Report No. 25969-MZ Mozambique Public Expenditure Review Phase 2: Sectoral Expenditures September 22, 2003 PREM 1, Africa Region Document of the World Bank CURRENCYEQUIVALENTS (as of September2003) CurrencyUnit: Metical (Mt);plural: Meticais US$1= Mt 23,400 FISCALYEAR January 1 to December31 Vice President: Callisto Madavo Director: Darius Mans SectorManager: PhilippeLe Houerou TaskManager: Peter G. Moll CONTENTS EXECUTIVE SUMMARY .................................................................................................................... i CHAPTER 1 17 CHAPTER 2 .. INTRODUCTION ................................................................................................... MACROECONOMICAND CROSS-CUTTINGISSUES ............................................. 18 A. 18 23 Update on the reform o f fiscal management ................................................................ Poverty and the intersectoral allocation o fresources................................................... The current macroeconomic context ............................................................................ B. D. C. 27 Cross-cutting issues insectoral spending..................................................................... 33 Thepolicy process. expenditure prioritization. and outcomes Chronic sectoral underspending Civil service reform .................... Progress with decentralization .............................................................. 38 CHAPTER3 A. . EDUCATION ......................................................................................................... 41 B. Introduction.................................................................................................................. 41 C. Key education sector issues.......................................................................................... 42 Expendituretrends........................................................................................................ 50 D. Poverty focus and incidence o f expenditures ............................................................... 57 F. E. Policy priorities for the education sector ..................................................................... Impact o f HIV/AIDSon the educationsystem............................................................ 59 60 CHAPTER4 . HEALTH ............................................................................................................... 63 A. Introduction.................................................................................................................. 63 B. Key health sector issues ............................................................................................... 63 D. C. 69 Impact o f HIV/AIDSon the health system .................................................................. Expendituretrends........................................................................................................ E. 76 Policy priorities for the health sector........................................................................... 77 CHAPTER 5 . ROADS ................................................................................................................. 81 A. Introduction.................................................................................................................. 81 B. K e y roads sector issues................................................................................................. 82 D. C. Expendituretrends........................................................................................................ 83 Policy priorities for the roads sector............................................................................. 88 CHAPTER 6 . WATER SUPPLYAND SANITATION ..................................................................... 90 A. Introduction.................................................................................................................. 90 B. 91 Trends inpublic expenditure inwater supply and sanitation ....................................... Key sector issues .......................................................................................................... D. C. Policy priorities .......................................................................................................... 96 100 REFERENCES ............................................................................................................................... 102 ANNEX1 ACTION PLAN, AND MILLENNIUM 108 ANNEXI1 DATA .. DEVELOPMENT GOALS ................................. ..................................................................................................................... 111 Figures Figure 2. Figure 1. 46 Recurrent Unit Costs by Program in2000 ................................................................. Percentage o f female teachers and girls drop.out. by province 2001......................... Figure 3. Figure 4. Govemment unitrecurrent expenditure by inputin2001 (in 1999 US$)...................52 53 Secondary teacher salaries inselected SSA countries................................................ 54 Figure5. Distributiono fbenefits from public spending ineducation by income decile...........58 Figure6. Distributiono fbenefits frompublic spendinginhealthbyincome decile.................74 Figure7. Inequityinbudget allocations acrossregions........................................................... 112 Tables Table 1. Basic macroeconomic indicators................................................................................. 19 Table 2. 20 Table 3. Macro framework, 1997-2005 ..................................................................................... Govemment finance, 1997-2003 (percentage o f GDP) ............................................... 22 Table 4. Feasible ceilings onrecurrent expenditures inselected sectors................................... 23 Table 5. Table 6. Recurrent and investment expenditure, by function, 2000 .......................................... Actual recurrent expenditure by functional classification, 1998-2002........................ 24 25 Table 7. Budgetedsectoral allocations (recurrent and capital accounts) relative to total spendingfor selectedcountries, 2000/01.............................................................................. 25 Table 9. Table 8. Summary o f Pressing Actions (was Table 7.1 inPEMR)............................................ 28 Table 10. Summary of PriorityActions (was Table 7.2 inPEMR) ............................................. 30 Table 11. 32 Ratio ofprivate sector to public sector compensation............................................... Summary of Actions for the MediumTerm (was table 7.3 inPEMR) ...................... Table 12. 36 Table 13. Efficiency indicators for education in2001............................................................... Completionrates ofEP1, EP2, ESGl and ESG2 in2001 ......................................... 44 Table 14. 46 Table 15. Selectedquality indicators by education subsector, 1997-2001 ................................ 48 Government and external education recurrent and capital expenditure in2000 (US$ Million) ................................................................................................................................. 50 Table 16. Education sector expenditures, 1995-2001(US$ Million) ........................................ Table 17. Yearly teacher salary and teacher salary/GDP per capita in2001 (US$) ..................51 54 Table 18. Regional distribution o fpopulation and benefits from education spending (in Table 19. thousand Mt.)........................................................................................................................ 58 65 Total actual health expenditwe at current prices. 1997-2000 ................................... Health expenditures and performance: aregional comparison .................................. Table 20. 70 Table 21. Table 22. 71 Healthexpenditure (actual) by economic classification............................................ Source of fundingby economic classification. 1997 and2000 ................................. Table 23. 72 Table 24. Projected health spendingup to 2010 ........................................................................ 73 Table 25. The cost ofHIV-relatedhealth services (percentageofGDP). 2000-2010 ...............75 Health service units per capita. selectedprovinces and regions ................................ 77 Table 26: Table 27. Road sector budgets. 1996-2002................................................................................ 84 Expenditureonroads. bymaintenance / investment. 1995 to 2001.......................... Table 28: Fundingsources (internal and external) for the road sector budget. 1996-2002 .......85 85 Table 29. Table 30. Budgetedvs.actual spending. by funding source. 1995 to 2001............................... 86 Table 31. Water and sanitation sector expenditures (actual) ..................................................... Transmission o f Road FundFuel-Based Revenues ................................................... 87 Table 32. 97 Table 33. Water supply and sanitation spending inrelationto the budget................................ 97 Table 34. Execution of the Govemment investmentbudget. 1999-2001 .................................. 98 Table 35. Water sources o f urbanpeople. 1997 (percent) ......................................................... 99 Table 36. 100 Mozambique's progress towards the MillenniumDevelopment Goals ..................108 Action plan............................................................................................................... Prices paidfor water inurbanareas. by location and source................................... Table 37. Table 38. 110 Table 39. 111 Health sector recurrent expenditures by level of care (percentage) ......................... Functional classification of expenditure (percentageof GDP) ................................ 112 ABBREVIATIONS AND ACRONYMS ADB African Development Bank ANE AdministraqZio Nacional das Estradas (National Road Administration) CFPP Centro de FonnaqZiode Professores Primarios: Teacher training system (grade 7 plus 3 years) CGE Conta Geral do Estado. See Republica de Moqambique, 2000- 2001. CPI Current Price Index CRA Conselho de RegulaqZo do Abastecimento de Agua (Regulatory Board for Water Supply) DfID Department for InternationalDevelopment DNA DirecqZio Nacional das Aguas (National Directorate for Water) DNCP DirecqZoNacional de Contabilidade Publica DNPO DirecqZio Nacional do Plan0 e Orqamento (National Directorate for PlanningandBudget) DPT Diphtheria, Pertussis and Tetanus ECMEP Empresa de ConstruqZioe de ManutenqZo de Estradas e Pontes, Enterprise for ConstructionandMaintenance o f Roads andBridges EFA Education for All EP1 Primaryschool (lower level) EP2 Primary school (upper level) ESER Education Sector Expenditure Review (2002). See References. ESGl General Secondary School (lower level) ESG2 General Secondary School (upper level) ESRP Economic and Social RehabilitationProgram ESSP EducationSector Strategic Plan ETSDS Expenditure Tracking and Service Delivery Survey EU EuropeanUnion FIPAG Fundo de Investimento e PatrimGnio do Abastecimento de Agua (Asset Holding andInvestment Company for the five cities' water systems) FRP Feeder Roads Program FTI Fast-Track Initiative FY FinancialYear GDP Gross Domestic Product GER Gross Enrolment Rate HNMS Highway Network Management System HSER HealthSector Expenditure Review (2002). See References. IGF Inspectoria Geral das Finanqas (Finance Inspectorate General) IMF InternationalMonetary Fund LCSP Low-Cost SanitationProgram MDGs MillenniumDevelopment Goals MPF Ministry o f Planningand Finance MTFF MediumTermFinancialFramework NGO Non-Governmental Organization NHS National Health System o f the Ministryo f Health NWDP National Water Development Program (I and 11) (projects hnded bythe WorldBank) O&M Operation andmanagement PARPA Action Plan for the Reduction o f Absolute Poverty (viz. PRSP). See Republic ofMozambique, 2001. PEMR Public ExpenditureManagement Review, 2001, by the World Bank. The first phase ofthe Public Expenditure Review, o fwhich this document is the secondphase. SeeReferences. PES Plano Econamico e Social (Social and Economic Plan) PESS Plano EstratCgico do Sector da Sacde (Health Sector Strategic Plan) 2001-2005-2010. See References. PRSP Poverty Reduction Strategy Paper PSIA Poverty and Social Impact Analysis PTIP Plano Trienal de Investimento Pcblico (Triennial Public Investment Plan) PTR Pupilper Teacher Ratio RBMMP Roads andBridges Management and Maintenance Program, sector-wide program, sometimes referred to as "Roads III" ROADSI11 See RBMMP ROCS Roads and Coastal Shipping projects 1and2, with World Bank anddonor participation RWS Rural water supply SDC Swiss Agency for Development Corporation (AgCncia Suiga para o Desenvolvimento e a Cooperagiio) SIP Sistema de Informaggo de Pessoal (personnel register o fthe civil service) SISTAFE Sistema Integrado de Administrag2io Financeira do Estado (integrated financial management information system) SPS Small Piped Systems SSA Sub-Saharan Africa SWAP Sector-Wide Approach UIMO (Brigadas de) Us0 Intensivo de MZo de Obra (labor-intensive brigades, associated with road maintenance) UTRAFE Unidade TCcnica para a Reforma da Lei da Administrag50 Financeira do Estado (Technical Unit for Reform o fthe Financial Management Law) UTRESP Unidade TCcnica para a Reforma do Sector Pcblico (Technical uws Unitfor Reformofthe Public Sector) UrbanWater Supply WB World Bank ACKNOWLEDGEMENTS This is the final (Gray Cover) report o fthe second phase o f the Public Expenditure Review undertaken by the World Bank andthe Ministry o f Planningand Finance o f the Government of Mozambique. Useful information and insight was providedby experts inmany departments inthe Ministry o f Planning andFinance. For the World Bank,the second phase o fthe Public Expenditure Review was managed by PeterM o l l (AFTPl). The primary authors o f this report are Peter M o l l and Antonio Nucifora (AFTP1); the team included Maria Teresa Benito-Spinetto (AFTPl), Abdelmoula Ghzala (AFTTR), Dipac Jaiantilal (AFTPl), Mary Mulusa (AFTHl), Catherine Revels (AFTUl), Alexandria Valerio (AFTHl) and Jane Walker (AFTUl). Contributions were made bythe Bank consultants Chris Finney, LarryHerman, Eduardo Macuhcua, Estrela Pol6nia and Manolo Shchez. Part o fthe funding for the consultants came from the Belgian Poverty ReductionPartnership Program Trust Fund. Assistance with editing, communication andbudget management was providedbyLigiaMurphy (AFTPl) and LurdesMalate (AFC02). Onthe Government side, the review was managed byMr.JosC Sulemane, National Director o fPlanningandBudget inthe Ministryo f PlanningandFinance. The team includedMr.Antonio Laice, the National Director of Treasury, andMr.Domingos Lambo, the Deputy National Director o fPlanning and Budget, as well as Mr. SCrgio Dista and Ms.Lidia Cabral. The team also included staff from the sectoral ministries (Education, Health, Agriculture, Water andRoads), and especial reference should be made to the contributions o fDr. Zefanias Muhate (Permanent Secretary, MINED), Mr. Virgilio Juvane (National Director o f Planning, MINED),Mr.Ant6nio Chambal (National Director o f Finance andAdministration, MINED), Dr. Frederico Sitoe (Deputy National Director for Agrarian Economics, Ministry o fAgriculture), Dr.Humberto Cossa (National Director, Ministry o f Health), Dr.AmCrico Muianga (National Director o f Water, Ministryo f Public Works), Mr.Carlos Fragoso (Chairperson o f the Administrugio NucionuZdus Estrudus), as well as that o f their staff. Furtheracknowledgements are cited inthe constituent papers on health, education, roads, water and expenditure tracking. Some o fthe work was done, under the direction o f the Ministry, by a syndicate o f consultant firms entitledFortulecimento du PZunzJicugio e Orgumento Sectoriuis (FoPOS), which included OxfordPolicyManagement, CESO Consultoria Intemacional and Austral Consultoria. The Expenditure Tracking and Service Delivery Survey (ETSDS) was done jointly by the Ministry o fHealth and the Ministryo f Planning and Finance, with OxfordPolicyManagement andAustral Consultoria as implementing agencies, andwith fbnding from DflD. Advice and comments were also received from a group o f donors involved in public finance, including Nick Highton (DfTD), Anton Johnston (SIDA), JosC Carlos Nunes (EU), MartinRoland (Embassy o f the Kingdomo fthe Netherlands), h i m Schwidrowski (IMF) and Thomas Thomsen (DANIDA). Their extensive knowledge o f public finance issues inMozambique was extremely valuable. Thepeer reviewers, Shiyan Chao (AFTHl), Kenneth Gwilliam(consultant), Jeffrey Hammer (DECRG), Julie McLaughlin (AFTHl), Alex McPhail (AFTU2), Alain Mingat (AFTHD), handRajaram (PRMPS), Helen Saxenian (LCSHH), Vinaya Swaroop (SASPR), andRichardVerspyck (AFTU2), provided valuable comments and suggestions at different stages o fthe report. The report was preparedunder the supervision o f PhilippeL e Houerou, Sector Manager for AFTP1. H e offered overall conceptual guidance, provided critical analytical advice and ensured quality control and management support. DariusMans, Country Director, supported the process and provided the major guidelines. i EXECUTIVESUMMARY 1. With careful economic management, substantial private capital inflows and support o f the intemational community, Mozambique has staged a dramatic recovery from the damage resultingfrom its long struggle for independence and its civil war. Since 1992,infrastructurehas beenbrought back to pre-war levels and incomes have risen considerably. But the country remains poor; infrastructure i s inadequate, there are serious unmet education andhealth needs, andpoverty issues need to be addressed directly. However, many o fthe "first-generation" reforms associated with market liberalization have already been implemented. The country now faces the prospect o f tightening macroeconomic constraints and an increasingneed for better prioritization and management o fpublic expenditures to eliminate absolute poverty. .. 11. This, the second phase o f the Public Expenditure Review (PER), covers aspects o f sectoral spending infour major sectors: education, health, roads and water. These sectors were selected because they account for the bulk o f government spending. Agriculture i s omitted because the backgroundwork had not been completed by the time o f going to press. ... 111. The principles underlyingthis evaluation are the standardones o f economic analysis: efficiency, equity andimplementability. Government intervention may raise efficiency by compensating for a market failure such as the inability to borrow (as inthe standardrationale for subsidized education andhealth services), or the inability o f entrepreneurs to charge for service (as inthe standard rationale for government provision o froads). Equity i s considered from the points o fview o fpoverty reduction, regional inequality and gender inequality. Implementability is a factor because there may be historical, political and sociological reasons why certainreforms are not adopted over longperiods o ftime. Inaddition, the policy maker mustweigh up whether the market failure exceeds the possible "government failure" inimplementation, stemming from a lack o fhuman capital or other sources. Since implementability is a large subject inits own right, it i s givenonly light treatment inthis study, the details beingleft for further work. Nevertheless inthis study a distinction is made between reforms which can be executed readily and those involving complexity, continued commitment and substantial technical resources (para. xlviii, p. xiv). 'Evenwith the ambitious reformsproposedinthis Review, only one o fthe MillenniumDevelopment Goals (MDGs) -that relatingto clean water inrural areas -is likelyto be metby 2015 (see the summaryof the MDGs inTable 37, Annex I.) .. 11 The macroeconomiccontext iv. Mozambique's recoveryfrom the devastatingjloods of 2000 is almost complete as the economy grew by 7.7 percent in2002 and i s expected to grow by 7 percent in 2003, driven by investment inmega-projects and recovery o f agriculture. Small and medium enterprises, however, have been negatively affected by very highinterest rates stemming from monetary difficulties inrecent years. Due to an expansionary monetary policy, the inflation rate rose to 22 percent at end-2001, compared to the PARPA target o f annual inflation o f 5 to 7 percent. Followinga tighteningo fmonetary policy, the inflation rate declined to 9 percent bythe end o f 2002 and i s expected to decline further over the next few years. v. Government revenues have grown significantly, from I I . 3 percent of GDPin I997 to 13.5percent of GDPin 2002, exceedingthePARPA target of 12.4percent. This, coupled with highlevels of donor support, allowed expenditures to expand from 23.5 percent o f GDP in 1997 to 32.2 percent in2002. But even with highlevels o f extemal grant and foreign net borrowing, domestic borrowing has swung from significantly negative to a positive 1.6 percent o f GDP in2002. vi. Thegrowth and current level of expenditures is not sustainable and the Government now needs to improve the quality rather than expand the quantity of its expenditures. Itwill be increasingly difficult to raise revenues, domestic borrowing i s already too large, and some experts speculate that extemal assistance (grants) might gradually fall from their present level o fmore than 10percent o f GDP toward the sub- Saharan African average o f 4 percent. Progress in fiscal management vii. ThePublic Expenditure Management Review of 20012noted serious deficiencies infiscal management, particularly inpublic accounting, cash management, and auditing. Public accounts omitted receitasprbprias (ministerial own receipts) and donor-funded expenditures. Cash management was inefficient and lacked transparency because largenumbers o f Government accounts were not being tracked. Intemal auditing was ineffective due to a lack of capacity for accounting. Progress was made as some o f the key recommendations o f the PEMR(2001) to improve fiscal management were implemented: A new FinancialManagement Law was issued in2001 and its regulationswere issuedin2002. The new Law provides for modernization o fthe entire fiscal systemand establishes the basis for anintegratedfinancial management infomation system, SISTAFE (SistemaIntegrado deAdministrap70 Financeira do Estado), which is being gradually introduced. The PEMR(2001) was the first phase ofthe PublicExpenditureReview. Thepresent documentis the secondphase. . ... 111 A newbudget classification system was introduced in2001 that will permit more accurate tracking o fpoverty-related expenditures. However, its implementation at the detailed level has been delayed untilthe introduction o fthe SISTAFE in 2004/05. viii. Some recommendations of the PEMR (2001) crucial to the reform process have not yet been implemented. Two o fthese are indispensable for the proper functioning o f the management information system (SISTAFE): . . Ministerial own receipts or "receitas prdprias", and donor-funded expenditures, are not beingreported infull inthe budget andbudget executionreports. The hundredso f accounts of the Government with the central bank and commercial bankshave not been inventoried or consolidated. T h e intersectoral allocation of resources ix. Budget allocations topoverty-oriented expenditures have risen. Current budget allocations to health and education rose from 4.1 percent o f GDP in 1998 to 6.3 percent by2002, on top o freal GDP growth o f35 percent duringthe period. These increases make Mozambique's social sector allocations some o fthe highest inthe region. These intersectoralchoices may well be appropriate inthe light o fMozambique's very high levels o fpoverty. x. Further reallocations infavor of health may be calledfor,j?nanced by cuts in non-priority areas. Ineach o f the four sectors studied indetail -education, health, roads andwater -there are needs for additional financing, without which the Government's medium term goals for the sector will not be attained. Given the poor state o f monitoring (see below, para. xii), absorptive capacity is an issue inall the sectors, particularly in health, with its weak links between inputs and outputs, andwater, where substantial under-executioni s the norm (see below, para. xiv). While the study didnot quantify the marginal welfare impacts o f competing sectoral investments, it i s hardto avoid the conclusionthat some further reallocations infavor o fhealth will be called for, merely in order to protect the gains made so far, since the health system's resources will be increasingly captured by HIV/AIDS patients with opportunistic diseases. Consideration may be givento makingcuts inthe sectors defined by the PARPA as "non-priority", and inparticular inthe rubrics ofdefense, embassies, the "non-priority" component of "security andpublic order", and recreation, culture andreligion. Commonthemes insector spending xi. Many o f the key findings ofthis review are specific to the four sectors involved - education, health,'roads and water -andmay be consulted inthe relevant sectors below. However, the review i s unifiedby the following issues common to all the sectors: . a problematic policy process, leading to poor monitoring and serious informational gaps; inadequate allocations and/or chronically low execution rates o fpoverty-oriented expenditures; ... iv anurgent need for civil service reform; the need for progress with decentralization; and the devastating impact o fHIV/AIDS. xii. Problems with thepolicy process lead topoor monitoring and serious informational gaps. Otherwise stated, there is a lack o f government leadership and prioritization inexpenditure management. Inturn this i s linkedwith a lack o f focus on outcomes, as was stressedby the first annual progress report on the PARPA (Republic o f Mozambique, 2003). The Government does not always have a detailed strategy for donor contributions to sector programs. Instead donors make proposals ina decentralized fashion and these tend to be accepted bythe authorities. The result is a mosaic o f programs based on different philosophies, o f differing quality, andwith widely differing cost structures (e.g. inschool construction, water point development). Although there are strategy documents inplace inthe key sectors, these often consist o fbroadly expressed ideals, without detailed medium-termprograms. Therefore the linkbetween expenditures, targets and outcomes i s not clear. xiii. The chronic lack of information on inputs, outputs, outcomes andflnancials makes assessment ofpublic expenditures difJult. The Government needs this information on a regular basis to facilitate allocation decisions. The annual progress report on the PARPA (Republic o f Mozambique, 2003) identified this weakness inthe policy process. The Government intends to respond by .. more carefully defining and monitoring PARPA objectives andtargets, restricting attention to a smaller list o fkey items for tracking; using the Plano EconGmico e Social andthe budget execution reports as its chief . instruments, for rational planning andmonitoring o fthe achievement o f outcomes; mounting regular broad-based discussions o fthese key monitoring documents in meetings termed the Poverty Observatory. This Review further recommendsthat . the cycle o f financial management incorporate regular elements o f expenditure review, partlyby creating in-house capacity andpartlybyhiringinskills as . needed. Indeedthe Ministry o fFinance is seeking, through this Review, to initiate this; the actionplano fthe PEMR(2001) becompleted, inter alia creating . mechanisms for reporting information on donor contributions andproperly implementing the detailed functional classification which was adopted in2001; donors increase their contributions channeled throughthe budget. This would require further reforms o fpublic finance, including thoroughgoing procurement reform, improved accounting and auditing, and steady implementation o f the integrated financial management information system, the SISTAFE. V xiv. There has been chronically low budget execution in somepriority sectors3, particularly water and health. Current expenditures incertain sectors (e.g. water, health) averaged considerably less thanthe budget allocations between 1999 and 2001. Among the causes, according to the ministries concemed, i s late arrival o fthe first o f the duode`cimos(or one-twelfths o fthe annual allocations). Another appears to be a lack o f capacity to operate the duode`cimosystem for releasing revenues to ministries. This system provides for replenishment o f funds against documentation o f the use o fthe funds inthe previous month. Since documentationofthepastmonth's expenditures is often delayed or incomplete, allocations to ministries tend to lag, sometimes badly, leading to chronic under-spending. It is expected that the gradual implementation o f the SISTAFE system will change this, but inthe meantime specific training i s needed inaccounting and the requisite procedures. xv. I n a time of budget stringency it ispossible to improve the quality of service delivery by rationalizing and reallocating spending. Some examples given inthe main text are: . to save resources by using local materials for construction ineducation; and fully fundingperiodic roadmaintenance to reduce rehabilitation costs; to reallocate subsidizedsewices to thepoor, including shifting the implicit subsidy on urbanpipedwater to rural and small-town and urban standpipe services, and further shifting expenditures away from largehospitals to rural and small-town facilities; and toprevent and treat HIV/AIDS and counteract its social and economic effects. xvi. The civil service is short of skilled, dedicated employeesas it is d@?cultto attract high-quality, motivated civil service stafJ Salaries are well below comparators (equivalently skilled private sector workers, or civil servants inneighboring countries), particularly at the upper skill levels. For technical specialists the differential i s about 50 percent. Various "off-budget" revenues are sometimes usedto supplement salaries. Such ministerial own receipts or receitasprdprias are large: in 1998 off-budget receipts o f the healthministry amounted to about 37 percent o f its current budget allocation. Thoroughgoing reform o f the civil service i s needed, including improved compensation andlinkingo fperformanceto promotion andpay. The Government is embarking on a pilot reform in2003 for the ministries o f education, health and agriculture entailing inter alia the introduction o fremunerationrelated to performance. Donor assistance maybe called upon to fund the decompression inthe initial years. xvii. The Government should continue to gradually decentralize administrative and jiscal responsibilities, and increase the amount of training inpublic administration at local level. A gradualist approach has been followed so far regarding the extension o f responsibilities to provinces, districts and municipalities. This gradualist approach seems appropriate inlight o f the dearth o f capacity inpublic administration outside Maputo. 