Report No. 49361-UG Uganda A Public Expenditure Review 2008 With a Focus on Affordability of Pay Reform and Health Sector May 30, 2009 Poverty Reduction and Economic Management Unit Africa Region Document of the World Bank ACKNOWLEDGEMENTS This report was prepared by Dino Merotto (Task Team Leader). Chapters 1 and 2 rely heavily on data collected by Paul Mpuga (AFTP2, Kampala), analysis by Charlotte Lundgren, and a fiscal sustainability analysis originally designed by Lars Sondergaard and run for this study by Rachel Sebudde (AFTP2, Kampala). Joyce Owusu-Ayim (Consultant) and Zhicheng Swift (AFTP2) helped collect and validate the public expenditure data which was compiled into a comprehensive multi-year database by Igor Khyfets (Consultant) and analyzed by Dino Merotto. Chapter 3 was edited by Lebohang Lijane (Consultant), based upon the stand-alone report prepared by a team in the World Bank’s Human Development Network headed by Pablo Gottret and comprising Julie McLaughlin, Peter Okwero, Ajay Tandon, and Susan Sparkes with inputs from Hans Hoogeveen (AFTP2). The work on efficiency, leakages and absenteeism in health draws heavily on work by Martina Bjorkman and Jakob Svensson (Consultants). The authors are grateful to Mr. Bwoch (Accountant General) and his staff, and particularly to Henry Daka and Ashe Kawoya Bulumba (Uganda Computer Services). Helpful support with data was provided by Paul Mwanje (Macro). The authors are especially indebted to Laban Mbulamuko (Budget Division) and Joyce Owusu-Ayim for their patient and knowledgeable feedback and for their valuable assistance with data collection and interpretation. The entire team is grateful to Arlette Sourou for support in producing the report and for administering contracts and Trust Funds each step of the way. Table of Contents ACKNOWLEDGEMENTS ................................................................................................... 3 EXECUTIVE SUMMARY .................................................................................................... 1 Chapter 1. ECONOMIC OVERVIEW AND TRENDS IN UGANDA’S BUDGET ........ 4 I. Macro Overview ............................................................................................... 4 II. Trends in Uganda Budgets from 2003/04 to 2006/07 ....................................... 7 III. Budget Variance: Who Overspends, Who Underspends and on What? ........ 19 Chapter 2. COSTING POLICY COMMITMENTS in PUBLIC SERVICE PAY REFORM AND UNIVERSAL SECONDARY EDUCATION...................... 27 I. Wage Reform Initiatives in Uganda ............................................................... 28 II. Civil service wages, staff structure and the rationale for pay reform ............. 30 III. Fiscal effects from selected pay reform scenarios .......................................... 34 IV. Summary ......................................................................................................... 38 Chapter 3. GETTING MORE HEALTH FROM THE BUDGET................................... 42 I. Background and Context................................................................................. 42 II. Health Status ................................................................................................... 44 III. Population and Fertility................................................................................... 46 IV. Assessment of Health Sector Performance: Health Care Utilization, Coverage, and Outputs .................................................................................... 47 V. Total Health Expenditures .............................................................................. 50 VI. Cost Drivers ................................................................................................... 54 VII. Waste and Inefficiency in the Health Sector................................................... 55 VIII. Options for Increasing Fiscal Space for Health in Uganda............................. 58 VIIII. Conclusions and Recommendations ............................................................... 63 ANNEXES Annex 1: Draft PFM Platform 1-Predictability and Stability ............................................ 68 Annex 2.1: Public Service Wage Tables and Wage Regressions ......................................... 71 Annex 2.2: Modeling the Fiscal Effects of Pay Reform ....................................................... 74 List of Figures Figure 1.1: Economic Performance ......................................................................................... 4 Figure 1.2: Composition of GDP Growth ................................................................................. 5 Figure 1.3: Government Revenue ............................................................................................ 6 Figure 1.4: Government Expenditure (as a percentage of GDP) ............................................. 7 Figure 1.5: Shares of Expenditure in Each of the Budgets ...................................................... 9 Figure 1.6: Expenditure in Each of Uganda's Budgets, 2003/04 to 2006/07 ......................... 10 Figure 1.7: The Gross Fixed Capital Formation Shares of the Development Budgets.......... 10 Figure 1.8: Economic Classification of Spending, 2003/04 to 2006/07 ................................ 11 Figure 1.9: Adjusted Economic Classification, 2003/04 to 2006/07 ..................................... 11 Figure 1.10: Economic Classification of the 2006/07 Budget, Excluding Debt and Arrears ................................................................................................................ 12 Figure 1.11: Budgets by Service Level (2003/04 to 2006/07) ............................................... 12 Figure 1.12: District Budgets by Type (2003/04- to 2006/07) .............................................. 13 Figure 1.13: District School Budgets (2003/04 to 2006/07) .................................................. 14 Figure 1.14: District Health Budgets (2003/04 to 2006/07) ................................................... 14 Figure 1.15: Referral Hospitals’ Budgets (2003/04 to 2006/07) ........................................... 15 Figure 1.16: Sector Composition of District Development Budget ....................................... 15 Figure 1.17: Agency Grants are a Significant Share of the Budget ....................................... 16 Figure 1.18: Allocation of Grants to Semi-Autonomous Agencies, 2003/04 to 2006/07...... 17 Figure 1.19: Arrears Payments Relative to Additional Spending on District Services ......... 18 Figure 1.20: The Structure of Arrears Payments ................................................................... 18 Figure 1.21: Under-Budgeting: Comparing the Estimated Budget for Constant Use with the Actual Budget (2003/04 to 2005/06) ............................................................ 19 Figure 1.22: Budget vs. Release Variance by Level .............................................................. 21 Figure 1.23: Budget vs. Release Variance by Budget ........................................................... 21 Figure 1.24: Budget vs. Release Variance by Economic Classification ................................ 22 Figure 1.25: Overspending By Economic Category (cumulative over four years) ............... 22 Figure 1.26: Underspending By Category (cumulative over four years)............................... 23 Figure 1.27: Net Four-Year Overspending by Budget Type ................................................. 23 Figure 1.28: Net Four-Year Overspending by Economic Classification ............................... 24 Figure 1.29: Net Four-Year Overspending by Level ............................................................. 24 Figure 2.1: Shares of Employee Costs in the Budget ............................................................ 29 Figure 2.2: Wage Shares in the Budget ................................................................................. 30 Figure 2.3: Numbers of Staff by Pay Scale: 2003/04-2006/07 .............................................. 32 Figure 2.4: Staff Costs by Pay Scale: 2003/04 to 2006/07 .................................................... 33 Figure 2.5: Uganda Budget Composition -- Baseline ............................................................ 35 Figure 2.6: Scenario 1: Nominal Doubling of Public Service Wages (all Staff) ................... 35 Figure 2.7: Scenario 2: Nominal Doubling of Wages of Managerial, Technical and ........... 36 Figure 2.8: Scenario 3: Selective Pay Increases Combined with USE Savings .................... 37 Figure 2.9: Education’s Share of Total Expenditure under Pay Reform Scenarios............... 38 Figure 3.1: Trends in Under-Five and Infant Mortality Rates for Uganda, 1960 to 2006 ..... 44 Figure 3.2: Health Expenditure as a Proportion of Total Expenditure and Catastrophic Health Expenditure in Uganda, 1996 to 2005 .................................................... 52 Figure 3.3: Official Development Assistance for Health to Uganda, 2000 to 2006 .............. 53 Figure 3.4: Elasticity of Government Health Expenditure with Respect to GDP in Uganda, ............................................................................................................... 59 Figure 3.5: Life Expectancy: Global Comparison of Ugandan Districts, 2006 ..................... 61 Figure 3.6: District Level Input and Output Indicators.......................................................... 63 List of Tables Table 2.1: Median Monthly Nominal Wages (Public and Private Sector), by Occupation (USh ‘000) .......................................................................................................... 31 Table 2.2: Median Monthly Nominal Wages--Public and Private Sector (USh ‘000) .......... 31 Table 2.3: Summary of Simulation Results; Spending on Education, Infrastructure and Wages as Share of GDP under Different Scenarios ........................................... 39 Table 3.1: Health Expenditure and Outcomes--Comparisons across Countries .................... 45 Table 3.2: Health Expenditure Indicators: Uganda & Comparator Countries, 2006 ............. 51 Table 3.3: Budgets at Uganda’s Frontline Public Health Facilities 2003/04 to 2006/07 ...... 51 Table 3.4: Estimating ‘Waste’ from Possible Sources, as Share of Budget Items ................ 56 Table 3.5: Very Approximate ‘Waste’ Estimates for FY 2006/07 ........................................ 56 EXECUTIVE SUMMARY This report is a further response to the call from Uganda’s leaders for ‘value for money’ in public spending. The report is the second in a series of three sector-oriented public expenditure reviews (PER) for the Government of Uganda (GOU) prepared by the World Bank as input into the Government of Uganda’s budget reform initiatives. The focus of this report is efficiency and efficacy of spending by the Government of Uganda with a specific focus on the health sector. The first report in the series was ‘Uganda Fiscal Policy for Growth’, published in 2007, with a focus on the education sector. For this financial year (FY09/10), a Public Expenditure Review with a focus on the roads sector is being prepared. This report comes in two volumes. Volume I is the main report with conclusions and policy recommendations which analyses composition of overall spending and budget implementation, budgetary consequences of ongoing pay reforms, and efficiency and effectiveness of spending in Uganda’s health sector. Volume II contains a full report on the health sector, and includes the background papers prepared. The analysis is designed to help Uganda’s authorities design a program of public expenditure measures to reduce waste, enhance efficiency, and improve service delivery with a special focus on the health sector. Key elements of such a reform program identified here include: ƒ Shift the composition of the budget in favor of service provision. Chapter 1 looks in detail at trends in budget composition and budget implementation in Uganda, outlining the challenges for service delivery revealed within recent budget trends. The analysis of allocations in relation to service delivery suggests that to get better value for money in public services, the Government needs to: a. Reverse underfunding of non-salary budgets of districts. b. Control employee-related costs in central ministries and Kampala. c. Launch a review of expenditures in ministries and agencies that focuses on reducing overheads and increasing efficiency, and as a result generate savings. d. Take a closer look at the end-use of ‘grants to semi-autonomous institutions’. e. Revise the presentation of the Budget: use economic classification to present the budget instead of the current classification of development spending versus recurrent spending. For example, there is a significant increase in re- current spending hidden in the Development Budget, which distorts an economic evaluation of the Budget. f. Renew efforts to improve the predictability of forward-looking sector ceilings in the medium-term expenditure framework (MTEF), including improving the predictability of funding in-year and reducing the reliance on supplementary budgets. 1 ƒ Analyze the affordability of new policy commitments and seek efficiency gains from them. Chapter 2 examines the costs and merits of pay reforms approved by the Cabinet in August 2006. The conclusion is that these pay reforms are unaffordable without further cuts in non-salary funding for public services, and may also be unwarranted given the level of wages available to most workers in the private and voluntary sectors. Given likely pressures for pay raises in advance of the elections, more work is urgently needed to design viable pay reforms. The report’s preliminary findings suggest that pay raises would need to be implemented selectively for certain grades rather than across-the-board, and should be preceded by reform of the Government pension scheme. ƒ Focus on efficiency savings in priority sectors. Like the previous World Bank PER in this series, “Uganda: Fiscal Policy for Growth”, this report also maintains the core conclusion that given reasonable assumptions about the availability of domestic revenues, grants and concessional borrowing, Uganda should seek efficiency gains in delivering public services while continuing to find savings from less productive sectors1. This PER focuses on resources and performance in Uganda’s health sector. Chapter 3 of this first volume summarizes the full report on health contained in Volume II of the PER. The inescapable conclusion of the report is that non-salary funding for public clinics and for district and referral hospitals in Uganda is insufficient for decent quality in the public health service. The analysis, based on global comparisons, suggests that health outcomes in Uganda are not commensurate with the relatively low public funding, relatively high private funding, and very high donor funding the sector receives. Lasting solutions to Uganda’s health sector predicament require a re-think of service management and the roles of the public and private sectors in delivering health services. This report concludes that improving the efficiency of Uganda’s public health spending is crucial for getting better results and identifies reform measures for improving health spending efficiency as a means to scaling up health service coverage, quality and impact. Service management priorities for the health sector would include: a. Reducing public health worker absenteeism and improving human resources management in health. a. Reducing drug leakages and waste in clinics and stores. b. Getting better value for money from health sector donors by improving the integration of off-budget donor funds into the planning and resource framework of the public health sector. c. Investigating the causes of the wide differences in performance across district health teams. 1 Relative to many countries, Uganda’s budget has quite a high share of productive spending. Including donor-funded projects, about 60 percent of Uganda’s non-statutory public expenditure in 2008/09 set out in the MTEF goes towards roads, health, education, agriculture, energy, and water. 2 d. Reducing service overheads and channeling more funds to non-salary items in clinics and hospitals. ƒ Prioritize sector efficiency reforms, target sub-sectors offering the most scope for savings. Chapter 2 concludes that the most significant budget policy issue in the Government’s efficiency drive would be to generate efficiency savings through the gradual roll-out of universal secondary education (USE). Failure to phase the implementation of the USE and tackle the biggest causes of inefficiency in secondary education (e.g. relating to teacher deployment) would most likely compromise the quality of the USE program2 and use up all the room in the budget for additional spending on infrastructure, other public services, and other policy priorities under the forthcoming National Development Plan. Reducing worker absenteeism in education, and especially health, is a pre-requisite for better efficiency and better service delivery in these important sectors. However, the gains on offer from health reforms are dwarfed by those on offer in education, particularly through the USE. This report does not deal directly with the technical content and sequencing of budget reforms or public expenditure reporting. The former has been discussed in the context of work on the Joint Budget Support Framework. That work is ongoing and should complement this report. This report also does not deal in detail with expenditure reporting, data integrity, and data gaps, although a lot of work on expenditure reporting went into and came out of the detailed work on the budget and expenditure database compiled with Uganda’s Ministry of Finance, Planning and Economic Development (MOFPED) and discussed in Chapter 1. The Public Expenditure and Financial Assessment (PEFA) will discuss reporting issues and accounting systems in depth, including at local government level. The next and last report in this series will analyze the mechanism through which resources are allocated to rural areas for road maintenance and road construction, and analyze the value for money of the resources allocated to the national road network. 2 Low quality secondary education is not in high demand in global markets. If faced with a serious trade-off between quality and universal access in education, the Government of Uganda would need to consider this issue carefully. 3 CHAPTER 1. ECONOMIC OVERVIEW AND TRENDS IN UGANDA’S BUDGET I. MACRO OVERVIEW 1.1 This chapter starts with a macroeconomic overview, followed by a series of observations on the composition and implementation of Uganda’s budgets from 2003/04 to 2006/07. The analysis of trends uses the a comprehensive annual data set of Uganda’s budget, which was compiled by the task team under this study and has been shared in full with the staff of Ministry of Finance, Planning and Economic Development (MOFPED). GDP Growth3 1.2 Real economic growth averaged 7.8 percent between 2000 and 2008. The economy remained robust through several exogenous shocks. A stable macroeconomic environment, strong export growth, high foreign direct investment, and increased private investment contributed to this growth.4 Figure 1.1: Economic Performance 20000.0 12 18000.0 GDP 10 (Billion 16000.0 Shillings) 14000.0 8 Growth 12000.0 10000.0 6 8000.0 4 6000.0 4000.0 2 2000.0 0.0 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 Real GDP Growth Rate Source: Uganda Bureau of Statistics and World Bank staff 1.3 Strong performance of the service and industry sectors has characterized recent GDP growth. Construction, in particular, which constitutes about 50 percent of industry, has grown at an average of 15 percent since 2000. Growth in construction slowed down in the last two years but remains strong at about 7 percent. Figure 1.2 shows the sector shares in GDP and real growth rates. In 2008, the service sector grew by 13 percent, with the financial services sector contributing the highest growth. These 3 This chapter uses the newly re-based National Accounts, Uganda Bureau of Statistics (UBOS), 2008. 4 Despite the impressive performance, growth has been constrained by several factors, especially inadequate infrastructure. 4 re-based GDP accounts figures suggest that the agriculture sector, which employs the largest proportion of Ugandans, has been declining over the years, accounting for just 15 percent of GDP in 2008 compared to 27.5 percent in 2000.5 Cash crops, particularly coffee, remain an important source of foreign exchange for the country and the strong performance of coffee exports contributed to overall economic growth. 6 A 27 percent rise in coffee export prices contributed to more favorable terms of trade since 2003. Figure 1.2: Composition of GDP Growth Sectoral Share in GDP Real GDP Growth by Sector 16.0 100.0 90.0 14.0 1 7 .0 1 5 .6 1 8 .3 2 0 .2 2 1 .1 2 2 .1 2 3 .1 2 3 .4 2 7 .5 80.0 12.0 70.0 10.0 60.0 5 0 .1 5 1 .6 4 9 .6 4 9 .0 4 9 .1 4 8 .6 50.0 8.0 4 8 .3 4 7 .3 4 4 .7 40.0 6.0 30.0 4.0 20.0 2 5 .3 2 4 .8 2 4 .5 2 4 .0 2 2 .8 2 2 .6 2 2 .2 2 2 .0 2 1 .4 10.0 2.0 0.0 0.0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2001 2002 2003 2004 2005 2006 2007 2008 Industry Services Agriculture Agriculture Industry Services Source: Uganda Bureau of Statistics and World Bank Staff 1.4 Private investment has also spurred economic growth. Private fixed domestic investment increased by an average of 17 percent--from USh 1233.8 billion (13 percent of GDP) in 2000 to USh 4404 billion (18 percent of GDP) in 2008. Foreign direct investment increased by an average of 17 percent, although the growth rate has been rising more sharply in the last five years, reaching growth of about 24 percent in 2008 and can been attributed to privatization efforts, infrastructure projects (including Bujagali), and oil exploration and development. Macroeconomic Environment & Fiscal Performance 1.5 Inflation between 2000 and 2008 averaged 5 percent. In 2008, annual inflation was 6.8 percent compared to 6.6 percent in 2007. The increase is attributed to rising food and fuel prices, which nevertheless contributed to a further improvement in export prices, as Uganda is a net food exporter. 5 There are strong reasons to doubt this low growth rate in agriculture, pending a new agriculture census. Despite a population growth which greatly exceeds this implied growth rate in agricultural value added, Uganda has continued to increase net exports of food, with no sign of increased hunger. Newly adopted crop varieties explain some of these exports, and should result in higher, not lower value added growth. 6 Exogenous shocks such as drought and floods contributed to a decline in agricultural production in Eastern Uganda in 2008. 5 1.6 Uganda’s external position deteriorated. The current account deficit before grants increased from US$870 million in 2007 to US$1237 million in 2008. This increase in the deficit arose from a widening deficit on the trade account. 1.7 Fiscal consolidation continued. The fiscal deficit declined steadily as a share of GDP from 13 percent in 2000 to 5 percent, where it has remained since 2006. 1.8 Government revenues have grown, with tax revenues contributing more than 80 percent to the revenue effort. In particular, collection of taxes has improved, with tax revenue increasing by an average of 17 percent between 2000 and 2008. Total tax revenue stood at 13 percent of GDP in 2008, up from 10 percent in 2000. 