmer-led, and targeted at subsistence farmers. NAADS is part of the Plan to Modernize Agriculture (PMA) and, in the larger framework, an element of the Poverty Eradication Agenda. It is a semi-autonomous agency with its own board and secretariat operating under the supervision of the MAAIF. NAADS initiated operations in 2001/02 with six participating districts involving 24 sub-counties but quickly expanded to cover 79 (of 80) districts and 710 sub-counties by 2007/08. Its expenditures increased accordingly. The delivery model is designed such that the district and sub-counties are responsible for implementing and supervising NAADS programs in their jurisdictions but grassroots farmer groups at the parish and village levels establish local needs and contract with private providers for the delivery of extension services. Those groups are supplemented by Farmer Fora at the sub-county and district levels. These players interact and coordinate with the agricultural production staff and other relevant offices at the district and sub-county levels. The initial agreement had the districts and sub-counties each contributing five percent of the costs of the NAADS programs in their areas. The Government of Uganda contributed eight percent and donors 80 percent. The farmers were to contribute two percent. As indicated in the table and figure below, even by 2003/04, NAADS accounted for essentially half of the agricultural transfers and more recently it represents about 90 percent (Table A1-2). The growth of NAADS not only augmented MAAIF funding to the districts but it probably constrained, if not replaced, other MAAIF programs (especially the pre- existing extension activities). 94 Table A-1 2. Composition of District Agricultural Transfers 2003/ 2004/ 2005/ 2006/ 2007/ 2008/ 2009/ 2010/ 2011/ 04 05 06 07 08 09 10 11 12 NAADS 47.1 58.0 73.9 77.9 81.7 83.5 94.1 91.6 89.4 Agriculture Extension 29.0 23.3 16.3 14.5 12.6 7.2 0.0 0.0 2.1 PMA NSCG 5.6 4.0 Production and Marketing 6.8 6.9 PAF monitoring 23.9 18.7 9.8 7.6 5.6 3.7 1.9 1.6 1.5 Notes: PMA NSCG refers to the Plan to Modernize Agriculture - Non-Sectoral Conditional Grant and PAF refers to the Poverty Action Fund. 45. After its initial phase, NAADS was redesigned to improve its efficiency. Phase I of NAADS, 2001/02 to 2007/08, got mixed reviews. 95 While noting problems, the World Bank (June 2011) pointed to 94 The sum of the five conditional grant programs reported (which exclude small development grants for agriculture) closely approximate the funding reported for the agricultural sector. 95 See, for example, Annual Report of the Auditor General for 2007/08 and 2010/11; Benin et al. 2007; EPRC 2011; ITAD 2008; World Bank June 2011; 114 the high economic returns identified in the NAADS impact evaluation. Some others have seen smaller benefits and have been less charitable in their assessments. One of the deficiencies was the failure of many farmers and some district and sub-county governments to co-fund NAADS activities as scheduled. Phase II of NAADS was subsumed (along with the National Agricultural Research Organization) under the newly established (as of 2011) Agricultural Technology and Agribusiness Advisory Services agency, which seeks to overcome the difficulties NAADS experienced during the subsequent five-year period. Funding for Phase II was increased by the central government, shifting its own share to approximately 75 percent and the development partners’ share to 25 percent. The districts and sub-counties are now expected to contribute 6 percent each and farmers 3 percent of the funds distributed. 115 Annex 2. Estimation of Fiscal Impacts of New District Creation 1. This annex presents analysis of the fiscal impacts of new district creation based on data supplied by Ugandan authorities. The Local Government Finance Commission draft report (LGFC 2012) presented information on the estimated cost of delivering services in an average district. “Administration� was one of the eight categories of services identified. As defined for that purpose, administration represented approximately 10.5 percent of total costs. Those estimates also appear in the local government set-up review (MoLG 2012). The Ministry of Local Government report also presents the three models of standard administrative staff establishments recommended for local governments. The information from those reports is used to examine the consequences of a division of a district into two entities on both the administrative and total costs of district services. 2. Some assumptions are required to generate the example. First, it is assumed that the new district is smaller than the mother district and that the population is divided such that one-third is located in the new district and two-thirds remain in the mother district. This post-division ratio is typical of actual events. Second, the (recommended) cost figures for 2007/08 from the LGFC report are used and it is assumed that the original district was twice the average size. So, after the change, the new district is two- thirds the average size and the mother district is one-third larger than average. Third, some decisions must be made as to the administrative structure left in place after the change. Three alternatives are considered. The results are presented in Table A2-1. 3. The pre-change situation is outlined in the first column of figures in Table 9. The district’s total costs are U Sh 69,715 million (2007/08 figures), administrative costs are U Sh 7,303 million (10.46 percent), and non-administrative costs amount to U Sh 62,412 million. In Case A, it is assumed that the non-administrative costs do not change and are simply split between the two districts in proportion to population. That is, it is assumed that the schools, healthcare facilities, water systems, roads, etc. in each are funded just as before. The differences in the three cases are in what happens to administrative costs. In Case A, it is assumed that, regardless of the size reduction in the mother district, its administrative organization does not change while, in the new district, a standard administrative structure is introduced. It is assumed that the Model 2 staff establishment (a total of 185 persons recommended, MoLG 2012) is put in place in the new district. These assumptions imply that the administrative costs in the mother district are U Sh 7,303 million and those in the new district are U Sh 3,651.5 million. Summing across the two districts, administrative costs now total U Sh 10,954.5 million, non-administrative costs remain unchanged at U Sh 62,412 million, and total costs reach U Sh 73,366.5 million. Thus, administrative and total costs have increased by U Sh 3,651.5 million for an increase in total cost of 5.24 percent. This simple “adding on� of a new administrative unit seems to be the assumption typically adopted when considering the cost of new districts. This approach (whether or not it is followed in practice) seems inappropriate. Even so, whether the increase in cost of 5.24 percent is large, small, or a bargain depends upon the benefits, if any, that result. 