3 The PARPA defined the "priority sectors" as education, health, infrastructure (including roads, energy, water), governance (including justice), agriculture, and macroeconomic management. vi The Government should act to expand the capacity and the role o f local administrations vis-&vis the central govemment, by introducingadequatetrainingprograms and gradually transferring fiscal and administrative responsibilities. Inthis context, it will be important to reform the current system o f dupla tutela o fprovincial directorates, since this system is notreadily compatiblewith theprocess o fdecentralization. xviii. Thesocial and macroeconomic consequencesof Mozambique's 13.8percent HIV/AIDSprevalence rate are substantial. HIV/AIDS could reduce per capita GDP growth rates by as much as 1percent annually inthis decade (Amdt, 2003), negatively affecting revenue growth. HIV/AIDS also has implications for expenditures. Inthe education sector, HIV/AIDS results inabsenteeism and deaths o fteachers, andthe need to subsidize due to orphaned children. The number o f orphans inlower primary school can be expected to increase from the current 10percent to 18 percent in2015. The cost to the orphans' guardians o f enrolling them was US$2.6 million in2002, and could rise to US$4.6 million by 2006. Inthe health sector, the costs o ftreating patients, carrying out prevention programs andreplacing andtraining staff will be considerable. To provide palliative care andtreatment o f opportunistic diseases to 20-30 percent o f the population with HIV would absorb as much as halfo f total healthexpenditure; adding anti-retroviral therapy for 10percent o f HIV/AIDS victims would cost more than another half. The Government is carrying out a detailed study o fthe likely costs to the health systemo fthe HIV/AIDS epidemic, but it is clear that the budgetaryimpact will be large. Education xix. Access toprimary education has improved dramatically in recentyears. The gross enrolment rate for the five years o f lower primary (EP1) has risen from 56 percent in1992to 106percent (95 percent for girls) in2002. Theincrease ingross enrollment in the two years o f upperprimary(EP2), from 13 percent to 28 percent over the same period, was less impressive, though important giventheir initial low base. The improvement at the primary level is partlydue to an increase indomestic expenditures (recurrent and investment), from 1.9 percent in 1997 to 3.1 percent of GDP in2001, and an increase inthe social demand for access to educational services. However this is still below the 5.1 percent average insub-Saharan Africa. Access to secondary schooling is lagging, with a gross enrolment rate o f only 8 percent. xx. The education system, however, remains highly inefficient. Drivenbyhigh repetition and dropout rates, the completion rate o f EP1is 36 percent, compared with an average o f 51percent inAfrica and 81percent inall developing countries. It takes 18 years o,finstruction to produce a primary graduate (EP1and EP2) insteado f the nominal seven. xxi. Schoolfees are a likely contributor to low completion rates. At the primary level, 18 percent o fthe unit recurrent cost is paidbyparents. This amounts to 12percent o f disposable householdincome o f the lowest-eaming decile. Muchofthe money i s used for cleaning and security services, and for books. Inthe 1997 household survey, fees were cited frequently byparents inrural households as a primaryreason for not sending children to school. Fees could be reduced at the primary level by a combination o f increased allocations to schools and improvemento f the exemption arrangements for vii students from needyhouseholds. Insecondary education, since the benefits from public spending are not equitablydistributed, the current highlevel o f cost recovery (about 30 percent) should be maintained, inparallelwith making bursaries available to deserving students. xxii. The decline in the quality of education needs to be reversed. The planned introduction o f a new curriculum will help to address some issues relatedto quality and efficiency. The magnitude o fthe effect, however, will depend on parallel programs to upgrade teacher subject knowledge, improve teaching practice, and ensure an adequate supplyofteaching guides andtextbooks to support the new curriculum andupgrade school infrastructure. The trendtowards hiring teachers without appropriate training qualifications shouldbe reversed, and a priority placed on improvingthe supply and qualityo fteacher pre- andin-service teacher education programs. Inthis context, it is recommendedthat the Government improve the quality o fteacher educationprograms offered at CFPP centers to ensure that all new primaryteachers have CFPP qualifications. Teacher educationprograms (length and curriculum) shouldbe restructured significantly to increase the supply o f qualified teachers. The CFPP system should be reformed to adopt alternative training modalities aimed at reducingresident time andincreasing practicaltraining xxiii. Reallocations within the education budget could improve efficiency while remaining within an envelope consistent with thePARPA. Savings canbe made by (i) reducing the number o f "ghost employees", as teachers on the EP1payroll outnumber those recordedby the schools by 20 percent; (ii) improving the use o f teachers' time in EP2; (iii) reducing construction costs; and (iv) reducing the subsidy on tertiary education. These savings would likely be sufficient to compensate for (i) increased training, (ii) more teachers, (iii) proposed scholarships for AIDS orphans; and (iv) the proposed the reduction inprimary school fees. xxiv. I t is unlikely that theMillennium Development Goal of universalprimary school completion will be attained unless drastic efficiency improvements are made. Projections indicate that important progress inraising completion rates inprimary schooling could be made given the present budgetary allocations, but the goal o f universal primary school completion will likely not be attained unless additional resources are allocated for recurrent andcapital expenses. It may be necessary to consider modifications inthe staffing profile and compensation structure o f primary and secondary school teachers to ensure the expansion is not only sustainable, but also fiscally affordable given the resource envelope. xxv. Merge the EPl and EP2 cycles into a singleprimary education cycle. The current distinction between EP1andEP2 embodies various inefficiencies and can be eliminatedwithout compromising the quality o f education. A mergedprimary school cycle would allow a substantial rationalization o f available resources across EP1and EP2 (notably teachers and classrooms), it would homogenize teachers' salaries andtheir qualifications across primary education (CFPP level), it would accelerate the school constructionprogram, andit would ultimatelyfacilitate the expansionto universal primaryeducation. ... VI11 xxvi. Improve cost efficiency in school construction andadopt a community based approach to primary school construction, particularly inrural andisolated areas. The Govemment should ensure that the construction cost per classroom does not normally exceed $10,000, including servicing and fumishing, and should revise the construction standards whichhave to be satisfied by contractors or communities. This would reduce the scope for rent seeking from school construction contracts and reduce the time it takes to complete works. In addition, the government should decentralize the school construction program, especially at the primary school level, to the provincial or district level, so as to ensure that schools are locatedwhere they are most needed. Health xxvii. Thehealth status of Mozambicans continues to bepoor, despiteconsiderable improvement in some key outcome indicators. The causes o f the poor health status include poverty, malnutrition, an environment which fosters harmful pathogens and parasites, lack o f cleanwater, poor hygiene and inadequate health services. Malaria i s the leading reported cause o f death and HIV/AIDS is becoming a critical problem. There has been little improvement inkeyhealth outcomes since the 1992 PER. In 1989, the infantmortality ratewas 137-150 per 1000livebirths(depending onthe source) andthe under-5 rate was 203; in2001 these were 129 and200 respectively. The intra-hospital maternalmortality rate fell from 230 per 100,000 live births in 1993 to 160 in2002. Life expectancy at birthwas 43-49 years in 1989 but only 42 years in2001. Part o f the decline was due to HIV/AIDS. xxviii. The Government's objectivesfor primary health care include better quality and improved equity of care. There has beenremarkable progress inthe coverage o fprimary health care since the 1992PER. The percent o f children fully immunized bytwo years o f age was 55 percent in 1995, reaching 82 percent in2001. Deliveries attended by a trained provider rose from 28 percent in 1995 to 45 percent in2002. The common pool for pharmaceuticalpurchases garners economies o f scale inprocurement and ensures efficient delivery o f drugs, at least as far as the provincial level. The overall output o fthe government health system - as measuredby a composite indicator of key outputs4-rose creditably by 59 percent between 1993 and2000. However, there continue to be serious problems. According to the Expenditure Tracking and Service Delivery Survey (ETSDS), 58 percent o fthe clinics surveyed in2002 lackedone or more essential drugs instock at the time ofinquiry. Consultationtimewas inadequate, averagingonly4 minutes,when consultations o f 10-15 minutes are needed. Various measures o f equity (e.g. the distribution o f health service byprovince) show little improvement since the early 1990s. xxix. Health spending increasedfrom US$4.6per capita in 1997 to US $7.5 in 2000 and US $10.7 in 2002, but the resultant outputs were less than expected. The health sector accounts for the largest share o f donor financing (exceeding that o f education), having received 11percent o f total grant funding between 1999 and 2002. However, the 4For the d e f ~ t i o ofthe composite indicator, see footnote 62 onpage 66. n i x increased spendingdidnot result ina substantial increase inoutput: the 66 percent increase inspending between 1997 and2000 resulted inoutput (as measured by the composite indicator) rising by only 13 percent, o f which 7 percent was due to increased service deliveredper staff member.5 It i s not clear that the additional funds were efficiently used, as indicatedby a 53 percent increase inthe unit cost o f service. xxx. The quality and quantity of health services can be raised by resource reallocation; even so, hard choicesabout carefor HIV/AIDSpatients lie ahead. Savings can be made for the Government at the tertiary hospital level by (i) raising efficiency (note, for example, that the bedoccupancy rate in2000 was only 59 percent) and by (ii) charging higherrates for more complex services on the basis o f ability to pay. The amounts forthcoming will only be known once the "off-budgets'' inthe health system havebeen clarified and regularized. Inaddition, on the assumption that health's share o f tax revenues remains constant6, there will, given the growth o f GDP and taxes, arise sizeable amounts o f domestic resources which could be usedfor expansiono f service and quality improvements. These additional amounts are roughly $7 million in2006, rising to $16 million in2010. Yet these amounts are insufficient to provide both for (i) extension o fthe reach o f the health systembeyond the current halfto two-thirds o f Mozambicans, and for (ii) even modest levels o f assistance for HIV/AIDS patients. Estimates indicate that it would cost $12 millionto provide palliative care for 30 percent o fH N / A I D S patients and $43 million to provide clinical treatment o f opportunistic infections, even without antiretrovirals. Hence difficult choices lie ahead. At the margin donor funds could be found to assist but inthe longrunthis would likely constitute a redistributionfrom other programsbecause aggregate grants are unlikelyto exceed their current dollar level o fbetween $350-$450 million annually. It i s hardto avoid the conclusion that further savings will need to be found inorder to provide at least a modest response to the ravages o f HIVIAIDS, provided that the efficiency o fthe use o f funds in the health sector can be raised. Suggestions as to where the needed savings couldbe found have been made above (para. x, p. iii). xxxi. The lack of detailedprioritization linking objectivesto spending, and based on analysis ofpast spending, is a limitation onprovision of health services and attainment of the Government's goals. The various policy and strategy documents, including the PESS, need to be consolidatedinto a consistent planninginstrument with clear prioritization linkedto resources. Such detailed prioritization is not feasible when program spending data are absent (e.g. spending, outputs andoutcomes o f the malaria vertical program are unavailable) and cannot be based upon analysis o fthe effectiveness o fpast spendingwhen the detailed functional classification adopted in2001 is not applied system-wide. Given the lack o fprioritization, andwith the tenuous linkbetween Comparingthe output increase o f 59 percent between 1993 and 2000 (see the previous paragraph), and the smaller increase of 13 percent between 1997 and 2000, it appears that the bulk o f the output improvement occurred inthe early nineties as the country was recovering from the impact o f the war. Assumptions: (i) tax revenues allocated to healthwill riseparipassu with GDP growth, (ii) maintaining current levels o f service will require increases inreal spendingparipassu with population growth, and (iii) the salary bill accruing to health increasesparipussu with GDP growth. Then the remaining funds are, in theory, available for deployment innew projects such as HIV/AIDS care and prevention. X the health service effort and some o fthe key outcome indicators, it is unlikely that the Millennium Development Goals for matemal mortality and under-5 mortality can be attained under present circumstances. xxxii. The userfee system is inequitable and lacking in transparency, and needs a completeoverhaul. Drugwastage i s common, overcharging i s widespread, health facilities charge widely differingrates, and the official exemptions for the needyare not consistently granted. The Expenditure Tracking and Service Delivery Survey (ETSDS, 2002) found that, although consultations are supposedto be gratis, 35 percent o f children were requiredto pay. This situation discourages the poorest from seeking care. A thoroughgoing review o f the entire policy and its implementation i s needed. The Government could also consider abolishing fees at Level Ifacilities altogether, because the contribution they make to the district budgets i s quite small, and little is retainedby the facilities themselves. Finally, all userfees should be reflected in the budget. Roads xxxiii. Progress has been made in the roadprogram since the 1992 PER, but road coverage is still thin. Intensive rehabilitation has resultedinan increase inroads ingood or fair condition from 10percent to 55 percent o fthe network. Most o fthe rehabilitation was o ftertiary andunclassified roads. Routine maintenance was increased from 4,000 M y e a r in 1994to 15,000 M y e a r at present. However, Mozambique's 32 kmo f roads per 1000 square kmis the lowest inthe Southern Africa regionwhere the median is 90 km. Interms o fmaintenance, too, Mozambique lags: 55 percent o fits roads are ingood or fair condition, against the regional average o f 71percent. xxxiv. Thebudgetfor road construction and maintenance constitutes 10percent of the total national budget and 20percent of the total investment budget. Roads spending comprises 36 percent o f the internally funded investment budget for the PARPApriority sectors, andhas double the share o f the next highest sector (education). Donors contributed via successive multi-donor programs, the ROCS 1and 2, the Feeder Roads Program, and most recently Roads I11(Roads and BridgesMaintenance Program, RBMMP). xxxv. Road maintenance expenditures have not increased in real terms since 1996 despitegrowing maintenance needs and a growing economy. Fundingo fmaintenance was inadequate because the real value o f the petroleum tax, which i s the major source o f . income for the Road Fund, fell by 40 percent between 1994 and 2000. Moreover, the authorities failed to transfer the portion of the tax that hadbeen allocatedby decree. In 2000, the total fuel tax was U S $36 million, but only U S $22millionwas transferredto the RoadFund. This was below requirements. The Roads I11program will succeed only ifthemaintenanceprogram (routine andperiodic) isfunded, asthe Government intends, byrestoringthe fuel tax to its real 1997 level or committing to alternative funding sources providing the same funds. xxxvi. Failure to carry outperiodic maintenance is undermining the roadsprogram. Only 18 percent o fplanned periodic maintenance, scheduled activities on larger assets, was executed duringthe 1990s. There are two reasons for this: only 57 percent o fthe fundingcommitted up to 1998 was actually provided; andperiodic maintenance is more xi difficult as it requires complex design and a longer procurement process. The Government needs to adhere to its financial undertakingsto fundmaintenance, even if this requires temporarily cutting back on the rehabilitation program andnew construction. xxxvii. Progress has been made in institutional development with the creation of a semi-autonomous roads authority (ANE). Roads administration needs to be autonomous, inorder to promote rational decision-making on the basis o ftechnical merit. The RoadFundshould be separated from ANE, as planned, andthe boards o f the ANE andthe RoadFundshould be structured to permit accountability. xxxviii. Theparastatal maintenance companies should be subjected tofull competition, and beprivatized. The parastatal maintenance companies still do much o f their work on force account, protected from competition. Furtherprivate sector development inthe form of open bidding andprivatization should help to reduce maintenance costs. xxxix. TheFeeder Roads Program and theprovincial distribution of rehabilitation and maintenance have given the roadsprogram a strongerpoverty reduction orientation. However, not enough has beendone to provide poorer districts with access to good-to-fair roads. The poverty reduction orientation should be strengthened by utilizing district-based poverty measures inex ante social impact analyses when allocating investment andperiodic maintenance budgets. xl. Roads expenditure data need a thorough overhaul. There i s a need for reliable, up-to-date data on actual externally f h d e d expenditures on roads, and reliable data are needed on unit costs o frehabilitation and maintenance so as to permit policy-relevant comparisons. Water supply and sanitation xli. Water supply and sanitation are criticalfor poverty reduction, but Mozambique has low coveragelevels -60 percent for water supply and43 percent for sanitation, including rural and urbanareas - among the lowest inSub-Saharan Africa. Spending in the sector has increased inrecent years, dueto donor assistance followingthe floods of 2000. Total public expenditure on water was $15 million in 1999, $28 million in2000 and $24 million in2001. Annual expenditure from locally generated sources (viz. Government) has consistentlybeen around $4.5 million, with the balance coming from donors. Water accounts for some 2.4 percent ofthe government-funded component o f the investmentbudget, well behindroads (20 percent), education (13 percent) andhealth (11percent). The sub-sectoral split is 56 percent on rural water supply (including small piped systems intowns), 34 percent on urbanwater supply and sanitation and 11percent on water resources. xlii. Significantprogress has been made in improving water supply in rural areas: coverage has increased from around 10percent in 1992 to around 35 percent in2002. However, sustainability i s low, with over one thirdofthe installations reportedly not functioning. xliii. Urban water supply coverage has decreased, as investment failed to keep up with increased demand resulting from population growth and urbanization. The number o f xii households with direct access to piped systems has fallen by 4 percent over the past ten years and quality o f service has deteriorated, with frequent interruptions and only 19 percent o f standpipes (the source o f water for most city dwellers) hlly functional. Over the past year, however, service has started to improve inthe five largest cities operated by a private operator. Progress has also been slow inmeeting the demand for improved sanitation facilities inbothrural andurban areas and more needs to be done to promote hygienic practices. xliv. Thepolicyfoundation for the sector is sound. The Government has made great strides inestablishing the policy foundation and institutional arrangements necessary for improved efficiency, creating an enabling environment for leveragingpublic expenditures through greater participation by communities, users andthe private sector inways that will improve sustainability. Further work is neededto bringdownthe costs o f investment requiredto improve coverage. xlv. To meet the PARPAgoals, and the Millennium Development goals, increased public spending and increased efficiency are needed. For rural water supply, it i s highly likely that the PARPAtarget o fincreasing coverage from the current level o f35 percent to 41 percent by 2005 can be met ifthe demandresponsive approach is followed to ensure sustainability. However, it is uncertain whether the PARPA targets o f increasing direct access to urbanwater supply from the 2001 level o f44 to 50 percent and increasing sanitation coverage from 43 to 50 percent can be met by 2005.7 The MDGs seek to reduce byhalfthe number o fpeople without access to improvedwater supply and sanitationby 2015. For this, the following reforms are needed: Rural water supply: an estimated 1400rural water points would have to be installed each year, up from 900-1300 inrecent years, backed up bythe demand responsive approach for sustainability; . an effective strategy should be developedfor small pipedsystems; investmentcosts must be brought down through increased competition; training inplanning andmonitoring should be given to provincial and district directorates responsible for water supply Urban water supply over 21,000 connections would need to be installed each year. However, the figure of 10,000 connections maybe more realistic giventhat the current level i s . some 2,500 per year. Inaddition, investment and operating costs wouldneed to bereduced, andservice levels improved through increased competition; tariffs shouldbe phased up to full cost recovery levels by 2008; and 'Withsubstantially increased investment it may be possible to reachthis objective by 2007. . xiii service to the poor should be improvedby privatizing standpost operation, allowing resale ofwater by individuals andtargeting subsidies to increase direct access by the poor. Sanitation . for urban communities, institutional arrangements need to be aligned so that . planningfor water supply and