1.9 Government debt declined to 25 percent of GDP, following a debt forgiveness of US$ 3.4 billion in 2007 under the Multilateral Debt Relief Initiative (MDRI). However, the nominal stock of debt increased by 30 percent in 2008. Figure 1.3: Government Revenue 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 2000 2001 2002 2003 2004 2005 2006 2007 2008 Direct Taxes Indirect Taxes Non-Tax Revenue Source: MOFPED and World Bank Staff 1.10 Prudence in macroeconomic management is also evident in the level of government expenditures. While in nominal terms government expenditures have been trending upwards (increasing from USh2.2 trillion in 2000 to USh4.6 trillion in 2008), real expenditure as a percentage of GDP has been declining (as shown by the trend line in Figure 1.3). Specifically, the share of GDP absorbed by government expenditures declined from 24 percent in 2000 to 19 percent in 2008, although the latter figure is a percentage point higher than it was in 2007. Current public expenditures comprise a large share of total expenditure, averaging 60 percent of total public expenditures. 6 Future Prospects 1.11 Uganda’s growth is expected to slow down to around 5 percent in 2008/09 due to the current global crisis, though medium-term prospects remain promising if infrastructure bottlenecks can be addressed. The global slowdown is expected to impact Uganda’s external position through reduced demand for Ugandan exports and deteriorating export prices. Export values are expected to fall by about 3 percentage points. The Ugandan shilling has been depreciating against the dollar. Uganda is a net importer and a stronger dollar has increased the cost of imports to the country as a large part of imports by Uganda are priced in US$. A decline in both foreign direct investment and remittances is expected to contribute to the deterioration of the balance of payments. Consequently, the current account deficit is expected to widen further--from 8 percent of GDP in 2007 to about 10 percent by 20107. Figure 1.4: Government Expenditure (as a percentage of GDP) 30% 25% 20% 15% 10% 5% 0% 2000 2001 2002 2003 2004 2005 2006 2007 2008 % of GDP 24% 20% 22% 22% 22% 20% 17% 18% 19% Source: MOFPED and World Bank Staff II. TRENDS IN UGANDA BUDGETS FROM 2003/04 TO 2006/07 Public Expenditure Data and Gaps 1.12 The analysis of trends in Uganda’s budgets uses a unique item-level public expenditure data set.8 Raw data were collected on the budgets, release of funds and 7 Global Economic Prospects, 2009. 8 This data set was developed by World Bank staff with the assistance of budget officers in MOFPED and the staff at the Accountant General’s Office in Uganda. 7 actual expenditures, from both the Integrated Financial Management System (IFMS) and legacy accounting systems currently active data in Uganda9. The process of cleaning the database with the assistance of MOFPED is still on-going; consequently, some of the data and tables presented here may be amended in the future. Nevertheless, the available information will still likely present a fair assessment of the most comprehensive and current disaggregated budget data set in Uganda today. 1.13 Comprehensive data on budgets, release of funds and actual expenditures is not available for all votes, programs and items. There remain large gaps in Uganda’s actual expenditure numbers. This is a systemic problem, not a problem with the data that was collected. The 2009 PEFA assessment will discuss Uganda’s public expenditure reporting gaps and what should be done to fix them. Actual expenditures are available only for some agencies and most ministries; although even for ministries the actual expenditures for some items are blank. No actual expenditure data is available for districts. The district conditional grants show only release data--not just in this data set but in the audited reports and financial statements of the Government (Schedule D of the Report and Opinion of the Auditor General to Parliament on the Public Accounts of the Republic of Uganda). Actual expenditures are not reported for most universities, for the regional hospitals, or for donor projects. Actual expenditures are also patchy for the development budget, and for many items in most agencies actual expenditures are identical to releases. The analysis presented here is therefore eclectic; it uses the best available data to answer each question. 1.14 Release data are more widely available, but there are gaps. These are especially evident on donor projects, where for some projects there are budget data but no release, and for others there is release but no budget. 1.15 Since the most reliable and comprehensive data available are budget data10, trend analysis in this chapter uses the budget data only. However, this budget trend analysis is complemented with an analysis of overspending and underspending using release data, which needs to be taken into account in interpreting the likely trends in release of funds and actual expenditures. In turn, this analysis is restricted to an assessment of only those items which have both budget and release data. 1.16 No attempt is made here to track outputs in relation to the expenditure trends analyzed. This was done in the chapter on primary education in the first PER, and is done to some extent for the health sector in Chapter 3 of this report. The database could be extended in this direction for future work, should reliable and consistent data for outputs become available. In the meantime, the Ministry of Finance is working on an exercise to relate spending to outputs in budget preparation. The aim of the analysis presented here is to use the data set to see if past strategic budget trends are consistent 9 It is important to note at the outset that to Uganda’s MOFPED, the term ‘outturn’ refers to cash released into accounts of ‘spending votes’ and other agencies. It does not mean actual expenditure. 10 Using budget data, the PER team was able to almost exactly replicate the approved budget estimates for the years 2003/04-2006/07, which was no small task. The budget data reveals the planned trend in spending in Uganda and can signal any gaps between policy and resource allocation. 8 with the objectives of the MOFPED’s reforms; that is, to achieve improved delivery of public services with emphasis on value for money. Structure and Trends in the Ugandan Budget, 2003/04 to 2006/07 1.17 Uganda’s budget estimates are presented in three sections--the Statutory, Recurrent and Development Budgets. Statutory expenditures are mandated by the Constitution and include the Statutory Agencies (Votes 101-108), as well as debt payments, various Presidential expenses, and court awards. The Development Budget comprises both domestic development and on-budget donor projects. The database has item and project level data for each of these budgets. 1.18 The donor-financed Development Budget (the ‘donor’ column in Figure 1.5) has declined in importance, while the share of the Recurrent Budget has increased. The donor development share declined from 29 percent to 22 percent between 2003/04 and 2006/07, whilst the Recurrent Budget increased from 38 to 43 percent during the same period (Figure 1.5). As Figure 1.6 shows, donor development spending also fell slightly in Ugandan Shilling (USh) terms over the period. In contrast, the recurrent budget increased by USh603 billion in the four-year period (Figure 1.6), absorbing almost two-thirds (64 percent) of the additional budget through that period--with domestic development capturing 27 percent and the Statutory Budget 12 percent. Most of the rise in the domestic development budget occurred in 2006/07, with a large share of this increase coming as emergency assistance for the energy crisis. Figure 1.5: Shares of Expenditure in Each of the Budgets Budget Shares 100% Recurrent 90% Development Statutory 80% Donor 70% 60% 50% 40% 30% 20% 10% 0% 2003/04 2004/05 2005/06 2006/07 Source: MOFPED and World Bank Staff 9 Figure 1.6: Expenditure in Each of Uganda's Budgets, 2003/04 to 2006/07 (% of GDP) 30% Statutory Development 25% Recurrent Donor 20% 15% 10% 5% 0% 2003/04 2004/05 2005/06 2006/07 Source: MOFPED and World Bank Staff 1.19 The capital share of the Development Budget declined between 2003/04 and 2006/07. Both domestic development and donor projects contain recurrent cost elements. Gross Fixed Capital Formation, as a share of the combined donor and domestic Development Budgets, declined between 2003/04 and 2006/07--from 58 percent to 42 percent. The capital share in the domestic Development Budget is slightly higher, but has fallen from 63 percent to 48 percent over the four-year period (Figure 1.7). The capital share of the donor Development Budget fell from 54 percent to 37 percent. Figure 1.7: The Gross Fixed Capital Formation Shares of the Development Budgets GFCF Shares of Development Budget 70% Development 60% Donor 50% 40% 30% 20% 10% 0% 2003/04 2004/05 2005/06 2006/07 Source: MOFPED and World Bank Staff 10 Figure 1.8: Economic Classification of Spending, 2003/04 to 2006/07 (USh Billion) Budget Trends - Economic Classification (% GDP) 30% Arrears + Financial Transfers / Grants Other Recurrent Employee Costs 25% GFCF 20% 15% 10% 5% 0% 2003/04 2004/05 2005/06 2006/07 Source: MOFPED and World Bank Staff Figure 1.9: Adjusted Economic Classification, 2003/04 to 2006/07 (USh Billion) Economic Classification (Assuming recurrent component in transfers is 80 percent) 4,500 GFCF 4,000 Other Recurrent 3,500 Employee Costs 3,000 2,500 2,000 1,500 1,000 500 - 2003/04 2004/05 2005/06 2006/07 Source: MOFPED and World Bank Staff 1.20 The falling budget share of capital expenditures was offset by rising shares of staff costs, transfers and other recurrent spending (Figure 1.8). Other recurrent expenditures took the lion’s share of the budget increases between 2003/04 and 2006/07. Assuming that 80 percent of the transfers and grants to semi-autonomous agencies in fact go to employee-related costs11, employee costs actually took almost two-thirds of the changes in the budget between 2003/04 and 2006/07 (Figure 1.9). Other recurrent 11 The actual figure for agencies is close to 96 percent employee-related costs. Employee-related costs in this analysis have been interpreted as all the direct and indirect costs which can reasonably be attached to employing staff. In addition to those 6 digit line items prefixed in the chart of accounts by the digits 22 (wages, salaries, allowances, social security medical, retrenchment costs), staff training, staff travel, hiring costs, workshops and seminars, and meals and drinks in employee-related costs are included. 11 spending took 50 percent of the additional spending, and GFCF took minus 19 percent, with arrears and debt-related payments accounting for 6 percent. 1.21 Excluding debt and arrears payments, the true share of employee costs in the budget was between 42 and 43 percent for 2006/07. Figure 1.10 shows that the share of staff costs in the budget in 2006/07 was actually double that of public capital investments. Figure 1.10: Economic Classification of the 2006/07 Budget, Excluding Debt and Arrears Composition of Budget 2006/07 (Excl. Arrears and Debt, Assumes Employee Cost Share of Transfers as Agencies) Other Recurrent, 38% Employee Costs, 42% GFCF, 20% Source: MOFPED and World Bank Staff Figure 1.11: Budgets by Service Level (2003/04 to 2006/07) Approved Budget by Service Level, 2003/04 to 2006/07 (% of GDP) 35% Agency Embassy 30% Ministry HQ District / Municipality 25% 20% 15% 10% 5% 0% 2003/04 2004/05 2005/06 2006/07 Source: MOFPED and World Bank Staff 1.22 Ministries dominated total spending throughout the period and took about 60 percent of the additional budget between 2003/04 and 2006/07. Ministry spending 12 accounted for about two-thirds of total spending throughout the period12. The fastest growing budget level however was for agencies, which saw their budgets grow by USh210 billion, or 53 percent, in four years. Whereas the Ministries saw their budgets increase by USh206 billion, the share of the budget going to districts remained constant, at 21 percent However, all the increase was from wages and salaries (Figure 1.11). Analysis: What Lies Behind the Trends? Are They Healthy For Service Delivery? 1.23 This section examines the components of spending by economic classification and by service level. The analysis suggests that the Ugandan Budget has (at least during the four years under review) deteriorated from the perspective of funding service delivery. Figure 1.12: District Budgets by Type (2003/04- to 2006/07) District Budget Breakdown (USh Billion) 1,200 Donor 1,000 Development Recurrent 800 600 400 200 - 2003/04 2004/05 2005/06 2006/07 Source: MOFPED and World Bank Staff 1.24 All of the increases in district budgets between 2003/04 and 2006/07 were in recurrent spending. Whereas spending on local level services rose as fast as total spending, the district development budget fell in nominal terms (Figure 1.12). 1.25 Rapidly deteriorating wage to non-wage ratios in health and education explain most of the increases in recurrent spending. Hiring more workers to expand access, and awarding them pay raises, seems to have eaten into sector non-salary budgets. The budgets in schools (Figure 1.13) and district health (Figure 1.14) for non-salary items (PHC building, SFG, district public health, and school non-salary grants) fell slightly in nominal terms. Meanwhile, access to these services increased markedly; in public health the increase in the period from 2002/03 to 2005/06 (inferred from Uganda National Household Survey data and Uganda’s demographics) suggests the numbers of people attending public health clinics probably increased by around 50 percent. 12 For consistency with the other graphs on spending, the Ministry total includes payments on debt and arrears, and agencies include statutory agencies. 13 1.26 Hospitals experienced the same overall trend of deteriorating wage to non- wage ratios. Hospital wages and arrears payments rose fastest between 2003/04 and 2006/07; hospital non-wage recurrent budgets for administration and maintenance fell in nominal terms (Figure 1.15). A donor-financed project to extend Butebika hospital in 2005/06 inflated the numbers for buildings, equipment and maintenance that year but, otherwise, capital spending in hospitals fell. The drugs and medical supplies budgets for hospitals saw a slight improvement during the period but budgets for other administration, running costs and utilities fell. Perhaps, as a consequence, the hospitals’ budgets for arrears increased. Figure 1.13: District School Budgets (2003/04 to 2006/07) District School Budgets, 2003/04 to 2006/07 600.0 School Salaries 500.0 School Construction (SFG) School Non-salary 400.0 300.0 200.0 100.0 - 2003/04 2004/05 2005/06 2006/07 Source: MOFPED and World Bank Staff Figure 1.14: District Health Budgets (2003/04 to 2006/07) District Health Budgets, 2003/04 to 2006/07 (Ush Billion) (During this period, UNHS data suggests the number of people seeking care from public health centers and dispensaries increased by over 50 percent) 120.0 District PHC Buildings District Public Health Non-Wage District Public Health Wage 100.0 80.0 60.0 40.0 20.0 - 2003/4 2004/05 2005/06 2006/07 Source: MOFPED and World Bank Staff 14 Figure 1.15: Referral Hospitals’ Budgets (2003/04 to 2006/07) Hospitals Budgets, 2003/04 to 2006/07 120,000,000 (USh’000) Buildings, Equipment, Maintenance Other Admin. & Running Costs Utilities Drugs and Medical Supplies 100,000,000 Arrears Personnel Related Costs (Incl. Travel) 80,000,000 60,000,000 40,000,000 20,000,000 - 2003/04 2004/05 2005/06 2006/07 Source: MOFPED and World Bank Staff 1.27 With health and education construction budgets falling, the district development budget now mostly funds the Local Government Development Program (LGDP), Water, and Uganda National Agriculture Advisory Services (NAADS). In 2006/07, LGDP accounted for 39 percent, water and sanitation for 25 percent, and NAADS for 23 percent--making up 87 percent of the district development budget13 (Figure 1.16). Figure 1.16: Sector Composition of District Development Budget Uganda: District Development Budget - Schools (423) (USh Billion) - Primary Health Care (422) 200 - LGDP (115) - NAADS (100) 180 - Water & Sanitation (156) 160 140 120 100 80 60 40 20 - 2003/04 2004/05 2005/06 2006/07 Source: MOFPED and World Bank Staff 1.28 Grants to autonomous institutions have increased between 2003/04 and 2006/07 and become a large share of the budget. The total grants and transfers over the period (mostly from ministry budgets) greatly exceeded (at 177 percent) the total 13 Includes donor funding. 15 subsidies for hospitals and district health. Grants and transfers constituted three-quarters of the direct transfers made to State-owned schools14. Figure 1.17: Agency Grants are a Significant Share of the Budget Agency Grants Compared To Social Service Delivery (Last 4 years' cumulative, USh Billion) 1,800.0 1,600.0 1,400.0 1,200.0 1,000.0 800.0 600.0 400.0 200.0 - Schools Front Line Health Agency / Grants District Classified Development Source: MOFPED and World Bank Staff 1.29 MOFPED should examine the procedure for monitoring and evaluating the efficiency of grants to autonomous institutions, to determine if it is as rigorous as that for funds from the Poverty Action Fund (PAF). Grants and transfers to autonomous or semi-autonomous agencies constituted three-quarters of the direct transfers made to State-owned schools. They were also, cumulatively, almost three times the classified budget (mostly defense), an issue which has often received critical attention from development partners. It is not necessarily a bad thing that the Government of Uganda supports autonomous agencies with transfers15 as it provides opportunities to improve performance orientation as long as the fiduciary procedures are the same as those for PAF. 1.30 Who receives these grants? Some of the beneficiaries are named in the Approved Budget Estimates, others are not. In Figure 1.18, grants are arranged into MTEF sectors. More than half the transfers (52 percent) do not go to direct public service providing sectors but to the residual category ‘other’--which comprises public administration, JLOS (Justice, Law and Order Sector), accountability, public sector management, etc. 14 Some residual items are excluded from frontline services but they would not significantly alter the conclusions. These items include materials and pharmaceuticals for clinics sent from the Health budget, and some support for private secondary schools sent from the Ministry of Education. 15 In fact, part of the scale-up in 2006/07 comprises emergency funding for the energy sector and transfers to the energy fund. 16 Figure 1.18: Allocation of Grants to Semi-Autonomous Agencies, 2003/04 to 2006/07 Who Got the Grants? Other Social Services, 58,518,247 Education , 269,804,380 (Incl. Makerere, capitation grant to secondary schools) Administration, JLOS, Security, 677,688,291 (Includes Parliament, Electoral Commission, URA, CHOGM grants from Foreign Affairs) Energy, 239,274,490 (Emergency package, transfers to Energy Fund) Other Economic Services , 62,939,186 Source: MOFPED and World Bank Staff 1.31 Repayment of arrears has significantly exceeded new funding for district services for each of the three years ending 2006/07. The increased importance of arrears payments as a share of the budget indicates that the Government is serious about arrears clearance in the short-term. Cumulatively, the Government spent USh279 billion in the three years between 2004/05 and 2006/07 on repaying arrears--this amounts to just over half the reported cash payments on servicing external debt repayment. However, the magnitude of arrears repayment is worth reflecting on because unlike external debt which is a deliberate and programmed source of budgetary financing on concessional terms, arrears payments are annual payments to make good on past unfulfilled obligations on somewhat unknown terms. The fact that the annual level of correction on past budget excesses exceeds new annual increments made available for district services (Figure 1.19) suggests that either there is an issue with how the budget is implemented (that is, an issue with the control environment) or that there may be systematic under budgeting. 17 Figure 1.19: Arrears Payments Relative to Additional Spending on District Services Uganda Spends More on Clearing Old Budget Excesses Than on Annual Increases in District Services (USh Billion) 160.0 Addition to District Services 140.0 Budget For Arrears 120.0 100.0 80.0 60.0 40.0 20.0 - 2004/05 2005/06 2006/07 Source: MOFPED and World Bank Staff Figure 1.20: The Structure of Arrears Payments Budgeted Arrears More Than Doubled (2004/5 to 2006/7 USh Billion) 160.0 140.0 Telephone Arrears 120.0 Water Arrears Pension Arrears 100.0 Utility Arrears Domestic Arrears 80.0 60.0 40.0 20.0 - 2004/05 2005/06 2006/07 Source: MOFPED and World Bank Staff 18 Figure 1.21: Under-Budgeting: Comparing the Estimated Budget for Constant Use with the Actual Budget (2003/04 to 2005/06) (USh Million) Example: Under-budgeting for Electricity 35.0 Electricity Budget 30.0 Budget for Constant Use 25.0 20.0 15.0 10.0 5.0 - 2003/04 2004/05 2005/06 2006/07 Source: MOFPED and World Bank Staff 1.32 Under-budgeting has created new arrears between 2003/04 and 2006/07. Rudimentary analysis (Figure1.21) points to systematic under-budgeting for electricity. If the structure of arrears payments in Figure 1.20 is representative of the stock of accumulated arrears, the dominance of pensions, utilities and contracts might also point to under budgeting. Since the objective of MOFPED’s budgetary reform is to improve the access, quality and efficiency of public services, Uganda’s authorities will have to stop accumulating new arrears. The next section in this Chapter assesses whether a picture emerges from the patterns of over- and under-spending in relation to the budget. III. BUDGET VARIANCE: WHO OVERSPENDS, WHO UNDERSPENDS AND ON WHAT? 1.33 This section considers whether the approved budget in Uganda is a guide to resources actually available for spending. It should be noted: 1. There are no reliable actual expenditure data for districts, agencies or donors, so release data are used, even though these are incomplete. 2. The accumulation of new arrears cannot be captured in-year using budget data on a cash basis; data on payments against old arrears may be available, and periodically MOFPED verifies the stock, but data on new annual arrears are rare16. 3. Supplementary Budgets are the mechanism through which expenditures in excess of the original budget are approved. However, some sectors also generate internal revenues or sell goods which, on occasion, they may use, thereby creating expenditure in excess of both the budget and release17. 16 Again, this is a topic which the PEFA assessment team will examine. 17 Prisoners being fed food from prison farms rather than the prison selling the food produced at the farms, is one example of spending in excess of releases. 19 4. From the perspective of a service manager, unless there is full flexibility to transfer funds from one line item to another, variance needs to be estimated item by item and needs to be ‘gross’ rather than ‘net’ at the sector level. The same method has been adopted in the analysis here. The absence of data should not be interpreted in variance calculations to zero variance--if there is no budget or release data, the expenditure item is excluded from the calculations. 1.34 Uganda does a good job of delivering the budget as planned, but recourse to budget supplements has increased markedly. As noted in the PER 200718, Uganda scaled up spending between 1998/99 and 2001/02, and has implemented a fiscal consolidation strategy since. However, under fiscal consolidation, new arrears have appeared and the Supplementary Budget has increased markedly as a share of the total budget. In 2006/07, the Supplementary Budget accounted for 10.1 percent of the approved estimates, compared with 5.5 percent of approved estimates in 2003/0419. 