4. A logical alternative in redistricting involves some adjustment to the administrative unit of the mother district. The loss of one-third of a district’s population (and area, likely) strongly suggests that a smaller administrative organization is needed to provide the same services and service quality. If there is no reduction in the mother district, then administrative services per person and per service should improve with the larger per-person and per-service administrative resources and costs. If so, the pros and cons of such a situation would require evaluation. Here, it is assumed that administrative service quality is constant and is achieved by the recommended establishments in the districts being considered. Case B.1 in Table 9 outlines the consequences of a possible adjustment in the administrative costs of the mother district. It is assumed that the administrative complement in the mother district is that of Model 3 (with 116 218 staff) and the administrative cost is then U Sh 4,303 million rather than U Sh 7,303 million. 96 No change is made to the new district from Case A, so the total administrative cost for the two districts is U Sh 7,954.5 or U Sh 651.5 million greater than when there was only one district. That implies an increase in total district government costs of 0.95 percent (from 10.46 to 11.41 percent). 5. The new district could well be, and typically is, somewhat smaller than the average district, and this has implications for staffing. In that case, an administrative establishment of Model 1 (having 147 staff) is taken as the administrative organization for the new district. That has a cost of U Sh 2,901.5 million, as outlined in case B.2. Again, assuming the mother district’s staff has been reduced, as in case B.1, the total administrative cost across the two districts is U Sh 7,204.5 million. This amount is less than U Sh 7,303 million in the pre-change situation, so redistricting implies a small savings in total costs (0.11 percent). Table A2- 1. Costs and Changes in Costs of Districts with and without Division: Three Cases Pre-Change Post-Change Districts Overall District Child/New Sum of Mother Change Due (consolidated District Mother and District (2/3) to Division district) (1/3) Child Case A. Only Add Administration Cost (No Change and Model 2 Staff Establishment) Total Cost 69,715 49,119 24,247.5 73,366.5 3,651.5 Administration Cost 7,303 7,303 36,51.5 10,954.5 3,651.5 Non-Administration Costs 62,412 41,816 20,596 62,412 0 Administration Cost as 5.24% a Percentage of Total 10.46% 14.87% 15.06% 14.93% (to 14.93%) Cost Case B.1. Adjust Administrative Cost (Models 3 and 2 Staff Establishments) Total Cost 69,715 46,119 24,247.5 70,366.5 651.5 Administration Cost 7,303 4,303 3,651.5 79,54.5 651.5 Non-Administration Costs 62,412 41,816 20,596 62,412 0 Administration Cost as 0.95% a Percentage of Total 10.46% 9.33% 15.06% 11.30% (to 11.41%) Cost Case B.2 Adjust Administrative Cost (Models 3 and 1 Staff Establishments) Total Cost 69,715 46,119 23,497.5 69,616.5 -98.5 Administration Cost 7,303 4,303 2,901.5 7,204.5 -98.5 Non-Administration 41,816 20,596 62,412 0 Costs 62,412 Administration Cost as -0.11% a Percentage of Total 10.46% 9.33% 12.35% 10.35% (to 10.35%) Cost Sources: LGFC (2012), MoLG (2012) in millions of shillings, 2006/07. 96 Model 2, at a cost of U Sh 3,651.5 million, is taken as suitable for an average district. The cost for Model 3, with 218 rather than 185 staff, is scaled up proportionately. 117 Annex 3. Determinants of Value-for-Money Econometric Exercise 1. This annex describes data sources, the construction of the efficiency measures and predicted savings, and the results of the econometric analysis referred to in Section 5. Data 2. The analysis in Section 5 draws on four main groupings of data. 3. Data on outcomes and district government performance are drawn from statistical abstracts issued by the Uganda Bureau of Statistics, the Ministry of Education and Sports, and the Ministry of Health respectively, as well as from the Uganda National Service Delivery Survey. The primary measures used in the analysis of 5.2 and 5.3 are data contained in the District League Tables created by the Ministry of Education and Sports and the Ministry of Health. 4. Data on spending levels are drawn from the World Bank’s BOOST database. The analysis focuses on data at the district level aggregating recurrent, capital, and development-budgeted spending. The results do not vary significantly if analysis is restricted to recurrent spending. 5. Data on governance are drawn from Volume 3 of the Office of the Auditor General’s Annual Report. Outcomes from the Ministry of Local Government Annual Assessment of Minimum Conditions and Performance Measures for Local Governments, and the Inspectorate of Government’s Report to Parliament were considered, but not used in the final analysis. 6. Data on control variables and other potential determinants are drawn from the National Service Delivery Survey, the Uganda National Panel Survey, and the United Nations Development Program Uganda Human Development Report 2007. 7. One data source needs further explanation. The Ministry of Education and Sports District League table composes rankings based on the average ranking across the three sub-components described in 5.1. Section 5 instead rescales each of the three sub-components using the following general formula: Actual value - Minimum Value Score = u 100 Maximum Value - Minimum Value 8. The variables are scaled in this fashion so that the scale is comparable to the Ministry of Health League Table. The score for the Education sector is then calculated as a simple average of the individual sub-component scores. The results are robust to other scaling and aggregation techniques. 9. In 5.1, the most up-to-date data (2011) are used. In 5.2 and 5.3, to match the data availability from different sources, data from 2010 are used. Where available, data from early years were used for robustness checks. For most sources, data were available for at least 75 districts. One notable exception was the UNDP Report, which for some variables, in particular measures of poverty, was restricted to 55 districts. Outliers were removed by eliminating data points more than ten times the interquartile range greater than or less than the mean. 118 Efficiency Measures, District Groupings, and Predicted Savings Section 5.2 constructs two measures of efficiency. 