sanitation is coordinated andservice is providedin the most efficient manner; for rural andperi-urban communities the Government's Low-Cost Sanitation Program should be scaled up with its focus on improved latrines; the Government should support increased participation by the private sector ininstallation of facilities andbyNGOs inhygienepromotion. xlvi. Water resource management issues need urgent attention and adequate budgetav allocations. Water resource management was not a major focus o fthis Public ExpenditureReview, as to do this would have required a large separate study. Nevertheless, due to its overriding importance inMozambique it should be stressedhere that substantial efforts will be required for strengthening o f water management capability. The Government is presentlyformulating an IntegratedWater Resource Management Strategy. Since Mozambique needs to be well equippedto negotiate agreements concerning its river basins with adjoining countries, provision for negotiation costs should be included infuture DNAbudgets. Finally, inview o f the severe flooding experienced in2000 and 2001, there i s anurgentneed for the government to develop a practical flood forecasting, warning andmanagement policy, strategy and program. The Expenditure Tracking and Service Delivery Survey (ETSDS) xlvii. The ETSDS o f2002 was a pilot surveyo fthe primaryhealth system to track the transfer of hnds from the central government, to provinces, districts and health posts. The hypothesis was that delays inbudget execution, as well as weak systems o fcontrol, with consequent scope for leakages and discretion inthe allocation o fresources, may adversely affect the quality and efficiency o f service delivery. Infact the ETSDS found that record-keeping was so poor that it was impossible to determine whether there was leakage o f funds between the different levels. It did, however, turnup evidence o f leakage o f drugs between the provincial and the district level. As mentioned above, the survey confirmed that the user charge systemis inequitable, lacks transparency and creates scope for staff to pocket the moneys collected. Finally, the survey found that only 80 percent o f staff were actually inpost at any one time. The survey, insum, revealed important information about the system's functioning andpointed the way to improvements. Several o f its recommendations have been cited above. Overall the ETSDS was a successful venture and should be repeated inthe health ministry and other ministries as a means o f improving service quality. xiv Key recommendations xlviii. Here follows a very brief bullet-form summaryo f selected keyrecommendations that emerge from the Public Expenditure review. (A detailed action plan, listing recommendations by sector, may be found inTable 36 inAnnex I, 108.) A distinction p. i s made between reforms whose implementation i s relatively straightfonvard given the political will, andreforms involving complexity inimplementation, requiringpersistent commitment andtechnical assistance as well as political will. Straightforward reforms . This PublicExpenditureReview shouldbe disseminatedwith a view to ... strengthening the capacity o f Government staff to execute future PERs themselves. Governmentaccountsat the centralbankand commercial banksshould be inventoried and consolidated. This will enable cash planningand save resources. The budget andthe budgetexecutionreportsshouldreflect, infull, the receitas prbprias inmany ministries and all donor hding. Completionratesat the primary school level shouldberaisedusing a combination o fprograms, including curriculum reform, in-service teacher education programs, reduction o f school fees (formal or informal) at the primary . level and decentralization o f several planningand resource allocation responsibilities to the districts. The correspondence o f the payroll andthe establishment (government .. employment roll) should be investigatedto identify possible "ghost employees" ineducation andelsewhere. Inroads, full fundingshouldbeprovided for periodicroadmaintenance,and periodic maintenance should be given a suitable institutional home. The urbanwater supply and sanitationprogramsshould be accelerated, and water tariffs should be raised gradually untilfull-cost pricing is reached in2008. Reforms involving complexity . The spending cycle shouldbelinkedto objectives,and monitoredusing reliable data. Usingthe PARPA as the vehicle, all the keyministries should establish explicit links between expenditure and objectives, andbetween these and outcomes. The monitoring process should inform subsequent rounds o f spendinginthe context o fthe mediumterm financial f'i-amework o fthe PARPA. Inthiscontext, public expenditure reviews will becomepartofthenormalwork . o fthe ministries. This would create a demand for reliable data on expenditures andprogram outputs. Civilservice reformisurgentlyneededto raise productivity by strengthening the linkbetweenperformance andcompensation, andto reduce the dependence o fthe ministrieson side payments. Inthe case o f education, salary moderation is called for, at least at the secondary level, inorder to permit the systemto expand. . xv Health policy documents shouldbe consolidatedina single planningsystem withprioritization linkedto resource allocation, and output monitoring shouldbe strengthened. This would require completing the plannedprogram classifier of expenditures, and implementingthe detailed functional classifier adopted by the MinistryofFinancein2001. 17 CHAPTER 1. INTRODUCTION 1. This is the second (and final) phase of the Public Expenditure Review (PER) for Mozambique. The first phase, initiated in2000 andcompleted in2001,andtermed the Public Expenditure Management Review (PEMR), dealt with the financial management system. It developed a large agenda o freform inall o f the parts o f the expenditure cycle: budgeting, execution, accounting, and auditing. Jointly with the Mozambican authorities, a final report was producedwhich included a time-bound actionplan. 2. This, the second phase of the PER, covers aspects o f sectoral spending infour major sectors: education, health, roads andwater. These sectors were selected because they account for 51percent o fgovernment spending and for 56 percent o fthe civil service, and are among the six "fundamental areas o f action" intheAction Planfor the Reduction of Absolute Poverty (termed PARPA, viz. Mozambique's Poverty Reduction Strategy Paper). This second phase o f the PER also provides an update to the reader about progress with the action plano fthe first phase, the PEMR. Finally, it reports briefly on a pilot expenditure tracking exercise carried out inthe specific case o fhealth, the Expenditure Tracking and Service Delivery Survey'. The PERi s ajoint product o f the Government and the Bank, each takingthe lead indifferent sectors. 3. The main objectives o fthe PER2ndphase are to examine allocative efficiency andcost-effectiveness, as well as the poverty orientation o f spending. Among the yardsticks used for examining the rate o f service deliveryare the targets set inthe Government's PARPA and also the MillenniumDevelopment Goals (MDGs). Although the PARPA (Republic of Mozambique, 2001) and the first annual progress report on the PARPA (Republic o f Mozambique, 2003) didnot take on the MDGs as the Mozambican Government's own goals, the MDGs can serve as a useful benchmark to permit comparison with other countries. 4. The scope o fthe in uiry is limited. Agriculture is omittedbecause, although some preparation was done ,the fullbackground paper by the Ministries o f Finance and 3 o fAgriculture was still under preparation at the time o f going to press. Conceming HIV/AIDS,researchwas done on the disease ingeneral, on its macroeconomic impact, and on its impact inthe educational sector, and some information was generated on its impact inthe health sector. But a major study on HIV/AIDS andits impact on the health sector, and measures to be taken, i s due to start during2003, ledby DNPO. It was not possible to reflect the results o fthis study inthe PER. * One o f the objectives o f the first phase ofthe PER was to do an expenditure tracking survey. This work, which was arranged by DNPO withDfID funding and execution by consultants, was substantially slowed down and is therefore reported on inthe 2"dphase o f the PER. See Finney (2003), and DNPO and DNEA (2003). 18 CHAPTER 2. MACROECONOMIC AND CROSS- CUTTINGISSUES A. THECURRENTMACROECONOMICCONTEXT 5. This section shows that Mozambique underwent a cycle o f swift monetary growth and inflation followed by traditional adjustment process o f demand restriction, the effects o fwhich are likely to beleaguer the economy for some time. Onthe fiscal side, the late 1990s saw a massive expansion inspending, but the perspective i s that spending be contained to 26 percent of GDPby2005. The major policy consequence o fthis discussion is that there will be a needfor concentration on improving the quality o f expenditures. 6. Mozambique's growth averaged a spectacular 9.4percentfrom 1997 to 2002 (Table l),driven bymegaproject construction, investment from neighboring countries, buoyant donor support, and healthy agricultural growth. The 2000 floods reduced growth to 1.6 percent but the economy recovered quickly, growing at 13.8 percent the following year. There i s probably some overstatement inthese numbers owing to the lack o f survey data inthe agriculturalvalue added estimates. Furthermore, there i s no room for complacency. Over the medium term, highgrowth rates, inparticular inthe agricultural sector, could be put at risk by constraints to further improvements inproductivity, such as land rights, transportation costs andvolatile internationalprices. 7. Thecountry is experiencing thefall-out from the easy money of the late 1990s. Reorganization of the banking systeminthe mid-1990s, together with tight money, resulted insingle-digit inflation up to 1999. Lax monetary policy (M2 growth o f 35 percent in 1999)induced a rush o finflation of 13 percent in2000 and as the exchange rate weakenedpari passu, inflation worsened to 17 percent in2002. Inresponse, monetarypolicy was steadily tightenedby increasing reserve requirements andby raising the bank rate. Inflation finally fell to 9 percent by the end o f 2002, at the cost o f extremely highinterest rates, e.g. prime o f over 35 percent at end-2002. The small and medium-sized firm sector i s inrecession, while growth is driven by large firms not dependent on Mozambican financial markets. There is an implication for government borrowing: further pressure arising from domestic financing at this time would arrest the recovery o fthe small and medium-sized firm sector. 19 Table 1. Basicmacroeconomicindicators I , 1997 1998 1999 2000 2001 2002 2003 P e r c e n t a g e , u n l e s s o t h e r w i s e s t a t e d Real GDP growth rate a 11.1 12.6 7.5 1.6 13.8 9.9 7.0 NominalGDP (Mttrillions) 40.5 46.9 51.9 56.9 73.9 92.8 107 'NominalGDP (US$ billions) 3.51 3.96 4.09 3.63 3.57 3.92 4.19 Inflation (period average) 6.4 0.6 2.9 12.7 9.0 16.8 8.5 ' InvestmentlGDP gross domestic savings/GDP 8.1 10.8 13.7 14.0 21.0 21.6 20.7 20.6 24.2 36.7 36.4 37.0 45.7 49.5 Interest rate (commercial lending rate) 24.4 19.6 19.0 22.7 25.5 lCurrentaccount deficitlGDP (excl.grants) -17.7 -18.9 -28.2 -28.7 -23.1 -29.2 -34.8 Exchange rate (000 M t / US$) 11.5 11.9 12.7 15.7 20.7 23.7 25.5 NF'V external debvexports (percent) 709.2 549.1b 212.0b 194.4 116.7 99.8 96.3 Sources: Mozambicanauthorities; World Bankstaffestimatesandprojections. "GDP in2002 was Mt 93 trillion or about $3.9 billion at the exchange rateo fMt 23,695 = $1. The sharp decreases inextemal debt in 1999-2001were due to the Highly IndebtedPoor Countries(HIPC) andEnhanced HIPC operations inthose years. The massivereductionis mainly abook entrybecauseMozambiquewas not servicingmost o f its debt inthe mid-1990s. The actual impactwas areductionin extemaldebt service fromabout $100million annually in the 1990sto $50 million after the EnhancedHIPC. 8. Theexchangerate isfullyflexible. Nevertheless the real exchange rate overshot during2001 owingto a loss o f confidence inthe currency, arisingpartly from problems inthe financial sector. As these problemswerepartially resolved, the Metical strengthened inreal terms during2002. 9. Mozambique has made important progress in tradepolicy. The top tariff rate was loweredto 25 percent during2003. Rates on capital goods andintermediates are between five and 10percent. Further reductions intariffs will follow as the SADC trade protocol is implemented. Management o f customs has beencontracted out, leading to increases in efficiency o f collection which have more than compensated for the decline inthe tariff rates duringthe 1990s. 10. A new arsenal of tax legislation is now ready." With a view to long-term fiscal sustainability, a value added tax was introduced in 1999. The value added tax is now the largest single taxation item (though the bulk o f it falls on imports). A large taxpayer unit was launchedin2001. A new income tax law was passed in2002, rationalizing corporate andpersonal income taxes, reducing the corporate tax from 35 to 32 percent, andbroadeningthe tax base. A newcode of fiscal incentiveswas passed, establishing standard concessions and transparent rules for foreign investors. 11. Dejkits after grants were relatively low until 2000, thanks to a prudent fiscal stance accompanied by substantial external assistance (see Table 2). Combinedwith a prudent monetarypolicy, especially inthe period 1996 to 1998, and a program o f structural reforms based mainly on privatization, tax and customs reform and trade liberalization, this resulted inlow inflation, highprivate investment andhighgrowth rates. At the same time, since 1998 there has been a shift inresources infavor o f health, education and agriculture, reflecting an increasing anti-poverty focus. Education, health lo For a thorough review of tax inMozambique, see Coelho et al. (2001). 20 and agriculture increasedtheir combined share intotal budgetary allocations from 29 percent in 1998 to 39 percent in2001. Table 2. Governmentfinance, 1997-2003 (percentageof GDP) 1997 _ _ _ 1998 A _ 1999 2000 2001 2002 2003 c t u a l - - - - Est. Proj. :otal revenue 11.3 11.4 12.0 13.1 13.0 13.5 14.4 'otal expenditure and net lending 23.5 21.6 24.7 29.4 31.4 32.2 28.6 Current expenditure 10.6 11.2 12.2 13.8 14.2 14.4 15.5 Compensation to employeesa 3.6 4.5 5.8 6.8 6.6 6.5 7.1 Goods and services 3.8 3.9 3.7 3.9 3.9 3.8 4.2 Interest o n public debt 1.3 1.o 0.6 0.2 0.6 1.2 1.1 Domestic 0.1 0.0 0.0 0.0 0.4 0.9 0.9 External 1.2 0.9 0.6 0.2 0.2 0.3 0.2 Transfer payments 1.9 1.9 2.1 2.9 3.0 3.O 3.1 Capital expenditure 11.9 9.8 11.6 13.7 14.1 13.7 13.0 Of which: locally financed 1.8 2.1 3.4 4.4 4.6 3.5 3.5 Netlending 1.o 0.6 0.9 1.9 3.1 4.0 0.0 Of which: locally financed 0.9 -0.6 0.0 1.6 3.1 2.4 -0.2 herall balance before grants -11.7 -10.5 -13.2 -16.7 -18.0 -18.7 -14.2 ;rants received 9.1 8.1 11.7 12.0 13.0 10.9 10.4 Project 4.8 4.0 5.4 6.7 8.1 7.5 6.7 Nonproject 4.3 4.1 6.3 5.3 4.9 3.4 3.7 >vera11balance after grants -2.6 -2.4 -1.5 -4.6 -4.9 -7.8 -3.8 Jet external borrowing 5.7 4.6 1.8 2.9 2.4 5.7 4.2 Jet domestic fmancing -3.1 -2.3 -0.3 0.8 1.9 1.6 -0.7 'ransfer o f HIPC assistance 0.9 0.7 0.5 0.3 c f e m o r a n d u m i t e m : 3xpenditure & net lending (US$ mil.) 825 856 1010 1067 1121 1261 1198 lources: Mozambicanauthorities; World Bank staff estimates andprojections. The figures for 1999 exclude, andthose for 2000 include, wage outlays that were payable in 1999but delayed until ,000pending re-certification o f civil servants. Bank of Mozambiquetransfer of assistanceunder the original HIPC by multilateral donors. 12. However,fiscal policy continues to suffer from high deficits before grants, which rose from 11percent o f GDP in 1998 to 17 percent in2000 and 19 percent in2002 (Table 2). Underpinningthese results was a substantial increase inexpenditures -which grew at 17 percent annually inreal terms from 1997 to 2002 - andwhich was not matchedby higherrevenue. The sharp increase in 1999-2002 was due inlargepart to bailouts o ftwo bankswhich could have been avoided ifthe privatizationprocess hadbeenbetter preparedand executed, and ifthe state had hlly divestedfrom the banks andpursued appropriate banking supervision. Itwas also due to an increase inthe civil service wage billof46 percent inreal terms between 1999 and200211, andto higher social spending made possible by HIPC debt relief. ''Thusthe billgrew by 13percent per year between 1999 and2002, well inexcess o fGDP growth. 21 13. Thefiscalposition of 2000-2002 cannot be maintained and, ifnot corrected, will threaten macroeconomic stability and growth. Untilnow, these deficit levels have only been possible because o f highlevels o f foreign grants. External assistance i s likely to remain highinthe short run, but inthe very long rungrants are likely to converge towards the average inSub-Saharan Africa, which is around 4 percent o f GDP (against 11percent o fGDPin2002 inthe case o fMozambique).12 Therefore, fiscal adjustment has become a priority o f Government policy. This effort will combine a relatively demanding revenue effort with measures to restrain expenditures. 14. Themacroframework envisages important changes on the revenue and spending sides. Table 3 presents the most probable scenario, on the basis o freasonable assumptions o f economic growth, revenue capability (see paragraph 10on page 19), and donor contributions (with grants averaging some US$350 million annually). The fluctuations ingrowth between 7 and 12percent reflect the megaprojects (withtheir contributions rising during construction), as well as other direct foreign investment, continueddonor support, and agricultural expansion. Revenueincreasedto 13.5 percent o f GDP in2002 and is programmedto reach 15.2 percent o f GDP in2005. For sustainability, revenue should reach 16-17 percent by 2010. Onthe expenditure side, spendingis programmedto fall, inlinewith the mediumterm perspective o fthe Government's Action Planfor theReduction of Absolute Poverty (PARPA), from the very highlevel o f 32 percent o fGDP in2002 to 29 percent in2003,26 percent in2005 and24 percent in2010. This represents a considerable reduction inpercentage terms - two percentage points per year -but inreal terms expenditures are still growing at 4 percent per year between 2002 and2010. Hence it i s not politically unrealistic. This shouldbe accompanied by a re-focusing o fpublic expenditures inpriority areas while improving the efficiency andpoverty incidence o fpublic spending. l2Note that this PERhas projections only as far as 2005, and does not attempt to make projections for the outer years on the basis o fthis assumption. 22 Table 3. Macro framework, 1997-2005 1997 1998 1999 2000 2001 2002 2003 2004 2005 - - A c t u a l - - Est. P r o j e c t e d loutput and prices Real GDP growth ratea 11.1 12.6 7.5 1.6 13.8 9.9 7.0 11.9 9.8 Inflation (period average) 6.4 0.6 2.9 12.7 9.0 16.8 8.5 6.0 5.0 Exchange rate (000 Mt/US$) 11.5 11.9 12.7 15.7 20.7 23.7 25.5 27.0 28.0 Money ( a s a p e r c e n t a g e o f G D P ) Money and quasi-money 18.6 18.6 22.7 29.5 29.5 28.3 29.1 34.5 39.8 Net foreign assets -2.2 -0.5 1.4 7.6 10.6 10.0 10.3 10.3 10.3 Net domestic assets 2.5 2.4 6.0 10.9 13.1 13.0 13.7 18.7 24.9 Credit to the government -11.1 -12.0 -10.8 -9.0 -5.7 -3.6 -3.8 -1.6 1.5 Credit to the rest o f the economy 13.6 14.3 16.8 19.9 18.9 16.5 17.5 20.3 23.4 Public Finance Total revenue 11.3 11.4 12.0 13.1 13.0 13.5 14.4 14.9 15.2 Total expenditure and net lending 23.5 21.6 24.7 29.4 31.4 32.2 28.6 26.6 25.9 Foreign financed dev. expenditures 10.0 7.6 8.2 9.3 9.5 10.2 9.5 8.9 8.5 Interest 1.3 1.0 0.6 0.2 0.6 1.2 1.1 0.8 0.7 Domestic non-interest expenditures 12.0 11.7 15.0 19.6 21.3 19.1 17.7 16.7 16.5 Overall balance before grants -11.7 -10.5 -13.2 -16.7 -18.0 -18.7 -14.2 -11.7 -10.7 Overall balance after grants -2.6 -2.4 -1.5 -4.6 -4.9 -7.8 -3.8 -3.6 -3.4 Financing Foreign financing 5.7 4.6 1.8 2.9 2.4 5.7 4.2 3.0 2.7 Domestic financing -3.1 -2.2 -0.3 0.8 1.9 1.6 -0.7 0.3 0.4 Transfer o f HIPC assistance 0.9 0.7 0.5 0.3 0.3 0.3 Sources: Mozambicanauthorities; World Bank staff estimates and projections. aGDP in2002 was Mt 93 trillion or about $3.9 billion at the exchangerate ofMt 23,695 = $1. BankofMozambiquetransfer ofassistanceunderthe original HIPCbymultilateral donors. Fiscal sustainability 15. A key contribution o fthis Review is to determine whether the spending projections madebythe respective sectors are feasible. The focus is on recurrent expenditures because ultimately these are limitedby government revenue. Startingwith the macro framework exposited inTable 3, the ceiling on spendinginthe six PARPA priority areas13i s derived, taking into account the intended increase ofthe priority sectors from 68-9 percent to 75 percent o f the total. Next the spending inthe biggest sectors - education, health and roads - for the most recent historical year (2001) is projected forward insuch a way as to keep their share intotal priority expenditures constant (Table 4). The conclusions are that the projections for roads inthe backgroundpaper (Herman, 2002), which are consistent with those o f the Roads I11program, are realistic, and have been maintained inthe table; l3The PARPA defined the following as priority (or "fundamental") areas: education, health, infrastructure (including roads, energy, water), governance (including justice), agriculture and macroeconomic management. . 23 the projections for health inthe draft sector expenditure paper (HSER 2002) . were overstated, with $187 million recurrent spendingprojected for 2005. These projections cap the figure at $170 million; the draft education sector expenditure review (ESER 2002) didnot present projections. The projections inTable 4 -with recurrent spendingrising to $135 million by 2005 - are on a path which i s well below that which would be required to put all children through primary school by 2015, the Millennium Development Goal. To achieve the MDGwould require external finance for a considerable part o frecurrent expenditures. Table 4. Feasible ceilings on recurrent expenditures in selected sectors 2002 2003 ~~ ~~~ ~~ 2004 2005 - - U S $ m i l l i o n s - - Current expenditures inPARPA priority sectors, v i ~edu- . cation, health, infrastructure (incl. roads, energy, water), governance, agriculture, macroeconomic 401 472 495 554 Feasible current expenditures ineducation, health and roads 269 312 332 373 Education 96 110 116 135 Primary 69 83 87 102 Secondary 26 27 29 33 Health 120 135 147 170 Roads 54 67 69 68 M e m o r a n d u m i t e m s Current expenditures ineducation, health and roads as: a percentage ofPARPA priority sectors 67 67 67 67 total recurrent spending 48 48 49 51 Sources; See macro framework, Table 3. B. POVERTY AND THE INTERSECTORALALLOCATION OFRESOURCES 16. DespiteMozambique's recent success, it remains one of thepoorest countries in the world. GDPper capita is $210 for Mozambique's population o f 18 million. Some 69 percent o f Mozambicans were below the poverty line in 1997; projections suggest that this figure hadfallenby2001.14 The keydeterminants ofpoverty are slow economic growthuntilthe 1990s; poor educational levels, particularly among women; high dependency rates; low agriculturalproductivity; lack of employment opportunities; and poor development o fbasic infrastructure inrural areas. Inthe agricultural sector, where poverty rates are higher, fewer than ten percent ofhouseholds, poor and non-poor, sell surpluses o fmaize, cassava or cotton. Poverty by non-income dimensions is serious: the health service reaches only two-thirds of the population; illiteracy is 60 percent; only 35 percent o f rural people have access to a protected water source. The target eight percent growth rate o fthe Poverty Reduction Strategy Paper (PARPA) would reduce the poverty l4See Simler and Harrower (2003). The projections were done using the household survey of 199617and the CWIQ survey of2001. 24 rate to 50 percent by 2010 and double the consumptionlevel o fpoor households in 12 years. 17. I t is hence appropriate that allocations topoverty-oriented expenditures have risen. Current budget allocations to health and educationrose from 4.1 percent o fthe GDP in1998 to 6.3 percent in2002 (Table 5) -this on top o fcumulative GDP growth o f 35 percent inreal terms inthe period. As is shown insubsequent chapters, with the exception o fupper secondary school andpossibly tertiary health services, education and healthbenefits are distributedprogressively comparedto the distribution o f consumption. Furtherrises in"priority spending" (largely educationandhealth) are anticipatedinthe PARPA-from 68-69 percent o fthe total (current andinvestment) budgetin2001 to possibly as much as 75 percent in2005. Table 5. Actual recurrent expenditure by functional classification,1998-2002 I t e m 1998 1999 2000 2001 2002 actual actual actual _ _ actual est. p e r c e n t a g e o f G D P - - Total recurrent expenditures 11.2 12.2 13.8 14.2 14.4 General Administration 2.3 2.6 2.2 1.8 1.8 Education a 2.0 2.5 3.2 3.0 2.8 Health 2.1 2.4 2.8 3.5 3.5 Agriculture 0.2 0.3 0.3 0.2 0.2 Roads 0.7 0.7 1.o 0.7 1.4 Water 0.1 0.1 0.1 Residual, plus all other sectors 3.9 3.7 4.2 5.0 6.I Source: Extracted from Annex I,Table 38. a Education includes primary, secondary and tertiary. The residual arises because the projected education, health, education, water and roads are subtracted from the ceiling recurrent expenditures. It also includes all sectors other than General Administration, Education, Health, Agriculture, Roads and Water. The increase to 6.1 percent of GDP in2002 does not result from fiscal pressures in one or other of the "non-priority" sectors but from the way the residual is computed. 18. The split betweenrecurrent and investment expenditures is best examinedfor the year 2000 as for that year, and that year only, hllinformation on donor funding is available (Table 6). Donor funding accounts for 46 percent o f all spendingon education, 70 percent inhealth, 75 percent inroads and 75 percent inwater. Much o f this is on the recurrent account, e.g. drugs inhealth, periodic maintenance inroads. 