1.35 The average overall gross variance was 16.2 percent between 2003/04 and 2006/07. This is calculated by adding underspending releases with overspending releases, and comparing the total to the budget20. The average was brought down by relatively low variance of just 13.2 percent in 2004/05. The aggregate gross variance in 2006/07 was 16.7 percent. 1.36 Variance is highest in the Development Budget, in ministries, and in transfers. Figures 1.22, 1.23 and 1.24 show the variance between budgeted and released expenditure over the four year period by budget type, level and economic classification. Figure 1.23 shows that there is significantly more variance compared with the budget in the ministries (about 30 percent) than there is in the districts (just 2.4 percent). Figure 1.22 shows that much of this variance seems to come in the Development Budget which, at 22.4 percent, has significantly wider variance than the Recurrent Budget (11 percent). With a variance of 21.2 percent, transfers are more than twice as volatile as any other economic category, with employee costs (9 percent) and capital (8.3 percent) coming next (Figure 1.24). Variance needs to be looked at as over- and under-spending. 18 World Bank (2007), “Uganda: Fiscal Policy for Growth”. 19 Total Supplementary/Approved Budget (including new resources + reallocations). 20 This calculation is complicated by data gaps. The items used here are only those for which there is both a release and a budget figure. The estimates, therefore, exclude donor projects for which there tends to be either no budget entry or no release data. 20 Figure 1.22: Budget vs. Release Variance by Level By Level: Four-Year Absolute Budget Variance Estimates (2003/04 to 2006/07 - Figures for items with Budget and Release Data Available) (Excludes Donor - for which release data is often zero) 35.0% 29.9% 30.0% 25.0% 20.0% 15.0% 10.8% 9.2% 10.0% 5.0% 2.4% 0.0% District Agency & Embassy Ministry Total Source: MOFPED and World Bank Staff Figure 1.23: Budget vs. Release Variance by Budget Four-Year Absolute Budget Variance Estimates (2003/04 to 2006/07 - Figures for items where Budget and Release Data Available) (Development excludes Donor - for which unreliable release data) 25.0% 22.4% 20.0% 15.0% 12.4% 11.0% 10.0% 5.1% 5.0% 0.0% Development Recurrent Statutory Total Source: MOFPED and World Bank Staff 21 Figure 1.24: Budget vs. Release Variance by Economic Classification Economic Classification; Four-Year Budget Variance (2003/04 to 2006/07 - Figures for which Release Data Are Available) Excludes Donor Projects, for which release not available: this reduces variance 25.0% 21.2% 20.0% 15.0% 9.0% 9.2% 10.0% 8.3% 8.0% 5.1% 5.0% 0.0% Employee GFCF Arrears + Transfers / Other Total Costs Financial Grants Recurrent Source: MOFPED and World Bank Staff 1.37 There is a pattern to overspending (Figure 1.25). The Recurrent Budget accounts for 55 percent of overspending over the past four years. Ministries make up 67 percent and agencies a further 19 percent, with district budgets making up just 14 percent of overspending. Emergency spending on the energy crisis in 2006/07 was the single largest overspending item. Excluding the latter item, wages, transfers/grants, and goods and services account for two-thirds of over spending, in roughly equal shares. Figure 1.25: Overspending By Economic Category (cumulative over four years) Overspending 2003/04 to 2006/07 Mostly Recurrent, Transfers Becoming Important 24% 27% Employee Costs GFCF Arrears + Financial Transfers / Grants 9% Other Recurrent 34% 6% Source: MOFPED and World Bank Staff 1.38 Underspending falls disproportionately upon the domestic Development Budget, and within it, upon capital spending. Capital spending made up 34 percent of underspending, compared with just 9 percent of overspending (Figure 1.26). 22 Figure 1.26: Underspending By Category (cumulative over four years) Underspenders 2003/04 - 2006/07: Mostly Capital Spending 24% 29% Employee Costs GFCF Arrears + Financial Transfers / Grants 5% Other Recurrent 8% 34% Source: MOFPED and World Bank Staff 1.39 The net overspending position appears unfavorable with regard to service delivery. Figures 1.27, 1.28 and 1.29 show that net variations in the budget tend to take from the Development Budget and, in particular, from capital spending. They only take marginally from the districts. Adjustments to the budget within year seem to have reduced rather than increased the impact of the budget on service delivery. Figure 1.27: Net Four-Year Overspending by Budget Type Net Overspend (Release vs Budget) USh Billion: 2003/04 to 2006/07 250.0 200.0 150.0 100.0 50.0 0.0 Statutory Recurrent Development -50.0 -100.0 -150.0 Source: MOFPED and World Bank Staff 23 Figure 1.28: Net Four-Year Overspending by Economic Classification Net Overspend - Economic Classification: USh Billion: 2003/04 to 2006/07 300.0 250.0 200.0 150.0 100.0 50.0 0.0 Employee Costs GFCF Arrears + Transfers / Other Recurrent -50.0 Financial Grants -100.0 -150.0 Source: MOFPED and World Bank Staff Figure 1.29: Net Four-Year Overspending by Level Net Overspending By Level (USh Billion 2003/04 to 2006/07) 180.0 160.0 140.0 120.0 100.0 80.0 60.0 40.0 20.0 0.0 District Ministry -20.0 Agency & Embassy Source: MOFPED and World Bank Staff 1.40 Vote level data show that few--mostly the central leadership and security- related ministries, are overspending. The State House, the Office of the Prime Minister, the Office of the President, and Defense, Prisons and Police account for just under 40 percent of overspending on goods and services. Defense, Police and Prisons together account for 25 percent of overspending on wages. Summary of Observations 1.41 Uganda’s budgets were not, until very recently, consistent with the Government’s policies as stated in the Poverty Eradication Action Plan (PEAP). (Until the 2008/9 budget scaled up road expenditures, infrastructure was being neglected). Most notably: • Non-salary budgets for clinics and schools have been falling in real per capita terms, relative to patient and student use, and the quality of district services is most likely suffering. 24 • Despite the emphasis on decentralized services in the PEAP, it is at the ministry and agency levels that non-salary budgets have increased. • Capital spending, largely in the infrastructure sector, had fallen. 1.42 Uganda’s fiscal rules have yielded mixed results for service delivery. The Government’s aim of consolidating the fiscal deficit and protecting priority PAF funding have helped maintain macroeconomic stability and protect aggregate PAF spending from mid-year release cuts. However, these fiscal rules have not protected non-salary PAF allocations in the budget in favor of quality and access in public service delivery. 1.43 The fastest growing MTEF `sector’ in the budget was public administration. This was contrary to announced policy intentions. The fastest growing expenditure items were employee-related costs and transfers to semi-autonomous agencies, some of which were not specified in the budget. Perhaps most surprisingly: • Over the four years between 2003/04 and 2006/07, Uganda budgeted almost twice as much for transfers to semi-autonomous agencies21 as for frontline health services (including salaries). 1.44 Spending on the clearing of arrears occurred throughout the budget. Ever since the Government audited and began budgeting to clear arrears, annual payments on arrears have exceeded the annual increases (in practice, all were salaries) for district services. Nevertheless, new arrears continue to emerge, and there is evidence of under- budgeting (for example, for electricity). 1.45 The trend in budget implementation looks worrisome for service delivery. Districts tend to get their budgets released as planned. Ministries and agencies overspend their budgets on average, most notably on the recurrent budgets for employee costs and transfers to semi-autonomous agencies. Regardless of who spends it, the Development Budget tends to be systematically underspent, particularly on capital items. Supplementary Budgets have increased as the means to authorize overspending. The main recipients of overspending (accounting for 40 percent over the four years under review) were the State House, the Office of the Prime Minister, the Office of the President, the Ministry of Defense, and Prisons and Police. Recommendations for Budget Reforms 1.46 Fiscal rules: In order to be consistent with the budget reforms being implemented to generate value for money in service delivery, MOFPED should, in developing a new fiscal anchor for the macroeconomic program, consider a range of fiscal guidelines and rules that could protect the composition of spending for infrastructure and service delivery. 1.47 Balancing the value for money drive: MOFPED’s budget reforms take service sectors as their main focus--seeking to link outputs to targets in health, education, roads, 21 Note: These are separate from the agencies which receive votes in the budget. 25 water, and agriculture sectors. The emphasis on seeking value for money from sector service delivery is valid, given the amounts being spent and the levels of waste and leakage. However, to avoid imbalance in these reforms--and in view of the rapid rise in spending on public administration, transfers to agencies, and ministries--the reforms should include an initiative to scrutinize efficiency at headquarters, agencies and subvention programs, and should differentiate policy functions from service overhead delivery costs. MOFPED should also clarify the extent to which efficiency `savings’ will be channeled back into sectors for better service delivery--for example, to finance broader access to secondary schooling, the pursuit of better health outcomes, etc. 1.48 Increasing transparency in arrears: The Accountant General should report annually to Parliament on Uganda’s outstanding stock of arrears, annual payments against it, and new arrears accrued during the year. Such reports should detail the items and votes incurring new arrears. To avoid confusion, arrears payments identified in the annual budget estimates under each vote heading, should be highlighted by vote, program and item. An assessment should be made annually of the causes of new arrears, by type of arrears. 1.49 Budget Predictability: Budget reforms must improve the in-year predictability of budget implementation. MOFPED should consider implementing the suggestions set out in Platform 1 of the matrix developed under Joint Budget Support pre-identification to improve predictability and budget stability. Annex 1 to this report presents the relevant measurers suggested to support improved predictability and stability in the budget in order to ‘remove barriers in public finance management (PFM) systems and capacities while reinforcing compliance with regulations and avoidance of leakage’ (that is, to achieve Platform 1). 26 CHAPTER 2. COSTING POLICY COMMITMENTS IN PUBLIC SERVICE PAY REFORM AND UNIVERSAL SECONDARY EDUCATION Introduction Than 2.1 Public service pay reform is needed in Uganda to strengthen public service delivery. Pay reform is a key step to improve the performance culture and the quality of public services as a whole22. Another key element in developing a performance culture is adapting human resource procedures and linking results-oriented management reforms to the budget.23 2.2 To contribute to deliberations on pay reform, this Chapter uses fiscal simulations to examine some requirements of an affordable public service pay reform in Uganda.24 This Chapter is split into three sections: a. A description of recent civil service pay reform initiatives in Uganda. b. The wages and staff structure of the public service and the rationale for wage reform. c. A fiscal sustainability template to simulate the impact on the budget of three different pay reform scenarios and measures to keep pay reform affordable. 2.3 The simulations focus on examining the fiscal effects and the availability of resources in the budget for three different pay reform scenarios. The first simulation models the short-term pay reforms set out in the pay policy approved by Cabinet in August 2006, implying at least a nominal doubling of wages for all pay scales over the medium-term. The second simulation models a nominal doubling of wages for managerial, technical and professional staff only. The third models these pay increases for selective staff but combines them with efficiency savings to illustrate that even under selective pay awards, efficiency gains will be needed for the budget to stay consistent with Uganda’s other fiscal policy objectives. The fiscal effects are analyzed using a fiscal sustainability model which was built under the 2007/08 PER. While the proposed pay reforms will have a significant impact on public pension liabilities, the lack of necessary data on staff ages and years of service has prevented any analysis of this 22 Remuneration is clearly important. Beyond it, recent global thinking on institutional economics focuses on the concept of ‘identity’ as a key in motivation. (See, for instance, recent works by Beasley and by Ackerloff and Kranton). Similarly, Renikka and Svensson (2003) “Working for God?” analyzed what motivates religious not-for-profit (RNP) health-care providers in Uganda. They found results consistent with the view that RNPs are driven (partly) by altruistic concerns, and that these preferences matter quantitatively. 23 “Review Of The Implementation Of Results Oriented Management (ROM) In The Public Service In Uganda”, May 2008, Atos Consulting. 24 This chapter was developed by Bank staff using extensive data provided by staff in the Ministry of Finance, Planning and Economic Development, Uganda Computer Services, Uganda Bureau of Statistics (UBOS), and Ministry of Public Service 27 impact at this stage. Details of the assumptions underlying the baseline scenario and the full results are set out in Annex 2.2 to this report. 2.4 This Chapter also looks beyond wages to estimate the share of total staff costs in the recurrent budget. The analysis of staff numbers and total staff costs is based on comprehensive data on the numbers of employees at different pay scales on the consolidated public service payroll, together with the detailed item-level budget database used under Chapter 1. Nominal staff costs as reported in the budget consist of gross salaries, allowances and other transparent monetary compensation for employees. However, in addition to these, public servants benefit from other perks which cannot be precisely detected when scrutinizing the budget; these include per diem for attending conferences and workshops, and for travel up-country and overseas. I. WAGE REFORM INITIATIVES IN UGANDA 2.5 The Government of Uganda has taken a number of reform initiatives over the last 10 years to strengthen civil service performance. Implementation of civil service pay reforms has been a central element. The Public Service Pay Reform Strategy adopted in April 2002 rationalized the public service pay structure into a single spine structure25 with the aim of removing inequalities in wages for similarly graded jobs. For efficient and effective performance of the public sector, the 2002 strategy also targeted pay increases to personnel with requisite technical, professional and managerial skills. The strategy did initially lead to significant pay increases for some of the targeted staff categories. 2.6 Overall progress against the original targets was hampered by a number of developments, including: ƒ Rapid growth of employment, especially in the public administration, social sectors and more recently in police and prisons. (Growth in social sectors has been driven by the expansion of service delivery--in health and education in particular). ƒ Wage provisions in the budget were insufficient to meet salary targets. ƒ In conflict with the 2002 reform strategy, pay increases were given selectively to certain staff categories—particularly to health workers and teachers. ƒ Real compensation levels have been blurred by the re-introduction of various non- salary allowances and benefits. To ensure transparency and effective wage bill control, allowances were consolidated into salaries in 1995 but were later reintroduced in order to compensate some staff categories for the slow gains made in real wages. 25 A pay spine consist of a hierarchy of pay or spinal column points, between which there are pay increments, and to which are attached job grades. 28 2.7 The most recent major initiative is the pay policy for public service institutions and Government agencies, which the Cabinet approved in August 2006. This policy takes a different stance from the one presented in the reform strategy adopted four years earlier. It still recognizes the need to enhance pay for certain categories--this time identified as scientists and researchers. However, it is not clear how this should be translated into salary targets for different pay scales. While the relative pay increases proposed vary between different pay scales, all pay scales are to receive around a doubling of nominal wages in the medium term26. Some specified officers and upper level management are targeted for additional increases that will triple their total nominal wage increase. The strategy also sets out long-term pay targets to be implemented over a ten-year period27. While the extended period would significantly reduce the real level of these increases, the targets for most pay scales imply a further doubling of nominal wages compared to the medium-term targets. 2.8 A substantial across-the-board increase of public wages is presently not affordable, given competing spending pressures. First, non-salary funding for service delivery needs to increase (Chapter 1). Second, the Government wants to scale up infrastructure investments that are crucial for sustaining growth. Third, wages already constitute roughly half of the recurrent budget, and employee costs more broadly comprise over 40 percent of the total budget. Figure 2.1 shows the share of employee costs in the total recurrent budget during the last few years. Figure 2.2 shows the wage and salary shares in the recurrent and total budgets. Figure 2.1: Shares of Employee Costs in the Budget 100% 90% 80% 70% Other Recurrent 60% 50% GFCF 40% 30% Employee Costs 20% 10% 0% 2003/04 2004/05 2005/06 2006/07 Source: MOFPED and World Bank Staff 26 Medium-term specified as FY2009/10. 27 Long-term targets are foreseen to be implemented between FY2007/08 and 2015/17. 29 Figure 2.2: Wage Shares in the Budget 60% Wages as Share Recurrent Wages as Share Total Non Debt 50% 40% 30% 20% 10% 0% 2003/04 2004/05 2005/06 2006/07 Source: MOFPED, MOPS, and World Bank Staff II. CIVIL SERVICE WAGES, STAFF STRUCTURE AND THE RATIONALE FOR PAY REFORM 2.9 Private sector wage comparisons provide good reasons to question the rationale for an across-the-board substantial increase in public salaries. In order to understand whether the public sector has the ability to hire and retain staff with the requisite skills, it is necessary to compare public sector wage levels with those in competing markets. While wage levels in neighboring countries can be relevant for highly mobile staff, appropriate wage comparisons for most public servants can be made with the Ugandan private sector. A complete comparative analysis of compensation levels in the public and private sectors is beyond the scope of this study. Data from earlier studies do not suggest unequivocally that wages in the public sector are too low. 2.10 A study undertaken at the request of Ministry of Public Service (MOPS)28 showed considerable pay disparities between the traditional public service29 and other public service institutions, and between the public sector and the wider market. The observed wage differences between private and public sector in the study were not uniformly in favor of either. The study showed that for 48 percent of the different level positions identified, the public service paid 100 percent or more (including allowances) than the median salaries in the market. 2.11 Another comparative study by UBOS suggests that only senior managers’ wages are lower in the public sector. The report by UBOS on labor market conditions in Uganda is based on UNHS data. It shows that wages for senior officials and managers in the public sector are lagging behind those in the private sector, but for other staff 28 Study on remuneration and pay policies for public service institutions and Government agencies in Uganda, conducted between March and June 2004 by the Law & Development Partnership limited on behalf of the Ministry of Public Service. 29 Government ministries and departments; the police and prison services; teaching, health and judicial services, and; local governments. 30 categories the relationship is the opposite. The study concludes that, overall, persons employed in the public sector earn more than those in the private sector. Table 2.2 shows median wages in the public and private sectors, sorted by education levels, indicating that the differences decrease with rising levels of education. Table 2.1: Median Monthly Nominal Wages (Public and Private Sector), by Occupation (USh ‘000) 2002/03 2005/06 Occupations Public Private Total Public Private Total Legislators, Sr. Officials and Managers 250.0 450.0 450.0 (120.0)* (181.0)* (120.0)* Professionals 120.0 150.0 140.0 300.0 170.0 250.0 Technicians and Associate Professionals 105.0 100.0 105.0 150.0 120.0 148.0 Clerks 100.0 100.0 100.0 110.0 72.4 80.0 Service Workers, and Shop and Market Sale 100.0 36.2 50.0 80.0 45.0 50.0 Agriculture and Fisheries Workers 70.0 20.0 20.0 58.1 27.2 27.2 Crafts and Related Trade Workers 250.0 63.4 63.4 180.0 84.0 90.5 Plant and Machine Operators and Assemble 70.0 80.0 80.0 94.0 90.5 90.5 Elementary Occupation 54.3 30.0 30.0 60.0 21.7 23.5 Source: UBOS, Report on the Labor Market Conditions in Uganda, December 2007 (UNHS data) *This particular figure is unreliable, given too few observations Table 2.2: Median Monthly Nominal Wages--Public and Private Sector (USh ‘000) 2002/03 2005/06 Public Private Total Public Private Total No Formal Schooling 30.0 18.1 18.1 35.0 18.1 18.1 Primary Schooling 70.0 30.0 30.0 60.0 27.2 27.2 Some Secondary 80.0 60.0 70.0 117.0 63.4 70.0 Schooling Post-Secondary 120.0 120.0 120.0 150.0 120.0 150.0 Education Source: UBOS, Report on the Labor Market Conditions in Uganda, December 2007 (UNHS data) 2.12 World Bank staff calculations from the UNHS III suggest that average hourly wages are higher in the public sector. A simple Ordinary Least Squares (OLS) regression which adjusts for age, civil status and skills--in terms of level of schooling completed, sector and geographic location--shows a public sector increment to hourly wages of about 33 percent, compared to the rest of the economy. The full results are provided in Table 2.1.3 of Annex 2.1. 2.13 Comparing establishment survey data from UBOS with data from the Ministry of Public Services (MOPS) and Ministry of Health (MOH) yields similar results. Teachers, doctors and nurses/midwives receive higher salaries, on average, in the public sector than in the private sector. These staff categories are important, both in terms of numbers and share of total costs for civil service staff, making it all the more likely that, on average, public servants receive a premium in the labor market. 2.14 Given these results, a selective approach to pay reform that targets skill categories in high demand is preferable to an overall wage increase. The comparative studies referred to above are subject to caveats over their categorization of grades, survey sampling frame, etc. Nevertheless, they indicate that the assumption that civil service pay in general is too low across the board for the public sector to hire and 31 retain staff is overly simplistic. Consequently, it is recommended that the Government conduct a more rigorous survey to assess skill shortages.30. 