10. The first is the ratio of outcome measure to the relevant sector spending, calculated as follows: Sector Outcome Measure Outcome per dollar of spending per capita Sector Spending Per Capita 11. The second is an implied efficiency measure that takes into account social conditions and income. It is calculated by regressing the outcome measure on spending per capita and social conditions, as follows: Outcomei D  E Spendingi  J Conditionsi 12. The results of this regression are then used to calculate the predicted level of an outcome measure for a district, given that district’s spending level and social conditions. The efficiency measure is then the ratio of the actual outcome to this predicted outcome. 13. The groupings used in Tables 1-4 are calculated using the following procedure: x Districts are ranked using the two value-for-money measures discussed above. x Districts that are in the top and bottom quartile for both measures are immediately considered. In addition, districts that are in the top and bottom quartile for one measure are inspected. x For the high-performing group, districts that do not perform above average in both measures are eliminated. x For the poorly performing group, districts that perform above average in either measure are eliminated. 14. While all efforts were made to ensure the robustness of these results, as with all relative ranking procedures of this nature the results should be taken as indicative only. 5.2 also calculates expected savings from improvements in efficiency. The approach taken is to consider the effect of shifting all districts up to the average ratio of outcomes-to-spending for the best performing districts. In this case, predicted spending for each district is calculated as follows: Outcome i Spending i Outcome - to - Spending Ratio for Efficient Districts 15. Savings are then the aggregate of the difference between actual spending and predicted spending for each district, excluding the districts in the efficient group. Econometric Results 16. Section 5.3 analyzes the determinants of value-for-money. The results are supported by a linear regression approach considering the relationship between the outcome-to-spending ratio and a set of 119 determinants. There are some natural concerns with this approach, which are further discussed after the results, but this approach was used to keep analysis relatively clear and transparent.97 17. The primary specification used is: Efficiencyi D  E Spendingi  J Governancei  ¦Q Controli  H i Results are reported in Table A3-1 (standard errors are robust). The efficiency measure used in Table A3.1 is the outcome-to-spending ratio; spending is sector-budgeted spending per capita (in 1,000s of U Sh); governance is a dummy that is 0 if the district received an unqualified audit in the Auditor General’s Annual Report and 1 if not; and income is a GDP Index for districts that ranges from 0 to 1, calculated by the UNDP in its 2007 Report. Table A3- 1 Econometric Results Education Efficiency Health Efficiency Spending -0.0777*** -0.0646*** -0.0658*** -0.836*** -0.848*** -0.725*** (0.0215) (0.0159) (0.0180) (0.215) (0.222) (0.245) Governance -0.390*** -0.355** -1.014* -0.948 (0.146) (0.144) (0.605) (0.579) Income Index 3.138* (1.738) Population 4.82e-06* (2.72e-06) Observations 79 74 71 67 64 64 R-squared 0.281 0.337 0.393 0.621 0.620 0.646 18. In the interest of space, we omit variables that are not statistically significant. Variables that were used in our analysis, but omitted here, include measures of staffing levels and absenteeism from the NSDS and UNPS, and measures of the capital stock, including the average distance for citizens to various public facilities. Note should be made that the absence of statistical significance for these variables does not imply that they do not play a role in district government value-for-money. It only suggests that there is insufficient evidence in the data to prove a significant effect. 97 More sophisticated approaches also face econometric difficulties, even with larger data samples. For further details, see for example Balaguer-Coll et al. 2007. 120 19. One concern in a regression with an outcome-to-spending ratio as a dependent variable and spending as an explanatory variable is that the negative relationship is a mechanical effect of the dependent variable being a ratio. To check this, a separate quadratic Tobit regression exercise was conducted with outcomes as the dependent variable and spending and spending squared as explanatory variables. This exercise supports the hypothesis of decreasing returns to spending, but is not reported in the interests of space, and because of instability in other variables in that setting. 98 Care should be taken, however, not to lean too heavily on the magnitudes expressed here, given this potential concern. 20. Robustness checks included specifications featuring interaction and polynomial terms, varying the weights and standardization of the outcome measures, and using different definitions of district government spending. In these robustness checks, the statistical significance of the relationships reported above varied, but the sign was consistent in most specifications. 98 Regressions using the implied efficiency measure as a dependent variable were also conducted. However, this approach has a different set of econometric problems, and the outcome-to-spending ratio approach was preferred in the interests of clarity and simplicity. 121 Annex 4. Value-for-Money Challenges faced by Uganda’s Local Governments: Lessons from the second LG PEFA Evaluation The Local Government Public Financial Management (LG-PFM) systems in Uganda were first evaluated using the PFM Performance Measurement Framework (PMF) developed by the Public Expenditure and Financial Accountability (PEFA) secretariat in 2005. The PEFA-PMF is an integrated monitoring framework that enables the measurement of country PFM performance over time according to a set of high-level indicators covering the areas of fiscal and debt management, budget formulation, budget execution, internal controls procurement, revenue collection systems, inter- governmental fiscal relations, accounting and reporting, auditing, transparency, and external scrutiny. A second Local Government PEFA evaluation was undertaken in FY 2013; eight districts and 10 municipalities were selected for evaluation. The criteria for choosing the entities in the sample selection are elaborated below. While the majority of local governments to be assessed were at the district level (eight), municipal councils were also included (two). Table A4-1 provides additional details on the sample. The key results of the Second LG PEFA evaluation are summarized below. Table A4-1. Sample Local Governments for the Second LG PEFA No. Local Government Criteria for inclusion in the sample Region Others in the methodology 1 Tororo District East PEFA 2005 LG 2 Mpigi District Central PEFA 2005 LG 3 Kasese District West PEFA 2005 LG 4 Kabale District West PEFA 2005 LG 5 Lira MC North PEFA 2005 LG 6 Gulu MC North Urban Council/Accrual Accounting 7 Buikwe District Central New District 8 Hoima District West Auditor General Report – Qualified with exceptions 9 Pader District North Auditor General Report – Qualifier