25 Table 6. Recurrent and investment expenditure, by function, 2000 ~ Recurrrent Investment Total - p e r c e n t a g e of G D P - - General Administration 2.2 1.7 3.9 Educationa 3.2 1.8 5.0 Health 2.8 0.6 3.5 Agriculture 0.3 2.9 3.2 Roads 1.o 0.7 1.6 Water 0.1 0.7 0.8 Other 4.2 5.3 9.5 Total 13.8 13.7 27.5 Source: Extractedfrom Annex I.Table 38. aEducationincludesprimary, secondary andtertiary. All sectorsless GeneralAdministration, Education, Health,Agriculture, Roadsand 19. Somefurther reallocation of resources towards the health sector may be needed. As Table 7 shows, Mozambique's educationandhealthspending surpasses, andwater spending falls short of, the relative levels o f countries at a similar level o f deve10pment.l~ Table 7. Budgetedsectoral allocations (recurrent and capital accounts) relative to total spending for selected countries, 2000/01 - Malawi Mozambique Uganda a s a p e r c e i t a g e o f t o t a l s p e n d i n g- Education 17.7 23.7 19.4 Health 9.8 11.4 10.8 Agriculture 11.1 4.0 4.4 Roads 3.8 5.2 13.7 Water 5.1 2.4 5.8 Sources: Budgetbooks for Malawi, MozambiqueandUganda. Such intercountry comparisons must be taken with extreme caution because o f differences inmeasurement andinpolitical, social andcultural factors. Inthis case the comparison might be taken to suggest that the reallocationstowards the social sectors have gone beyond those o f other countries andthat it may be time for an increased focus on infrastructure. More pertinently, the analysis inthe sectoral chapters shows that: . ineducation, reallocationswithinthe sector, andefficiency improvements, will probably be sufficient to achieve the key reforms, but even so the goal o f . universal primaryschool completion is unlikely to be attained by2015 (para. 104, p. 61); inhealth, reallocationswithin the sector will notbe sufficient to provide even modest levels o f assistance to HIV/AIDS patients (para. 142, p. 76) ; l5Itwas decided to use budgeted resources for this inter-country comparison rather than actuals. This is because the data on actuals o f the other countries is o fmixed quality, frequently not accounting fully for donor-funded spending. Whereas the budgets generally reflect donor fhding quite well. . 26 inroads, the desiredprogram is adequately fundedbarring one element: the Government will needto fundperiodic maintenance fully, by makingan . appropriate budgetary allocation, which could be funded by raising the petroleum tax to its previous level inreal terms (para. 172, p. 86); inwater, reallocationsarecalled for, butthese willnotbe sufficient to accomplish the Government's goals inurbanwater (para. 200, p. 94) and in sanitation (para. 202, p. 94). the called-for civil service reform with a major salary decompression (para. 40, p. 36) can probably be attained within the projectedbudget envelope o f 7 percent o f GDP, so that no additionalreallocations from Government (or donors) are needed.l6 20. The decision as to which o fthe competingneeds - education, health, roads, water -shouldreceiveadditionalresourcesisnotaneasyone,becauseallthesectorsmake important contributions to economic growth andto poverty reduction. This study has not sought to quantify the marginal welfare impacts o f competing sectoral investments. Giventhe poor state o fmonitoring (para. 35, p. 33), absorptive capacity is an issue inall the sectors, particularly inhealth, with its weak links between inputsand outputs (para. 115, p. 67), andwater, where substantial under-execution o f the budget is the norm (Table 33, p. 98). This said, it maybe that spending will have to increase inthe health area merely inorder to protect the gains made hitherto, because the health posts and clinics will be increasinglybeset by HNpatients with opportunistic diseases. Whereas inthe other three sectors the existing allocations will be sufficient to preserve existing gains andmake modest progress, provided intra-sectoral reallocations are done and provided the increased fuel tax moneys are applied to the periodic maintenance o froads. At the margindonor funds couldbe found to assist, but inthe longrunthis would likely constitute a redistribution from other programs because aggregate grants are unlikely to exceed their current dollar level o fbetween $350-$450 million annually. It is hardto avoid the conclusion that further savings will need to be found inorder to provide at least a modest response to the ravages o f HIV/AIDS. 21. Furtherexaminationwill beneededto determine where the additional savings shouldbe found, as this Review did not seek to examine indetail any sectors other than education, health, roads, water and agriculture. Nevertheless some suggestions can be made as to potential candidates for expenditure cuts: l6Accurate costing for a significant salary decompression has not yet been done because the mode and the extent o f decompression have not yet been determined. However, it i s likely that the civil service reform could be attainedw i t h the projected budget envelope o f 7 percent o f GDP (the current figure), over the next four or five years, because (a) projected GDP growth of about 7 percent (inreal terms) provides considerable room for maneuver, (b) the civil service program ought to freeze salaries at their current real levels, so that all the growth inthe compensation envelope can be allocated to salary increases as the reformprocess gets under way; (c) the civil service reformprocess will entail substantial redeployment o f staff, reducing the cost o f separations, (d) given the presence o f HIV/AIDS, the attrition rate has increased, providing more openings for redeployments; (e) it appears that substantial resources are going to "ghost employees", at least inone major sector, and that, ifthis fiding i s confirmed, these resources could be redirectedto f h d the salary decompression. 27 (a) among the PARPAnon-priority sectors, which encompass the remaining 40 percent o f the budget, the PARPA identified activities which, while not "fundamental" 17, are nevertheless "complementary activities" to the fundamentals, including policies for sustainable growth (transport and communications, technology, environmental management), social welfare programs, sectoral policies that contribute to income generation (business development, fisheries, mining, industry, tourism), andprograms to reduce vulnerability to natural disasters. Consideration could be givento reducing spendingon these "complementary activities" by, for instance, holding their allocations constant inreal terms or applying a modest percentage decrease throughout inrecognition o f the fact that these are not the keypriorities; (b) outside o f the "fundamental" and"complementary" areas are some natural candidates which could be considered for reduced spending: defense (budgeted spendingfor 2002 was $35 million or 3.4 percent o f the overallbudget), . embassies (budget on the current account for 2002 was $20 million or 6 percent o fthe central government's share o fthe current budget) . the "non-priority" part o f the rubric "security andpublic order" ($20 millionor 2 percent o f the overall budget)" recreation, culture andreligion, some o f which mightbe turned over to the private sector ($13 million or 1.3 percent o f the overall budget). c. UPDATEONTHE REFORMOFFISCAL MANAGEMENT 22. Introduction. This, the 2"dphase o f the PER, i s focused primarilyon sectoral expenditures andit will not attempt to re-examine these aspects indetail. Instead the aim is to update the reader on the actions taken arising from the agreed-upon action plans in the PEMR(2001). This will be done by going through the action plan, briefly explaining the objective o f the action, and notingwhat has been done. The details on the background to each o fthe actions may be sought inthe PEMR document (World Bank, 2001a). 23. ThePEMRfound that thefiscal management system bore serious deficiencies, particularly inpublic accounting, cash management and auditing. To name the key issues: Public accounting covered only a quarter o f Government spending, ignoring receitasprbprias (ministerial own receipts) and donor-funded expenditures. Cash management was inefficient andlacking intransparency because there were large l7The PARPA definedthe "fundamental sectors" as education, health, infrastructure(including roads, energy, water), governance (including justice), agriculture, and macroeconomic management. `'TheOrcamento do EstadoPara o Ano 2002 (Maputo, Outubro 2001) has M t 1,807 bnfor "Seguranqa e ordem publica" inthe overall functional classification of the budget (Table 9, p. 20). Ithas Mt 1,350 bnfor the same rubric inthe listing o fpriority expenditures inTable 11onpage 20. The difference is Mt 456 bn or about $20 m,or 1.9 % o f total budget o fMt 23,867 bn. 28 numbers o f Government accounts which were not beingtracked. Internal auditingwas ineffectual due to a lack o f capacity and funding. 24. The authorities moved boldly to modernize the system in2001 bypassing a new law, the Lei da Administrap70 Financeira do Estado, which inprinciple modernizes the entire fiscal management system. This bold step was taken inorder to correct a set o f difficult legacies from the past: single-entry accounting; a multiplicityo f government accounts renderingcash management impossible; and limitedbudget coverage inthat receitasprbprias (ministerial own receipts) were tolerated and donor-funded expenditures not recorded. The key reform o f the new law was to provide for an integratedfinancial management information system, entitled the Sistema Integrado de Administra@o Financeira do Estado (SISTAFE). Inorder to prepare the way for the SISTAFE, a three-phase program of actions was agreed uponwith the authorities: "pressing actions" (reproduced inTable 8 below) to be done duringthe course o fthe preparatorywork o f the PEMR, "priority actions" (Table 9) to be done during2002, and mediumterm actions for the period2003-2005 (Table 10). 25. Most of the short term `pressing actions were executed (Table 8). A new, more " detailed functional classifier was introduced into the budget, consistent with UN guidelines, with a view to tracking poverty-related expenditures more accurately. However, its implementationwas incompletebecause only the broad categories were used. The detailed classifier, which will permit tracking o fpoverty-related expenditures, was not implementedbut it i s expected that this will be done inthe course o f the introduction o fthe information management system SISTAFE. The budget, which previously hadbeen done inreal terms, owing to the very highrates o f inflation o f the early 1990s, is now being done innominal terms. The regulations o f the new financial management law were completed. Restrictions were introduced on banks accounts held bypublic institutions inthat they hadto be authorizedbythe Direcgiio Nacional do Tesouro as a co-holder, and they had to be closedthree months after the budget year to which they applied. Table 8. Summaryof PressingActions (was Table 7.1 in PEMR) Area Recommendations Time Status in Frame 2003 Budgetformulation Submit 2002 budget using new budget functional End-2001 Partially classification. done (only broad, not the detailed, classifier) Formulate the budget incurrent prices, starting with the 2002 budget. Done Legalframework Draft the implementationregulations ofthe new public March Done finance management law (Lei da AdministraGEo 2002 Financeira do Estado). Cash and asset Instructions on bank accounts ofpublic institutions: End-2001 Done management authorized by DNT, DNT a co-holder, closed by DNT on March 31ofnext year Public accounting 2002 budget execution mustbe consistent with the 2001 Partially new budget functional classification. II done. 29 26. Progress with the `priority actions "for 2002 wasgood but is incomplete. These are dealt with inthe following paragraphs. To take the first line o f Table 9, a significant share o f receitasprdprias should have been included inthe 2003 budget. Progress to date covers only about halfo f the receitas prbprias. An 11-strong list o f items inhealth andpublic works are still excluded, not to mention the education, agriculture and other mini~tries'~.Tax expenditures have not yet been included inthe Budget; these were to be included inthe 2003 Budget (line 1o f Table 9). 27. Budget execution reporting has improved, but most donor-funded expenditures are still excluded (line 2 of Table 9). The quarterly budget execution reports now present all actual expenditures usingthe newbudget h c t i o n a l classification, but only inbroad categories; the detailed classifier was not implemented. However, the revised appropriation o fNovember 2002 was not reportedinthe budget executionreport ofthe first quarter of 2003. A major step forward was made inthe budget execution report o f the first halfo f 2002 when for the first time tables were presented reporting donor-funded expenditures on the investment account. Yet as of late 2002 the reported executed expenditures were only 8 percent of the budget plan. 28. Double entry accounting needs to be introduced (item4 inTable 9). The "complementary period" after the end o fthe fiscal year, duringwhich payments may be made for commitments duringthe financial year, was reduced from three months to two. Some training indouble entry accountingwas done, but double entry has not yet been introduced throughout the system. The proposed shift from the duode`cimo system of replenishment to modified accrual accounting is probably too ambitious for the time being. Untilsuchtime as the SISTAFE system is implemented, it would bemore prudent to improve the replenishment system, through appropriate training, and make it more flexible. (See the further discussion at footnote 21, p. 31, andparagraph 39 onpage 36.) 29. Much remains to be done in cash management (item 5 inTable 9). The multiplicity of Government accounts with the central bank and the commercial banks needs to be inventoried. Although "globalizing accounts" have been created with a view to movingto a single interlinked account arrangement, this has little meaninguntilsuch time as all the accounts are interlinkedandthe Ministry o fFinance is able to readoffin real time the amount o f cash held. l9Some progress was made inthat the "receitas consignadas" inthe Budget increased from Mt48.3 bnin 2001 to Mt 67.6 bnin2003 (inreal 1998 Mt),viz. an increase of40 percent. This amount ($6 m) would be increased by another $3mby adding injust the identified receitasprdprias inthe ministries o fhealth and public works. Of a list o f 11receitas prdprias inthese ministries, turned up by a Government-sponsored study(Austral Consultoria, 1999), none hadbeenincluded inthe 2003 budget. They are: 1.H C M Serviqo de clinica especial. 2. H C M Serviqo de atendimento especial. 3. DirecqHo Nacional da Saude: Venda de medicamentos pel0 Serviqo Nacional de Saude. 4. Centro Regional de Desenvolvimento Sanitaria: Alojamento e habitaqgo pagos pelos estudantes. 5. ProduqHode material didactico. 6. ComissHo Central de AvaliaqHo e AlienaGHo de Imoveisde habitaqgo do estado. 7. Laboratorio de Engenharia de Maputo: Venda de ensaios laboratoriais. 8. Fundo de Foment0 de HabitaqHo (FFH): Percentagem na venda de imoveis do Estado. 9. FFH: Juros. 10. FFH: Venda de casas construidas com crkdito. 11. ARA-Sul Laboratorio de Engenharia de Moqambique: Venda de iguas brutas. Probably the other ministries would account for at least another $3m. 30 30. Much remains to be done in internal control and auditing (item 6 inTable 9). The internal audit department(IGF) was to have its ownbudgetline, but this was not done. It was also to be strengthened with new hiringbut this i s still under consideration. Thebudgetary allocation for the Administrative Tribunal was increased by 10percent in real terms between 2001 and 2003. It was proposedthat the Administrative Tribunal set uppartnerships withprivate auditing firms andtwinningarrangements with foreign supreme audit institutions, but as of 2003 such arrangements as hadbeenmade have had only a low level of impact. Table 9. Summary of PriorityActions (was Table 7.2 inPEMR) Area Status in2003 1.Budget coverage Little ac- complished (see text) Not done 2. Reporting (i) done, not (ii) done, not (iii)partially done Some partial reporting started 3. Legalframework Action plan & regs. done, regulations and according to a time-bound action plan. being imple- mented 4. Public accounting Done Some training launched 5. Cash and asset Not completed management incommercialbanks. Done Create globalizing bank accounts for revenue and Not completed expenditures inBM. Done 6. Internal control Not done and auditing reforms to raise IGF's capacity. Increase budgetary allocation infavor o f the Done (10% in Administrative Tribunal (TA) real terms) Grant the TA the ability to set its own salary scale. Not done Not effectively done 31 31. Themedium termprogram offiscal management reform needs to be addressed. The elements inthe mediumterm reformprogram of the PEMR (2001), reproducedin Table 10, are still valid and can be addressedwhile the SISTAFE i s being implemented. Few of them have been addressed so far. It i s proposed that the authorities seek to adhere to the original timetable. Among the major elements here are: 0 the enhancement o f the role ofthe MediumTerm Financial Framework.20 Makingthe MTFFinto anoperationalinstrumentwill aidbudgetingand execution inthe sectors that are studiedinthis PER. As i s stressed inthe subsequent chapters, the lack o f a medium-term perspective results inpurely incremental budgeting (viz. X percent increase over last year's) with no real link with activities or outputs or outcomes; 0 the introductiono f double-entry accounting, under the cash basis2'; 0 cash planningandbudgetingare among the corner stones o f the conceptual business model underlying SISTAFE andan integral part o fthe design o fthe system. Introducingmonthly and annual cash plans will enable more rational use o f resources, resulting inconsiderable savings because it will be possible to hold less (non-interest-earning) cash overall andresort less frequentlyto using treasury bills. This will be done while the SISTAFE is being installed. Once the SISTAFE i s runningthe development of such plans will be strengthened; 0 reports on domestic and external debt, lending, cash flows, and tax expenditures alongside the Conta Geral do Estado will enable a comprehensive view of state finances which i s one o f the ultimate objectives of the new FinancialManagement Law22; 0 auditing the state accounts (Conta Geral do Estado) within 12months o fthe close of the financial year will enable Parliament to exercise closer control o fthe expenditure process. Whereas presently the long delaysz3inissuance ofthe documents render the documents irrelevant. 2oNote that the DirecqBo Nacional do Plano e Orqamento issued a study, "A experihcia com o Cenario Fiscal do MBdio Prazo e opq8es para seu futuro desenvolvimento" (2003). 21A modified&basis o f accounting will be appliedwith the implementation o f SISTAFEthrough the introduction o f a budget credit systemagainst which commitments willbe registeredand accounted for when a transaction is incurred. The release o f credits will serve as virtual cash allocations and will determine and control the ceiling o f spending per budget line item. Inthe meantime, while the SISTAFE systemis being implemented,the replenishment ("duod&cimo") systemneedsto be made more flexible and training should be givento enable the lowest level units to operate withinit efficiently. See also the discussion at paragraph 39 onpage 36. 22Note that underthe cashbasis o f accounting, as definedby the Public Sector Committee o fthe International Federationo fAccountants (IFAC), it is standard to include information about accruals, as memorandum items, not ledger entries. The following are normally reported: financial assets receivable, long term debt, contingent liabilities and guarantees, and other liabilities. 23The Conta Geral do Estado for 2001became available to the public inearly 2003. 32 Table 10. Summary of Actionsfor the MediumTerm (was table 7.3 inPEMR) 4rea Recommendation rime Statusin2003 Frame Budgetformulation Medium Term Financial Framework (MTFF): Reinforce 1003 !Jot yet done. the MTFFby (i) integrating it inthe decree for the revisedregulations of the SISTAFE and (ii) it amaking public document. Planning documents: (i) Eliminate the Three-Year Not yet done Investment Plan (PTIP) as a stand-alone document and treat investment expenditures w i t h the normal budget formulation process. (ii) Develop the Economic and Social Plan(PES) as the key instrument to monitor and program the implementation o f the PARPA. Public accounting Launchthe introduction o f double-entry accounting. 2004 Now expected in2004 Free the accounting department (DNCP) of all activities 2003- Not yet done not related to accounting and reporting. 2004 Reporting Develop and make available on a regular basis, financial 2003- Now envisaged reports inaddition to the Conta Geral do Estado, starting 2004 for a later phase with (i)a report on short andmedium-term external and o f the domestic debt, (ii) a report on lending and on-lending, SISTAFE. (iii) oncashflowsand(iv)areportontax reports expenditures. Cash and asset Introduce a treasury single account simultaneously with Sept. management the new integrated financial management information 2003 system. Improve financial planning with the introduction o f 2004 Now expected annual cashplans, budget implementation plans and in2004 monthly cashplans. Tax administration: (i) the current mechanism o f Extend Not yet done VAT collection through the banking system to other taxes after assessingits feasibility bybanks operating in Mozambique. (ii) Introduce, whenever possible, a single document (Documento Unico) for the collection o f all taxes. Internal control and Ensure that the state accounts (Conta Geral do Estado, 2004- To be done in auditing CGE) are audited within 12 months after the endofthe 2005 2004 fiscal year, starting with the 2003 CGE. Launchbudget evaluation function. 32. An ambitious timetable has been setfor the installation of the SISTAFE. During 2001 and 2002, the technical unit for reform o f State Financial Management (UTRAFE, "Unidade Ttcnica para a Refonna da Administraqiio Financeira do Estado") was set up and the regulations for the new law were prepared. At the same time planningwas done for the installation o f the information management system SISTAFE. A budget o f $27 million was developed, and fundingfor the bulk o f it was secured from donors including the World Bank. The timetable (UTRAFE2002, p.12, UTRAFE2003, p. 2) anticipates procurement (viz. determination o f the IT architecture and account component package) inthe secondhalfof2003, roll-out ofthe system inthe Ministries ofFinanceand Education inthe last quarter o f2003, and extension to the remainingsector ministries in 2004. The focus will initially be on the introduction o f a single treasury account in2003 for all line ministries and the improvement o f accounting. After 2004 the focus will shift 33 to internal control, auditing, accounting for state property(PatrimBnio) and debt management. 33. During2001 the IMFandthe Bank engagedthe government, inthe context o fthe HIPC, inan overall planfor public expenditure management reform which brings together the different elements -PEMR (2001), the SISTAFE, and items inthe Fund's Poverty Reduction and GrowthFacility (PRGF). This was termed the Public Expenditure Management Country Assessment and Action Plan (Fiscal Affairs Department et al., 2002). It developed 15 benchmarks o fthe quality o fpublic expenditure management. A March2002 Boardpaper reportedthat, as with most HIPC countries, Mozambique scored poorly, with 5 benchmarks met. It i s not necessary to report the details here because the key problems were identified inthe PEMR (2001) and the update above conveys the actions which the Government is taking to right the situation.24 D. CROSS-CUTTINGISSUESINSECTORAL SPENDING 34. Several specific themes came up repeatedly duringthe work on the five sectors (education, health, roads, water, agriculture). It is economical to treat them all ina consolidated fashion here: ....a policy process which underplays prioritization andmonitoring o f outcomes, leading to poor statistics; chronic underspending incertain sectors; the need for civil service reform; and the need for progress with decentralization. Another common theme i s the impact o f HIV/AIDS. Since there are differences inits causality andthe extent o f its impact across sectors, this discussion i s placed inthe sectoral chapters - ineducation sector on page 59, andinhealth on page 76. The policyprocess,expenditureprioritization, and outcomes 35. Problems with thepolicy process lead topoor monitoring and serious informational gaps. Mozambique faces the dual problems o f a lack o f government leadership and a lack o f prioritization inexpenditures. Inturnthis is linkedwith a lack o f focus on outcomes, as was stressedbythe first annual progress report on the PARPA (Republic o fMozambique, 2003). Like some other countries inreceipt of large amounts of external finding, Government does not consistently direct expenditures inline with nationally determined objectives. The policyprocess does not always consist o f formulating a strategy andthen having donors contribute to specifiedparts o f it. Instead donors make proposals and, ina decentralized fashion, projects are taken on. The result i s a mosaic o fprograms based on different philosophies, o f differing quality, and with 24 In addition, DfID undertook a fiduciary risk assessment covering several o f these areas (Lernstrup, 2002). 34 widely differing cost structures (e.g. inschool construction, waterpoint development). The symptoms o fthis fundamental problemmanifest themselves inseveral forms: I Outcomes are frequently not measured or reported, and ingeneral have scant bearing on the policy process. It is extremely difficult to obtain data on donor-funded expenditures, and particularly actuals. Hence it is difficult to obtain information on unit costs andit is virtually impossible, without a dedicated PER-type o f exercise, to compare the relative efficiency o f differingapproaches andprograms. There are few data on programs. Basic information such as their geographical location, the intendedbeneficiaries, disbursements, measures o f their output and impact, etc. are hard to come by. This was true to a greater or lesser extent of all the sectors studied- education, health, roads and water. The financial projections inthe MediumTerm FinancialFramework (MTFF) and latterly the PARPA are only loosely relatedto actions undertaken. I Government strategies sometimes consist o fbroadly expressed ideals and statements o f intention to produce precise analyses andrecommendations (see esp. the PESS, Health Sector Strategic Plan). 36. Consequentlythe sector-wideprograms failed in their informational aims. A sobering discovery o f this public expenditure review i s that none o f the sector-wide programs ineducation, health, roads or agriculture automatically provided data linking expenditures, outputs and outcomes. Similarly the review o f the policy process by Chichava et al. (2001) found that the mechanisms o f collection and dissemination o f data within government are insufficient. While most sectors produce vast amounts o fdata, the key elements for public expenditures and for decision-making were often absent. Specialized and arduous efforts were requiredto obtain basic expenditure data (especially donor-funded expenditures) inthese sectors (as well as in~ a t e r ) . There ~ ~ ~ ' ~was a conspicuous lack o f data on ministerial own receipts (receitasprbprias) ineducation, health and agriculture. Inhealth, there were no breakdowns o f spendingby level (e.g. hospitals vs. healthposts), location (capital, town, rural) or vertical program (e.g. malaria). 37. It isproposed that the Government seek to correct thepolicy process by using the PARPA toprioritize and expand its leadership. The Government has already started this process by initiating work on more precisely defining the targets o f the PARPA and bringingthe "priority ministries" closer to the process o fpursuing these goals. During 2003 the Government initiated a series o f "Poverty Observatories" at whichthe key documents -particularly the PARPA andthe Annual Progress Report on the PARPA- 25 Reliable data on donor-funded expenditures were not available for education for any recent year other than 2000; they were not available for health for 2001; those on roads were made available, but are not reliable for any recent year. 26 For details on sources, and the nature o f the gaps ininformation, the reader i s referred to Table 38 on page 111. 