2.15 Ugandan civil servants are concentrated towards the lower end of the pay scale, reflecting the dominance of teachers and nurses on the payroll.31 In June 2006/07, there was over 227,000 staff on the consolidated public service payroll held by MOFPED32. As shown in Figure 2.3, the lion’s share of these public servants can be found at the lower end of the pay scale. This is largely due to the fact that the education sector makes up about two-thirds of pay roll staff, and teachers are skewed towards the lower pay scales at `U7 Upper’. Teachers account for three-quarters of the U7 Upper spine. Also included in this pay scale are nurses and midwives, but they only make up an estimated 6 percent of the remaining posts in U7U. Figure 2.3: Numbers of Staff by Pay Scale: 2003/04-2006/07 160000 140000 120000 f staff 100000 ro 80000 mbe 60000 Nu 40000 20000 0 R R F D E D ICE U E IC L O TE 2U 3U 4U 5U 6U 7U 8U 1S 2L 3L 4L 5L 6L 7L 8L 1E 1S 1E FF F U U U U U U U L A U U U U U U U U U U U A T O IC S T T Pay scale IE LI O O N IF C E P S P 2003/04 2004/05 2005/06 2006/07 Source: MOFPED, MOPS, and World Bank Staff 2.16 The cost structure of the payroll is heavily tilted towards the lower end of the pay scale, and in particular towards teachers. Whereas higher wages for higher pay scales redistribute some payroll costs toward the middle and upper pay scales, Figure 2.4 on staff costs shows a very similar distribution to Figure 2.3 on staff numbers. According to Bank staff estimates, the U7U scale takes up 46 percent of the total wage bill33, of 30 Uganda could usefully pilot the Bank’s new Actionable Governance Indicators on Public Service & Human Resource Management. 31 See Annex 2.1 for detailed numbers on personnel and wages by pay scale. 32 The public service payroll held by Uganda Computer Service at the Ministry of Finance does not include staff employed at the following public service entities: missions abroad, non-civil defense, universities, most governmental agencies, and some public commissions. 33 Calculation based on payroll numbers, multiplied by average base wages for different pay scales. This calculation thus excludes other staff-related costs. 32 which education staff make up 35 percentage points. Taking into account all pay scales, education sector wage costs make up 61 percent of total public sector wage costs34. 2.17 Efficient teacher deployment, combined with judicious pay rises (especially under USE), is central to affordable pay reform in Uganda. Political commitment to expand education, combined with Uganda’s rapid population growth and youthfulness, will greatly increase the number of secondary students in the coming years--especially if the quality of primary education improves. This will clearly reduce the scope for any significant cost savings. However, as noted in the previous Public Expenditure Review35, absenteeism in the education sector is high (as it is in health). Better utilization and deployment of available teachers and health workers could at least limit the need for further increases in the numbers of these personnel in the short-term, freeing up resources for non-salary spending. Figure 2.4: Staff Costs by Pay Scale: 2003/04 to 2006/07 400,000 350,000 300,000 250,000 Ush, Mn 200,000 150,000 100,000 50,000 0 S S ER IF O F E D O ER U L E 8U 2U 3U 4U 5U 6U 7U 2L 3L 4L 5L 6L 7L 8L 1S 1E 1E 1S IC EC AL A T C U U U U U U U U U U U U U U U U U U IE FI FF C ST Pay scale LI OT D N TI SP PO 2004/05 2005/06 2006/07 Source: MOFPED, MOPS, and World Bank Staff 34 Here defined as staff included on the consolidated payroll held by Uganda Computer Service. 35 “Uganda Fiscal Policy for Growth”, Public Expenditure Review 2007, World Bank 33 III. FISCAL EFFECTS FROM SELECTED PAY REFORM SCENARIOS 2.18 The aim of the simulations set out below is to provide general guidelines for pay policy and efficiency work, and the Government’s drive for ‘value for money’. The work uses a RMSM-X36 based fiscal projections model developed by the Bank under the 2007 PER and calibrated using the old National Accounts series37. It does not attempt to model actual budgets, but instead to guide policy makers as to the relative orders of magnitude of alternative policy decisions; in this case, it assesses the fiscal trade-offs that need to be taken into account in designing pay reforms and implementing USE. A more complete description of the model can be found in Chapter 4 of Volume II of the PER for 2007 Assumptions underlying the baseline used here are set out in Annex 2.2 of this report. 2.19 The simulations suggest that in the absence of efficiency savings from USE, even without public service pay reforms, the increased demand for teachers under USE will reverse the momentum towards infrastructure spending in the 2008/09 budget. Different pay reform scenarios were simulated using the projections model, of which three are presented here. The baseline scenario assumes no public service reform, but has wages for all staff categories continuing to increase in line with past trends since 199738. For public service staff numbers the baseline has the stock of teachers rising in line with USE, with other staff categories increasing in line with the past trend. No efficiency savings are assumed. Figure 2.5 illustrates this scenario, showing the share of infrastructure in public spending being squeezed to a lower share of GDP over the period. It is assumed that health expenditure will grow in line with past trends, a scenario that may be overly optimistic given the pressures in the health sector arising from the fast growing population and increasing demand for improved health services. 36 RMSM-X stands for Revised Minimum Standard Model Extended and uses flow of funds model in which each income flow in one sector has an expenditure flow in corresponding sector. 37 The new National Accounts that raised GDP growth by about 2 percentage points imply that even though expenditure as a share of GDP is maintained at or below 22 percent of GDP, there is room for higher expenditure than had been the case in the original estimates where GDP was lower. 38 An annual average increase of 5.1 percent in real terms. 34 Figure 2.5: Uganda Budget Composition -- Baseline 25.0% 22.7% 20.0% 15.0% (share of GDP) 10.0% Health 5.0% Education 0.0% 2008/09 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 2016/17 2017/18 2018/19 2019/20 2020/21 2021/22 2022/23 2023/24 2024/25 2025/26 2026/27 Education Health Infrastructure-intensive programs/votes Economic services and agriculture Accountability, justice, etc Pension Unconditional grant to districts and urban Interest payments Arrears repayment Total Source: World Bank Staff estimates Figure 2.6: Scenario 1: Nominal Doubling of Public Service Wages (all Staff) 25.0% 22.7% 20.0% 15.0% (share of GDP) 10.0% 5.0% 0.0% 2008/09 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 2016/17 2017/18 2018/19 2019/20 2020/21 2021/22 2022/23 2023/24 2024/25 2025/26 2026/27 Education Health Infrastructure-intensive programs/votes Economic services and agriculture Accountability, justice, etc Pension Unconditional grant to districts and urban Interest payments Arrears repayment Total Source: World Bank Staff estimates 35 Figure 2.7: Scenario 2: Nominal Doubling of Wages of Managerial, Technical and Professional Staff 25.0% 22.7% 20.0% 15.0% (share of GDP) 10.0% Health 5.0% Education 0.0% 2008/09 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 2016/17 2017/18 2018/19 2019/20 2020/21 2021/22 2022/23 2023/24 2024/25 2025/26 2026/27 Education Health Infrastructure-intensive programs/votes Economic services and agriculture Accountability, justice, etc Pension Unconditional grant to districts and urban Interest payments Arrears repayment Total Source: World Bank Staff estimates 2.20 The across-the-board public service pay increases approved by the Cabinet would reduce the room in the budget for infrastructure spending. Figure 2.6 replicates the medium-term pay targets for public service staff outlined in the pay policy approved by Cabinet in August 2006. While the relative pay increases outlined in this policy vary between different pay scales, all pay scales would receive at least a doubling of their nominal wages. The simulation in Figure 2.6 assumes a nominal doubling of all pay scales shared equally across FY2009/10 and FY2010/11. However, whereas the approved pay policy foresees a further nominal doubling of wages in ‘the longer term’, this scenario has instead assumed that after the initial two-step nominal increase, wages would revert to increasing in line with their past trend. The simulations assume no measures to increase efficiency, meaning the same trend in staff numbers prevails as in the baseline scenario. According to the simulation here, the Cabinet approved wage increases would result in infrastructure spending being gradually crowded out to 0.2 percent of GDP by 2027/28. In real terms, the spending on infrastructure will decline from the peak reached in 2008/09 of USh 1,389 billion to just USh 215 billion in 2026/27. By 2017/18, education spending as share of GDP would reach 10 percent. The conclusion is that this scenario would leave no room for the scaled-up infrastructure spending required to avoid growth in Uganda from slowing down. 2.21 Even selective wage increases would leave insufficient room for a scale-up of infrastructure spending. In Figure 2.7, scenario 2 simulates a selective wage increase with the nominal pay increase (doubling) from the scenario 1 limited to managerial, technical and professional staff39. Otherwise, scenario 2 follows the assumptions of the base case and scenario 1. Figure 2.7 shows selective pay raises leave more room for spending on infrastructure. However, infrastructure investments would fall to within the 39 In terms of pay scales, this has been defined as the upper ‘U’ levels of pay scale U5 and above. 36 range of 2 percent to 3 percent of GDP, which would be inadequate to remove infrastructure bottlenecks to sustain high growth in Uganda. Figure 2.8: Scenario 3: Selective Pay Increases Combined with USE Savings 25.0% 22.7% 20.0% 15.0% (share of GDP) 10.0% 5.0% 0.0% 2008/09 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 2016/17 2017/18 2018/19 2019/20 2020/21 2021/22 2022/23 2023/24 2024/25 2025/26 2026/27 Education Health Infrastructure-intensive programs/votes Economic services and agriculture Accountability, justice, etc Pension Unconditional grant to districts and urban Interest payments Arrears repayment Total Source: World Bank Staff estimates 2.22 A selective approach, together with efficiency savings in USE and a better utilization of public service staff, is necessary to protect budgetary space for infrastructure. Scenario 3 in Figure 2.8 combines selective pay reform with efficiency savings in education. In the education sector, these include: (i) an increase in the student teacher ratio; (ii) faster adoption of double shifting; (iii) increased participation of private schools in delivering USE; and (iv) gradual increase of the share of diploma teachers towards the 3:1 diploma-to-graduate teacher policy ratio40. These measures will contain total costs for education at close to 5 percent of GDP throughout the projection period. This scenario also assumes that the number of non-education public sector employees stays constant for the rest of the projection period and that construction unit costs for schools are reduced. The efficiency savings under scenario 3 create space to maintain infrastructure spending at above 5 percent of GDP through the projection period. 2.23 In sum, without efficiency measures in education, neither a selective pay reform nor full roll-out of universal secondary education will be affordable without an expansion of the deficit or a squeeze in expenditures on health and infrastructure. Figure 2.9 shows the share of education spending under the different pay reform scenarios. The roll-out of universal secondary education, combined with an across-the-board increase in public service wages without any efficiency measures, would increase the share of education to one-half of total public spending. The corollary of this would be infrastructure spending declining to as little as 1.1 percent of the total budget. Even a selective pay increase would increase costs for education to a level which 40 Further details on the projected efficiency measures can be found in Annex 2.2 to this report 37 would leave insufficient room to increase infrastructure spending, and thus needs to be combined with efficiency measures. Figure 2.9: Education’s Share of Total Expenditure under Pay Reform Scenarios 60.0% 50.0% 40.0% % 30.0% 20.0% 10.0% 0.0% 2006/07 2008/09 2010/11 2012/13 2014/15 2016/17 2018/19 2020/21 2022/23 2024/25 2026/27 Baseline scenario Scenario 1: USE rollout wit h minimal efficiency measures + doubling of all wages Scenario 2: Select ive pay reform t hrough doubling of managerial, t echnical and professional ranks + USE rollout wit h minimal efficiency measures Scenario 3: Select ive pay reform + USE rollout +efficiency savings in educat ion and bet t er ut ilizat ion of public service st aff Source: World Bank Staff estimates IV. SUMMARY 2.24 Pay reforms aimed at assuring the recruitment, retention and motivation of staff must be implemented within an affordable resource envelope. The ability to attract, hire, retain motivate, and manage well qualified public servants who, in turn, manage and work with integrity is central to well functioning public service the world over. Recognizing that the public service in Uganda is poorly motivated, it has long been argued that public service pay reform is needed ‘to incentivize’ public servants. The key challenge is how to achieve this affordably while instilling a performance culture and reducing absenteeism. This is all the more important in Uganda, given the need to find additional finance for district services (Chapter 1) and infrastructure41. 2.25 A large general pay rise is unaffordable in Uganda given public sector employment levels and competing budget demands. The number of public servants grew by 21 percent between 2000/01 and 2006/07, with the education staff numbers growing at 31 percent. Over this period, the wage bill increased by 58 percent in real terms, with public sector average real salaries increasing by 30 percent. Wages alone took up half of the Recurrent Budget from 2003/4 to 2006/07. Broader employee-related costs constituted about 42 percent of the total budget, including development spending. Uganda’s young and expanding population and a welcome increasing demand for public services are building strong demand-side pressure for more expenditure on public 41 World Bank (2006) “Uganda: Investment and Behavior Change for Growth”, Country Economic Memorandum. And World Bank (2007) “Uganda: Fiscal Policy for Growth”. 38 services. At the same time, Government is seeking to increase spending on infrastructure to maintain growth, and is planning to expand access to secondary education. Table 2.3: Summary of Simulation Results; Spending on Education, Infrastructure and Wages as Share of GDP under Different Scenarios 2008/09 2010/11 2012/13 2014/15 2016/17 2018/19 2020/21 2022/23 Baseline Education 5.4% 5.4% 5.4% 6.3% 6.9% 7.7% 7.2% 6.7% Infrastructure 5.2% 3.4% 3.8% 3.5% 3.3% 3.0% 3.0% 3.2% Wages 5.2% 5.3% 5.6% 6.0% 6.3% 6.8% 7.2% 7.5% Nominal doubling of public service wages for all staff Education 5.4% 7.1% 7.3% 8.5% 9.4% 10.6% 10.4% 10.2% Infrastructure 5.2% 1.8% 2.0% 1.4% 1.0% 0.3% 0.4% 0.4% Wages 5.2% 8.7% 9.4% 10.1% 10.8% 11.8% 12.6% 13.5% Nominal doubling of wages for managerial, technical and professional staff (selective) Education 5.4% 5.7% 5.8% 6.7% 7.4% 8.2% 7.7% 7.2% Infrastructure 5.2% 3.0% 3.4% 3.0% 2.8% 2.4% 2.7% 3.0% Wages 5.2% 6.1% 6.5% 6.9% 7.4% 7.9% 8.4% 8.8% Nominal doubling of wages for selected categories combined with efficiency measures Education 4.7% 4.8% 4.5% 4.8% 5.0% 5.2% 5.2% 5.0% Infrastructure 5.6% 3.7% 4.3% 4.4% 4.6% 4.7% 4.5% 4.8% Wages 4.9% 5.5% 5.5% 5.5% 5.5% 5.5% 5.5% 5.5% Source: World Bank Staff estimates 2.26 The simulations in this report show that a selective approach, together with efficiency savings in the education sector in particular, is needed to protect budgetary space for infrastructure. Even without public service pay reform, the increased demand for teachers under USE will limit infrastructure spending. A general increase in public wages will aggravate this effect. The results from the simulations, in terms of spending trends for education and infrastructure, are summarized in Table 2.3. 2.27 Key findings and recommendations are: a. A substantial increase of wages is not affordable. Staff costs are a very large share of the budget. With mounting pressures on service delivery and the need to create fiscal space for infrastructure, Government should re-think its wage policy. b. The rationale for an across-the-board increase in wages is questionable. The results from comparative studies of public service wages relative to the private sector are inconclusive. They indicate that, while wages for some categories of public staff are below market equivalents, just as many are higher in the public sector than in the private sector. For example, an estimated 60 per cent of total staff on the consolidated payroll can be found in the U7 pay scale which is predominantly teachers, but also includes nurses and midwives. Wage statistics indicate that both teachers and nurses receive higher wages in the public than in the private sector. Regression analysis of wage data from the Uganda household survey provide evidence of higher wages overall in the public sector than in the private sector. While it is recognized that wage increases are warranted for some categories of staff, the above findings put into question the rationale for an across- 39 the-board increase of wages as a means to attract and retain skilled and motivated employees in the public service as a whole. Since private real wages are not increasing for some skills sets, the Government should not take lightly the claim that public wages need to increase because they have not increased recently. c. Pay-increases need to be selective if they are to be affordable. Categories to be targeted include scarce skills, for which salaries in the public service fall short of those offered by competing private employers and those available in neighboring countries. Politically motivated and selective pay enhancements, as practiced in the past, should be avoided. d. The roll-out of USE must generate efficiency savings in teacher deployment for pay reforms to be affordable. The planned increase in publicly hired secondary teachers under USE means that even without any public service pay reforms the budgets for infrastructure and service delivery may be squeezed. e. The potential efficiency gains in USE dwarf any efficiency savings that could be reasonably achieved in the health sector. Whereas Chapter 3 identifies chronic leakages in health which can also generate budgetary savings, the amount of efficiency savings on offer as part of the full implementation of USE are far greater. Setting strategic priorities for the efficiency drive is important. f. Selective pay reforms must be accompanied by efficiency savings--both in education and through better use of public staff in general. Given the number of public sector workers in education and health, better understanding of the determinants of absenteeism need to be factored into pay reforms and performance measures. g. Reducing total staff numbers is probably not feasible, given existing policy targets and the increased demand for expanded health and education services42. However, as noted in the 2007 PER43 and in Chapter 3 of this report, absenteeism in education, and especially health, is high. Better utilization and deployment of available staff budgets may limit the need to further increase staff numbers in the short-term44, but longer-term growth in the public service is inevitable given Uganda’s demography. 2.28 More empirical work is needed on staff incentives in Uganda’s public sector. Before more detailed pay reforms are agreed upon (including the proposal from MOH for the public sector to supplement the pay of private health workers), further analysis is needed to establish the extent to which pay increases are warranted by pay grade or professional groups, given comparisons with other labor market options available to workers. This work would look at public service retention and staff turnover rates (particularly for new entrants), public vs. private wages, and in-kind earnings relativities for different skills and experience. It would also have to assess under-filled and over- 42 Altogether, the education and health sectors make up three-fourths of the total staff numbers on the consolidated payroll. 43 “Uganda Fiscal Policy for Growth”, Public Expenditure Review 2007, World Bank 44 Studies on incentives and staff time use may be needed to fully understand the causes of staff absenteeism and deployment. 40 applied-for vacancies in the public sector. Each of these will provide more micro evidence of the appropriate extent of pay increases needed in different cadres of the public service. Further macro-fiscal simulations could be run once more micro work is completed on suitable pay levels. 2.29 More work is also needed to improve human resource management for service delivery in key sectors. Any wage reform needs to be accompanied by measures to improve human resource management, personnel planning, further strengthening of payroll, and the results orientation of public servants’ contracts. 2.30 It will be important to sequence pay reforms with reform of the public sector pension scheme. The proposed pay reforms are likely to have a considerable impact on pension liabilities in the Government terminal benefits pension scheme (that is, based upon years of service and non-contributory). While the necessary information on age and years’ of service to analyze these effects is presently not available to the Bank staff, it will be necessary in future analyses to model the impact of pay reforms on pension liabilities. A prudent sequence for reform would be to change the public pension scheme before any wage increases. 41 CHAPTER 3. GETTING MORE HEALTH FROM THE BUDGET 3.1 This chapter responds to the concern expressed by Uganda’s leaders that the ‘fiscal space’ for health needs to be increased to improve health, nutrition and population outcomes. The questions investigated here are: • Is there a severe shortage of funds for health in Uganda? • Can health, nutrition and population outcomes best be improved by increasing budget ceilings on health expenditures to accommodate additional external grants `in the budget’45? • Would increased financing for health translate into better results? • Can outputs be delivered for the same financing by reducing inefficiencies? 3.2 This chapter summarizes a broader health report conducted under this PER. It assesses the performance, needs and scope for additional funding for the health system in Uganda. It concludes that additional funding is unlikely to meet the rising needs of the sector, suggesting that there is both a need for and scope to identify efficiency gains. The chapter ends by identifying the main components of inefficiency in public health spending in Uganda, and makes recommendations to improve allocative and technical efficiency in the health system. 