Disclaimer 10 Jinja District East IFMS LG and big size Summary of Value-for-Money Challenges Identified by the LG PEFA (2013) Budget Credibility: Variance of aggregate expenditure outturn compared to the original approved budget among the 10 LGs is unequal, but variance affects budget credibility for most entities. Variance in the composition of expenditure outturn compared to the original approved budget is also unequal among entities. It is high for four LGs, which indicates a low performance; fairly high (average performance) for another four entities; and very low (very good performance) for just one LG. Aggregate revenue outturn to original approved budget differs significantly for virtually all the entities. This indicates a very low performance in the forecast of internal revenue, an element that affects the preparation of a credible budget. The stock and the monitoring of expenditure payment arrears for most of the entities are overall high and/or ineffective, respectively; this status further affects the credibility of the budget. Comprehensiveness and Transparency: Budget classification meets international standards for practically all of the 10 LGs assessed. This is undoubtedly an element that strongly supports comprehensiveness and transparency in Public Finance Management (PFM). On balance, the evaluation for the 10 entities indicates that the budget is comprehensive and budget information is accessible to the public. However, fiscal risk oversight remains very weak for most of the entities. 122 Policy-Based Budgeting: Overall, the 10 LGs under review have adopted a multi-year perspective in fiscal planning and budgeting. However, multi-year budget estimates are not always consistent with budget ceilings. In addition, for costed sector strategies of districts/MCs, links between investments and forward recurrent costs are weak. Predictability and Control in Budget Execution: Government entities, taxpayer obligations, and liabilities are fairly transparent, and taxpayers have access to some information on tax liabilities and administrative procedures. However, existing measures for taxpayer registration and tax assessment do not work well because controls in the taxpayer registration system are very weak and penalties are ineffective. Moreover, the mode of tax payment collection is not effective. Arrears are not registered in most LGs and they are not collected in any district/MC. The availability of funds for committing expenditures is not predictable in any of the 10 LGs. In practice, districts/MCs do not undertake cash-flow planning and/or it is of very poor quality, with limited reliable information provided to budget entities with respect to actual resources available for commitments. The management of cash balances is satisfactory in only about half of the evaluated entities. Procurement systems operate adequately for all the 10 LGs. The legal and regulatory framework is transparent, fairly comprehensive and based on competition. Due to the isolation of some districts/MCs, the public does not always have access to complete, reliable and timely procurement information. It should be pointed out that there is no independent administrative procurement complaints system. Controls for non-salary expenditure are ineffective and/or weak for about half of the entities. In these cases, risks are not assessed and the controls do not manage the risks. In addition, existing rules are generally not respected. Accounting, Recording, and Reporting: Overall accounts reconciliation for accounts under district/MC control is regular and timely, although the quality of the exercise is better in some entities than in others. In-year budget reports are well elaborated in 90 percent of cases. Annual financial statements are of fairly good quality and timely in 90 percent of cases. In fact, statements are, for the most part, comprehensive and are submitted for external audit on time. In addition, annual financial statements are presented in a consistent format with some disclosure of accounting standards, though they do not comply with IPSAS. External Scrutiny and Audit: External audits by the Office of the Auditor General are undertaken promptly and the timeliness of submission of the audit reports to the legislature remains good for all 10 entities. The scope of audit is good and audit standards are met. However, there is little or no follow-up on recommendations that are presented in the audit reports. Legislative scrutiny of the budget law is satisfactory for most (eight) LGs. In fact, the scope of legislative scrutiny by the entity’s Council is good, and the legislative procedures are well established. Moreover, the length of legislative scrutiny is adequate. However, there are significant delays in the scrutiny of audit reports by the legislature and in the timely reporting of their recommended actions for most entities (eight). Overall, this function is considered to be weak. In general, arrangements for the scrutiny of public finances for the 10 LGs under review by the Supreme Audit Institution and by the Council are adequate. Follow-up by the Executive is not operational, however. Foreign aid: Only six entities receive donor aid. Donors involved do not typically provide information on aid and aid disbursements prior to the entity’s budget preparation, nor do donors generally rely on the national system. 123 Fiscal transfers from higher-level government (HLG): The central government respected its commitments on transfers to the LGs in terms of the amounts (aggregate transfers) it has programmed to transfer (2008/09–2009/11). This is the case for a majority (six) of the entities. However, the central government did not respect its commitments on transfers to the LGs with respect to the amounts of earmarked grants. Regarding the in-year timeliness of transfers from HLG (compliance with timetables for in-year distribution of disbursements agreed within one month of the start of the fiscal year), and based on the information available, the assessment of this dimension demonstrates that transfers did not arrive on time (or early enough) to allow efficient service delivery by the districts. 124 Annex 5. Assessment of a Proposal for the Creation of Regional Governments in Uganda I. Background 1. In November, 2012, the Ministry of Local Government (MoLG) published a document titled “Synthesis Report: Reviews of Local Government Set-Up and Local Government Structures.� The chief thrust of this document, which is drawn from two background studies undertaken by consultants commissioned by MoLG, is to propose the formation of a system of regional governments (RGs) in Uganda as an intermediate layer between the central government and local (district) government system. This Annex provides a brief assessment of the proposals in that document. II. Regional Governments A. Proposal 2. In February 2006, the Uganda Constitution was amended to introduce a provision that “Two or more districts may cooperate to form a regional government….