35 would be discussed with the sectoral ministries and with key stakeholders outside the Government. To support this process, the following could be added: . An element o fpublic expenditure analysis andreview shouldprogressively be integrated into the cycle o f financial management. This may entail analyzing a handful o f sectors per year, and may incorporate the hiring-ino fthe necessary skills. Intime, such review could become a constitutive element o f the budget cycle, as has been done insome countries o f the region (see Moon, 2003), although this i s likely to require considerably more capacity than is available at present. The review o f the policy process by Chichava et al. (2001) recom- mendedthe creation o f a cadre o f public policy analysts. The Ministry o f . Finance i s seeking, through this PER, to create such in-house capacity, but much remains to be done; The actionplan o fthe Public Expenditure Management Review (2001) should be . completed, creatingmechanisms for reporting information on donors' contributions; Civil service reform is needed, linkingcompensation to performance, and providingappropriate levels of compensation(see p. 36). The present environment o f inadequate compensation creates undue incentives at the ministerial and sub-ministerial level for seeking donor-funded projects as these usually provide some level o f benefits to the operatingunits. This inturn becomes the driver o fproject selection, rather than the desired "bird's eye view" which determines the portfolio inthe light o f the overall goals and outcomes. At the same time as civil service reform there should be a general tighteningup of the conditions for ministerial own receipts (receitasprbprias); for ifthere i s . proper compensation there is no need for extra-budgetary transfers to support hiringandstaffretention. The donors shouldincrease the share o f their contributions that is channeled directly through the budget. Inorder to increase their confidence inthe fiduciary system, the Government should pursue the program o freform o fpublic finance - thoroughgoing procurementreform, improved accounting and auditing, andthe establishment o fthe informationmanagement system SISTAFE. Chronicsectoralunderspending 38. Some sectors underspend regularly, especially the water andhealth sectors, andin some years inthe roads sector. The water sector spent an average o f 63 percent o f its current budget allocation between 1999 and 2001. The health sector spent 86 percent o f its recurrentbudget in 1999 and 80 percent in200027. There are considerable intra- province differences inexecutionrates inthe health sector, e.g. inZambCzia, execution rates for the 1999 budget range from 98 percent to 52 percent (HSER 2002 p. 66). ''See also the study by Harding (2003). 36 39. There were two main causes o funderspending: (a) the lack of capacity to operate the duode`cimo allocation system, which i s essentially one o freplenishment o f accounts against rendering o f accounts (viz. proofo f correct use o f the funds, orprestaqio de contas) inthe previous month; and (b) the timingproblem since late arrival o f the first duode`cimosputs some sectors (e.g. education) inacute difficulties because their greatest needs arise at the beginningo f the year. Insome ministries the first duode`cimoarrived between five and seven months late inrecent years. The replenishment arrangement - standard inthe region -is good because it creates a powerful incentive for accountability, helpingto limit overspending. Specific training is neededinaccounting andthe use o f the duode`cimo system to makeit work without undue delays. The timingproblem can readily be addressed by having the relevant sectoral ministries agree with the Ministry o f Finance to align the time patterno f advances more closely to the time pattern o f expenditures. (Eventually this arrangement ought to change as coverage o f the SISTAFE increases, gradually including more operational units inthe "single treasury account": for underthis arrangement no advances will begiven, but operationalunits will make commitments, payments for which are executed by the central treasury.) Civil service reform 40. I n most sectors, salaries are well below comparators (equivalently skilled private sector workers, or civil servants inneighboring countries), particularly at the upper skill levels and for technical andprofessional staff. Doctors, for instance, are paid about 50 percent less than inLesotho or Botswana (HSER, 2002). A study in2002 revealed very large differentials betweenpublic sector and private sector compensation (Table 11). After taking into account subsidies andbenefits, theprivate sector premium was 160 percent for nurses and 210 percent for professionals such as accountants. Numbers o f a similar magnitude were turned up by a study in2000 (Sulemane andKayizzi-Mugenva, 2001): 200 percent for senior managers, 280 percent for professionals, and 580 percent for unskilledworkers. Differentials o f this magnitude are well inexcess o f the non- monetary benefits conferred bypublic service such as security o f employment. Table 11. Ratio of private sector to public sector compensation Occupationalgroup Ratio: privateto public compensationa Nurses 2.6 Professionals 3.1 Source: Pesquisa Salarial(2002), Tables 5 and 6. a =salary+ subsidies +benefits. Economists, managers, lawyers, auditors, accountants, engineers, IT experts. 41. Poor remuneration undermines the civil service. It is difficult to attract high- quality motivated staff, and staff are induced to use various "off-budgets" or dual employment arrangements to supplement their salaries. For instance, some ministries operate funds to which clients contribute inexchange for services, but which are not recorded on the budget. A Government-commissioned study on off-budgets(Austral, 1999) found that substantial funds are not captured by the budgetprocess and that many o fthese lack a legal basis -the healthministry in 1998 collected $3.8 million(37 percent 37 o f its current budget allocation) and the public works ministry$0.8 million (107 percent o f its current budget allocation), inthe form o fvarious fees. None o fthis hadbeen moved on-budget by 2002. Some part transferred to staff inthe form o fvehicles, training courses and other rewards, as i s done also with the budget category "other goods and services" inhealth (see HSER 2002). Over-charging beyond the official user fees occurs inthe healthsector.28 Certainministriespayunder-cover subsidies of80-100percent of the base salary (PesquisaSalarial, 2002, p. 31). 42. I n one sector - education - certain salaries are well above comparators. The salaries o f secondary school teachers are approximately double the salaries incomparable countries, inper capita GDP terms. Salary moderationmay be needed to accomplish the expansiono f the system so as to achieve the poverty reduction and development objectives o f the PARPA. 43, An investigation is urgently needed of the correspondence of salary payments and thepersonnel register. Evidence inthe education sector suggests that about 15-20 percent o f employees are "ghost workers" (see the education chapter). Civil service compensation increased by 46 percent inreal terms between 1999 and 200229,rising as a percentage o f GDP from 5.8 in 1999 to 6.5 in2002. Civil service employment didnot increase inanything like this measure. No breakdown o f the increasedpay is available - e.g. to explain the components o f salary increases, increases indeath anddisability payments due to HIV/AIDS, benefit changes, and increases inpersonnel. Indeed even the number o f civil servants is not knownwith accuracy, estimates ranging from 106,500 (inthe year2001)30to 121,562 (inthe year 2000)31. Thenumber ofworkers inthe personnelregister (SIP, Sistema de Informaq5o de Pessoal) i s also not known with accuracy: it had 83,777 persons in2001 according to MAEstaff32,but 99,158 persons in 2001 according to UTRESP, (2002). Given the lack o f clear information, the deficiency o f the SIP compared to actual employment, the unexplained increases inremuneration, andthe alleged ghost employees incertainministries, it is urgentthat a thorough investigationbe launched. It could be that considerable savings could be made through checking the correspondence o fthe personnelregister andthe salary bill. 44. Thepublic sector reform program will introduce salary reform once the basic personnel reforms are done. The Public Sector Reformprogram, supported by the World Bank and other donors, got under way in2002. Three ministries -health, education and agriculture -will serve as pilots for the reform process which will reorganizethe ministries' employment configuration interms oftheir fundamental objectives, and also linkperformance to remuneration. Oncethesebuildingblocks are inplace, the salary reform will be introduced. Since the costs will be considerable, it maybe that donor support will be sought initially, on a decrementingbasis untilthe government is able to 28See HSER, 2002, p. 73, and the discussion at paragraph 120on page 68. 29See footnote 11. 30UTRESP, 2002, p. 5. 31Sulemane and Kayizzi-Mugenva, 2001, p. 9. 32A breakdown of SIP employees by "carreira" suppliedto the Bank and the FundinFebruary 2001. 38 shoulder the burden itself; though ifthe investigationo f the personnel register called for above secures sufficient savings. The timingo fthe reforms is uncertain, as i s the magnitude of the salary adjustments andthe component o f donor assistance, and so no attempt i s made here to reflect the salary increases inthe fiscal projection^.^^ Progresswith decentralization 45. ThePEMR (2001) highlighted the important role of decentralization and deconcentration for the improvement o f fiscal management and service delivery ina country as vast anddiverse as Mozambique. So far, the strategy o freform o f the state has beenpursuedboth interms of decentralizationinthe urban zones (granting municipalities greater political, administrative and fiscal autonomy), and deconcentration inthe ruralzones (delegation ofsome responsibilities, while retaining fiscal control, to provinces and districts). From a fiscal perspective, this seems an appropriate model in the specific case o fMozambique as most "taxable" economic activities are concentrated incities andtowns. A very gradualist approachhasbeenfollowed so far regardingthe extension of responsibilities to provinces and districts. This gradualist approach seems appropriate especially inlight o fthe dearth o f capacity inpublic administration that is especially acute outside Maputo. However, the PEMRalso argued infavor o f re-thinking and expanding the capacity andthe role o fprovinces and districts vis-&vis the central government. Inthis context, the PEMRunderlined the importance o f reforming the current system o f dupla tutela o fprovincial directorates bothto their sectoral ministry andto the governors, since this system is not readily compatible withthe process o f decentralization. 46. Administratively, Mozambique is divided into 10provinces and 128 districts. The only form o f elected local government is the 33 urbanmunicipalities. The municipalities enjoy a significant degree o f administrative and fiscal autonomy, andtheir mandate includes economic and social development, basic sanitation, public services, health, education, culture, leisure, and sport, policing, and urban infrastructure, construction, and housing. The legal framework provides for the gradual transfer o f functions and revenues over time as municipalities are ready to assume them. As yet, however, most municipalities have come to exercise authority inonly a limitednumber o f sectors, notably policing and sanitation. 47. I npractice, the administrative system remains highly centralized. Some progress has beenmade with deconcentration to the provincial and district levels inall o f the four sectors analyzed inthis study, namely education, health, roads, and water and sanitation. Further,inthe roads andthe water andsanitationsectors, decentralizationo f responsibilitiesto the provincial level has beenapproved. This remains on paper, however, due to the lack o f capacity at the provincial level andthe fact that fundinghas not yet beendecentralized. Here follows a briefreview on the current plans for deconcentration or decentralization ineach sector. 33See footnote 16, p. 26. 39 48. In education, a certain amount of deconcentration is already effective. The provincial directorates are responsible for managing some 25 percent o f the total number of sectoral projects. N o provision has beenmade, however, to increase the autonomy o f provincial and district administrations, not even concerning simple responsibilities such as settingthe school calendar to adapt to local needs. 49. In the health sector, deconcentration at theprovincial level is alsofairly advanced, with the Provincial Directorates for Health playing a significant role inthe intra-provincial allocation ofresources. An increase inthe extent o f decentralization is envisaged for the future. The Strategic Plan approved by the Council o fMinisters in2001 advocates the intention to improve organizational performance through a gradual process o f decentralizationo f resources, decision-making power, andplanning and management functions within the Ministry o f Healthto the provincial branches. Inthis framework, the Government will prepare a `Health Sector DecentralizationProgram'. Inpractice, however, this has not progressedbeyond general discussions o f (i) the possible separation o fthe Ministryo f Health as a funding andregulatory agency, and the N H S as a service provider; (ii) further decentralizationo fbudget management responsibility to the provinces; and, (iii) the extent o f autonomy o f large facilities, particularly the Maputo Central Hospital. 50. Efforts are being made to deconcentrate (and later decentralize) the operations of the water and sanitation sector. As from 2002, responsibility for National Directorate o f Water (DNA) local procurement andthe payment o f contractors and suppliers has been transferred to the provinces. Public works capability at the district level i s being built up. This is, however, a slow process, due to shortages o f trained staff. Nevertheless, the DNAis clearly movinginthe rightdirection, to the eventualbenefit o frural water supply and sanitation service provision. 51. I n roads, muchprogress remains to be made in devolving responsibilities. The Government's policy is to increasingly shift responsibilities to provincial, municipal and local authorities. Responsibilities for routine maintenance on all classified roads (including national roads), as well as responsibilities for periodic maintenance and rehabilitation on tertiary roads, have formally been devolved to the Provincial Departments o fRoads andBridges (DEPs). These are organs o f provincial government, accountable to bothprovincial and central Ministryo f Public Works (MOPH) authorities. However, technical andmanagement capacities at the provincial levels are weak and almost all provincial hnding for roads management i s still allocated via the Road Fund, which i s controlled bythe National Roads Administration (ANE). As a result, planning and contract management support continues to beprovidedthrough ANE regardless o f the formal separation o f authority. Inaddition, provincial or lower level (district, municipality or community) advisory road commissions are nonexistent, and there does not appear to be any effective provision for their establishment or operations. Although a variety o f donor-sponsored technical assistanceprograms are aimed at strengthening provincial capacities, it would be unrealistic to expect that road-sector management will be effectively decentralized for some time to come. 52. The Government should continue to gradually decentralize administrative and fiscal responsibilities, and increase the amount of training inpublic administration at local level. The Government should act to expandthe capacity and the role o f local 40 administrations vis-a-vis the central government, by introducing adequate training programs and gradually transferring fiscal and administrative responsibilities. Inthis context, it will be important to reform the current system o fdupla tutela o fprovincial directorates, since this system i s not readily compatible with the process o f decentralization. 41 CHAPTER 3. EDUCATION A. INTRODUCTION 53. The Government's objectives inthe education sector are to provide universal primary schooling andto improve the efficiency and quality ofteaching.34To achieve these objectives expenditure on education has increased by 50 percent inreal terms over the last 5 years. Bothrecurrent and capital expenditures have remained substantially skewed infavor o fprimary education, although there has been a small increase inthe percentage allocatedto secondary andhigher education.35The increase inexpenditures has beenmatched by a doubling of enrolment rates at all levels o f education, including most notably an increase fi-om 56 percent to over 100percent between 1992 and2001 in Gross Enrolment Rates (GER) for primary education. This increase reflects an extraordinary rate o f expansion inthe provision o f education inMozambique over the last few years. 54. Household education has a large positive impact on the level of consumptionper capita. The importance o f ensuring that children from poorer households attend school i s underscored by the large role o f low educational attainment as a determinanto fpoverty. OLS regressions using data fi-om the 1996/97 Household Survey indicate that households withbetter educated adults have significantly higher consumption per capita, andthat in general this difference i s greater inurbanareas than inrural areas (Handa et al., 1998). The average level o f consumption per capita o f urban households with some primary education i s higher by 7 percent compared to households with no education (10 percent for rural households). Consumption per capita i s higher by 27 percent ifthe highest level o f education inthe householdi s EP1(16 percent inrural households). It is higherby 57 34 These objectives are consistent with the Government's Education Sector Strategic Program (ESSP), currently under implementation, andwith the educationMillennium Development Goals o f universal enrolment and completion o f primary educationby 2015. 35 The educationsystem inMozambique is structured as a 5-2-3-2 system: 5 years o f lower primary school (Ensino primhrio 1 or EP1, Grade 1-5), 2 years o fupperprimary school (Ensino primhrio 2, or EP2, Grade 6-7), 3 years o f lower secondary school (Ensino secundirio geral 1, or ESG1, Grade 8-10) and 2 years o f upper secondary school (Ensino secundbrio geral2, or ESG2, Grade 11-12). Technical and vocational education is structured into 3 tiers: elementary, basic and medium. Teacher training programs for primary Grade 10 +2 years o ftraining). Secondary school teachers are trained at the Pedagogical University.The school is dividedinto 2 levels: basic (CFPP, for Grade 7 + 3 years o ftraining) and middle (IW, for educationsystem i s almost entirely public at the primary and secondary level (less than 2 percent o f primary schools and 4 percent o f secondary schools are private or semi-private) and therefore no distinction i s made betweenprivate andpublic schools inthe rest o f the chapter. Higher educationinstitutions are underthe tutelage o fthe Ministryof Higher Educationand Technology. There are currently tenhgher educationinstitutions inMozambique, half o f which are public. 42 percent ifthe highest education level inthe household EP2 (36 percent inrural households), and by 97 percent ifthe highest education level inthe household is secondary or higher (57 percent inrural households). The results highlightthat household education has a large positive impact on the level o f consumption per capita. The results also highlight the importance o f female education, particularly inrural areas, possibly due to the observation that women inMozambique are responsible for much o f the agricultural work, which i s the primary source o f subsistence inMozambique. 55. While significant progress has been attained towards universal access to primary education, little progress has been made inimproving school completion rates which remain extremely low, particularly for girls and inpoorerprovinces. The education system remains very inefficient when compared to neighboring countries, with the average actual cost per primarystudent three times its expected level. Giventhat hrther increases inresources are not available, improvements inefficiency and significant reductions inunit costs will be necessary to finance the planned expansion to universal primaryeducation enrollment andcompletionby2015. Furthermore, even larger improvements inefficiency will be requiredinthe future to allow the expansion o fupper primaryand secondary education. Additional concerns arise from the wide gendergap in education (especially at the post-primary level), the regional disparity inpublic funding for education, the highshare o f costs borne by households, and the increasing impact o f HIV/AIDS. 56. This chapter reviews the Mozambique education sector with a view to identifying the critical issues for the optimum use o fpublic expenditures. The chapter begins by discussing the key trends inaccess, performance, efficiency and quality. It then discusses the allocation ofpublic expenditures, its impact on poverty, assesses the likely impact o f HIV/AIDS, and concludes bysuggesting policy priorities. B. KEYEDUCATION SECTORISSUES Studentflows: 57. I n spite of signijkant gains in access toprimary education, enrolment in upper primary (EP2) and secondary education (ESGI and ESG2) remains limited. Gross enrolment rates (GER)36have increased very rapidly at all levels o f education, with the biggest improvement for EP1(from 56 percent in 1992 to over 100percent in2001). About one million children aged 6-11(37 percent o fthe age bracket) remain out o f school, h~wever.~' EP2 enrolment rates doubled, from 13 percent in 1992 to 28 percent in 2001, but remain low. GER are also extremely low for secondary education, at 8 percent and2percent for ESGl and ESG2, respectively. The enormous gap betweenthe 36 The GER is defined as total enrolment at a neducational level regardless o f age, expressed as a percentage of the school-age population ina given school year; therefore, it is sometimes higher than 100 percent. 37 Itshould be noted that this number includes both children who never attended school and children who were at school but then dropped out prematurely after the first few years. As will be discussedbelow, the latter tends to be the most serious problem for primary education inMozambique at present. 43 enrollment rates for EP1 and EP2 implies that significant progress is needed to achieve the Government's goal o funiversalprimary school completion (7 years) by 2015. 58. Geographical disparities in education access are large and continue to increase. There are great provincial variations inadult illiteracy, with higher average levels inthe Northernprovinces (72 percent in 1997) compared to the Central provinces (63 percent) andlowest inthe Southern provinces (39 percent). The GERinEP1varies substantially across provinces, rangingin2001 from below 90 percent inseveral Central and Northern provinces to over 130percent inSouthern provinces. The GER for EP2 also varies substantially. Since recent growth inthe GER has been highest inthose provinces that already hadhigher GER, provincial inequalitieshave infact beenwidening. Completion rates o f EP1 follow a similar pattern. 59. Significant gender differences in access to education persist. The literacy rate in 1997was 57 percent for men and26 percent for women. GER o fboth sexes inprimary education increased at about the same pace between 1997and2001, so the absolute gender gap inEP1 enrolment has remained about the same. On average, GER for girls are about 23 percent lower than for boys, but the gender gap ineducation differs substantially across provinces. The gender gap inenrolment rates i s about 10percent inthe Southern region, but reaches 35 percent insome Central andNorthern provinces. Student repetition and completion rates: 60. School completion rates remain extremely low. The average repetitionrate in primaryeducation (EP1 andEP2) is veryhigh(25 percent) comparedto neighboring countries such as Tanzania (2 percent), Zambia (3 percent), and Malawi (18 percent) (figures for 1995, UNESCO 2000). The average drop-out rate for primaryeducation (EPI and EP2) i s also high, at around 13 percent. Such highrepetition and drop-out rates result inextremely low completion rates. Despite improvements inrecent years, only 36 percent o f students admittedto Grade 1complete the EP1 cycle.38This compares with 81 percent inTanzania and 70 percent inZimbabwe. The completion rate is progressively smaller at higher levels o f education, with only 24 percent o fthe student cohort graduating from EP2, and 1percent graduating from ESG2 (Table 12). This compares with completionrates inZambia o f 63 percent to Grade 7 (EP2 equivalent) and 18 percent to Grade 12 (ESG2 equivalent) (figures for 2000, Oxford Policy Management 2002). 38Completionrate i s definedas the number o f students passing the relevant grade divided by the number o f children o f official graduation age inthe population. 44 Table 12. Completionrates of EP1, EP2,ESGl andESG2in2001 Average Girls EP1 36% 32% EP2 24% 20% ESGl 9% 6% ESG2 1% 1% Source: Plano EstratCgico de EducagHo, Ministryo fEducation, 2002. 61. Drop-out, repetition and completion rates vary significantly acrossprovinces. Drop-out rates for primary education inCentral andNorthern provinces are roughly double those inSouthern provinces. Similarly, gross repetition rates for primary school (EP1 and EP2) vary from 19percent to 30 percent across provinces. 62. Signifwant gender differences alsopersist in drop-out and completion rates. The drop-out rate is higher for girls compared to that o fboys, which results ina large gender gap by the end o f EP1(Table 12).39The completion rate for EP1i s approximately 40 percent for boys and 32 percent for girls. Similarly, the completion rate for EP2 i s 28 percent for boys and20 percent for girls. The gender gap continues insecondary education. 63. Theprecise cause of the high drop-out and low completion rates is unknown and needs to be investigated with an in-depth study. It i s not clear to what extent highdrop- out i s due to demand side factors (direct costs, opportunity costs, poverty, etc.) or to supply side factors (incomplete schools, distance to school, language of instruction, qualityo f services offered, gender o fthe teacher, school calendar and hours o f instruction, etc.). The precise causes o fthe low completion rates are multifaceted and are not well understood. These need to be investigated with an in-depthstudy. 