3.3 The chapter is divided into five sections. The first section provides the background and context for the sector. The second section assesses the performance of the sector in terms of key outcomes and improvements in accessibility, availability and utilization of health services. The third section looks at resource mobilization and the use of resources for health in Uganda, before tracing the main drivers of spending. The fourth section looks at the sources and extent of waste and inefficiency in the sector. The final section discusses the realities of ‘fiscal space’ in the context of Uganda, before considering the various options for getting better results. I. BACKGROUND AND CONTEXT Overview of the Administrative and Health Care Systems 3.4 The Government of Uganda has operated under a decentralized framework since 1997. With regard to the health sector, ownership of public health facilities (health centers and general hospitals) and responsibility for delivering health services was transferred to the local governments at the time. 3.5 The health system in Uganda comprises public, private-not-for-profit (PNFP) and private-for-profit (PFP) providers, as well as traditional and complementary 45 The role that off-budget aid plays is particularly important since global development aid designated specifically for health has dramatically increased in recent years. 42 practitioners. National and regional referral hospitals report to the central government, and the general hospitals together with health centers (Types II – IV) report to the local governments. The districts are further divided into health sub-districts, which are administered at the health center IV level. 3.6 The 2006 health facility inventory by the Ministry of Health reported 3,237 health facilities countrywide, of which 71 percent were public, 21 percent PNFP and 9 percent PFP. Public and PNFP facilities almost doubled between 2000 and 200646, increasing from 1630 to 2960. The increase was principally driven by the construction of new health centers by the Government in its drive to improve access to health services. Over 90 percent of PFP facilities are clinics that provide outpatient curative services. Although the health infrastructure has expanded, most facilities are not fully functional because they lack equipment and staff and suffer from poor maintenance. 3.7 Uganda faces a serious shortage of health personnel in the workforce. For every 100,000 population in the country, there are eight physicians, 55 nurses and 16 midwives47. This shortage is further compounded by mal-distribution, with a predominant urban bias in favor of the central region. This region hosts only 27 percent of the total population, yet 64 percent of all nurses and 71 percent of physicians work here.48 Health Policy, Strategy and Key Issues 3.8 The strategic goals for health care are to reduce morbidity and mortality from major illness and reduce disparities in health status among the population through the delivery of an essential public health package. The strategy employs a sector wide approach (SWAp) with seven areas: (i) mobilizing funding and improving its utilization; (2) better recruiting and deploying of staff; (3) improving supply, distribution and rational use of essential drugs; (4) increasing access; (5) implementing the health sub-district concept; (6) strengthening public private partnerships for health; and (7) improving environmental health and sanitation. 3.9 The pace of reform in the health sector has slowed. The health sub-district concept, which underpinned the reform agenda, has only been partially implemented. The majority of HC type IV remains incomplete. Implementation of key programs on reproductive health, environmental health and sanitation, and village health teams is lagging. 3.10 Uganda must improve its capacity to program external funds if it is to absorb and benefit from the Global Health Initiatives and reduce the current funding gap. Both the Global Fund for the Fight Against AIDS, Tuberculosis and Malaria (GFATM) and the Global Alliance for Vaccine Initiative (GAVI) have, at different times, suspended support to Uganda because of management concerns. 46 Ministry of Health, “Health Facility Inventory”, 2006. 47 World Health Organization (2007), Statistical Information, Geneva. 48 Ministry of Health, “Human Resource for Health Policy”, April 2006 43 3.11 Uganda is not realizing the full benefits of public-private partnerships for health or the recent recruitment drive, because of mal-distribution, high rates of absenteeism and rapid turnover of health workers. Remote and hard to reach districts, in particular, face significant obstacles in attracting and retaining health workers. Health workers from the PNFP facilities continue to migrate to Government employment because of relatively better salaries in the Government, and those who remain behind have been lobbying for public sector salaries. The balance of salary to non-salary funding at the district level is heavily skewed towards salaries. II. HEALTH STATUS 3.12 In 2006, life expectancy in Uganda was about 51 years. Life expectancy in the country had been declining since the 1970s as a result of civil strife and the impact of HIV/AIDS, but has been improving steadily since the late 1990s and is now about the Sub Saharan Africa (SSA) average. 49 3.13 Uganda’s under-five and infant mortality rates have been improving, albeit more slowly than the changes in Uganda’s economic performance and health funding would predict. In 2006, Uganda’s under-five mortality rate was 134 and its infant mortality rate was 78 per 1000 live births.50 At the rate of change shown in Figure 3.1, Uganda will not meet its 2015 millennium development goal (MDG) for under-five mortality of 75 per 1000. Figure 3.1: Trends in Under-Five and Infant Mortality Rates for Uganda, 1960 to 2006 Trends in under-five/infant mortality, 1960-2006 250 224 lityrate 0 204 0 Under-five mortality 2 185 rta 179 175 o 170 tm 160 0 156 n 5 fa 1 145 /in 133 . 134 e Infant mortality r-fiv 121 0 107 e 0 101 d 100 100 1 n 93 92 U 85 . 78 50 1960 1970 1980 1990 2000 2006 Year Source: WDI 3.14 There are significant health inequalities related to geography and socio- economic status in Uganda. The latest Demographic and Health Survey (DHS) 49 Population-weighted average for sub-Saharan Africa, excluding South Africa. 50 These are estimates reported in the World Development Indicators (WDI), 2008. The latest Demographic and Health Survey (DHS) for Uganda estimates under-five mortality to be 137 per 1000 and infant mortality to be 75 per 1000. The estimates differ slightly because WDI reports data that are adjusted for intertemporal comparability. For details, see UNICEF-WHO-World Bank-UNDP (2007), “Levels and Trends of Child Mortality in 2006: Estimates Developed by the Inter-agency Group for Child Mortality Estimation,” New York. 44 indicates that urban under-five and infant mortality rates were much lower than those in rural areas.51 Kampala had the lowest levels and internally displaced persons (IDP) camps in the northern region the highest levels of under-five and infant mortality. Wealth and higher maternal education were correlated with lower childhood mortality. Information from the DHS surveys from 1995, 2000, and 2005 reveals that inequalities in under-five and infant mortality rates have persisted, declining only slightly in 2005 relative to previous years. The modest declines in under-five and infant mortality rates observed in 2005 appear to have come largely from improvement in health outcomes for the lowest wealth quintile and, in percentage terms, the decline in under-five mortality for the poor was greater than the corresponding decline in the infant mortality rate for the poor. 3.15 Uganda’s level of under-five mortality was about average relative to its income level, but low relative to the total spending on health. Life expectancy was slightly below average and under-five mortality rates were just about average relative to Uganda’s income level in 2006. However, several countries with comparable and even lower per capita health expenditures have child mortality rates that are significantly lower, suggesting that Uganda’s health system could be performing better in reducing infant mortality (see Table 3.1). Table 3.1: Health Expenditure and Outcomes--Comparisons across Countries Total health Government health Under-five Infant expenditure expenditure per mortality mortality Country per capita US$ capita US$ (per 1000) (per 1000) Bangladesh 13 5 69 52 Cambodia 30 8 82 65 Ethiopia 7 4 123 77 Ghana 35 13 120 76 India 39 8 76 57 Kenya 29 14 121 79 Lao PDR 22 5 75 59 Madagascar 9 6 115 72 Nepal 17 5 59 46 Senegal 40 13 116 60 Tanzania 18 11 118 74 Uganda 25 7 134 78 Vietnam 46 15 17 15 Zambia 49 23 182 102 Sub-Saharan Africa 28 13 156 93 Low-income 30 8 98 67 Source: World Bank Staff estimates 3.16 With 435 deaths per 100,000 live births, Uganda has one of the highest rates of maternal mortality in the world.52 Things have improved slightly since the 1990s when maternal mortality exceeded 500 per 100,000 births. If Uganda is to attain the 51 Uganda Bureau of Statistics (UBOS) and Macro International (2007), Uganda Demographic and Health Survey 2006, Calverton, Maryland, USA 52 UBOS and Macro International (2007). 45 MDG target of a three-fourths reduction over the 1990-2015 period, it must reduce its maternal mortality rate to 131 per 100,000 live births by 2015. Poor access to quality maternal care services53, particularly emergency obstetric care, blocks attainment of improvements in maternal mortality rates in Uganda High fertility, teenage pregnancy and the lack of birth spacing are also challenges to reducing maternal mortality. III. POPULATION AND FERTILITY 3.17 Uganda has the third highest fertility rate in the world which has remained broadly unchanged since 1960. In 2006, Uganda’s total fertility rate (TFR) was 6.7 births per woman--the sixth highest fertility rate in the world, just behind Liberia (6.8 births) and Niger (7 births per mother). In comparison, sub-Saharan Africa had an average of 4.9 births per woman in 2005. Uganda’s high TFR correlates with high infant and maternal mortality rates. Rural women, poor women, and women with little education have the highest TFR in Uganda.54 3.18 Uganda has one of the youngest populations in the world, with over half the total population under 15 years of age. In 2005, the median age of the population was 14.8 years.55 Progressions to 2025 indicate that the population pyramid will remain relatively stable in Uganda. Even in 2025, the percent of Ugandans less than 15 years old is projected to be approximately 49 percent of the total population. This would place a large burden on child and adolescent health care. The dependency ratio was also relatively high at 112.4 in 2005. This high ratio mostly signifies many unsupported children--including orphans who have to depend on others. The dependency ratio is projected to decrease slightly to 103.5 by 2025. 3.19 Communicable diseases account for a large proportion of deaths in Uganda and the prevalence of HIV/AIDS, which alone accounted for 24 percent of all deaths in 2002, remains significant. HIV/AIDS, malaria, and respiratory infections were prominent causes of mortality, as were diarrheal diseases and childhood-cluster diseases in 2002 when the last assessment of the burden of disease was conducted. With HIV prevalence estimates at 6.4 percent, Uganda has come a long way from an 18 percent prevalence rate in the early 1990s. Current estimates indicate that over 100,000 new HIV infections take place yearly. Recent evidence reveals a worrying trend of apparent reversal in the uptake and practice of safer and preventive sexual behavior in the general population. 3.20 These three diseases are also the top three causes of the overall disease burden in terms of disability-adjusted life years (DALYs) lost.56 DALYs due to non-communicable diseases (NCDs) and accidents are also prominent. NCDs and unintentional diseases, 53 UNDP (2007), Millennium Development Goals: Uganda’s Progress Report 2007, Kampala: United Nations Development Program. 54 World Bank (2005), Improving Health Outcomes for the Poor in Uganda, Washington, DC: Africa Region Human Development Working Paper Series, 42-44. 55 May, J, P Mpuga, and JPM Ntozi (2007), “Uganda’s High Population Growth: A Challenge to Improved Livelihoods,” Working Paper, Washington, DC: World Bank. 56 Disability-adjusted life years (DALYs) are a metric for combining healthy time loss from morbidity as well as premature mortality. 46 including neuropsychiatric conditions and nutritional deficiencies, comprised 30 percent of the overall disease burden in 2002. Malaria is the most common cause of death in children under five in Uganda, accounting for about one-third of the under-five deaths, followed by prenatal and early neonatal conditions, meningitis and pneumonia. IV. ASSESSMENT OF HEALTH SECTOR PERFORMANCE: HEALTH CARE UTILIZATION, COVERAGE, AND OUTPUTS 3.21 Utilization of health care facilities has been rising in Uganda, increasing from below 60 percent in 1996 to almost 88 percent in 2006.57 Government facility utilization rates have increased since the abolition of user fees in 2001. Somewhat surprisingly, private utilization also increased following the abolition of user fees at Government facilities in 2001. One explanation could be that drugs are not always available in public clinics. Those seeking care in the public sector reported to the UNHS III survey that they paid similar fees for treatment to those who went private. Reasons for non-use of health care include cost (32 percent) and distance from health facilities (10 percent). The top reason offered for not attending public clinics by households in the 2004 National Service Delivery Survey was lack of drugs (30 percent). 3.22 Even though private clinical facilities are fewer than public health facilities in Uganda, more people seek health care in the private rather than public sector. An analysis of the 2006 National Household Survey showed that about 46 percent of the population that sought health care did so at private clinics, compared with about 35 percent that went to mostly Government facilities. Conversely, 76 percent of health facilities in Uganda are owned or managed by the Government, while only 24 percent of the health facilities are private or NGO-managed. With respect to coverage, distribution of facilities across the country closely mirrors the distribution of the population, with a few exceptions. 3.23 Uganda has a mixed record with trends in coverage. Indicators such as DPT3 immunization rates have risen significantly. In 2006, 84 percent of children 12-23 months old had DPT3 immunization in Uganda, a performance that is reasonably good relative to other countries. Skilled birth attendance, an important factor in reducing maternal mortality, is around 42 percent in Uganda, lower than the 51 percent average for sub-Saharan African countries. This has remained virtually unchanged since the beginning of the 1990s. Contraceptive use in Uganda is on the low side relative to other comparator countries. Access and Availability of Health Care Services 3.24 Access to health facilities has been steadily increasing. In 1999, the average reported distance to a health facility or hospital was 5.56 kms, with 75.4 percent of respondents reporting that they lived within 5 kms of a health facility or hospital.58 By 57 This includes both inpatient and outpatient utilization rates. The household survey data do not consistently differentiate between the two over time. 58 It is important to note that the average distance calculations are based on information provided by individuals who reported utilizing health care. It does not take into account segments of the population that 47 2006, the average distance had dropped to 4.11 kms and 82.5 percent reported living within 5 kms of a health facility or hospital. Interestingly, richer quintiles reported themselves as being further away from health facilities than the poorest quintiles. The reported distance to health facilities was highest in the western region (5.0 kms) and lowest in the northern region (3.5 kms). 3.25 Over two-thirds of health facilities provide basic packages. On average, curative care for sick children and STI services are available in all facilities. The availability of other services ranges from 65 percent for growth monitoring to 88 percent for child immunization.59 ‘Health center IVs’ provide the most comprehensive services to patients, while ‘health center IIs’ have more limited capabilities. The distribution of observed consultations across the four levels of health facilities mirrors the service availability patterns. 3.26 Nevertheless, despite these improvements, access to health services remains problematic. In the 2006 UDHS, 86 percent of all women reported encountering at least one serious problem in accessing health care. 65 percent of women reported lack of money to pay for treatment as a constraint to seeking treatment. Other problems included distance (55 percent), having to take and pay for transport (49 percent), concern over availability of medications (46 percent), concern that there would be no provider (27 percent), and not wanting to go alone (27 percent). There are regional variations in these numbers, with 61.9 percent of women in Kampala citing at least one problem in accessing health care, and 90.8 and 94.8 percent of women in the southwest and north respectively reporting having more than one problem in accessing health care. 3.27 Access to prevention and treatment of malaria must improve. Malaria is highly endemic, with 63 percent of the population exposed to high transmission levels and 25 percent exposed to moderate transmission levels. In 200506, half of the population that fell sick in Uganda reported malaria or fever as the major sickness.60 Only 34 percent of the households interviewed in the 2006 DHS survey reported having mosquito nets, with 15 percent having more than one. There is a large urban-rural divide, with 60.6 percent of urban households having at least one mosquito net compared with only 29.4 percent of rural households. Approximately 99 percent of health facilities report offering malaria treatment; however the ability to access these services is not universal and drugs often stock-out. The 2006 UDHS found that in the two weeks prior to the Survey, 41 percent of children had fever. 61 percent of the children with fever received anti-malarial drugs but only 29 percent received the drugs within the recommended 24 to 48 hours of the onset of the fever.61 did not utilize available health care due to distance. Therefore, this estimate might be lower than the situation on the ground. 59 Uganda Ministry of Health, Uganda Bureau of Statistics, and Macro International Inc. (March 2008). “Uganda Service Provision Assessment Survey 2007.” 60 Uganda Bureau of Statistics (UBOS) and Macro International (2007), Uganda Demographic and Health Survey 1995, 2000 and 2006, Calverton, Maryland, USA: UBOS and Macro International Inc. 61 Ibid. There are also frequent reports that stock-outs force physicians to incorrectly treat malaria, leading to increased resistance by the parasite against treatment. 48 3.28 About 30 percent of the health facilities, mostly hospitals and health centers reported having HIV/AIDS testing facilities. Beyond the initial testing availability, approximately six in ten health facilities have HIV/AIDS care and support services.62 Given the increasingly large presence of HIV/AIDS related NGOs, most of them funded through US President’s Emergency Plan for AIDS Relief (PEPFAR), this percentage is not necessarily the result of the Uganda Government’s direct intervention. In 2007/08, PEPFAR’s contribution was estimated at 76 percent of the total HIV/AIDS expenditure of US $295 million63. 3.29 Uganda has made great progress in treating tuberculosis (TB). The current focus is first on diagnosing those with TB, and then on targeting treatment. In 2006, there were an estimated 80,000 new cases of TB, however only 41,927 were notified. The Government is committed to offering counseling and testing services to larger numbers of TB patients at the district level. In 2005 this ratio was 40 percent, and the Ministry of Health set a target to 60 percent in 2006, which they were able to exceed to reach the reported rate of 75 percent. Equity of Access 3.30 The poor in Uganda were less likely than the other quintiles (except the richest) to report health problems and seek health care when sick. About 37.7 percent of the population in the poorest quintile reported health problems and 15.8 percent did not seek care. Among the richest, about 36.8 percent reported health problems but only 7.9 percent did not seek care. About 35.1 percent of those not seeking care among the poor gave “too expensive” as the reason for not seeking care as opposed to 16.8 percent of the richest who did not seek care for the same reason. Not surprisingly, the poorest quintile was most likely to seek care in Government health units (34.4 percent) and the richest quintile was most likely to seek care in private health facilities (58.8 percent). 3.31 The patterns in health care utilization in Uganda suggest that Government health centers are pro-poor but Government hospitals are not. Data from the 2006 National Household Survey show that among the population that sought health care in Government health centers, 28.9 percent were from the poorest quintile and only 10.4 percent were from the richest quintile. With regard to Government hospitals, however, the two richest quintiles were responsible for almost 50 percent of the overall utilization patterns. The poorest quintile accounted for only 16.9 percent of the utilization patterns * The Uganda Service Provision Assessment Survey 2007 does not differentiate between Central 1 and Central 2, therefore for comparison purposes an average of Central 1 and Central 2 is taken from the UDHS 2006. 62 HIV/AIDS care and support services are defined as the provision of curative or preventive care for illnesses that may be HIV/AIDS related, such as treatment of opportunistic infections including TB, STIs, and malaria; the provision of, or referrals for counseling for social support services for help in living with HIV/AIDS. 63 Uganda AIDS Commission. Report on Implementation of National HIV and AIDS Strategic Plan FY 2007/08. 49 in Government hospitals.64 The patterns were similar for NGO health units versus hospitals, with the poor mostly using the former and the higher income groups largely benefiting from the latter. V. TOTAL HEALTH EXPENDITURES 3.32 Uganda spent a greater share of its GDP on health than comparator countries, but the Government’s share in health spending was lower than the sub- Saharan African average in 2006. Uganda’s total health expenditure per capita was about US$25 in 2006. In 2006, health spending was financed from three main sources: external funding (29 percent); Government (30 percent); and out-of pocket (38 percent). This total was slightly higher than the average for sub-Saharan Africa, but slightly lower than that for all low-income countries (see Table 3.2). Excluding aid flowing into the health sector, Government spending comprised only 5 percent of overall public spending, much lower than the Abuja Declaration 15 percent target. Government Health Expenditures 3.33 In nominal terms, the health sector budget (including aid as a percentage of the total Government budget) grew from 7 percent in 1997/98 to 12 percent in 2001/02. It has remained fairly constant at this level since but the increases are not reflected in real figures. However, the picture changes dramatically in real per capita terms. In particular, Uganda’s Government health spending per capita rose in nominal terms from over USh 10,000 in 2000 to over USh 17,000 in 2005, and then declined back down to about USh 12,000 in 2006. In real terms, per capita Government health spending in 2006 was less than in 2000. 3.34 There has been an increase in the share of the health budget for district health services with a concurrent decline in the share for Ministry of Health headquarters. Consistent with the Government policy of decentralized service delivery, between 1997/98 and 2006/07, the share of district budgets increased from 16 percent of the total health budget to 35 percent. 