� (178 (1)). The Amendment also established a basic procedure through which an RG can be established, and identified five regions where, subject to 178 (1), RGs shall have been deemed to have been established. The Fifth Schedule to the Constitution provides further detail on the composition, functions, etc. of the RGs. 3. The “Synthesis Report� responds to these constitutional provisions by providing fairly detailed proposals regarding the basic nature and character of RGs in the context of the Regional Bill, which was drafted subsequent to the 2006 constitutional amendment. The core elements of these proposals are summarized below. Overall, were the system to be implemented, it would constitute a fundamental revision of the sub-national governance and service delivery system in Uganda. x RGs will form a fully constituted governmental layer of 16 regional units, with an elected Assembly, other basic elements of a political superstructure (Chairman, Cabinet, etc.), and an administrative and bureaucratic establishment, including a CEO and staff, for the execution of their powers and functions; x The number of district governments will expand from the current 111 to 153, with all current counties being reconstituted as districts; x The RGs will be the highest political authority within the regions and shall have political, legislative, administrative, and cultural functions. In terms of broad functional allocations, with some important exceptions, 99 districts will largely retain their functional responsibilities, such as primary education, 100 primary health, water supply, etc., except that RGs would be given key powers in the execution of these functions, including the recruitment, posting, and management of most district staff; x RGs will have an oversight and supervisory role regarding the districts and will intermediate all relationships between district governments and the central government; x District organizational structures will remain largely intact, except in those areas where a current sub-function (e.g. in Health, the management of level III and IV clinics) is to be transferred from the district to the regional level. The RG structure will largely mimic the district structure and, at the administrative level, is organized along similar functional and operational lines; 99 Health is the most important of these, where the document suggests that responsibility for level III and IV clinics, as well as district hospitals, be transferred from districts to RGs. 100 This includes those areas, such as secondary education, where the districts currently have de jure but not de facto responsibility. The detail of these proposals may be found on pages 21-27 of the report. 125 x The CEO and Heads of Department of the fegions will be recruited by the PSC. The PSC will also continue to appoint the CAO and Deputy CAO of all of the Districts. All other RG and district staff appointments will be made by a Regional Services Commission to be formed by each Region and appointed by the PSC. 101 DSCs will be abolished; x The funding of RGs is left open-ended. There is some broad attempt to calculate the aggregate cost of (part of) the structure, and some suggestions are made concerning the sorts of fiscal mechanisms that could be introduced to fund it, but no further fiscal details are provided. B. Observations 4. Development and technical rationale. The� Synthesis Report� contains no statement or analysis of the institutional, organizational, functional/service delivery or fiscal problems that the RG proposal is intended to address, and there is no justification for the structure on developmental or technical grounds. In other words, it is not clear whether, or how, the structure is likely to make local service delivery and governance in Uganda more effective, efficient, and equitable. 5. Arguably, because the proposal responds to a constitutional provision, there is no need for such a justification: if the Constitution envisages RGs, they should be introduced regardless of their impact or cost. Two points are of relevance here: x On the face of it, it appears that section 178 of the Constitution is an enabling rather than a mandating clause. 102 At (1) the section states that “two or more districts may co-operate to form a regional government� (emphasis added). There is no statement to the effect that any districts are obliged to do so. If RGs are in fact optional, it is reasonable to expect that they should only be formed if they are likely to make a positive contribution to the delivery of the sort of service and governance that are delivered at the local level in Uganda (and which are also, broadly, referenced in the Constitution). However, as the brief assessment provided below indicates, this seems unlikely; x Moreover, if RGs were to be introduced, they should be designed with due regard for their potential impact on local service delivery and governance. While the degree of design freedom in the construction of RGs may be limited by Schedule 5 of the Constitution—which provides for an elected assembly, Ministers, certain broad functional responsibilities, and so on—it appears that there is still considerable scope to develop an RG structure that would have fewer negative impacts on local service delivery than the one proposed in the “Synthesis Document.� For example, to anticipate a comment made below, there is no need for RGs to assume all responsibilities and powers over district staff, a measure that is likely to have serious consequences for the ability of districts to deliver effectively. In the absence of any rationale or explanation to the contrary, the RG structure currently proposed appears to undermine, rather than strengthen, the existing system of decentralized service delivery in Uganda. 6. Fiscal impacts. The Synthesis Document provides various figures for the estimated aggregate costs of subnational administration following the establishment of RGs. The basis for the numbers and the methodologies presented in the report are unclear and it is difficult to establish their veracity. Moreover, because neither Regional nor District government structures are fully detailed in the document, it is not possible to come to any definitive conclusion about what these costs are likely to be. However, some broad preliminary observations can be offered: 101 Discussion with MoLG officials and consultants, 31/02/13. 102 It should be stressed that this part of the note has not been drafted or reviewed by a constitutional legal expert, and thus represents a tentative, informal observation only. Certainty on this would require further review. 