64. On the demand side, the direct costs of schooling are an important cause of the high drop-out rate. The results o fthe Questionbrio de Indicadores Basicos de Bem-Estar (QIBB)indicate that the three mainreasons for school drop-out are that school istoo expensive (29 percent), that studyingis not useful (29 percent), and that the school is too distant (11percent) (InstitutoNational de Estatistica, 2001).40 This finding i s inline with the results o f an ADB survey (African Development Bank, 2002), where household poverty was the mainreason cited for not attending school: 38 percent o f the 6-12 year olds and 27 percent o f the 13-17 year olds not enrolled inschool said school was "too expensive". Similarly, inthe results of the 1996/97 household survey fees were cited frequently byrural households as a primaryreason for not sending childrento school (Handa et al. 1998; also see the analysis presented inSection Dbelow). These results suggest that one o f the main motives for low school enrolment rates and drop-out rates i s the highcost o f schooling related expenses (such as matriculation fees, school supplies, 39Note that the bulk of the gender disparity is concentrated inthe first two classes o f EP1. 40A further reasonexplaining drop-out rates for girls is teenagepregnancy andpremature marriage, sometimes as early as age 12. 45 uniforms). The importance o f direct costs is also suggested bythe finding that at the vocational education level, dropouts usually obtainjobs and frequentlyreturn to school later (Austral Consultoria Lda, 2002). 65. On the supply side, the high average distancefrom the nearest school and the large number of incomplete schools appear to be important. The extent to which high drop-out rates are also due to supply side factors i s not fully clear. As mentioned, the results o fthe QIBBindicate that 11percent o f drop-out students give the reason that the school is too distant (Instituto National de Estatistica, op.cit.). Information from the 1996/97 household survey indicates that the average distance to school i s 4.5 km and that the average time needed to reach an EP1school often exceeds one hour. The large numbero fincomplete schools is also likelyto play a significant role inthe highlevel o f drop-outs. As many as 23 percent o fprimaryschool pupilsin 1998 attended schools which didnot offer the complete cycle o f classes, andso it canbe expected that many studentsonly attend school for the available classes (Mingat et al.,2002). Inaddition, other factors are also likely to play an important role. The `quality' o f the teaching (language o f instruction, curriculum content, dedication or behavior o f teachers) may not be perceived as relevant enoughto the needs o f the locality. The school calendar maynot match that o f agricultural activities inthe community. Inbrief, the available information is not sufficient to determine which factors account for low retention rates inthe various cycles o f study. A more rigorous analysis o f the impact o f the various demand and supply factors i s needed. 66. One reasonfor the higher drop-out and lower completion rates of girls might be thefailure to hire enoughfemale teachers.Just 27 percent ofprimaryschool teachers are women, ranging from only 9 percent inCab0 Delgado to 62 percent inMaputo City. Recruitment o f female teachers has been shown to increase completion rates o f girls, with an elasticity o f completion o f girls with respect to the proportion o f female teachers o f about 0.4 for low income countries (Kengue andMingat, 2002). Inthe context o f Mozambique, this implies that increasingthe proportion o f female primary teachers from the current 27 percent to the African average o f 32.4 percent (in 1990), might increase the retentionrate of girls from 32 percent to about 35 percent. Increasingthe hiringo f female primary teachers from the current 27 percent to 50 percent might increase the retention rate o f girls to about 41 percent. The available observations byprovince support the possible negative correlationbetweenthe percentage of female teachers andthe drop- out rate o f girls (Figure l),although this correlation i s also likely to be a result o f different levels o f income. 46 Figure1. Percentageof female teachers and girls drop-out,by province2001 701 -E 1 60 42am 40 50 -m 30 91 (u E 20 LL 10 0 0 5 10 15 Girls drop-out (%) Source: Education Statistics 2001, Ministry of Education. Efficiency of resource use: 67. Theeducation system is exceedingly inefficient and wastes a large amount of resources which could be used to expand the number of graduates. Due to the highdrop- out andrepetitionrates, it takes an average 18 years ofresource inputs to produce one primary school graduate inMozambique insteado fthe prescribed 7 (Table 13). This compares with Zambia's 10 years ofresource inputs. Current survival rates implythat producing a primary school graduate costs US$613 insteado fthe plannedUS$23241.To reduce this problem, the government is considering some form o f semi-automatic promotion within cycles o f learning inprimary schools, which would also help reduce the current practice o f repeatedly failing students to encourage betterperformance. Table 13. Efficiencyindicatorsfor educationin2001 Primarv (EP1 and EP2. made 1-7) Inpuh'output ratio (school years) 18 Theoretical cost per completer (US$) 232 Actual cost to produce one Grade 7 student(US$) 613 Primary and secondary (EP1, EP2, ESGl and ESG2, grade 1-12) , Input/output ratio 231 Theoretical cost per completer (US$) 1234 Actual cost to uroduce one Grade 12 student(US$) 23728 Source: Education Sector Expenditure Review 2002, MinistryofFinance and MinistryofEducation. 41The cost ofproducing a primary graduate should be (US$24.6 per EP1 studenttimes 5 years) + (US$54.6 per EP2 student times 2 years) =US$232 47 68. The allocation of teachers acrossprovinces and schools does notfollow efficiency criteria and needs to be adjusted over time. There i s evidence that teacher deployment, both across provinces andacross school within provinces, varies widely without reference to efficiency criteria, andthat there is ample room for improvement inthe allocation ofteachers inprimary and secondary education alike (Mingat et al., op.cit.). This could be done over time takingadvantage o fthe natural attrition and new recruitment o f teachers. Inaddition, the allocation o fteacher time betweenEP1andEP2 also appears to be inefficient. Teacher per class ratio for EP1in2001 was 0.7, indicating that double or triple shifts are not uncommon (Table 14). Onthe other hand, the teacher per class ratio inEP2was 1.3, suggesting that teachers at this level do not spend all their time teaching. This raises questions about the efficiency o fteachers' usewithinprimary education (particularly giventhe significantly higher salaries o fEP2teachers - see below). Using teachers more efficiently (i.e. ensuring that teacher time is utilized fblly) could decrease both the teacher per class ratios andthe pupilper teacher ratios (PTR), without increasing unit costs. Quality of education: 69. The curriculum and the system of student examination are out of date and ineffective. The current curriculum is outdated andthere i s a lack o f a national assessment system to monitor student learning. The Ministry o f Educationwill introduce a new primary level curriculum beginningin2004. The new curriculum is innovative inmany respects. Itprovides a space forlocally developed andrelevant content, introduces bilingual education inkey subjects inthe early grades, increases the number o fhours o f instruction, introduces new pedagogicalmethodologies to improve teaching practice and strengthens school-based student assessment. Concrete plans to establish a learning assessment system that would serve to measure student learning rather as a mechanism to assign places inhigher levels are not yet forthcoming, however. 70. Thenumber of teachers increasedfor all education levels between 1997and 2001, but lessforprimary education. Inspite o fthe recent gains, the number o fteachers inprimaryeducation remains below requirements, especiallyinEP1. Inlinewithteacher numbers, the pupil-teacher ratio (PTR) for EP1was 67 in2001, up from 62 in 1999, and well above the recommended maximum o f 40-45 (Table 14)42.The relatively highand increasingPTR inEP1suggests that the quality o f lower primary education has declined inrecent years. Inaddition, the largevariation inPTRwithineachprovince (e.g. between42 and 120inZambCzia) may indicate wide variations inthe quality o f education, and suggeststhat there are inefficiencies inthe allocation ofteachers. 42Compare also the ESSP goals inMinisterio da Educa@o, 1998,pp. 41-42. 48 Table 14. Selected quality indicators by education subsector, 1997-2001 Percentage increase Percentage untrained Pupils per Teacher per Pupils per inteachers teachers teacher class class 1997-2001 1997 2001 2001 2001 2001 EP1 31 30 39 67 0.7 48 EP2 67 21 35 40 1.3 51 ESGl 111 18 33 35 1.5 50 ESG2 84 8 24 25 1.8 46 Source: EducationSector Expenditure Review 2002, Ministry of Finance and Ministry of Education. 71. The recent increase in the number of teachers has been accompanied by a decrease in theproportion of trained teachers. On average, the current teacher supply covers less than 50 percent o fthe demand for teachers inthe country. The increased demand for teachers has beenmet by hiringteachers without appropriate training or qualifications. The proportion o f teachers without pedagogicaltraining has increased from 30 percent in 1997 to 39 percent in2001 inEP1 and from 21 percent to 35 percent inEP2 (Table 14). Theprovisionofadequate in-servicetraining, particularly for those teachers who lack pedagogicaltraining, have weak knowledge o f the subject, or do not have appropriate teaching qualifications, i s therefore crucial to prevent a reduction inthe quality o f education. The need for in-service training will become even more important to ensure the proper implementation o f the new curriculum. 72. There is a need to increase the output ofprimay education teachers with at least minimal qualzfzcation. Since the expansionineducation is intendedto accelerate, there i s a need for innovative approaches to teachers' training. The current policy o f adopting the I M A P certification (Grade 10plus 2 years training) as the standard for EP1teachers in unlikelyto be attained (due to the smallnumber o fGrade 10 graduates). Itis also not likelyto be sustainable given the current teacher salary structure for MAP graduates. The Government shouldtherefore concentrate its efforts onrestructuringand introducing innovative ways to expand the CFPP system (Grade 7 plus 3 years training), which already provides the bulk o f the supply o f new teachers. This would increase the annual supplyofteachers with at least minimalpedagogicalqualifications andwith a more affordable salary cost and no significant reduction inquality.43 73. Poor infrastructure and the lack of books and teaching materials are the major sources of dissatisfaction for those attending school. The results o f the QIE3Bsuggest that more than half o f the primary education students (57 percent) are not satisfied with the level o f service they receive (Instituto National de Estatistica, op.cit.). The extent o f dissatisfaction varies widely across provinces, from 27 percent inNiassa to 90 percent in Inhambane. The major causes o f complaint relate to poor infrastructure (64 percent), especially inrural areas, and lack o fbooks and materials (53 percent), especially inurban areas. The extent o f dissatisfactioninsecondary school is smaller (41 percent) and again 43The available evidence from Mozambique, albeit limited, shows that the gains in EP1 students' performance when an EP2 (IMAP)teacher teaches them are not very significant. 49 it is mainly due to the lack o fbooks andteaching materials (57 percent), andpoor infrastructure (33 percent). These findings indicate that the challenge of expanding the education sector are not limitedto ensuring an adequate supply o f qualified teachers, but also include improving the quality of schooling infrastructure andthe availability o f books andmaterials. Tertiary Education 74. This PERdoes not go into detail about the higher education system. The bulk o f public expenditure on educationinMozambique i s dedicated to primary and secondary education. However, there are significant disparities inaccess to tertiary education, and efficiency andquality remain poor. These issues are addressed inthe Higher Education Policy, which is currently under implementationwith substantive financial support from the World Bank. 75. Access to higher education has expanded, but remains very limited. Over the past decade the numbero f HigherEducation Institutions (HEIs) has increased from three to ten, andthis has resulted ina subsequent increase intotal student enrollment from 3,750 in1990to over 12,000 in2001.44Thistrendindicates that ahighsocial demand for post- secondary education exists inthe country. Despite the increase inenrollments over the last decade, the GER intertiary education inMozambique i s still low at about 1percent. However, this i s comparable to the tertiary GER for the majority o f countries with similar levels o f GNP not only inSub-Saharan Africa, but also inother parts o fthe world. In addition, there i s a great disparity inaccess to higher education among those with different geographic, gender andeconomic backgrounds, which is exacerbated by inequities that prevail at lower levels o f education. Only 40 percent o fthe students in HEIsare from the Northemand Centralprovinces, whichrepresent 75 percent o fthe total p ~ p u l a t i o nSimilarly, the proportion of female students enrolled inpublic HEIs has . ~ ~ remained constant at around25 percent since 1993. 76. Thehigher education system has become increasingly inefficient. The intemal efficiency o f the higher education system canbe estimated bylooking at admissions as a proportion o f enrollment. The admission to total enrollment ratio declined from nearly 20 percent in 1992 to 16 percent in 1999, indicating a decline inthe efficiency o fthe system (students are takingmore and more years to complete their studies). One o f the key factors is the highlevel of student repetition. Similarly, the percentage o f studentswho graduate within the prescribed time was between 7 percent and 10 percent for public HEIsin 1999. This situation suggeststhat there is aneedto re-evaluate andperhaps restructure the current degree requirements to increase the proportion o f studentswho 44Although public HEIs still account for the bulko f studentenrollment (76 percent), private HEIs are gaining ground andthe Government has supported policies aimed at increasingprivate sector investment in lugher education, whether by encouraging public-private partnerships or by providing a conducive environment for new private HEIs to flourish, especially indisadvantaged regions o fthe country though theprovincialscholarship program. 45The high concentrationo fstudents fromthe far SouthandMaputo City canbe explainedby the geographical concentration o f HEIs inthe Southern region coupled and the limited availability o f scholarships and boarding facilities for students from distant provinces. 50 complete all degree requirements intime. This is being done underthe higher education project with a view to boosting graduationrates, increase the internal efficiency o f the system and reduce unit costs. 77. Courses are often outdated and of low quality. Many courses and programs offered at public HEIs are outdated, o f limitedrelevance and quality anddo not respond to the demands o f a fast growing economy, the specific needs inindividual provinces, or emerging sectors. Furthermore, the style o f learning i s often rote learning, but problem solving and innovative skills are required by employers (The 100 Biggest Companies in Mozambique, KPMG, 2000). c. EXPENDITURE TRENDS 78. A signijkant share of the education budget is externallyfinanced, but data are not readily available. A breakdown o f data on external financing is available only for the year 2000 and indicates that about US$41million(26 percent) o fthe total education budget o fUS$153 millionwas externally financed (Table 15). Total recurrent expenditure was about US$119 million (15 percent from external sources), andthe balance o f US$34 millionwas capital expenditure (64 percent from external sources). However, these numbers do not include receitas prbprias, which in 1997 amounted to a considerable 6 percent o frecurrent cost for EP1(results o f 1997 household survey; mainly own fees collections by schools). The lack o f accurate data on fees collections imposes a serious constraint on the analysis carried out inthis chapter. Table 15. Governmentandexternaleducation recurrentandcapitalexpenditure in2000 (US$ Million) Tech. Teach Literacy Primary Secondary Education education and other Total Percent Recurrent expenditures Government 70.6 16.1 8.1 5.0 1.0 100.8 85 External assistance 8.7 3.5 2.8 2.4 1.1 18.4 15 Total recurrent 79.3 19.6 10.8 7.5 2.1 119.2 100 Percentage 66 16 9 6 2 100 Capitalexpenditures Government 8.4 1.8 0.7 0.9 0.0 11.8 36 External assistance 18.0 0.6 1.9 1.5 0.3 22.2 64 Total capital 26.6 2.6 2.7 2.4 0.3 34.0 100 Percentage 77 8 8 7 1 100 Total expenditures Government 78.9 18.1 8.7 5.9 1.0 112.6 74 External assistance 26.7 4.0 4.7 3.9 1.4 40.7 26 Total 106.0 22.2 13.5 9.8 2.4 153.2 100 Percentage 69 14 9 6 2 100 Source: Conta Geral do Estado, Ministry of Finance, and Ministryof Education. 51 79. Total education expenditure has increasedsubstantially in recentyears, rising from 3.9 to 5.8percent of GDP between 1997and 2001. Education expenditure has increasedby over 50 percent inreal terms inthe last five years, raising total per capita education expenditure fromUS$4.1 to US$6.3. This is the result o fbothhigher donor fundingand increased government-funded spending. Itis important to highlight, however, that while Government-funded education spending increased from 1.9 to 3.1 percent o f GDPbetween 1997 and 2001 (Table 16), it remains below the SSA regional average o f about 4 percent o f GDP andbelow the international benchmark o f 3.5 as stipulated bythe EducationFor All Fast Track Initiative (data for 1998-2000, Bruns,et al., 2003). Table 16. Educationsector expenditures,1995-2001 (US$ Million) 1995 1996 1997 1998 1999 2000 2001 Total Government recurrentexpenditures 246.1 272.4 370.0 444.6 498.1 497.8 436.8 Ministry of Education 89.9 115.1 114.5 123.9 n.a. 153.2 195.1 Government-finded expenditure 30.1 43.0 56.1 73.8 97.6 112.6 98.5 Recurrent account 28.0 39.0 50.5 67.3 86.9 100.8 92.9 Capital account 2.1 4.0 5.6 6.5 10.8 11.8 5.6 Donor-fundedexpenditure 59.7 72.1 58.4 50.0 n.a. 40.7 96.6 Ministry of Higher Education n.a. n.a. 20.5 33.8 n.a. n.a. n.a. Government-funded expenditure 7.3 9.7 10.7 13.8 18.2 19.0 15.0 Recurrent account n.a. n.a. n.a. n.a. 14.9 16.5 13.1 Capital account n.a. n.a. n.a. n.a. 3.4 2.6 1.9 Donor-fundedexpenditure n.a. n.a. 9.9 20.0 n.a. n.a. n.a. Total domestically-funded expenditure o n 37.5 52.7 66.8 87.6 115.8 131.6 113.5 education Total externally-funded expenditure o n education 59.7 72.1 68.2 70.0 n.a. 40.7 96.6 Total expenditure o n education (domestically- 97.2 124.8 135.0 157.6 115.8 172.3 210.1 funded+externally-funded) Recurrent domestically-funded spending o n education intotal Government recurrent 14.4 16.4 15.4 17.6 20.4 23.6 24.3 spending (%) Domestically-funded spending o n education as .6 .9 ,9 2.2 2.x 3.5 3. % of GDP Total Expenditure on education as % of GDP 4.1 4.4 3.9 4.0 n.a. 4.6 5.8 GDP (IMF 2002 data) 2392 2842 3438 3893 4090 3750 3610 Exchange rate (Meticais/US$) 8890 11294 11546 11850 12691 15689 20707 Source: Conta Geral do Estado, Ministryo f Finance, and Ministryof Education, and Mingat et al. 2002 52 Recurrent expenditures: 80. The allocation of government recurrent expenditures across education levels has remainedfairly stable in recent years. The share o frecurrent resources going to primary educationhas remained constant between 1999and 2001 at about 61percent. However, within primaryeducation, resources havebeenreallocatedtowards EP2, which has increased from 11to 13 percent o f total recurrent expenditures. The share o f secondary education (ESG1 andESG2) has increased from 13 to 16percent. The recent increase in fundinghas not benefited teacher training, with the proportiono frecurrent funding for training primary school teachers remaining constant at 3.3 percent. 81. Recurrent unit costs are high compared to international standards, especiallyfor upperprimary and secondary education. Interms o f GDP per capita, Mozambique's education is more expensive than inneighboring countries. While spending on lower primarystudents (EP1) is somewhat below the SSA average of 13percent (Colclough andAl-Samarrai, 2000), spending on upperprimary school students (EP2) is muchhigher at 23 percent o f GDP per capita. Similarly, spending per student on secondary education corresponds to 88 percent of GNP per capita, which compares to 27 percent inMalawi, 11percent inZambia and 32 percent inZimbabwe. Figure2. RecurrentUnit Costs by Programin2000 E I 900 1 1 800 0Privatefees & books 700 E 600 . = 500 0 0 (r)400 3 300 200 ZOO 0 EPI EP2 ESGl ESG2 Tech. Teacher educ. educ. Source:Education Sector Expenditure Review 2002, Ministryof Finance and Ministryof Education. Notes: 1.Private costs estimates based on 1997 householdsurvey. 2. Administrative expenditures allocatedproportionally across programs. 82. Thebreakdown of government recurrent expenditures by education level highlights large cost differences in teacher salaries and in non-teacher salaries at different education levels. Most o fthe differences inrecurrent unit costs stem from considerable differences inthe teacher salary costs per student. For instance, the four-fold increase inunit expenditures from EP1to ESGl i s associated with an increase inthe teacher salary costs per student from US$32 to US$111per student. The proportion of non-teacher salaries also appears to be much larger inEP2 (16 percent) as compared to EP1(5 percent). Expenditure on non-teacher salaries is even higher insecondary 53 education (20 percent) and teacher education (32 percent). It i s questionable whether such a mark up inspending on non-teaching staff is necessary. Figure3. Governmentunitrecurrentexpenditureby inputin 2001 (in 1999US$) Administration Source: EducationSector Expenditure Review 2002, Ministry o fFinance and Ministry o f Education. 83. Thepay differential between teachers at different education levels is excessive, with upperprimary and secondary school salaries well above regional averages. Salaries o fEP1teachers are about 4.4 times GDP per capita, above the average 3.6 times o f GDP per capita inSSA (Table 17). The differential between EP1 and EP2 i s 109percent, between EP1 andESGl i s 426 percent and the differential betweenEP1 and ESG2 is as much as 562 percent. This compares with a pay differential between primary and secondary school teachers not exceeding 100percent inmost SSA countries. 46 Secondary school salaries, which are double the regional average inper capita terms (Figure4), are unsustainably highand are animpediment to the expansion of secondary education. 46Itis ofinterestthat evidence frommanyAfrican countries suggests that setting primaryeducation teachers' salaries at around 3.5 times per capita GDP provides a good balance betweenrecruiting and maintaining teachers with adequate credentials and morale while allowing the universal coverage ofbasic education without overburdening public finances (Mingat, 2002). 54 Table 17. Yearly teacher salary and teacher salary/GDP per capita in 2001 (US$) EP1 (min.salary) 870 Teachersalary/ GDPper capita 4.4 EP2 (min.salary) 1820 Teacher salary/ GDPper capita 9.2 ESGl (min.salary + 30% bonus) 4574 Teacher salary/ GDPper capita 23.2 ESG2 (min.salary + 40% bonus) 5757 Teacher salary/ GDPper capita 29.2 Source: Education Sector Expenditure Review 2002, Min.ofFinance and Min.of Education. Figure 4. Secondary teacher salaries in selected SSA countries4' 2a 30- Mozambique 2 % 8 25 2 a20 $; rC3 15 A Eg .a r;s10 4 5 - 4 4 4 4 t w 84. Large differences exist between the number of teachers in thegovernmentpayroll and those reported by schools. The number o fteachers recorded on the payrolli s as much as 20 percent higherthanthose reportedby schools inthe year 2000 and 14percent in 2001. This discrepancy suggests the needfor an in-depth study and may indicate the opportunity to make easy and significant cost savings. Capitalexpenditures: 85. Theallocation of capital expenditures is concentrated onprimary education, and school construction inparticular. Capital expenditure is heavily concentrated on primary education (80 percent in2000), with most o fthe fbndingoriginating fiom external financing (about three-quarters in2000). Capital spending on secondary educationis about 6 percent, technical education about 8 percent andteacher education about 7 47Benin, Burkina Faso, Cameroon, CGte-d'Ivoire, Madagascar, Mauritania, Mozambique, Niger, Rwanda, Tanzania, Togo 55 percent. Extemal capital fbndingi s very heavily skewed towards primary education. Government capital expenditures also focus on primary education, but to a lesser extent thaninthe caseo fexternal financing.48 Over the last three years government capital expenditure has been reallocated from primary and secondary to technical education. In terms o f activities, the breakdown o f government capital expenditures indicates that the majority (62 percent in2000) was spent on construction and rehabilitation o f schools. The other major category inthe capital budget is equipment. 86. Thecost of school construction varies by as much asfour hundredpercent, without apparent reason. The cost o fbuildinga classroom usingpermanent materials (cement and varies from US$5,000 to US$20,000, partly as a result o f different models o f donor assistance andpartlybecause o f limited competition inschool construction contracts. It has been shownthat excellent facilities canbe built for a cost o f about US$lO,OOO per classroom (including furniture, water and sanitation and auxiliary facilities). This amount should be viewed as a ceiling for construction costs, and should be implementedjointly with a revised set o f construction standards. 