64 Participation incidence is only indicative; a full benefit incidence analysis should be done to assess the incidence of Government health spending in Uganda. 50 Table 3.2: Health Expenditure Indicators: Uganda & Comparator Countries, 2006 Country/Region GNI per Total health Total health Govt health Govt health capita expenditure expenditure expenditure expenditure (US$) per capita (% of GDP) (% of total health (% of overall (US$) expenditure) Govt budget) Bangladesh 450 13 3.1 36.8 7.4 Cambodia 490 30 6.0 26.1 10.7 Cameroon 990 51 5.2 28.1 8.6 Ethiopia 170 7 4.9 60.4 10.6 Ghana 510 35 6.2 36.5 6.8 India 820 39 4.9 19.6 3.4 Kenya 580 29 4.6 48.2 6.1 Lao PDR 500 22 3.6 20.8 4.1 Madagascar 280 9 3.3 63.9 9.6 Nepal 320 17 5.7 30.5 9.2 Senegal 760 40 5.4 31.5 6.7 Tanzania 350 18 5.5 59.2 13.3 Uganda 300 25 7.2 26.9 10.0 Vietnam 700 46 6.6 32.4 6.8 Zambia 630 49 5.2 46.8 10.8 Sub-Saharan 466 24 4.8 43.9 8.3 Africa Low-Income 591 27 4.6 28.2 5.2 Source: World Development Indicators and World Health Organization 3.35 All of the increase in district funds was through the conditional grant to primary health care salaries (PHC). Non-wage spending on public clinics at districts fell even in nominal terms. The budget for PHC wages increased by two- thirds in nominal terms between 2003/04 and 20067/07, from USh 44 billion in FY 2003/04 to USh 75 billion in FY 2006/07. On the other hand, non-wage recurrent spending declined each year from USh 33.5 billion in FY 2003/04 to USh 33.5 billion in FY 2006/07. Grants for district PHC buildings fell too (Table 3.3, and Figures 1.13 and 1.14). Table 3.3: Budgets at Uganda’s Frontline Public Health Facilities 2003/04 to 2006/07 District Budgets, (USh 000) 2003/4 2004/05 2005/06 2006/07 District Public Health Wage 44,673,668 68,847,077 72,305,800 74,619,609 District Public Health Non-Wage 33,518,106 32,893,609 32,472,425 33,523,452 District PHC Buildings 9,202,800 6,095,000 5,921,011 6,095,693 Referral Hospitals Budgets, (USh 000) 2003/04 2004/05 2005/06 2006/07 Personnel-related Costs 24,821,265 37,073,227 36,959,543 38,160,996 Buildings, Equipment, Maintenance 27,026,606 23,279,361 49,962,998 12,104,977 Drugs and Medical Supplies 7,351,151 6,898,322 8,276,437 8,206,578 Arrears 23,000 950,179 326,983 4,744,597 Utilities 2,102,694 2,093,340 2,155,287 2,204,762 Other Admin. & Running Costs 4,501,923 7,200,915 6,298,494 5,586,417 Source: Public Expenditure Database, World Bank Note: (i) District budgets exclude drugs and medical supplies from headquarters (budget for which has fallen in this period). (ii) Personnel-related costs in hospitals include travel. (iii) Referral hospital totals include donor assistance; Africa Development Fund assistance for remodeling Butabika Hospital in 2005/06 accounts for the increase. 51 3.36 Over 85 percent of the health sector budget is earmarked. In addition to existing earmarks for wages and PHC conditional grants, new funding from the Global Health Initiatives is earmarked to specific diseases. In the past two fiscal years, FY06/07 and FY07/08, new funding from the Government was provided for already earmarked specific activities Private and Household Spending 3.37 Out-of-pocket health expenditures--at 9 percent of total household expenditures and 37 percent of total health spending--are a prominent source of financing for the health sector. In comparison to other countries, out-of-pocket expenditures in Uganda were lower than the average for sub-Saharan Africa and for low- income countries. Detailed analysis of the National Household Survey data from 2006 indicate that, on average, almost 28 percent of sampled households faced health expenditures that could be deemed “catastrophic”, that is, health expenditures that were more than 10 percent of total household consumption expenditure. Those residing in the western region spent the largest share of their incomes on health, and also had the highest incidence of catastrophic health spending (38.1 percent). Moreover, catastrophic health expenditure among the poor has increased even after the elimination of user fees in 2001 (Figure 3.2). Figure 3.2: Health Expenditure as a Proportion of Total Expenditure and Catastrophic Health Expenditure in Uganda, 1996 to 2005 Health's proportion of total expenditure Catastrophic health expenditure 30 30 User fees eliminated User fees eliminated 25 25 Richest quintile 20 20 15 15 10 10 Poorest quintile Richest quintile 5 5 Poorest quintile 0 0 1996 1998 2000 2002 2004 2006 1996 1998 2000 2002 2004 2006 Year Year Source: UNHS (various years) Development Assistance for Health in Uganda 3.38 External funding is a prominent source of health expenditure in Uganda, financing about 28.5 percent of total health spending in 2006. External funding in Uganda was much higher than the average for sub-Saharan Africa and for low-income 52 countries in general. In 2006, total Government spending on health was US$ 184 million, while external resources amounted to US$ 194 million.65 3.39 Driven by the off-budget part, development assistance for health (DAH) has been steadily increasing in Uganda in recent years to a high of US$ 460m in 2006. This reflects inflow support from the US Government (PEPFAR and PMI) which is largely off-budget. While on-budget DAH trends have remained stable, off-budget spending on health as a percentage of total off-budget spending increased from 9 percent in FY 2005/06 to 14 percent in FY 2007/0866. 3.40 The planning, budgeting and monitoring of on- and off-budget projects and programs is weak. This weakness, despite the existence of a health SWAp, raises concerns about the reliability of donor budgetary information in the Budget Framework Paper (BFP) and MTEF. The weaknesses arise partly from how the concept of on- and off-budget is applied where the aid architecture for health has dramatically changed. As a result, there are cases of donor project support which ought to be on-budget but are excluded and, conversely, of donor support that ought to be excluded but is included on- budget. A survey by the Ministry of Health in 2005/06 revealed that 56 percent of project funds were spent on non HSSP inputs, most of them on overheads or costs unrelated to health. Figure 3.3: Official Development Assistance for Health to Uganda, 2000 to 2006 Donor assistance for health, 2000-2006 600 Donor assistance for health in Uganda Ugandan Shillings (millions) 400 Total 200 On-budget Off-budget 0 2000 2001 2002 2003 2004 2005 2006 Year Source: MoF 3.41 The picture that emerges is of relatively high donor funding getting average health results in Uganda. Total health spending is relatively high compared to other countries. Public and out-of-pocket spending is relatively low. However, aid is considerably higher compared to other countries, accounting for the overall relatively high spending. Outcomes are just average given Uganda’s income level, and the change in outcomes over the past decade could be interpreted as being disappointing given improvements in access, education, water and sanitation, and average incomes. More 65 World Health Organization data. 66 FY 2007/08 Budget Speech Table 7 53 analysis is needed to disentangle the role of public health services in achieving the improvement in health outcomes. VI. COST DRIVERS Growing Population 3.42 Uganda’s high fertility rate will continue to drive health expenditures upwards. Even in the most optimistic scenario, Uganda’s population will keep on growing at a rate of over 3 percent in the next generation. The young population and high TFR will place a huge demand for maternal and child health services, as well as adolescent services, for Uganda to achieve and sustain its service targets. 3.43 Uganda has also started to experience a rise in non-communicable diseases associated with ageing67. Health care services in the country for non-communicable diseases are under-developed. Uganda must continue expanding maternal and child health services while developing the capacity of the health system to cater for non- communicable diseases. Service Standards and Norms 3.44 The Government has committed itself to service standards and continuing to expand health infrastructure. Together with the demand for construction of new district hospitals, these commitments have serious implications for the Development and Recurrent Budgets. The introduction of the ‘health sub-district’ concept has implications for staffing, equipment, infrastructure, and operating costs. There are already serious concerns about the status of existing health infrastructure in terms of completion, functionality and condition. According to the Ministry of Health, less than 40 percent of ‘health clinic IVs’ are fully functional, only 33 percent of general hospitals are properly equipped, and only 30 percent of health staff currently reside in institutional accommodations. Rising Unit Costs of Treatment 3.45 The unit costs of delivering health services are increasing. Even without the expansion of coverage, the unit costs of existing services are rising due to medical supplies inflation, administrative wages, and resistance to treatment regimens. 3.46 Growing resistance to the existing treatment for malaria (and more recently for TB), is forcing Uganda to adopt more expensive treatments. Although replacing existing anti-malarials with artemisinin-based combination therapy (ACT) does not increase demands on implementation capacity, it drastically increases the unit costs of treating each case of malaria. The drug cost per treatment of an adult malaria case using Artemisinin-based combination therapy (ACT) is estimated at US$ 1.80, which is more than three times the cost of treatment with a chloroquine (CQ) and sulphadoxine- pyrimethamine (SP) combination. 67 Especially cancer of the cervix and breast, diabetes and cardiovascular diseases. 54 Global Pressures to Increase Health Expenditure 3.47 Uganda’s health sector is subject to global pressures to expand services and adopt new technologies. Over the past 15 years, the world health community has created several global imperatives. With respect to adopting new technologies, these are potentially more cost-effective (for example, pentavalent vaccines), but they are all more expensive. 3.48 The Global Alliance for Vaccines and Immunization (GAVI) is expected to contribute US$ 205 million to support Uganda’s immunization program between 2002 and 201568, of which US$ 175 million will go towards procuring pentavalent vaccines. While bringing in substantial health gains, the adoption of pentavalent vaccines has tripled the cost of procuring vaccines in Uganda. Uganda is expected to gradually absorb the cost of immunization. HIV/AIDS Remains a Major Driver of Health Spending 3.49 The HIV/AIDS epidemic is the most important of health cost drivers and has the most significant cost implications. According to the costing of the National AIDS Strategic Plan, Uganda will need to mobilize between US$ 263 million and US$ 361 million from 2007/08 to 2011/12 to fund its national HIV/AIDS program. Uganda’s efforts are likely to be unsustainable unless it improves prevention; already 350,000 persons need antiretroviral therapy and over 100,000 new infections occur annually. VII. WASTE AND INEFFICIENCY IN THE HEALTH SECTOR 3.50 Waste in public spending is a significant and deep problem in the health sector. This report assesses waste in public spending in the health sector by reviewing: (a) studies tracking the flow of funds; (b) questionable expenditures69 from audit reports by the Office of the Auditor General and Public Procurement and Disposal of Public Assets; (c) personnel behavior and management practices; and (d) procurement of medicines and supplies. Tables 3.4 and 3.5 provide very approximate but quite startling results, after adding up the possible values of leakage in the health budget. At an upper limit, up to 42 percent of the public health budget for facilities might not be reflected in patient care at public health centers. Absenteeism: Personnel and Management Practices 3.51 Staff absenteeism is the greatest source of waste in the sector, and is compounded to a lesser extent, by ghost workers and drug leakage. During unannounced visits to a sample of Government health facilities in 2006, 52 percent of health workers who were supposed to be working were absent70. An earlier study71, 68 GAVI Fact Sheet on Uganda, June 2008 69 In this paper, ‘questionable expenditures’ refers to expenditures that are reported in the annual Auditor General’s reports, on account of which the Auditor General gives a ‘qualified opinion except for…’ or a ‘qualified opinion, disclaimer’. 70 Bjorkman and Svensson (2007), Efficiency and Demand for Health Services: Survey Evidence on Public and Private Providers of Primary Health Care in Uganda 55 using a survey from 2002/03, found 37 percent of health workers to be absent. Assuming absenteeism was on average 40 percent as suggested in the recent DHS facility survey, absenteeism alone could have cost the Government of Uganda around USh 45 billion in 2006/07. Regarding ghost workers (those who are inactive but inappropriately appear on the payroll), the payroll audit of 2005/06 found that 1.5 percent of health workers were ghost workers. This suggests that a further USh 1.6 billion (1.5 percent of wage spending in FY 2006/07) may have been spent on ghost workers. Table 3.4: Estimating ‘Waste’ from Possible Sources, as Share of Budget Items Problem area Source 1990s 2001 2002 2003 2004 2005 2006 Ghost workers Payroll clean-up exercises .. .. .. .. 1% "Questionable expenditures" Auditor General's Reports .. .. 4% .. 5% 4% 2% Conditional Transfer to PHC leakages PETS (PHC Transfer) 14% 10% .. 14% 14% 14% NGO PHC Grant Leaks PETS PETS .. .. .. ... 21% 20% 16% Medical Supplies 93% Health Worker absent from facility WDR 2004 methodology .. .. .. 37% .. .. 52% Table 3.5: Very Approximate ‘Waste’ Estimates for FY 2006/07 Problem area: USh. billion Ghost workers 1.14 "Questionable expenditures" 2.09 Conditional Transfer to PHC leakages 3.21 NGO PHC Grant leakages PETS 2.84 Medical Supplies (assuming Bjorkman & Svensson results of 93% leakage) 22.25 52% Health Worker absent from Bjorkman & Svensson * salaries 59.13 Total 90.66 Share of Total Budget 2006/07 42% Waste and Leakage in the Pharmaceutical Sub-sector 3.52 Pharmaceuticals are the second major health expenditures after staff. Drugs are exposed to risk of waste. Waste in pharmaceuticals may be through direct drug 71 Chaudhury, Nazmul, Jeffrey S. Hammer, Michael Kremer, Karthik Muralidharan, and F. Halsey Rogers. 2006. "Missing in Action: Teacher and Health Worker Absence in Developing Countries." Journal of Economic Perspectives 20(1): 91-116 56 leakages (theft), poor procurement and supply management leading to expiry of drugs in stores, and poor prescription practices. 3.53 Essential drugs in both National Medical Stores72 and Joint Medical Stores are less expensive than the international reference prices by 85 and 72 percent respectively73. 3.54 Third party procurements have grown, straining the capacity of National Medical Stores (NMS). In 2005, the value of third party stock (USh 17. 2 billion) exceeded the value of NMS stock trading (USh 5.5 billion) by three to one74. Eighty four percent of warehouse space is being rented by NMS to store third party commodities at a cost of US$ 33,000 per month. While NMS clearly needs more storage space, it can reduce current warehousing space through better coordination of third party procurements and management of business workflow processes. 3.55 The value of expired drugs written-off has increased. Between 2005 and 2007 alone, drugs valued at US$ 2.4 million were written-off, 82 percent of them are third party related procurements. It is estimated that Uganda will continue to lose drugs through write-offs worth over USh 1.25 billion, for a total stock level estimated at USh 50 billion for NMS. 3.56 Leakage of drugs occurs between NMS and the facility level. Integrity surveys and media reports show drug leakage from facilities. A study of the flow of drugs and stock-outs across facilities (in a community monitoring project) found that stock-outs were significantly less common in the treatment group than in those facilities that were not monitored.75 3.57 Some 14 percent of the district PHC grant did not reach the lower levels in FYs 2003/04, 2004/05 and 2005/0676. Leakages were also found in grants to NGOs although, unlike the PHC grant, there had been improvements in the proportion of the funds reaching the NGO health facilities, increasing from 79 percent to 86 percent between FY2003/04 and FY2005/06. 3.58 The Auditor General’s report for FYs 2003/04 to FY 2005/06 with respect to Government Financial Accounts for the health sector, found cases of questionable expenditures with respect to payments-- no supporting documentation, unrecovered advances and payments for undelivered goods or services, etc. The highlighted problems related to areas of financial mismanagement, improper procurement, breach of 72 National Medical Stores is mandated to supply drugs to Government providers; meanwhile Joint Medical Stores supplies the PNFP providers. 73 Uganda Medicine Pricing Survey Report. April 2004. 74 Policy Review of the Role of the National Medical Stores in the Public and Private Health Care System in Uganda, 2006 75 Bjorkman and Svensson (2007) used clinical and stock records in facilities to assess the correlation between NMS records on what was provided, and clinic records of what was used and in stock. Their R- squared statistic was 0.07 (7 percent explanatory power in the correlation between what was sent and what was used or in stock). 76 Ministry of Health. Assessment of Public Expenditure on Health in Uganda, 2007. 57 regulations, and misuse of facilities by staff among others. The lack of supporting documentation represents a breakdown of controls in the payment system, casts doubt on the authenticity of the payments and suggests possible fraud. The Auditor General’s reports for FY 2004/05 and 2005/06 further highlight a serious laxity in requiring individuals to account for advances and not refunding funds borrowed from projects, which weakens accountability and may encourage the misuse of public resources. VIII. OPTIONS FOR INCREASING FISCAL SPACE FOR HEALTH IN UGANDA Elasticity of Government Health Expenditure to Economic Growth 3.59 The estimated elasticity of Government health spending to GDP for Uganda depends on whether donor funds are included. Government expenditure in total and for health rises with GDP. Estimates from 2000 to 2006 suggest that the elasticity of Government health expenditure with respect to GDP, when donor funds are included, is about 1.44 in Uganda, implying that a growth of 1 percent in GDP is associated with a rise in Government health spending of about 1.44 percent on average (Figure 3.4). This compares favorably with the global average of 1.2. However, for domestic-financed health spending (excluding external grants), the elasticity is only about 0.95 (Figure 3.4). Fiscal space for health derived from economic growth projections in Uganda would depend on global funding rising with Uganda’s GDP (which seems unlikely), or on the extent to which Uganda’s domestic resources can rise even faster to substitute for global funds and keep the health budget rising with GDP. 3.60 If the elasticity with respect to GDP does not fall significantly, Uganda can expect robust endogenous increases in Government health spending given its economic growth prospects. Once the impact of the global crisis rescinds, e.g. the IMF projects economic growth in Uganda could continue at 6-7 percent per year in real terms at least for the next 5-7 years.77 Assuming that Government health spending will respond in the same way to growth as it has over 2000-2006, these economic growth projections would imply a doubling of nominal Government per capita health expenditure over the period 2007-2015 (from USh 19,453 to USh 41,214), going from 3.1 percent to 4.1 percent of GDP. 77 IMF (2007), Uganda: Article IV Consultation, Washington, DC: International Monetary Fund. 58 Figure 3.4: Elasticity of Government Health Expenditure with Respect to GDP in Uganda, 2000-2006 Government health expenditure vs GDP, 2000-2006 800 2006 2005 600 Government health expenditure (Billions of Ugandan Shillings) Total 2004 400 2002 2003 2000 2001 2006 2004 2005 200 2003 2002 Excluding donor funds 2001 2000 8000 10000 12000 14000 16000 18000 GDP (Billions of Ugandan Shillings) Source: MoF; Log scale 3.61 Another source of fiscal space for health expenditures would be from increasing the overall Government budget, while keeping the health’s share in the budget unchanged. In 2005, WHO estimates indicated that Uganda’s Government health expenditure was about 10 percent of the overall budget. 3.62 One constraint to increasing Government spending is Uganda’s revenue- generation capacity. Although the IMF projects that revenues as a share of GDP in Uganda will increase from about 11 percent in 2005 to 14 percent in 2012, it is not clear whether or not this can translate into increases in the overall Government budget as the grant-financed proportion of the budget is projected to decline over the same period.78 3.63 Can Uganda increase the share of its Government budget devoted to health? Reprioritizing health spending at the expense of other sectors seems unlikely. It is not clear which other sector budgets can feasibly be cut in order to increase allocations to health. Government policy has emphasized fiscal consolidation, whilst agriculture, energy, roads and USE are each identified as priorities in the coming years. As a share of its budget, health spending in Uganda is slightly higher than the average for low-income countries as well as for sub-Saharan African countries. At least for its income level, the amount Uganda spends on health is about the norm as a share of the budget (but slightly lower in per capita terms). 3.64 Can Uganda increase fiscal space for health using additional external financing? Globally, development assistance for health has risen significantly over the past few decades, from an estimated US$ 2.5 billion in 1990 to US$ 13 billion in 2005. Given the increased availability of global funds, can Uganda increase its dependence on external funds for generating fiscal space for health? Much of Uganda’s health financing is already external. The WHO estimates that about 35 percent of total health spending in 78 IMF (2007). 59 Uganda is externally financed. This is more than double the average in sub-Saharan Africa and more than five times the average for low-income countries. It seems unlikely that Uganda could absorb additional external funding for health, given that it is already quite heavily dependent on donor financing. Furthermore, external funds come with their own set of aid management problems: they can be unpredictable and unsustainable and can create contingent liabilities for the Government in the event that donor funding declines in the near future. 3.65 The composition of external assistance for health for Uganda is skewed towards HIV/AIDS. Although the prevalence of HIV/AIDS is relatively high in Uganda, there are other diseases such as malaria that are responsible for a significant burden of disease and that have not received as much attention from external sources. Furthermore, tying the flow of aid to a particular disease such as HIV/AIDS has the potential of distorting the allocative efficiency of spending. 