126 x The report claims that the aggregate costs of district administration would decline under the overall arrangement being proposed (i.e. increase from 111 to 153 districts). Analysis undertaken by the Bank indicates that every time a new district is created, aggregate staff positions (i.e. the staff requirement for both the new and old districts) increase by around 35 percent, even if both districts shift to Model 1 from a pre-existing Model 3 for the old district. Given that the structures proposed at the district level appear not to differ substantially from what is currently in place, substantial savings (indeed, any savings at all) at the district level seem unlikely, even if the new districts were to be formed at the Model 1 level (or, indeed, at something substantially smaller than Model 1); x One figure the report clearly provides is that “accommodation of the current financing requirements related to setting up the regional tier requires that Government allocates an additional Shs. 390 billion annually to finance this set up.� Thus, all other things aside, the report estimates the annual equivalent of around US$156 million to fund the administrative costs of the regional tier. So far as can be established, this seems to exclude the cost of the political superstructure, 103 which would add tangibly to this figure. 7. In short, introduction of the RGs as proposed is likely to add very substantially to the costs of subnational government in Uganda, which already suffers from a very substantial funding gap (“Review of Local Government Financing,� LGFC, October 2012). In the context of current facts and trends— where only around half the existing district administration salary bill is funded—intergovernmental fiscal transfers to LGs have been declining in real, per capita terms; and, as a result of new district creation, funds have been diverted away from development to cover administration and political establishment costs; this is clearly problematic. Arguably, the key fiscal priority in subnational governance in Uganda is to resolve the current and deep disconnect between the resources that are made available annually and the financing needs of the existing system. The RG proposal in the document appears to aggravate rather than ameliorate this problem. In addition, the introduction of fiscal mechanisms and flows to fund RGs (local own revenue sources etc.) is likely to cannibalize those currently going to districts. 8. Human resources. At the moment, only around 55 percent of current staff positions in district administrations are filled, and shortages are particularly severe at senior levels. In part, this results from the fiscal disconnect outlined above, but it also arises from the aggregate lack of sufficiently qualified personnel prepared to work at the salary levels offered under current Public Service arrangements. This has a debilitating effect on local service delivery and governance as, regardless of whether any given district structure is optimally efficient, vacancies create bottlenecks and the functioning of districts becomes compromised. Because they will draw on the same limited pool of qualified people, the creation of RGs, and the proliferation of additional districts, is likely to exacerbate this problem. 9. Functional powers and arrangements. 104 Broadly speaking, the RG proposal keeps the functional responsibilities of the districts intact. One of the chief exceptions to this is in health, where responsibility for operating level III and IV clinics and district hospitals is transferred to the RGs. In all functional areas, however, the report proposes to vertically dis-integrate the powers that are exercised by the districts in pursuit of delivering these services. For example, while districts will still be responsible for primary education, certain primary health functions, district roads and so on, it abolishes all control that districts have over appointing and managing the staff that are meant to undertake these activities. The proposal also appears to place most planning, investment, and construction activities for local infrastructure in the hands of the RGs, leaving maintenance and operation to the districts, and suggests that in areas such as 103 Confirmed during meeting with MoLG officials and consultants, 31/02/13. 104 For purposes of brevity this note does not focus on the implications of the report for municipalities, which are less dramatic than for those of districts. However, it may be pointed out that the overall impact on municipal governments would be similar in areas such as powers and functions to that on districts. 127 procurement, districts should become subject to the supervision of the regions, with the regions assuming direct control of some—unspecified—procurement activities “of higher magnitude.� 10. Two observations may be made here: x First, with respect to health, the proposal amounts to a recentralization of certain sub-functions away from districts to the RGs. No economic or other justification is provided for this and, on the face of it, there doesn’t appear to be a convincing reason to do this. Moreover, consultations with MoH indicated a preference to retain the current functional/sub-functional allocation, with MoH providing greater support to the district to play their current health facility management roles; x Second, the international experience of vertical dis-integration of powers across governmental within areas of functional responsibility is not good. Once local government entities that are responsible for any given function lose key powers related to the execution of that function, basic accountability is lost and incentives for the effective and efficient delivery of services rapidly weaken. This has long been characteristic of local government systems in countries such as India and Pakistan, and there is a strong consensus in the international literature that it should be avoided. Equally, placing infrastructure planning and development (hence capital investment) decisions in the hands of one level of government, and maintenance and operations (hence recurrent costs) responsibilities in another undermines incentives that the development agency (i.e., RG in this situation) has to worry about long-term operating cost issues in infrastructure design, and makes it very difficult for the operating agency (i.e., district government) to budget coherently for the future. It also tends to undermine local ownership and responsiveness to bottom-up planning. Overall, service delivery sustainability is weakened, and the scope for ongoing conflict between different levels of government is widened. Finally, imposing a procurement approval system whereby one level of government approves the procurement decisions of another (within specified thresholds for existing procurement methods, which is a separate matter) can create rent-seeking opportunities and project execution bottlenecks. Internationally, countries are tending to move away from this sort of system. 