87. There is a need to accelerate the school construction program and to optimize the location of future schools. Well targeted construction o fnew schools will be critical to sustain the expansion o f enrollment rates (Handa and Simler, 2000). As many as 3800 new classrooms per year will have to bebuilt over the next 15 years inorder to sustain the expansion inprimary education (EP1 and EP2). This expansion rate i s far beyond the current capacity o f the central Government o f less than 2000 classrooms per year. A possible approach, usedinvirtually all neighboring countries, would be to encourage active participation bythe communities, and allow the location, construction and maintenance o f new facilities to be decided at the local level (regional or district authority), on the basis o f the funds allocated by the Ministryof Education." Villages mightberequiredto pick uppart ofthe cost, through the use o flabor for constructionand maintenance. Fiscalframework andjhancing of higher education 88. Theshare ofpublic spending on higher education should be reducedfurther by reducing unit costsper graduate student and increasing the extent of cost recoveryfor those who can afford it. Allocation o fpublic expenditure to tertiary educationhas been decreasing from about 20 percent in 1995 to current levels o f about 14 percent (Table 16). Therefore, although less than 1percent o f the total populationis enrolled intertiary education, higher education is consuming almost one sixth o f overall public spending for education inMozambique. Barring a few exceptions, tuition fees inpublic institutions are very low (below US$lOO per year) while students inprivate institutions contribute, at 48However, interms o f expenditureper student (i.e. Government unit capital expenditure) the allocation o f Government expenditure remains heavily skewed towards technical education and teacher training, which are as muchas 30 and 169 times the unit expenditures for primary school students. 49Localmaterials are generally cheaper than using `permanent' material (cement and bricks), but they are o f lower quality and durability, andrequire more maintenance. 50The Government should set centrally the quality standards for the construction o f new schooling facilities and provide technical assistance, however. 56 a minimum, ten times more. Given the inequitable distribution o fhigher education expenditures, andthe highreturns facing individuals who invest into a higher education degree (interms o f higher salaries), the Government should increase the level o f fees to cover most o f the recurrent costs o ftertiary education. Existing constraints incredit markets inMozambique, however, implythat poor studentsmay be unable to borrow the resources required to finance their studies. To overcome these constraints the Government should expand current provisions for student loans, inparallel with making bursaries available for deserving students.The public higher education institutions already operate several student scholarship schemes (partly supported by IDA) that provide financial assistanceto needy students, andthese schemes should be expanded to ensure that deserving students can attend tertiary education. Planned expenditures: 89. Achieving universalprimary education by 2015 will require significant additional expenditures, reductions in costs and improvements in eficiency. The Government's objectives o f achieving universalprimaryschooling, and o f improving the efficiency and qualityofthe education system will require significant resources interms of additional teachers and schools. For instance, the number o f EP1teachers and EP1 classrooms would have to almost double, and the number o f EP2 teachers andEP2 classrooms would have to increase almost four-fold. Preliminarycalculationsbased on conservative assumptions5' indicate that the amount o fresources requiredwould amount to an additional 2 percent o f GDP annually, raising the domestically fwnded contribution to education expenditure from 3 percent o f GDP (see Table 16) to 5 percent. Such an expansion would require that the share o f education inGovernment expenditure increase from the current 21 percent to about 35 percent o f the Government budget, at the expense o f other sectors; obviously this is not a realistic scenario. According to the projections made under the Educationfor All Fast TrackInitiative (EFA FTI), evenwhen assuming significant efficiency improvements and an increase inexternal assistance, the financing gap would remain extremely large.52It i s clear, therefore, that sharp improvements inthe efficiency ofthe education sector will be imperative to ensure a significant reductions in unitcosts, butnot sufficient to achieve the Government goals (and the educationMDGs). 51Average real GDP growth rate o f 6 percent p.a.; education sector will continue to receive about 21 percent o f the budget (or about 3 percent of GDP); about 60 percent o fthe education budget will continue to be spent on primary education; constant teacher salaries inreal terms; plannedimprovements in GER, completionrates and PTR. 52Mozambique i s one o f the 18 countries eligible to participate within the framework o f the Educationfor All Fast Track Initiative (EFA FTI), which was adoptedbythe international community in2002 to help countries mobilize the resources necessary for an acceleration o f progress towards the EFA and the education Millennium Development Goals. The Government is currently finalizing its proposal to the FTI financing agencies and expects to begin its implementation in2004. The Government's FTIproposal includes the introductiono f a new curriculum for basic education in2004, the development o f a teacher education strategy, the emphasis on gender equity, community support for education and the prevention and mitigation o fHIV/AIDS. A central feature o f the proposal are strategies designed to progressively reduce student repetition and drop-out to a minimum, ensure regular student progress inlearning and achieve universal completion, while ensuring fiscal and institutional sustainability. 57 90. Themerger of EPl and EP2 cycles into a singleprimary education cycle would allow a substantial rationalization of availableresources. At present there are several differences between the EP1andEP2 cycles. At the EP1level, teachers normally have at least Grade 7 plus 3 years o f training (CPFF) andthey are multi-discipline teachers (one per class) with an average salary o f 4.4 times o fper capita GDP. Teachers with 7 years o f education or less andno training receive an average salary o f2.8 times per capita GDP.About halfo fthe EP1teaching force i s inthe unqualifiedcategory, an issuethat merits immediate attention andremedy. Currently the Government policy is to upgrade the entry qualifications for primary school teachers from 7 to 10 years o f education and offer a 2-year instead o f a 3-year program. The new polisy i s beingimplemented at the IMAP centers, which havebeenreceiving the bulko fexpenditure inthe last 5 years. There are two issues o f concernwith the current policy. First,the Ministry statistics show that despite a growing supply o f IMAP graduates, few graduates enter the teaching profession andwhen they do, they tend to chose positions inpost EP1levels. This has negative implications for the supply o fnew teachers inEP1schools, particularly since little new investments are being made inCFPP centers. Second, the MAP graduates have teach only 2 or 3 subjects (there are several teachers per class) andhave an average salary o f 9.2 times per capita GDP, which i s more than doubled o fthe average o f 4.4 for CFPP graduates. While there is no comparative information inMozambique on the relative effects o f different types o f qualifications (CFPP and IMAP)on student learning, evidence from other countries indicates marginal gains when primary school teachers go beyond 11years o f education. While it i s clear that the current policy to ensure primary school teachers have IMAP level qualifications i s costing the Government twice as much, it is less clear what the additional effect is on student learning Itwould be more more efficient and fiscally affordable to switch to a single primary school system with multi- subject teachers throughout the 7-year cycle. D. POVERTYFOCUSAND INCIDENCEOFEXPENDITURES 91. Theregional distribution ofpublic expenditures in educationperpetuates the regional differences in literacy rates. The distribution o f education resources (monetary benefits) across regions i s heavily skewed, with Maputo City with only 6 percent o f the population receiving almost a thirdo fthe total benefits (Heltberg et al., 2001). This result i s caused inlarge part by higher enrollment in secondary education inMaputo City, compared to anywhere else inthe country. However, while the rest ofthe Southern region (i.e. excluding Maputo City) also receives a share o fbenefits slightly higher than its proportiono fpopulation, the Northern region and especially the Central region are significantly under-served interms o fper capita benefits. 58 Table 18. Regionaldistributionof populationandbenefitsfrom education spending(in thousandMt.) Share ("h) North Center South Maputo city (excluding Maputo) Population (% households) 32.5 42.6 18.8 6.1 Educationbenefits ("h) 18.8 26.2 22.7 32.2 Source: Heltberg, Simler and Tarp, 2001, based on 1996197householdsurvey. 92. Education spending is equitably distributed at lower levels of education, but is very inequitable at higher levels. The distribution o fEP1benefits almost overlaps with the 45-degree line, indicating that the distribution o f spending on EP1i s very equitable (Heltberg et al., op.cit.). The benefits from EP2 spending are less evenly distributed, withthe poorest halfofthe populationreceivingabout 35 percent o fthe benefits. At higher levels o f education, the distribution o f spendinggets progressively less equitable, withthe poorest 50 percent only receiving about 5 percent o fthe benefits at ESG2level (Figure 5). Figure5. Distributionof benefitsfrompublicspendingineducationby income decile 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Proportionof population, poorestto richest . Source:Heltberg et al. (2001), Figure 6. 93. Students andparents pay a signijkant proportion of the recurrent costs at all levels of education, includingprimary school. As much as 18percent o f EP1 and 23 percent o f EP2recurrent unit costs are coveredbyparents (school fees and books; data from 1996/97household survey). Inthe case o f EP1, this corresponds to 12percent o f household consumption expenditures o f the lowest-earning decile per child, and 10 percent ofhousehold consumption expenditureso fthe second-lowest earning decile. For EP2, this cost becomes prohibitively high, correspondingto 34 percent and 29 percent o f household consumption expenditures per child for the lowest and second-lowest income deciles, respectively. This suggests that the highlevel of costs borne directly by the household acts as a serious disincentive towards sendingchildren from poorest 59 households to primary school. Inaddition, the level o f fees varies widely from one village to another, but there is only a very weak correlation between income levels o f municipalities andschool payments. This implies that school fees and other school expenditures are often a deterrent to sendingchildren to school inpoorer provinces. E. IMPACT OF HIV/AIDS ONTHE EDUCATION SYSTEM 94. One of the major constraints to the achievement of universal enrolment and completion is the impact of the HIV/AIDSpandemic on the education sector. Further investments inthe sector will need to consider the impact o fHIV/AIDS on the teaching force and on the student populations by accounting for teacher absenteeism and attrition, the need for further teacher training to offset these losses, as well as the cost of enrolling orphans. 95. Costs of absenteeism due to HIV/AIDS are incurred when HIV-related illness in infected teachers keeps themfrom their teaching duties. Combining the information on the typical impact o f the illness with the salary levels, it i s possible to estimate these costs at US$1million in2002, increasing to US$1.5 million in2006. It is estimated that total teacher attritionwill be about 7,450 between 2002 and2006. Ofthese, about 2,700 teachers will be lost to HIV/AIDS.53 The unit cost o fteacher training implies that the additional costs o fteacher attrition due to HIV/AIDS can be estimated at about US$2.3 million in2002, rising to US$2.7 million in2006. 96. It is also estimated that HIV/AIDS will have a direct impact on the number of orphans. The number o f orphans is expected to increase fkom the current 10percent in EP1 and 11percent inEP2, to 18 percent and 25 percent in2010, respectively. The cost to enrol andmaintainthis number o f students under bursary or other programs to reach orphans and vulnerable children (estimated bythe Government at US$8 per student year)54,i s calculated at US$2.6 million in2002, risingto US$3.6 millionin2004 and US$4.6 million in2006. 97. The total cost of HIV/AIDS on the education sector is estimated at US$7m in 2006, equivalent to a 5percent increase in total education spending. Incorporating the cost o fteacher absenteeism, attrition andtraining, as well as the cost o f enrolling orphans andvulnerable children, the total cost amounts to US$4millionin2002, increasingto US$7 million in2006. Given the magnitude of these costs, a comprehensive education sector response to mitigate the impact o f HIV/AIDS among school children, youth, teachers and administrative staff will need to be operationalizedas a matter o fpriority. The Ministry o fEducationhas started this process by launchingthe `Strategic Plan for HIV/AIDS' that sets out a timetable to develop andimplementpolicies to (i) provide 53For simplicity it i s assumed that the AIDS prevalence among teachers is the same as for the rest of the population. 54This estimate is likely to be far too low, however, since it onlyjust covers the average level of school fees. Inpractice orphans will have to benefit frombursaries to support living expenses as well to be able to attend school. 60 support and care for education sector staff, (ii) provide substitutes for teachers who are absent through illness, (iii)ensure that orphans andvulnerable children have access to education, and (iv) address HIV/AIDS prevention inthe new curriculum together with extra curricular activities to support prevention inall EP1 and EP2 schools. F. POLICY PRIORITIES FORTHE EDUCATION SECTOR 98. This chapter has highlighted several weaknesses inthe performance o fthe education sector and has identified eight critical areas for policy improvement. 99. Investigate the causes of the low completion rates. The evidence available on the general causes o fthe low completion rates inMozambique is scant andthere i s a need for a further in-depthstudy. Inthe meantime, it is recommendedthat the authorities move aheadvigorously with the implementationo f the new curriculum which plans to intoroduce three learning cycles withautomatic promotion within cycles. The implementation o fthe curriculum will begin to address internal efficiency issues evident inthe highrepetitionanddrop-out rates. Available internationalevidence suggeststhat the girl drop-out rates are negatively correlated with the proportion o ffemale teachers. Therefore, given the extent o f the existinggender gap inprimary education, the hiringo f female teachers should be pursued andi s likely to lead to improvements ingirls' retention and completion rates. 100. Merge theEPl and EP2 cycles into a singleprimary education cycle. The current distinction betweenEP1andEP2 embodies various inefficiencies and can be eliminated without compromising the quality o f education. This reform would allow a substantial rationalization of available resources across EP1 and EP2 (notably teachers andclassrooms), itwouldhomogenizeteachers' salaries andtheir qualifications across primaryeducation, itwould accelerate the school construction program, byeliminating the needto have separate schools for lower and upper primary school cycles, andit would ultimately facilitate the expansionto universal primary education. 101. Improve cost efficiency in school construction and adopt a community based approach to school Construction. The Government should ensure that the construction cost per primary school classroom does not normally exceed $10,000 (inclusive o f furniture, water and sanitation, and auxiliary facilities), and should revise the construction standards which have to be satisfied by contractors and communities. This would reduce the scope for rent seeking from school construction contracts andreduce the time it takes to complete works. Inaddition, the government should expand the capacity to buildnew schools by decentralizing the school constructionprogram to the regional or district level andinvolves the community inthe construction, rehabilitationandmaintenance o f schools, particularly inthe primary level and inrural andisolated areas. This would also ensure that the location o f new schools occurs where is most needed (demand driven), and that the scarce available funds are usedefficiently. 102. Improve the equity of education spending. The highlevel o fprivate costs bomeby the householdreduces the school enrollment o fthe poorest children and i s a contributing factor for highdrop-out rates andpoor completion rates. The official or unofficial use o f school fees inprimary educationneeds to be studied inorder to ensure it i s reduced and 61 restructuredin such a way that it does not hinder the access o fthe poor to education. This could be done through a combination o fincreased allocations to schools and improvement o fthe exemption arrangements for students from needyhouseholds. At the same time, the distribution o fbenefits from public spending insecondary educationi s highlyinequitable, andthe shareofthe educationbudget going to secondary education should be reduced further, infavor o f primary education. The level ofcost recovery in secondary education should be maintained, inparallel with making bursaries available to deserving students. 103. Reverse the decline in the quality of education by introducing the new curriculum and expanding the CFPP teachers training. Extending access to education should not be at the expense o f educationquality, and several measures can respond to weaknesses identifiedinthis report. The plannedintroduction ofanew curriculumwill help to address some issues related to quality and efficiency. The magnitude o fthe effect, however, will depend on parallel programs to upgradeteacher subject knowledge, improve teaching practice, ensure an adequate supply o f teaching guides and textbooks to support the new curriculum and upgrade school infrastructure. The trendtowards hiring teachers without appropriate training qualifications should bereversed, and a priority placed on improving the supply andquality o fpre- and in-service teacher education programs. In this context, it is recommendedthat the Government improves the quality o fteacher education programs offered at CFPP centers to ensure that all new primary teachers have CFPP qualifications. Teacher education programs (lengthand cuniculum) should be restructured significantly to increase the supply o f qualified teachers. The CFPP system should be reformed to adopt alternative training modalities aimed at reducing resident time and increasingpractical training. This would allow an increase in the annual supply o f teachers with at least minimalpedagogical qualifications and a more affordable salary cost. The recommendation above to improve efficiencies inteacher allocation will also help improve qualityby reducing PTRs. 104. Reallocations within the education budget could improve eficiency while remaining within an envelopeconsistent with the PARPA. Savings canbe made by (i) identifyingand removing all any "ghost teachers" from the payroll, as teachers on the EP1payroll outnumber those recordedbythe schools by20 percent; (ii) rationalizing the deployment o f teachers across provinces and schools, improving the use o fteachers' time inEP2, andreallocatingteachers from EP2towards EP1; (iii) reducing construction costs; and (iv) reducingthe subsidy on tertiary education. These savings would likely be sufficient to compensate for the proposed scholarships for AIDS orphans andthe proposedreduction inprimaryschool fees. 105. Develop a comprehensiveeducation sector actionplan on HIV/AIDS. The HIV/AIDS epidemic will have wide implications for the education sector, mainly as a result o f the large number of orphaned children. The Government needs to operationalize its `Strategic Plan for HIV/AIDS' into a comprehensive education sector response to mitigate the impact o f HIV/AIDScrisis. 106. Improve data collection on education spending. Although the availability o f data on indicators o f access, performance and quality ofthe education sector i s very good, the reporting on external financing, other donor fundingmade available directly to the education sector (off-budget aid from donors), and particularly on the collection of local 62 fees (receitusprbprius), i s weak. This lack o f comprehensive expenditure data imposes serious limitations on the analysis o f education expenditures and needs to be resolved. 63 CHAPTER 4. HEALTH A. INTRODUCTION 107. Despite progress inmedical coverage inrecent years, Mozambique's key health indicators show a mixedpicture against the findings o f the last PERin 1992. Infant mortality andunder-5 mortality -usually emblematic o fthe state o fhealth ingeneral - have improved slightly or stagnated, depending on the source. In 1989, the infant mortalityrate was some 137-15055per 1000 live births, falling to 129by2001 (at best a 14 percent improvement in 12years). The under-5 mortality rate was 203-21956in 1989 and 200 in2001 (at most a 9.5 percent improvement in 12 years). Some indicators have improved markedly, e.g. the intra-hospital maternalmortality rate fell from 230 per 100,000 live births in 1993 to 160 in2002. Yet others deteriorated: in 1989 life expectancy at birthwas 43-49 years5'; in2001 it was 42 years, some of the reduction beingdue to HIV/AIDS. Mozambique's indicators also continue to lagbehindthose o f its neighbors. Malaria i s the leading cause of death. HIV/AIDS prevalence was 13.8 percent in2002 andhas become one o fthe main causes o f death for the working age group. These poor indicators cannot be attributed entirely to low health spending levels because these have been above 2.5 percent o f GDP inmost years since 1992. Neighboring countries with similar levels o f spending exhibit better health indicators (Table 19 on page 65). This chapter o fthe PERwill, among other things, investigate the various factors that influence the outcome indicators and check to what extent there i s a linkbetweenhealthspending andoutcomes. Ithas beenobserved inother countriesthat there is only a tenuous linkbetweenhealth spending and health status (Filmer et al., 2000). 108. This chapter first describes the background to Mozambique's "lost decade" in health outcomes. Itproceeds to examine the allocation and effectiveness o fpublic expenditure inhealth. It concludes byproposing further reprioritization and a series o f reallocations inaccordance with the efficiency and equity goals o f the Government's health strategy. B. KEYHEALTHSECTORISSUES 109. Health care is overwhelmingly provided by thepublic sector, organized under the title National Health System (NHS). It has 1,224 facilities and 16,098 beds and a staff o f 17,890 workers (HumanResource DirectorateAnnual Report). Ofthe medical staff, 60 55 137 from World Bank (1992); 150 fromWDI. See similar numbersinINE-DHS(1998). 56203 fromWorld Bank (1992); 219 fromDHS-INE(1997). 57Life expectancy: 49 inWorld Bank (1992); 43 inWDI for 1990. 64 percent are trained. Inthe past decade, private care provision has expanded, mainly in the major cities; the services are priced for middle-class customers. As o f 1997, the Ministry o f Health accounted for 70 percent o fhealth spending, NGOs for 23 percent and for-profit private operators 7 percent. Non-profit care is provided increasinglybyNGOs, some with donor backing, but their contributions are significant only inMaputo City and Nampula, where 31were identified ina survey in2001.58 110. There are two further sets o f providers: the network o f community facilities and the traditional healers. The former i s managedby community financed village health workers, who receive technical support and drugs from the National Health System. The volunteer workers (socorristas) are not paid. Due to very limited training and access to only a restricted set o f drugs, they are not very effective health care providers. The communities apparently know this, and frequently bypass the community facilities. To convey an idea o f their relative importance, some 3 percent o f clients at rural N H S clinics had previously visited a community health post, and hlly 76 percent had visited a traditional healer. Onthe other hand, a beneficiary assessment (Swiss Agency for Development Cooperation, 2000) found that only 3 percent ofhouseholds surveyed went to traditional healers, though inNampulaprovince the figure rose to 6 percent. Dueto the lack o f information about these two sets o fproviders, this PER does not focus on them other thanto reinforcethe plea o f the Health Expenditure Review (Yates and Zorzi, 2000, p. 4) that their role be assessedformally and that ways o f expanding their role in the health care process be investigated. 111. Thehealth status of Mozambicans is extremelypoor, due to poverty, lack o fclean water, poor hygiene standards, malnutrition, an environment which fosters harmful pathogens and parasites, the after-effects o f war, environmental disasters5' and inadequate health services. As inmany sub-Saharan African countries, the disease burden arises mainly from infectious andparasitic diseases, particularly malaria, pneumonia, measles, tetanus, tuberculosis, gastrointestinal illnesses and HIV/AIDS. Malariai s still the dominant cause o f morbidity: according to a survey o f well-being, the QlBB(Questionho de IndicadoresBisicos de Bem-Estar), 18percentofthe population had been illinthe fortnight precedingthe survey, o fwhom 46 percent hadmalaria and 13 percent diarrhea (Instituto Nacional de Estatistica, 2001). Table 19 shows that Mozambique's infantand under-5 mortality rates, andmaternal mortality rates, exceed those ofMalawi, Tanzania andZambia andthe sub-Saharan African averages. '*On the role ofNGOs, see also Pfeiffer (2003) andAgQncia Suiqa para o Desenvolvimento e a ''Cooperaqiio(2001). From 1965 to 2001, 14major floods, nine major droughts and four major disasters from typhoon landfalls occurred. Most were associated withoutbreaks of cholera and diarrheal diseases. 