3.66 Uganda might be unable to absorb additional external funds into the health sector but there is room to increase the efficiency of external resource flows. Some of the starkest absorptive capacity constraints in Uganda are evidenced from its human resources for health numbers--0.08 doctors per 1000, 0.55 nurses per 1000, and 0.16 midwives per 1000, based on 2004 data. These numbers are far lower than the minimum 2.3 health workers per 1000 recommended by the WHO. Further, the potential volatility and unpredictability of external funds make it difficult to plan for the health budget based on commitments, more so when the greater share of the sector’s budget is externally financed. In Uganda, disbursements of official development assistance for health have been steadily rising between 1996 and 2006, but commitments of funds have been quite volatile. 3.67 The introduction of social insurance is theoretically an option for raising additional revenues for the health sector. Ghana and Kenya have introduced social health insurance (SHI) to supplement government financing of health care. However, given the existence of large informal sectors, as well as the administrative and managerial challenges, it is not clear how feasible and effective the introduction of SHI is for generating additional fiscal space for health in poor countries.79 3.68 Uganda has considered phasing in SHI in Uganda, with initial plans to cover only formal sector workers. The plan is for an initial coverage of all formal sector employees, which is a very small percentage of Uganda’s overall labor force. Estimates suggest that only 300,000 Government employees and 100,000 private sector employees constitute the formal sector in Uganda.80 The SHI plan is to have a mandated 4 percent contribution by employees, combined with a matched 4 percent employer contribution. It 79 Wagstaff, A (2007), “Social Health Insurance Reexamined,” Policy Research Working Paper No. 4111, Washington, DC: World Bank. 80 New Vision (2007), “Insurers Wary of National Health Scheme,” October 24, 2007. 60 is important to note that the latter would require additional fiscal space on the part of the Government as it is the employer of a large proportion of formal sector workers.81 3.69 The best option for generating more health outputs in Uganda would seem to be through improved efficiency of Government spending rather than increasing Government spending. Although the argument is sometimes made that improvements in efficiency by themselves require some initial increase in outlays, it is important to assess if there are differentials in the efficiency of the Ugandan health system--which could be linked to health service management--by comparing performance across districts and/or regions within the country. 3.70 There is great variation in health outcomes within Uganda--almost as wide as across the world itself. Though average life expectancy in Uganda is about 51 years, some districts such as Kapchorwa and Bukwa have life expectancy rates that are over 60 years, and other districts such as Kitgum have life expectancy rates that are lower than 30 years. This suggests the possibility that some of this variation is related to allocative or technical efficiency of health service delivery. Similar variations in infant and child mortality rates can be also observed across Uganda. Figure 3.5 puts these variations in a global perspective. Figure 3.5: Life Expectancy: Global Comparison of Ugandan Districts, 2006 Life expectancy: Global comparison of Ugandan districts, 2006 Japan 80 70 Bolivia India Life expectancy Bangladesh Senegal Yemen, Rep. 60 Kapchorwa Bukwa Cambodia Kotido Kaabong 50 Bushenyi Adjumani Koboko Arua Oyam Apac 40 Swaziland 30 Kitgum Uganda districts Other countries Source: UNDP & WDI 3.71 More analysis using DHS data may be needed to determine the importance of non-health sector factors on health outcomes in Uganda82. The Ministry of Health produces a District League Table (DLT) that ranks districts based on the performance of their health systems. Based on the DLT, the top performing districts include Gulu, Jinja, and Mbarara.83 There is practically no rank correlation between the rankings of districts based on the DLT and the rankings based on health outcome measures such as life expectancy or with general human development measures such as the Human 81 Estimates from a World Bank mission in 2007 indicated that this would amount to USh 69 billion, 8 percent of the total Government wage bill of USh 860 billion. 82 For instance, nutrition, water and sanitation, mother’s education, climate, local custom, etc. 83 Uganda Ministry of Health (2007). 61 Development Index (HDI). This lack of correlation may be due to the fact that final outcome measures such as life expectancy or more general welfare measures such as the HDI are not a function of the performance of the health system at the district level per se but are related to other factors. 3.72 While there is little correlation between DLT rankings and outcome indicators, there is a correlation between the rankings and output indicators. Those districts in Uganda that do better than average on DPT3 coverage also tend to do better than average in the percentage of deliveries that occur in Government or NGO health facilities. Furthermore, the bottom ten districts in the DLT tend to be below average on both indicators and the top ten districts tend to be above average. This provides a viable tool to assess a district’s relative success in the provision of health services to its population, possibly through managerial success. 3.73 Is there a relationship between input and output indicators? Public sector health workers per capita and total hospital beds per capita are used to measure input. As shown in Figure 3.6, both DPT3 coverage and deliveries in facilities are strongly and positively correlated with the number of hospital beds per capita in a district. The better performing districts tend to have more positive input and output indicators than the poorly performing districts. However, and significantly, this relationship does not hold with respect to health workers per capita. Better performing districts do not have more health workers per capita than those that are performing poorly. While this is an interesting outcome, it is inconclusive since this report only examines public sector workers. This outcome raises questions about the efficiency and effectiveness of the public sector workforce and, along with high absenteeism, provides a starting point for further examination. 3.74 The wide variation in input, output and outcome indicators across districts in Uganda raises questions about why some districts are more successful than others. The strong relationship between output indicators and DLT rankings shows that the better performing districts are able to provide more services. The inability to translate these outputs into outcomes is cause for concern. Furthermore, the positive relationship between DLT rankings, output indicators, and hospital beds per capita reveals a link between output and input indicators. The lack of significance of health workers per capita in relation to the output indicators also shows that there are efficiency concerns with regard to the public sector health workforce, which will be investigated in a subsequent study. 62 Figure 3.6: District Level Input and Output Indicators DPT3 Coverage vs. Health Workers, 2007 DPT3 Coverage vs. Hospital Beds, 2007 DPT3 Immunization Rate DPT3 Immunization Rate 150 150 100 100 50 50 0 .0005 .001 .0015 .002 0 .001 .002 .003 Health workers per capita Hospital beds per capita Source: MOH, UNDP Source: MOH, UNDP Note: Green circles represent top ten districts Note: Green circles represent top ten districts and red circles represent bottom ten districts and red circles represent bottom ten districts % of Deliveries in health facilities % of Deliveries in health facilities Deliveries vs. Health Workers, 2007 Deliveries vs. Hospital Beds, 2007 0 20 40 60 80 0 20 40 60 80 0 .0005 .001 .0015 .002 0 .001 .002 .003 Health workers per capita Hospital beds per capita Source: MOH, UNDP Source: MOH, UNDP Note: Green circles represent top ten districts Note: Green circles represent top ten districts and red circles represent bottom ten districts and red circles represent bottom ten districts 3.75 Summing up, Uganda’s health policymakers must identify a combination of efficiency savings and re-prioritization to sustain progress towards health targets. Preliminary health sector modeling work carried out under this study suggests that the resources needed to meet Uganda’s health targets are unattainable. Given existing staffing norms, program unit costs, future cost drivers, the increased demand for health services, and future population growth, health policymakers face tough decisions on prioritization, even if they are able to improve efficiency. Wages and drug purchases will put insurmountable pressure on the health budget. Barring reforms to human resource management policy, the wage bill could grow to over 50 percent of the health sector budget, crowding out financing for drugs, interventions and infrastructure. HIV/AIDS interventions are projected to take up the lion’s share in the budget for drugs and preventive interventions, which could crowd out funding for malaria and immunization programs. Difficult choices and prioritization will have to be made between meeting the staffing norms or raising the coverage of some of the key interventions necessary to treat and prevent disease. VIIII. CONCLUSIONS AND RECOMMENDATIONS 3.76 Whilst health outcomes are improving, Uganda’s health system continues to face enormous challenges. Demand is rising faster than the funding available for treatment. District level services have seen real per capita non-salary spending halve in four years. Health outcome indicators are not progressing fast enough to meet the MDGs. Rapid growth in population, which will not slow for the foreseeable future, and increasing unit costs will continue to strain existing resources. 63 3.77 Uganda clearly needs to increase public health spending for non-salary costs at clinics and hospitals. However, limited opportunities for additional public funding seem to exist, and those opportunities available will fall far short of the needs of the health system. 3.78 Efficiency gains will be needed and can be found. Leakages and waste significantly reduce the proportion of Government spending that actually reaches patients. A better allocation of resources and measures to improve management and accountability could generate technical efficiency gains. 3.79 How can Uganda get more value for money through better allocative efficiency? Uganda achieves average outcomes with more than average spending. Uganda is a significant outlier in maternal health and other reproductive health indicators, even compared to poorer countries. However, some programs such as DPT3 immunizations and TB have shown marked improvement in coverage. This suggests that at the margin, more resources or efficiency measures may be needed for the non- responsive programs. Performance in health outcomes also varies greatly by district, suggesting that efficiency gains can be found in these cross-district variations. Given relatively high private funding, opportunities may exist to exploit public private partnerships in health, for instance through contracting. 3.80 Absenteeism and drug leakage point to serious performance management and accountability problems that could render scaling-up of spending ineffective in improving health care. Costs and demands for health care are rising but real financing, especially non-salary funding at clinics and hospitals, has been falling, implying that funding should be increased. However, the level of waste that exists in the health care system suggests otherwise. 3.81 The absence of a relationship between unit costs and output indicators at the facility level indicates that there is potential to increase the efficiency of resources use. Off-budget donor funds could be an important distortion to health service delivery, particularly if they entice scarce qualified workers from the public facilities. A review of human resource systems and their management was started under this study and should be an important first step in efforts to improve value for money in health. 3.82 Better human resource management would in any case be needed to absorb capacity constraints which make it difficult for Uganda to increase planned expenditure. Without improving human resource capacity, it is not evident that additional resources can be utilized efficiently in Uganda’s health sector. Providing additional resources beyond the absorptive capacity of a sector has negative consequences, as exemplified by Uganda’s experience with HIV/AIDS. Without addressing the labor shortage and without increasing the efficiency of the existing staff, additional resources for health may lead to further inflationary pressures. 64 Recommendations Improve efficiency and absorptive capacity for the health sector 3.83 The most pressing priority is to utilize the existing funding for health more efficiently. Improving efficiency is an essential step in making the case for increased fiscal space from other sources. This includes not just increased Government financing, but also external financing and to some extent other domestic resources. Strengthen programming of development assistance for health. 3.84 Clarify the concept of on-budget project support and commit to its consistent application during the budgeting process. This will enable the Government to correctly capture accurate on-budget donor support, plan and set realistic targets, and facilitate regular monitoring and reporting of funded projects. This exercise should also consider how to capture HIV/AIDS funding which falls under sectors other than health, to avoid bundling these expenditures under the health budget. 3.85 Establish a system to capture and monitor large inputs provided through off- budget donor support. Whilst this process is expected to be complicated, consideration should be given to start with those inputs that can be monitored through the existing Government systems, which include: drugs and medical supplies, medical equipment and vehicles. The Ministry of Health’s Comprehensive Procurement Plan captures most of the medicines and supplies provided to the sector. To inform the Government of its future liability, the costs can be adjusted to reflect what Uganda would pay if it were to purchase the inputs from its own funding. 3.86 Pursue available avenues for improved donor coordination and harmonization. The recent initiatives by the US Government/USAID in the development of multi-year compacts, and the new The International Health Partnership (IHP+) initiative are both clearly worth pursuing. There is need to design clear procedures under which Uganda is prepared to receive earmarked funds and for districts and health facilities to receive direct external support. The support to districts and health facilities should not crowd out original levels of service delivery and/or cause undue differences in the wage levels. Improve management and performance of the health workforce 3.87 Remove obstacles that impede basic personnel management functions in the health sector. Delays in recruitment, payroll entry, confirmation, promotion, etc., are major de-motivating factors. 3.88 Institute measures to attract and retain staff, especially in rural and remote areas. This could start with rolling out the allowance scheme for hard-to-reach rural areas which is already approved by the Government. Priority should be directed to providing institutional accommodation for health workers serving in remote and rural areas. In addition, instead of providing across the board salary increases the Government 65 could emphasize the provision of special allowances to health workers serving in clinical positions. 3.89 Align sector performance to outputs. This will require reinforcing management functions in hospitals and district health services, changing from paying for inputs to paying for defined outputs and providing more autonomy to health managers who will be accountable for the agreed results/outputs. In going forward with giving more autonomy to regional referral hospitals, it is recommended that the Government take the opportunity to seriously explore the possibility for output based health financing. Experience from other countries shows that salary increases alone will not lead to increased performance; other human resource issues such as absenteeism and staff de-motivation need to be tackled at the same time. 3.90 Collect more information on the health workforce and labor market. Available evidence relates to loss of PNFP workers to the Government. There is ample opportunity to pursue studies to understand better the labor market for health workers, including the impact of regional dynamics on Uganda’s health workforce. Drug procurement and logistics management 3.91 Implement the recommendations of drug procurement and logistics management reviews.84 These recommendations remain valid and deserve to be implemented. Uganda must (a) harmonize procurement of third party commodities; (b) clarify and agree on the roles of the main stakeholders involved in drug procurement and logistics management; (c) review the various Acts that have a bearing on procurement to reduce procurement restrictions imposed on NMS and ensure it becomes competitive in the liberalized environment; and (d) review the financing of drugs through PHC and Credit Line to avoid duplication and reach agreement on the drug financing through NMS, Credit Line and PHCCG. Recommendations to reduce the growing pressure to increase health spending 3.92 In setting sector standards and norms, consider their financial implications, especially when adopting new interventions. Even when funding is provided by an external party, consider the financial implications as this will inform the government of the future contingency liability. 3.93 Periodically monitor the cost of selected essential health services in terms of unit costs. The Government may start with monitoring the costs for (a) treating one patient over a year with ART, (b) treating a patient with ACT, and (c) fully immunizing a child. 3.94 Develop a policy to guide construction, expansion and maintenance of public health facilities. The policy should include criteria for the Government to adopt facilities built through community efforts. 84 See “Consultancy Services for the Review of the Role of National Medical Stores in the Public and Private Health Care System in Uganda” (August, 2006). 66 3.95 Expand family planning services. Uganda can afford to provide public contraceptives to its population to meet the growing unmet demand. In doing so, Uganda will catch up with its neighbors who already provide over 60 percent of contraceptives through the public sector compared to Uganda at 30 percent. 3.96 Study and adopt global initiatives that can reduce costs for health care. The initiative by the Malaria Medicines Venture suggests that there is a possibility to improve the availability of ACTs in the private sector that could save the Government substantial funding. 67 ANNEXES Annex 1: Draft PFM Platform 1--Predictability and Stability Area of activity Suggested actions Problems to be addressed by Link to the the initiative thematic areas Review outturns for line Agree the line item regularly Instability of the budget as a Improved items typically under chronically under budgeted. basis for service delivery. predictability and budgeted (utilities, rents Appears to be: stability etc) and apply not less • Utilities. Undermining of predictability than the same amount in • Rent. of resources actually available the cash plan for 2008/09 • Some areas of supply to service deliverers. without good reason; also (such as pharmaceuticals) apply to line item Review outturns for 2007/08 in Disruption of the link between estimates for 2009/10 these areas. resources and expected onwards. Discuss with budget holders any outputs. significant difference. Unless clear reason, bring budget up to at least level of actual spending on previous year. Use mid-year review to consider sources of funding for the update: • By offsetting reductions within the vote. • By reductions in other votes • Through additional revenue becoming available. Once brought up to date, roll the adjustments into future budgets. Reinforce adjustment and realism by greater use of pre- payment mechanisms. Enforce timely Issue advice on best practice Weak budget implementation Improved compliance with with regards to work, planning undermining overall predictability and procurement and work procurement and cash plans. cash planning. stability plans; use as basis for Provide related costing and improving predictability programming training. Instability of the budget as a Flexibility to of in-year releases. Identify incentives to be given to basis for service delivery. achieve service a selected group of budget targets. Intensive capacity holders that commit to produce Weak ownership of their building for production of good quality plans on time, such budgets by some accounting Improved realistic work and as: officers. awareness and procurement plans. • Greater virement rights resource within groupings of Cumbersome fragmentation of management Improve ownership of economic items. budget. capacity budget by those • Greater certainty of responsible for service releases in accordance Finger pointing between delivery by giving more with the phasing shown in accounting officers and flexibility in allocating their plans. MOFPED as to causes of resources within budget Build those incentives into the budget instability. limits. performance contracts issued to the selected accounting officers More management using existing mechanisms 68 Area of activity Suggested actions Problems to be addressed by Link to the the initiative thematic areas freedoms and flexibilities, available to MOFPED. but initially selectively Develop criteria for judging based on proven quality of work plans, etc., and competence/ compliance. inform spending bodies of those criteria. Apply those criteria to identify initially perhaps 2-3 examples of good practice that will be given the greater freedoms/flexibilities. Follow up to review quality of plans in the light of experience. Review continuation of the incentives. Share examples of good practice with all spending bodies so that they have an objective. Strengthen budgetary and To distribute examples of good Unrealistic budgeting and Improved policy analysis at budget analysis and practice and budget implementation awareness and MOFPED and key sectors recognize the individuals planning. resource through training and involved. management targeted recruitment Weak linkages between capacity Invite people such as Afritac or resources and expected results. CABRI to suggest a curriculum in policy analysis and to offer a Weak challenge function in course. MOFPED. Develop internal guidance on aspects of budget analysis, including check lists of issues to consider when analyzing budget submissions, requests for supplementaries and new initiative proposals. Hold international forum to discuss modern budget analysis. Improve capture in budget Develop a more comprehensive Instability of the budget as a Improved and budget protocol setting out procedures, basis for service delivery. predictability and implementation of donor timescales and consequences for stability flows. Government of: a) Under identification of Unrealistic budgeting and flows. budget implementation b) Lack of predictability planning. as to timing. Identify past problem areas and Undermining of predictability seek help from donor of resources actually available community as a whole to tackle to service deliverers. the problems. Report back to donors on Disruption of the link between comparison of declared amounts resources and expected and timing at time of budget and outputs. actual performance. Start to build the capacity Improve cash planning Use of controlling pace of Improved for managing short-term techniques based around budget implementation as the predictability and 69 Area of activity Suggested actions Problems to be addressed by Link to the the initiative thematic areas debt as a way of avoiding improvements in work planning only lever for managing short stability the need to manipulate and programming by spending term and temporary resource budget releases on bodies. flow fluctuations. account of short term and temporary resource flow Instability of the budget as a fluctuations. basis for service delivery. Undermining of predictability of resources actually available to service deliverers. 