11. Transition and capacity. Introducing RGs would entail a fundamental disruption of the current institutional system of subnational governance and service delivery in Uganda. Moreover, it would take a substantial period for RGs to develop the capacity to be able to play their proposed oversight and supervisory roles effectively. Evidence drawn from the experience of new district creation in Uganda suggests that service delivery deteriorates when new districts are created. In this context, it is quite possible that the introduction of RGs as proposed, while unlikely to lead to service delivery improvements in the medium to long term, would nonetheless be accompanied by a deterioration of service delivery in the short to medium term. It may also be observed that clauses 178 2 (a) and (b) of the Constitution create a very onerous procedural threshold for the creation of any single RG. In effect, either one-third of the councilors of one district in a new RG area, or one-third of the sub-county councils in any one district in that area, can veto the creation of an RG. In this context, it is quite possible that actual formation of RGs could become a protracted, difficult and conflicted process. Again, this is likely to negatively impact service delivery for its duration. 128 Annex 6. Assessment of a Proposal for Enhancing Own-source Revenues of the Districts Background 1. Despite its size and substantial responsibilities, funding from its own-revenue sources provides only a small portion of local government revenues in Uganda (about 10 percent). (Levels of the own- revenue share as low as 3 percent are reported, but the data support 9 percent to 11 percent.) The major sources of local own-revenues are the local services tax (LST, an income tax), a hotel tax, property taxes, business licenses, and user fees. Squeezed by diminishing real per capita central government transfers and the loss of the Graduated Tax in 2005, there are pressures to expand local own-source revenues. 2. The LGFC has proposed measures to improve the effectiveness of existing own-source revenues and also to add new sources of local government revenue. If the proposed revenue enhancements were realized, own-source revenues would increase from about one-tenth of local budgets to one-quarter (and perhaps even one-third). 3. There is a large difference in the per capita own revenues between the rural districts and the urban local governments (excluding Kampala). The urban to rural ratio increased from 2.4:1 to 9.3:1 with the increased number of independent towns and municipal councils in 2010. The International Context Local Government Finance in Industrialized Countries 4. Local government is fiscally important in industrialized countries. However, the magnitude of local governments there varies widely. Differences result mostly from the extent of local governments’ responsibilities for social services; that is, usually schooling and sometimes healthcare and social protection. Responsibilities for “core� local services (e.g., roads, water and waste, community amenities) are quite consistent and are a relatively modest and uniform portion of GDP, typically 4 percent to 6 percent. Local governments usually rely upon one major source of taxation, either the property tax or the income tax, but about one-quarter of the countries utilize a mixture of property, income, consumption, and some other taxes. While taxes provide about 42 percent of local revenues on average, non-tax own revenues (e.g., fees, charges, licenses, permits, fines) are important at 22 percent. The remainder comes from transfers that, especially in the property and mixed-tax reliant countries, support social services. Local Government Finance in African and Latin American Countries 5. Local government in developing countries is not well documented but information is presented for Africa and Latin America. Local government is smaller in developing countries than in industrialized countries, both as a share of general government and as a share of GDP. Also, local government in Africa tends to be smaller than in Latin America. There is little comparative information on either expenditure responsibilities or revenue sources. However, some responsibilities for schooling and for healthcare seem common and information from a small sample (including transition countries) suggests that own-source revenues make up about one-half of total revenues, with taxes at 38 percent and fees, etc. at 11 percent. An international review of property taxes reveals that they amount to about 0.53 percent of GDP in developing countries and generate 18.3 percent of local revenue. Overall, although typically fiscally smaller than in industrialized countries, local governments in developing countries have similar responsibilities and revenue sources. 129 Ugandan Local Government: Revenues and Revenue-Enhancement Proposals A Comparative Perspective 6. Local government in Uganda is comparatively large, accounting for about 20 percent of general government expenditures. At that level, it is among the largest in developing countries and comparable in size to local governments in many industrialized countries. Outlays on social programs, predominately for schooling and healthcare, account for 65 percent of local budgets. The local funding of those expenditures, however, is notably small, with the latest data at 11 percent. Furthermore, of that, non-tax own revenue provides 60 percent of own-source revenue and taxes the smaller share at 40 percent. Total own-source revenue amounts to 0.43 percent of GDP, a level well below the 0.7 percent average across African countries. The Revenue Enhancement Proposals: Fiscal Implications 7. The LGFC has proposed measures to increase local governments’ own-source revenues by improving the effectiveness of existing own-source revenues and also by adding new revenue sources. 105 If the increases of the proposed revenue enhancements materialized in full, own-source revenues would become about 2.5 times larger and would increase from about one-tenth of local budgets to one-quarter (and perhaps even one-third). (Some uncertainty as to whether the projected revenues are presented in real or nominal terms results in a range in the magnitudes of the fiscal impacts. Where magnitudes are reported, they are lower-end estimates.) Such an increase would also raise local own-revenues from 0.42 percent to about 0.7 percent of GDP (a level equal to the average among African countries). The Revenue Enhancement Proposals: Observations and Assessments 8. The magnitude of the expansion in self-funding has numerous consequences and implications. Major points to note follow: x Linking the benefits of local services with their costs by means of user charges and local taxes is important because it provides an important means of funding services, it constrains demands to and focuses demand on valued activities, and it enhances the accountability of local representatives and officials to local voter-taxpayers. Contributing to the financing of local services and enhancing the downward accountability