70 Annex 2.1: Public Service wage Tables and Wage Regressions Table 2.1.1: Total Number of Employees, by Salary Scale Grade 2003/04 2004/05 2005/06 2006/07 Not Stated 3,268 880 406 186 Political Officers 148 147 140 159 Specified Officers 37 34 33 32 U1EL 172 184 198 213 U1EU 204 269 308 327 U1S 26 29 31 31 U1SE 407 468 467 481 U2L 480 531 553 550 U2U 450 507 522 621 U3L 935 981 1,344 1,582 U3U 897 1,031 1,148 1,313 U4L 2,479 2,827 2,388 2,466 U4U 3,720 4,030 4,000 4,291 U5L 4,663 4,861 4,284 3,993 U5U 5,446 6,456 7,194 8,218 U6L 3,317 3,573 3,613 3,594 U6U 3,238 3,606 3,568 3,763 U7L 4,847 5,579 4,503 1,393 U7U 28,852 30,637 27,769 31,270 U8L 3,371 3,591 2,887 3,112 U8U 10,113 10,712 9,684 10,121 Grand Total 77,070 80,933 75,040 77,716 Table 2.1.2: Non-Education, Average Yearly wage, in UHs Grade 2004/05 2005/06 2006/07 Political Officer 21,532,044 23,171,356 30,265,278 Specified Officer 31,016,881 29,738,424 35,330,025 U1 Special 24,963,304 24,178,475 25,001,621 U1 Senior Exec 19,434,346 19,017,746 19,317,426 U1 Upper 16,094,672 15,371,035 16,008,106 U1 Lower 14,847,757 15,090,539 15,350,159 U2 Upper 13,902,581 13,545,008 14,076,516 U2 Lower 11,374,681 11,414,119 11,878,220 U3 Upper 9,996,612 9,905,767 10,263,226 U3 Lower 8,418,245 8,542,524 8,627,535 U4 Upper 7,920,295 7,937,632 8,370,279 U4 Lower 6,006,390 6,107,232 6,273,150 U5 Upper 5,731,293 5,788,652 5,856,511 U5 Lower 3,568,476 3,601,241 3,758,908 U6 Upper 3,289,107 3,352,853 3,302,888 U6 Lower 2,438,118 2,489,175 2,704,055 U7 Upper 3,168,394 2,742,970 2,736,727 U7 Lower 2,573,032 1,838,714 1,886,578 U8 Upper 1,551,249 1,612,876 1,782,252 U8 Lower 1,333,788 1,465,292 1,787,283 71 Table 2.1.3: Education Sector: Average Yearly Wage (UHs) Grade 2004/05 2005/06 2006/07 U1SE 14,887,271 15,119,534 11,435,091 U1 Upper 15,074,261 15,553,702 15,406,676 U1 Lower 14,613,994 14,927,320 16,293,447 U2 Upper 11,457,708 11,457,708 9,055,764 U2 Lower 11,318,737 11,326,049 12,140,326 U3 Upper 19,185,930 14,626,908 9,548,472 U3 Lower 8,266,057 8,253,873 9,076,246 U4 Upper 6,269,755 6,192,898 6,416,290 U4 Lower 5,874,007 5,689,265 6,432,422 U5 Upper 4,544,930 4,321,607 4,951,478 U5 Lower 4,054,037 3,424,724 3,860,850 U6 Upper 2,746,687 2,657,196 3,078,524 U6 Lower 2,408,648 2,408,648 2,481,068 U7 Upper 1,871,770 2,009,394 2,707,393 U7 Lower 1,861,011 1,620,896 1,636,750 U8 Upper 1,504,899 1,142,541 1,501,232 U8 Lower 1,046,704 1,068,148 1,391,388 Invalid PS 2,758,676 2,494,509 2,655,320 72 Table 2.1.4: World Bank Wage Regressions –OLS Weighted (from UNHS 2005/06) OLS National Determinant OLS Rural wage OLS Urban wage wage coef t coef t coef t Female -0.490*** -4.327 -0.576*** -4.372 -0.464*** -3.046 Person Is Married -0.044 -0.517 -0.046 -0.463 -0.069 -0.529 Interaction: Married, Female 0.354*** 2.820 0.457*** 3.373 0.282 1.473 Person Is Divorced/Separated -0.254 -1.637 -0.270 -1.450 -0.226 -0.883 Interaction: Divorced, Female 0.456** 2.002 0.570** 2.073 0.268 0.742 Person Is Widow/Widower -0.022 -0.100 -0.121 -0.577 0.119 0.305 Interaction: Widow, Female 0.396 1.271 0.647* 1.685 0.127 0.275 Age In Completed Years 0.071*** 4.590 0.073*** 4.410 0.077*** 2.654 Age In Completed Years, Squared -0.001*** -3.491 -0.001*** -3.764 -0.001* -1.919 Some Primary 0.343** 2.467 0.262 1.452 0.437** 2.240 Completed Primary 0.520*** 3.116 0.372* 1.925 0.710*** 2.880 Some Lower Or Technical Secondary 0.456*** 3.170 0.505*** 2.593 0.507** 2.559 Completed Lower Or Technical 0.676*** 4.696 0.447** 2.262 0.881*** 4.225 Secondary Some Or Completed Upper Secondary 0.978*** 5.762 0.664*** 2.723 1.219*** 5.895 Post-Primary Specialized Training 0.921*** 5.915 0.849*** 4.112 0.999*** 3.979 Post-Secondary Specialized Training 1.217*** 8.091 1.103*** 5.186 1.344*** 6.400 Degree And Above 1.952*** 11.155 1.616*** 7.091 1.992*** 8.769 Missing Education 0.469 1.451 0.147 0.340 0.830* 1.904 Mining, Construction 0.149 1.596 0.106 0.851 0.218 1.593 Trade -0.173 -1.525 -0.065 -0.380 -0.142 -0.932 Hotels -0.367** -2.242 -0.365 -1.255 -0.392** -2.204 Transport, Communication, Finance, 0.168* 1.663 0.060 0.505 0.211 1.306 Real Estate Public Admin, Health, Educ. Services -0.199** -1.968 -0.246* -1.749 -0.109 -0.668 Other Services, Ios -0.122 -1.217 -0.250* -1.849 0.047 0.324 Temporary Or Casual Worker In Main -0.247*** -3.780 -0.135 -1.300 -0.341*** -4.078 Wage Employment Government Employee 0.330*** 3.875 0.389*** 3.822 0.402** 2.110 _Cons 3.986*** 12.634 3.964*** 10.530 3.034*** 3.388 Number Of Observations 1,579 731 848 R2 0.464 0.496 0.482 note: *** p<0.01, ** p<0.05, * p<0.1 district fixed effects not reported reference category education: no schooling or less than P1 reference category sector: manufacturing/utilities reference category marital status: never married excludes agricultural laborers corrects standard errors for clustering and stratification weighted using hh sampling weights 73 Annex 2.2: Modeling the fiscal effects of pay reform This annex explains in greater depth the key assumption and results from the modeling of public service pay reform and universal secondary education in order to inform the costing of policy commitments. The modeling exercise extends the previous Fiscal Sustainability Analysis model85 by modeling various scenarios of pay reform. In addition, it examines the affordability of these reforms by assessing the impact of greater depth efficiency measures in education and other sectors. In the health sector, it is assumed that the increasing fiscal pressures arising from the fast rising population and the implied demand for more health services will be met within the assumed trend growth in health expenditure. The results illustrate that with the education sector remaining the largest driver of the Government budget, the Government needs to deepen efficiency measures within this sector to ensure quality on one hand while making education more fiscally feasible and affordable. It further suggests that the Government will need to actively seek reduction in the wage bill--also driven largely by the education sector--by using the civil servants more efficiently. BASELINE SCENARIO The baseline scenario shows that even without public service pay reform, the increased demand for teachers as USE gets rolled out, will limit fiscal space to maintain the momentum in infrastructure spending as set out in the 2008/09 budget and needed to sustain growth at desired levels. Within this baseline scenario, the analysis suggests that: • The education budget rapidly expands on account of the growth in numbers of students entering the system and also rolling over into the free secondary level of education, which started a gradual implementation in the 2006/07 financial year. Whereas efficiency measures already considered by the Government will curb some of the rapid growth in expenditure, education sector spending will increase to an average of 30 percent of the budget over the projection period (2008/09 to 2024/25), from 16.5 percent in 2007/08. • In line with the past trends in the numbers of non-education civil servants and their wages, the wage bill, which accounted for about 27 percent of the budget in the base period, is expected to increase its share by about 11 percentage points by the end of 2024/25. This implies that even without a pay reform, the wage bill will increasingly diminish the room for maintaining the momentum of infrastructure investments. • The health sector, where expenditure is assumed to grow in line with historical trends, will be contained at about 10 percent of the budget through the projection period. This is far lower than what is required to meet the target within the Health Strategic Plan--of reaching US $ 28 per capita, from the US $ 7 recorded for 2006/07. 85 See Chapter 4 of PER 2007: Uganda Fiscal Policy for Growth, Public Expenditure Review 2007, World Bank 74 • In contrast to the Government’s current policy objective of raising spending on the infrastructure intensive programs in the medium-term, spending on infrastructure is squeezed to a lower share of GDP during the projection period. • As shown in Annex Figure 2, from an economic perspective, whereas the wage bill presents the biggest source of pressure to the budget, this again is driven by the education wage bill, specifically through the increase in teacher salaries. With the number of teachers doubling from 150,000 on the payroll in 2006/07, the wage bill will more than quadruple over the same period. ASSUMPTIONS: The baseline scenario, which on one hand mimics the recent policy stance of the Government, as provided in the 2008/09 budget and MTEF, also uses historical trends as a guide to making projections. 1. The macro picture: From the macro point of view, whereas growth is modeled endogenously through its linkage with investment, the simulation is limited within the Government’s macroeconomic framework by using a cap on fiscal expenditures of 22 percent of GDP, in line with LTEF in PEAP. This is also the average attained over the past 10 years, with the exception in the immediate MTEF period where this was exceeded in some years. Revenue is modeled to increase only modestly in view of the slack growth in collections realized over the tax reform period and the difficulty of instituting new measures to widen the tax base. This implicitly incorporates the Government’s fiscal consolidation strategy, given the resultant declining fiscal deficit from above 7 percent to about 4 percent by 2027. The baseline assumption is still to finance this deficit with grants and concessional loans in equal proportions. This conservative approach to financing is in line with the Government’s debt strategy of limiting new borrowing to concessional financing, except for infrastructure. 2. Budget composition: The baseline scenario assumes infrastructure spending to start off at a much higher level of 5.2 percent in 2008/09, which is almost double the level observed historically. Even then, as a residue, the share of infrastructure spending is depicted to decline as the other components’ (that is, education and wage bills) share increases. Projections for the education sector are based on the Ministry of Education financing model. However, some adjustments have been made to make these more realistic (see item 4 below). The wage bill for the non-education sectors is modeled as a product of civil service numbers and their wages. Overall, the staff numbers excluding education grow annually by about 5.0 percent with no specific efficiency improvements in the utilization of civil servants. In line with past trends, their wages increase by an average of 5.1 percent in real terms per year. The health sector is allowed to grow in line with past trends, which in turn contains its share at about 10 percent of total budgetary outlays through the projection period. 3. Macro-budget linkages: The fiscal framework is linked with the macroeconomic framework to ensure macro-economic consistency and also provide indications on the implications of budget composition to growth. The assumptions are summarized in Box 1 below. 75 4. Education: The roll out of USE that started in 2007/08 has had significant implications for the education sector budget. In the projections, which are based on the Ministry of Education financing model, the targeting suggested by the Government that only those students who have passed with a minimum aggregate of 28 points at primary level are eligible for USE is adopted. This already reduces the numbers of students expected to enter USE. Whereas this proposal raises issues with respect to equity (that is, not all students who pass primary are allowed to enter the free secondary), it incorporates an incentive system for students to perform at minimum standards to access the free education, and hence addresses the quality of secondary schooling from entry. It is on account of this factor that the rate of transition from primary to secondary level is not expected to go beyond 79 percent even by 2018. The Government’s assumption with regard to gradually increasing the student-teacher ratio from 34 to about 40 (the international norm) over the projection period is also adopted. The projections are more pessimistic about the improvements in dropout rates, particularly at S4 level, where it is assumed that the rate is to decline from about 60 percent in 2007/08 to 20 by 2024/25. Furthermore, it is assumed that the composition of the teaching staff will not change immediately to attain the desired 3:1 policy ratio for diploma: graduate teachers. Consequently, this is assumed to remain at 1.2:1 through the projection period. On the Government’s contribution to universal secondary schooling in Uganda, the assumption is that it remains at about 40 percent, although the Government has indicated a gradual reduction of this share through more private sector participation in delivering secondary education. The baseline scenario assumes no double shifting. 76 Annex Box 1: Key Macroeconomic and Fiscal Assumptions for the Baseline Scenario Real GDP growth is driven by the composition of fiscal expenditures and is hence endogenously determined. It is estimated within a production function with constant returns to scale with physical and human capital, while the total factor productivity is assumed at trend growth of 0.7 percent per annum. Private sector investment is assumed to grow at a trend rate of 15 percent, but in addition accelerates in line with public capital per worker, which is determined as a function of public capital expenditure and growth in labor force. Fiscal policy is based on the fiscal consolidation strategy adopted by the Government. • Domestic revenues are closely linked to the macro evolution, assuming modest benefits from increased effort – income tax to increase from 3.6 percent of GDP to 5.4 percent by 2026/27; trade revenue to maintain the same share of 25 percent of imports. • Total expenditure is fixed at 22 percent of GDP beyond the MTEF period, reflecting an average of 10 years and the Government’s commitment to fiscal consolidation. The deficit before grants is assumed to decline from more than 7.8 percent of GDP in 2007/08 to 4.1 percent by 2026/27. • With grants tapering off to about 2 percent of GDP over the forecast period, non-interest expenditures (primary expenditures) are projected to stay around 20 percent of GDP. The resultant primary deficit path envisages a gradual decline from 9 percent in 2004/05 to about 5 percent by 2026/27. • While the fiscal deficit is assumed to be retrospectively funded with external loans and grants, the resultant gap between current funding and the respective deficits has to be financed through domestic borrowing. Nominal interest rates on domestic public debt are assumed to stabilize at about 11 percent (about 6 percent in real terms), slightly higher than the average of 8.7 percent over the past 10 years. The interest rate on foreign debt is expected to decline to about 0.6 percent, from the average of 0.8 percent over the 10 years to 2005. Inflation: Domestic inflation which has remained above the target of below 5 percent for 3 consecutive years is expected to remain above 5.0 percent until 2008 on account of the effect of the energy crisis, before declining to the target of 5 percent for the rest of the forecast period. Foreign inflation based on trading partners inflation, is assumed to gradually decline from 5 percent to 3.5 percent by the end of the forecast period. Nominal Uganda Shilling exchange rate is assumed to change in line with changes in inflation differentials between Uganda and its foreign trading partners, for the real exchange rate to remain constant. The current account deficit (excluding interest payments) in terms of GDP gradually rises from 2.8 percent in 2004/05 to about 6 percent of GDP over the forecast horizon. 77 Annex Figure 1: Uganda Budget Composition - Baseline 25.0% 22.7% 20.0% 15.0% (share of GDP) 10.0% Health 5.0% Education 0.0% 2008/09 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 2016/17 2017/18 2018/19 2019/20 2020/21 2021/22 2022/23 2023/24 2024/25 2025/26 2026/27 Education Health Infrastructure-intensive programs/votes Economic services and agriculture Accountability, justice, etc Pension Unconditional grant to districts and urban Interest payments Arrears repayment Total Annex Figure 2: Uganda Budget Composition –Economic Perspective -- Baseline 25.0% 20.0% 19.0% (share of GDP) 15.0% 10.0% 5.0% 0.0% 2006/07 2008/09 2010/11 2012/13 2014/15 2016/17 2018/19 2020/21 2022/23 2024/25 2026/27 Wage bill Development, non-infrastructure Development, infrastructure Non-wage Arrears Total 78 SCENARIO 1: NOMINAL DOUBLING OF PUBLIC SERVICE WAGES FOR ALL STAFF This scenario replicates the medium-term pay targets for public service staff outlined in the pay policy, approved by the Cabinet in August 2006. While the relative pay increases outlined in this policy vary between different pay scales, all pay scales would receive at least a doubling of their nominal wages. This scenario assumes a nominal doubling of all pay scales shared equally across FY 2009/10 and FY 2010/11. However, whereas the approved pay policy foresees a further nominal doubling of wages in “the longer term”, here it is instead assumed that after the initial two-step nominal increase, wages would revert to increasing in line with their past trend. It is assumed that no measures to increase efficiency are implemented, meaning that the same trend in staff numbers prevails as in the baseline scenario. According to this simulation, the Cabinet approved wage increases would result in infrastructure budget getting gradually crowded out to 0.2 percent of GDP, by 20026/27. In real terms, spending on infrastructure will decline from the peak reached in 2008/09 of USh 1,389 billion, to USh 215 billion in 2026/27. the conclusion of this scenario is it would leave little room to scale up infrastructure spending. Annex Figure 3: Scenario 1: Nominal Doubling of Public Service Wages – all Staff 25.0% 22.7% 20.0% 15.0% (share of GDP) 10.0% 5.0% 0.0% 2008/09 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 2016/17 2017/18 2018/19 2019/20 2020/21 2021/22 2022/23 2023/24 2024/25 2025/26 2026/27 Education Health Infrastructure-intensive programs/votes Economic services and agriculture Accountability, justice, etc Pension Unconditional grant to districts and urban Interest payments Arrears repayment Total 79 SCENARIO 2: NOMINAL DOUBLING OF WAGES FOR MANAGERIAL, TECHNICAL AND PROFESSIONAL STAFF This scenario simulates a selective wage increase with the nominal pay increase (doubling) from the scenario 1 limited to managerial, technical and professional staff. This category of staff has been defined as the upper ‘U’ levels of pay scales U5-U1. The result is that such a selective pay reform would allow for relatively larger room for spending on infrastructure, compared to full scale pay reform. However, fiscal space for infrastructure investments will barely survive in the range of 2 percent to 3 percent of GDP, which still seems inadequate to remove infrastructure bottlenecks to sustain high growth in Uganda. Annex Figure 4: Scenario 2: Nominal Doubling of Wages of Managerial, Technical and Professional staff 25.0% 22.7% 20.0% 15.0% (share of GDP) 10.0% Health 5.0% Education 0.0% 2008/09 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 2016/17 2017/18 2018/19 2019/20 2020/21 2021/22 2022/23 2023/24 2024/25 2025/26 2026/27 Education Health Infrastructure-intensive programs/votes Economic services and agriculture Accountability, justice, etc Pension Unconditional grant to districts and urban Interest payments Arrears repayment Total 80 SCENARIO 3: SELECTIVE PAY REFORM COMBINED WITH EFFICIENCY SAVINGS, IN EDUCATION AND OTHER SECTORS This scenario combines selective pay reform with efficiency savings in education and more efficient utilization of public service staff. (i) The efficiency savings under this scenario 3 create room for maintaining infrastructure spending at above 5 percent of GDP through the projection period. The assumptions made for the efficiency measures include: In the education sector: 1. A faster move in increasing the student-teacher ratio. This measure is critical but requires a curriculum review to shift from using teaching load as the criteria for enrolling teachers to student-teacher ratio. Furthermore, this measure would be supported by an increase in the number of core subjects that can be taught by each teacher from 1.2 to 1.8 to increase the teaching load. 2. A faster adoption of double shifting--adopting double shifting for some schools where enrolments have reached the appropriate threshold, to maximize utilization of available physical facilities. This would implicitly raise the student classroom ratio without necessarily adversely affecting quality. It is assumed that schools will gradually adopt this policy in such a way that the overall number of streams in each classroom increase to about 1.8 by 2020. 3. Increased private participation in delivering of secondary education. Contribution of the Government to USE, currently assumed at 40 percent of all schools, is assumed to decline to about 18 percent by 2025/26. This will relieve the Government of the need to construct school buildings and supply equipment. The Government will only provide a capitation grant for students enrolled under USE in privately owned schools. 4. Gradual increase of the share of diploma teachers towards the 3:1 diploma-to-graduate teacher policy ratio. Biasing teacher qualification towards diploma, especially for the lower secondary classes where the quality of teaching is not compromised by the difference between graduate and diploma qualification, is a critical efficiency measure. T he share of diploma teachers, currently at 24 percent, would be raised to 75 percent by 2018 and maintained there. With the cost of graduate teachers three times higher than that of diploma teachers, this will reduce teacher costs by about 40 percent. For the rest of the sectors, this scenario assumes the following: 1. The number of non-education public sector employees is assumed to remain constant for the rest of the projection period, implying a more efficient utilization of the public employees. 2. While it is assumed that all infrastructural investments will still be undertaken, this will be done at lower cost. The assumption is that the unit costs of construction (including that for classrooms and teacher houses) will be lower by 30 percent compared to the baseline. 81 Annex Figure 5: Scenario 3: Selective Pay Increases Combined with USE Savings 25.0% 22.7% 20.0% 15.0% (share of GDP) 10.0% 5.0% 0.0% 2008/09 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 2016/17 2017/18 2018/19 2019/20 2020/21 2021/22 2022/23 2023/24 2024/25 2025/26 2026/27 Education Health Infrastructure-intensive programs/votes Economic services and agriculture Accountability, justice, etc Pension Unconditional grant to districts and urban Interest payments Arrears repayment Total Annex Figure 6: Budget Composition - Economic Perspective - Scenario 3 25.0% 20.0% 19.0% (share of GDP) 15.0% 10.0% 5.0% 0.0% 2006/07 2008/09 2010/11 2012/13 2014/15 2016/17 2018/19 2020/21 2022/23 2024/25 2026/27 Wage bill Development, non-infrastructure Development, infrastructure Non-wage Arrears Total 82