of local decision makers and the participation of citizens in local government seems particularly relevant in the Ugandan context. x Even if own-source funding is low by conventional standards and there are legitimate reasons for expanding own-source revenues, increasing taxes (and charges) two to three times presents challenges. In particular, local residents will expect compensating improvements in local services. An implication is that the central government should not expect to reduce transfers upon which local governments now depend and leave the local governments no better off financially. x There are large rural-urban differences that need to be recognized. On average, the rural districts obtain about 6.2 percent of their revenues from own revenues while the municipal councils generate about 30 percent. If the proposed increases in own revenues are distributed across municipal class as now, (even) a 2.5-fold overall increase would imply that the districts would fund about 16 percent of revenues from own sources and the municipal councils about 75 percent. Concerning the municipal councils, two questions arise. Will a 75 percent level of self-funding 105 Some other recommendations of the study included review of the legal framework for the existing local taxes, development of Local Revenue Policy, supporting the formation of Local Revenue Databases in the local governments, and some measures aimed at enhancing collection of local revenues. 130 (and potentially more in some cases) be acceptable to municipal residents? Even if acceptable, would it be desirable when social services are such a large part of local government budgets? x Schooling and healthcare, the main social programs at the local level, represent more than half of local government spending (about 40 percent in the urban localities and 60 percent in the districts). High levels of self-finance can be expected to result in large differences in schooling and health services between urban and rural authorities and among rich and poor local governments (i.e., those with high and those with low own-revenue potential). x National goals and objectives for education and health could call for continued central funding of school and healthcare or, if own-source funding results in inter-authority disparities, an effective equalization program. National interests and national standards might be important considerations also in some other local government areas of responsibility. Many of the anticipated impacts of expanded self-funding depend upon how central grants evolve. x A logical standard for self-funding is that local governments meet from local taxes and charges the cost of “core� services (e.g., roads, water and waste, community amenities). The funding of social services, and any others of national interest, are typically more complicated and conventionally involve inter-governmental transfers of some form or another. x While the (per capita) own-revenue capacities will differ among districts, the overall projected level of self-funding for districts (at 16 percent) is at a level that is unlikely to pose complications such as those in the municipal councils. In fact, such an increase is expected to be an important benefit, and particularly for the districts losing urban communities as the number of independent towns and municipal councils expands (although the actual net fiscal impact to a district of the hiving-off of urban centers in not known). x Little attention has been paid to local finances (either revenues or expenditures) or to the implications of the proposed revenue enhancements beyond the level of aggregate local government. As noted, the overall numbers mask large and important differences among the categories of local governments (not to mention among individual districts, towns and municipal councils). Those consequences and implications deserve closer examination, discussion and consideration. Comments on Specific Revenue Enhancement Proposals 9. The proposed own-revenue enhancements focus on three revenue sources. Those are greatly expanded revenues from the existing Local Services Tax (LST), new taxes on residential properties, and the introduction of user fees for school and health services. Each merits some comment. x Local Services Tax (LST). Enhancement of the LST should involve reduction in the number of exemptions and tax bands, and expansion of the tax base, notably to cover commercial farmers. Further, the documentation indicates that the LST overlaps to a very large degree with the PAYE, the national income tax. That overlap implies that piggybacking the LST on the PAYE could eliminate local administration and compliance costs while generating almost the same revenue. Piggybacking might even encourage local government to use the LST. x Education and Health Fees. New school and health user fees are proposed and are projected to generate one-quarter of the enhanced own-revenues. If earmarked to support those services, the education fees would contribute about 10 percent to school funding and health fees about 50 percent to health expenditures. Both, but notably health fees, would add substantially to current resources. An implication is that differences in own-source revenue potentials could create (or aggravate) inter-jurisdictional disparities in health services, unless they are corrected through appropriate adjustment of transfers from the central government. 131 x Residence Property Taxes. Two new taxes on residential property are proposed: a property service tax and a residence tax. Together, they are expected to raise 18.4 percent of the enhanced own-source revenues. Introducing these two taxes, particularly by encompassing owner-occupied housing, would broaden the tax base and improve tax equity. Through them, more citizens will be exposed to the tax-service linkage and contribute to the costs of local services. Even with this expansion in property taxes, local property taxes would amount to about 0.25 percent of GDP (up from 0.13 percent), which is still well below the 0.52 percent average of developing countries. Conclusion 10. Own-source revenues make a small, or at most a modest, contribution towards the expenditures of local governments in Uganda. Hence, the proposed expansion in the role of own revenues is a positive development. Certainly there is scope for enhanced self-funding. Own revenues represent presently about 0.42 percent of GDP in Uganda. The proposals, if meeting the projected results and proposed schedule, would see them grow to 0.72 percent, which is the average level observed in African countries. However, analysis at the level of local government in aggregate fails to recognize the large dissimilarities that exist among the classes of local government (and even the rural-urban differences). Those differences are important to the impact and success of the proposals and they need to be recognized, analyzed, and discussed. 132