65 Table 19. Health expenditures and performance: a regional comparison Mozam- Malawi Tanzania Zambia Sub-Sa- bique haran Africa Public healthexpenditure as a % of GDP 3.4 2.8 1.3 3.6 2.4 Malnutrition (wt. for age, % under 5 years) 26 25 29 24 n.a. Populationper physician 43,583 45,737 n.a. 10,917 21,970 Life expectancy at birth(years) 42 39 44 38 47 Infant mortality rate per 1000 live births 129 103 93 115 91 Mortality rate under 5 per 1000live births 200 193 149 187 162 Maternal mortality rate (/lOO,OOO live births) 1,083 620 (PESS) - 649 (PESS) - (PESS) Sources: World Bank, World Development Indicators, unless otherwise indicated. "PESS" i s the Plano Estratdgico do Sector du Suzide (Health Sector Strategic Plan) 2001-2005-2010(see Republica de Mogambique, Council of Ministers, 2001). Most o fthe indicators refer to the year 2000, this being the latest year for which data are available. As remarked above, infant mortality has fallen only a little inthe past decade. There are also large geographical variations ininfant mortality rates, from 60 per 1000 inMaputo City to 180 inZambCzia. Much o f this is associated with malnutrition. Almost one in every two children inrural areas is malnourished. Global studies indicate that malnutrition i s implicated inaround 50 percent o f childhood illnesses. While food availability i s a necessary condition to reduce malnutrition, there are more important causal factors: lack o f knowledge about appropriate feeding practices, poor maternal health and nutrition, and lack o f diet diversity. Primary health care can potentially make an important difference even without significant changes inincomes and food output. The fact that there has beenbut little improvement inthe infant and the under-5 mortality rates prompts the question whether the primaryhealth care services have hadthe needed impact inchangingbehavior inappropriate ways. 112. The causes o fthe keyhealth outcomes, as remarked above, are multi-factorial. In Mozambique's case, the key factors have generally improved: (i) Foodavailabilityhasimprovedbysome31percentonacountry-widebasisinthe past decade, andmost o f this i s due to increased output o f cereals by smallholders6'; (ii) Povertyhasdeclinedinbothurbanandinruralareas,asemergesfrom comparisons o fthe household survey o f 1996/7 with the CWIQ survey o f 2001 (Simler andHarrower, 2003). Thenumber o fhouseholdowning a radio rose from 27 percent in 1996 to 49 percent in2001, andthe number owning a bicycle rose from 12percent in 1996 to 27 percent in2001. (iii) Access to clean water has improved considerably inrural areas-from about 10 percent in 1992 to 35 percent in2001 (see chapter 6 o fthis PER). The impact onhealtho fother factors such as natural disasters are harder to evaluate, and it maybethat the floods o f2000 hada largenegative impact on outcomes; but on the 6o Agriculture value added grew 4.9 percent per annum 1990-2000,while population grew 2.2 percent pea., making an increase of 31percent infood availability. 66 other hand, Mozambique has sufferedmanynatural disasters inthe past, such as droughts. andit is not clear that the recent ones have beenmore damagingthan those o f a decade ago. The puzzle, then, i s why, despite the improvements inseveral health- relevant factors, as well as the increased effort by the health services, the headline health outcomes improved so slowly. Onthis point this chapter o f the PERi s deliberately agnostic, owing to the complexity o f the subject. Inthe remainder o fthis chapter, an attempt is made to evaluate some aspects o fthe health service to see whether there are inefficiencies that could berectified so that the health service has a greater impact on outcomes inthe future. Note that this chapter o f the PERdoes not argue that the health services are to blame for the disappointing health outcomes. Rather the point is that the linkbetweenhealthoutcomes andhealth services istenuous, andthat it consequently behooves the authorities to seek efficiency improvements so as to have a greater impact inthe future. 113. Therehas been a substantial increase in the coverage ofprimary health services. The percent o f children fully immunized by two years of age was 55 percent in 1995, risingto 82 percent in2001. Immunizationrates improved-withmeasles coverage risingfrom 39 percent in 1980-85 to 57 percent in2000, andDPT coverage rising from 29 percent in 1980-85 to 61 percent in20006'. Deliveries attended by a trained provider rose from 26 percent in 1993 to 45 percent in2002. Consultationsper inhabitant more thandoubled, risingfrom 0.36 in1993 to 0.77 in1999. Taking allthe effortstogether (consultations, pre- andpost-natal care, etc.), the overall output of the government health system rose by 59 percent between 1993 and2000, particularly inthe northern region, owing to large investments andredeployment o fhealth staff. On a per inhabitant basis, "standard delivery units" (usinga set o f coefficients to weight the sum o fthe different types o f care) 62 increased from 2.4 in 1993 to 3.3 in 1999, or a 37 percent increase. 114. Thecommonfund for drugprocurement hasproven a success. A common fund, financed by 8 donors, was started in 1997 to rationalize pharmaceuticalprocurement. Expendituresrunto $15-20 million annually, o f which the Government shoulders a small ($3-4 million) but growing share. Usinginternationaltendering, the pool attains economies of scale. The central health authorities have administered it competently, with assistance fi-om the Swiss Development Corporation, which has providedthe added benefit o frelatively good data sets on many aspects o f health delivery. The bulk o f the drugs are distributedvia the "classical route" o frequisitionsbythe hospitals, whence they are distributedto the clinics; about one-tenth are distributed inthe form o fdrug "kits", as part o f the EssentialDrugProgram, direct to clinics andhealth posts. A cost recovery systemhas been inplace since the 1970s. Officially, an average o f some 2.5 percent o f the cost i s recovered. However, the administration o f the cost recovery setup, with its well-intentioned exemptions, is widely perceived as unfair andprejudicial to the poor (see below inparagraphs 119 and 120 onpage 68). There will be a need inthe 61There may be an issue of definition. By Ministerial records, D P T coverage was 82 percent by 2001 (HSRP Indicators, 2002). By World Development Indicators records, DPT coverage was 61 percent. 62The formula is: SDU = 12*deliveries + 9* consultations + mother & child consultations. bed occupancy days + 0.5* immunizations + external 67 future to "indigenize" the management o fthe common fund and the data collection that accompanies it. 115. But the increased coverage and health care effort has not, in general, had the anticipated impact. Three distinct groups o f linkages can be identified: 0 Ina few areas, there is a clear linkbetweenthe care effort andthe outcome. An example i s the increased rate o f clinic-attendedbirths, which contributed to the sharply reduced maternal mortality rate. 0 Ina few other areas, one wouldnot expect theincreasedcare effort to havehad muchimpact, e.g. the reduction inlife expectancy stemming from HIVIAIDS (because o f the prohibitively highcost o f anti-retrovirals). 0 Overall, as pointedout above, the link betweenthe care effort and expenditure on the one hand, andhealth outcomes on the other, i s tenuous. The improved vaccinations and immunizations, andthe overall increased effort (consultations, pre- andpost-natal care, etc.) didnot do much for the infant andunder-5 mortality rates. This maybepartially due to the impact o f HIV/AIDS on infant and under- 5 mortality (manybabies bornwith AIDS die early, and families afflicted with AIDS providepoorer care for children). The following paragraphs suggest some clues about factors diminishing the quality o f care and limitingaccess63,and which may contribute to the explanation for the tenuous linkbetweenthe increasedhealtheffort andthe disappointing outcomes. These include: the distance from health facilities, low staff qualifications and general lack of staff, poor morale and staff remuneration, lack of drugs, and inappropriate use of user fees. 116. There continue to beproblems of access due to distance, the cost o ftransport, and the lack o f access roads. The SDC-fundedbeneficiary assessment was revealing on this point64. When asked about difficulties inusing the health facilities, between 13 percent and 57 percent o f respondents said the high cost was a problem, between 2 percent and 32 percent referred to bad treatment, between 14percent and 80 percent referred to transport difficulties, andbetween 2 percent (Tete) and 30 percent (Maputo City) referred to "lack of medicines". 18 percent o f Tete respondents, 15 percent of Inhambane respondents and 13 percent o fNampula respondents took more thanthree hours to reach the health facility (ibid, p. 72). Furtherprogress with the road rehabilitation program could be one o fthe keyhealth interventions o f the future. 117. Staffqualijkations and numbers do not match the caseload. As o f 2002 there were 17,890 public health sector staff (HumanResource Directorate Annual Report), their numberhaving increasedby 4 percent since 1998. There are 43,584 inhabitants per 63The beneficiary assessmentby SDC (Swiss Agency for Development Cooperation, 2000) examined a large number ofpotential barriers and found, using a logit model (p. SS), that socio-economic status, educational status, distance from health facility, andprovince (possibly proxying for illegal user fees) were statistically significant indetermining access. 64Swiss Agency for Development Cooperation, 2000, p. 68, Table 3.14. It should be noted that the survey has limited representativeness as it covered only four provinces. 68 doctor, similar to Malawi's 45,737 but greatly inexcess o f Zambia's 10,971 or sub- Saharan Africa's 21,970 (see Table 19 onpage 65). Surveys indicate that consultation times were very short (average 4 minutes), that triage practice was poor, that health worker compliance with protocols was poor, andthat diagnosis andtreatment was were not always appr~priate~~.Although the SDC beneficiary assessment found (p. 12) that 87 percent o frespondents were satisfied with the care that they received, staff shortages proved to be a great concem, e.g. over halfo f the users o fhealthposts inInhambane raisedthis as a serious concem @. 11). When asked what improvements were needed, 36 percent overall referredto availability o f staff (p. 69). 118. Staffsalaries, especially at the senior level, are low. Particularly at the senior levels, salaries are low compared with border wage rates. A hospital doctor in Mozambique is paid $3,850, compared with $9,000 inLesotho, Botswana andZimbabwe (HSER, 2002, p. 73). A senior healthmanager is paid $3,800, compared with $10,500 over the border. With markets for physicians andregistered nurses rapidly becoming integrated Africa-wide, this i s an unsustainable situation. The low wage rates are neither conducive to attracting the top level skills, nor to improving staff productivity or the quality o f care. 119. Drug availability continues to be aproblem. There have beenimprovements in the adequacy and predictability o f medical supplies due to the allocation o f a significant share o f a growing budget to pharmaceuticals. However, there is considerable evidence that drug availability didnot improve at the rate planned. Christie and Ferrara's (1999) surveyinquiredwhy respondents who were illdidnot seek health care at a facility. 38 percent responded that the facility was too far, 35 percent said they lackedthe money, and 6 percent said the facility lackedthe drugs. A beneficiary assessment (Swiss Agency for Development Cooperation, 2000) documented a poor state o f drug availability in many health centers; approximately two-thirds o f the surveyed health facility users identified the availability o fmedicines as the area needing the greatest improvement and attention. The RationalDrug Study by SDC (200 1) found problems with availability of antibiotics, and noted that the availability o f drugs was no better thanin 1996 when comparable data hadbeen collected. Inthe QIBB/CWIQ survey in2001, respondents were asked whether they went for a medical consultation inthe past fortnight. 11percent did(INE2001, p.36), but o fthese, 39 percent complainedo fthe longwait, 35 percent o f a lack o f drugs, and 19percent o f the cost. Inthe Expenditure Tracking and Service Delivery Survey (see para. 136), 58 percent o fthe clinics surveyed had suffered a stockout o f at least one essential drug inthe previous six months, and only 66-76 percent had all essential drugs instock at the time o f asking. 120. Irregular use of user charges is a disincentive to receiving adequate care. There i s a need for a transparent cost recovery system andrules o f exemption. There i s much anecdotal evidence o f illegal drug selling. This is illustratedby a quote from the SDC beneficiary assessment (p. 76): "As one focus group participant inMaxixe District, 65 URC study ofthe management of quality (1998), Rational drug studyby SDC (2001), Baseline survey for evaluationo f IMCI(2001). 69 Inhambanestated it, "qyoupay 1000Mts. there are no drugs, [but] $you have 40- 50,000 Mts, drugs appear immediately". Perturbingis that while users say they are forced to pay, healthworkers believe that they provide free care as needed (SDC beneficiary assessmentp. 80): in22 out o f the 30 health facilities examined, where health workers indicated that patients get treatment even ifthey do not have money, it was found that at least 20 percent o fusers o fthose facilities indicated that they were refused treatment. There is a perceptionthat there "are no free services". Fee information i s not readily available, making way for negotiation over the price. Widely differingprices are charged. While child consultations are supposed to be free, rates o f 500-1000 Meticais are often quoted, and sometimes rates significantly higher than these are quoted66. The ETSDS found in2002 that, although consultations are supposed to be gratis, 35 percent o f children were requiredto pay. More surprise inspections o f facilities may be called for, as well as clear postingofprice and exemption schedules at all facilities. 121. It is unlikely that theMillennium Development Goals will be attained. The MDG for matemal mortality i s that it fall by %by 2015, viz. from 1083 per 100,000 live births to 271, This i s unlikelyunder existing circumstances: even ifoverall matemal mortality were to fall at the samerate as intra-hospital maternalmortality, this would bringit down to only 644 by 2015. Another MDGi s that under-5 mortality fall bytwo-thirds, viz. from 200 in2001 to 67 in2015. Takingthe "best" rate o freductionbetween 1989 and2001 as a guide, projecting current trends delivers an under-5 mortality rate o f 180 in2015. Even this is optimistic due to the possible intensification o f AIDS. c. EXPENDITURETRENDS 122. Health expenditure has risen sharply. As i s shown inTable 20, total public expenditure on health, domestically and externally financed, grew by 21 percent annually inrealterms between 1997and2000, rising from 2.3 percent ofGDPin1997to 3.4 percent in2000. Otherwise expressed, it was the equivalent o fUS$5.0 per capita per annum in 1997 to $7.5 in2000, and is estimated at $10.7 in2002. 123. Health expenditure is largely externallyfinanced. This considerable increase in spending was financed by increases inthe government budget allocation from about 7.7 percent o f domestically financed expenditure in 1997 to 8.8 percent in2000, as well as by increases inaid -grants, loans, andthe health component of general budget support. The domestically and extemally financed contributions grew at roughly the same rate, such that external fundingremainedbetween 69 percent and 70 percent throughout the period. Expenditures are projected to rise to $15.9 in2005 andto $21.3 in2010. 124. A study carried out byMinist6rio da Saude e SDC (2001) managed to obtain information from 18 bilateral donor agencies, 5 development banks and 4 UNagencies supportingthe health sector. Ofthese, the largest supporters were USAID, the World Bank, Switzerland, Norway, theNetherlands, Denmark, the EuropeanUnionandthe UK. 66SDC beneficiary assessment (p. 13). 70 Disbursements by all donors inthe health field totalled some US$ 80 million in2000, of which three-quarters took the form of grants. Table20. Totalactualhealthexpenditureat currentprices, 1997-2000 Item 1997 1998 1999 2000 Total public expenditureinthe healthsector Nominal (Mtbn) 929 1,141 1,476 1,974 Real (constant 2000 M t bn) 1,117 1,312 1,649 1,974 inpercent ofGDP 2.3% 2.4% 2.8% 3.4% Contributionfrom different sources of finance in billion Meticais Statebudget, viz. &omtax revenues 258 316 401 571 Aid General budget support, th" state budget 73 90 220 345 Grants and loans to health sectora 574 708 817 1007 Household contributions 24 28 37 51 as percent of total expenditures in health State budget, viz. from tax revenues 28% 28% 27% 29% Aid 69.7% 69.9% 70.3% 68.5% General budget support, thru' state budget 7.9% 7.9% 14.9% 17.5% Grants and loans to health sector 61.8% 62.0% 55.4% 51.0% Household contributions 2.5% 2.4% 2.5% 2.6% 100% 100% 100% 100% Per capita total expenditureinthe healthsector Incurrent Meticais 57,793 69,384 87,636 114,502 InUS dollars 5.O 5.9 6.9 7.5 Share of domestic state budget allocatedto health 7.7% 6.5% 7.7% . .. 8.8% Source: HSER 2002 Tables 4.1 and 4.2, unless otherwise stated. aSource: MinistCrio da Saude e SDC, 2001. Source: from estimates o f recordeduser fees and charges inAustral e MinistCrio do Plano e Finanqas (1999). These are probablyunderstated. 125. Thecontribution of externalfunds to capital expenditures declined from 82 percent in 1997 to 61 percent in2000 (see Table 21). This reflects a shift indonor support frompost-war reconstruction towards a greater emphasis on consolidatinghealth infrastructure and improving the volume, coverage and quality ofhealth services. 126. Recurrent cost implications of capital spending are now high on the agenda. The Ministry has produced a draft investmentplan (2002) which proposes restrictions on new investments to the most under-served areas andto facility upgrades where there are clear efficiency gains to be achieved. The recurrent cost implications have been assessedand included inthe Cenririo de Despesas eFinanciamento do Sector Saude de Me`dio Prazo 2001-2005 (MinistCrio da S a ~ d e2001a). 71 1 Table21. Source of fundingby economic classification,1997and2000 1 9 9 7 2 0 0 0 Item State Extemal State Extemal budget fundsa budget fundsa Total publichealthsector expenditure(current Mtbn) 331 486 917 864 Recurrent expenditure 301 353 793 668 Salaries and personnel-related costs 137 107 373 196 Pharmaceuticals 17 155 62 265 Other goods & services and recurrent costs 147 92 358 207 Capital expenditure 30 132 124 196 Contribution frommainsources of finance(%) Recurrent expenditure 46% 54% 54% 46% Salaries and personnel-related costs 56% 44% 66% 35% Pharmaceuticals 10% 90% 19% 81% Other goods & services and recurrent costs 62% 39% 63% 37% Capital expenditure 18% 82% 39% 61% Source: HSER (2002) Table 4.3, inturn from CGE (1998-2000), Ministry o f Healthand Management Sciences for Health(1999) and MinistQio da Saude e SDC (2001). aExcluding health's share o f general budget support implicit inthe State budget. 127. Thefastest growing recurrent spending category, "other goods and services", is being diverted into staffbenefits. Between 1997 and2000 this category grew by 97 percent inreal terms (see Table 22). Anecdotal evidence suggests that much o f this constitutes transfers inthe form o fvehicles, training courses, seminars, workshops, inlandor abroad, andother rewards. As notedinthe chapter on cross-cuttingissues, higherandmiddle levelhealth sector workers arepaidwell below their market value. Underthese circumstances dual employment arrangements, illicit fee charging and a failure to exempt indigentpatients from charges are inevitable. The extent o f the resultant inefficiencies may be considerable because o f the extent to which expenditures on fuel, utilities, transportation and maintenance related to priority service delivery are being compromised. 128. The share o fpharmaceuticals inexpenditure declined from 26-30 percent in 1997/8 to 22 percent in2000 (see Table 22). Nevertheless real spending on pharmaceuticals increased by 58 percent over the period 1997-2000. 72 Table 22. Health expenditure (actual) by economic classification Item 1997 1998 1999 2000 Growth 1997- in billions of current Meticais: Total expenditure 817 1028 1,331 1,781 Recurrent expenditure 654 876 1,112 1,461 Salaries and personnel-related costs 244 353 461 569 Pharmaceuticals 172 261 267 327 Other goods & services and recurrent costs 239 262 384 565 Capital expenditure 162 152 219 320 in billions of constant 2000 Meticais: Total expenditure 982 1,182 1,487 1,781 81% Recurrent expenditure 787 1,006 1,242 1,461 86% Salaries and personnel-relatedcosts 293 405 515 569 94% Pharmaceuticals 207 300 299 327 58% Other goods & services and recurrent costs 287 301 429 565 97% Capital expenditure 195 176 245 320 64% Comvosition of recurrent expenditure: Salaries and personnel-relatedcosts 37% 42% 39% Pharmaceuticals 26% 24% 22% Other goods & services and recurrent costs - 37% 35% 39% 100% 100% 100% 100% Source: HSER 2002, Tables 4.4 and 4.5, intum from CGE 1998-200; Ministtrio das Finanqas e SDC (2001) and others 129. Theshare of staff costs in recurrent expenditures isfairly low, at 40percent (see Table 22). Inlarge part, this reflects the wage compressionreferred to above. The increase instaffbetween 1997 and 2000 was 3 percent, so that nearly all o fthe increase o f spendingon this item (94 percent, inreal terms, between 1997 and 2000) went into compensation - salaries, benefits andtraining. Civil service wage scales were decompressed to some extent in 1999 and this explains the bulk o fthe increased compensation. Despite the increase inwages, as o f 2000 the staff cost per worker was only $20067per month all told. Low salaries have often tempted professionalhealth workers to engage insecondary income generating activities, andthis may help explain the lack o f impact of the salary reform on productivity. 130. Thesalary decompression of 1997-2000failed to induce signiJicantproductivity gains in terms of quantity or quality. Between 1997 and 2000, health service provision rose by 13 percent, while service quality appearsto have remained low. Services delivered per staff member rose by 7 percent (or just over 2 percent per year) while remunerationrose by 87 percent. The unit cost o f service rose by 53 percent. From the point o fview o f civil service reform, the major increase inspending was not a success, in that by usingthe increased salary bill to reward the competent and by reassigningthe incompetent, greater productivity increases could have been obtained, either inthe form of increased outputs or improved quality. Inthe next round o f civil service reform-in 67Source: HSER 2002, Table 4.6. 73 which the health ministry is one o fthe pilot ministries - it should be ensured that salary reform is more effective. 131. Further expansions of health spending areplanned. The projections inTable 23 are mostly from the Health Sector Expenditure Review (HSER, 2002), which are based on (i) increasing resources for the health sector, consistent with the PARPA; (ii) increasing external financing as forecasted by the sector's mediumterm expenditure framework (MinistCrio da Sahde, 2001a); and (iii) GDPgrowth o f 8 percent. This scenario leaves Mozambique almost as dependent upon foreign assistance as currently (66 percent in2010 versus 69-70 percent currently). By 2010 spendingwould be about $20 per capita. 132. However, theseprojections are inconsistent with the recurrent budget ceilings developedby this Review. The methodology andrationale for determining ceilings on recurrent budget items was explained inparagraph 15 on page 22 (Table 4). Inthe case o fhealth, this recurrent budget ceiling i s reproduced inline 7 o f Table 23. The implication is that the medium-term projections of Table 23 will have to be reducedby about 9 percent -unless additional donor assistance can be found and applied to the recurrent budget. Table 23. Projected health spending up to 2010 Item 2000 2001 2002 2005 2010 actual est. est. proj. proj. 1 Total public expenditurein the healthsector 2 Nominal (Mtbn) 1,974 2,782 3,458 6,389 12,342 3 Real (constant2000 Mtbn) 1,974 2,588 2,945 4,701 7,116 4 in percent of GDP 3.4% 3.8% 4.0% 5.1% 5.2% 5 Total recurrent expenditure 6 fromHSER2002 (US$ m) 93 110 120 185 7 feasible ceilingfrom Table 4 (US$ m) 93 110 120 I70 8 Contributionfromdifferentsources offinance 9 as percent of total exDenditures in health 10 Statebudget, viz. fromtax revenues 29% 27% 29% 31% 32% 11 Aid 12 Generalbudget support 18% 19% 18% 9% 5% 13 Grantsand loans to healthsector a 51% 52% 52% 58% 61% 14 Householdcontributions 2.6% 2.1% 2.2% 2.2% 2.2% 15 Per capita total expenditureinthe healthsector 16 IncurrentMeticais (`000) 114 158 191 328 620 17 InUSdollars 7.5 7.6 7.8 11.4 20.3 18 Share of healthindomestic state budget 8.8% 9.7% 9.9% 11% 13% Source: HSER2002 Tables4.10 and4.11, unless otherwise stated. aSource: Ministerio da Saude e SDC, 2001 (for 2000 and2001). Source: fi-omestimatesofrecordeduser fees and charges inAustral e MinistCriodo Plan0e FinanCas(1999). These are probablyunderstated. 133. For those in receipt of health care, health service use isfairly equitably distributed, particularlyfor preventative health care. A benefit incidence study using the 1997 household survey showed that the distribution o f services from preventative health care (bothvaccinations and antenatal care) are very close to the 45-degree line, indicating that service use is highly equitable inthe case o fvaccinations (Figure 6 reproduced from 74 Heltberget al., 2001). The benefits from curative health care6*are less evenly distributed, with the poorest half o fthe population receivingabout 35-40 percent o f hospital and health center usage.69 Figure6. Distributionofbenefitsfrompublicspendingin healthby incomedecile i Consumptionper capita Vaccination participation Antenatal participati a 0 Proportionof population,poorest to richest 1 134. Equity of access to basic services improved modestly, but much remains to be done. A measure o f equity i s to compare the amount o f services7' per inhabitant on a provincial basis (Table 24). Inthe northem provinces these grew at generally faster rates than inthe initiallybetter-served southem provinces, so that the high/low ratio fell from 7.7 to 5.9 between 1993 and 2000, indicating a modest i m ~ r o v e m e n t . ~ ~ 68 All individuals who accessedhealthservices (hospital, health center, or other medical facility) during the monthpreceding the interview. 69 In most countries consumption i s more evenly distributed than hospital usage. Figure 6 indicates the opposite for Mozambique. This anomaly stems from: (i) respondents' conksion among "health post", "clinic" and "hospital", and (ii) the focus o f the survey question onpublic facilities, omitting private hospital use by the better off. 70 See the definition o fthe composite indicator o f output at footnote 62 onpage 66. 71 The 1992PERnotedthat the Government's objective was to increasethe share ofrecurrent expenditures at Levels Iand I1(primarycare) and thus improve the equity o fthe system. This share fell, however, from 64 percent in1982to 44 percent in 1989 and to 42 percent in2001 (for details see Annex 11, Table 39, p. 112). Inpreparatory work for the PESS it was assumed that the ratio could be raisedto 47 percent by 2005 (MinistCrio da Saude, 2001b, table 6). The tertiary hospitals in2000 consumed 45 percent o fthe finding for goods and services. Another measureo f equity i s the ratio o f the goods and services budgetper capita for the most favored (Maputo City) andthe least favored (Zambkzia) provincehegion. This ratio fluctuated without trendbetween 3.92 and 9.51 between 1994 and 2001 -see Annex 11, Table 39 on page 112. 75 Table 24. Health service units per capita, selected provinces and regions Health service unitsaper capita 1993 2000 increase (YO) C. Delgado (lowest in2000) 1.6 2.4 45 Zambtzia (lowest in 1993) Maputo province + city +H C M (hlghest in 1993) 1.5 2.5 70 11.4 13.9 29 Mozambique 2.7 3.6 35 Ratio: lughest / lowest 7.7 5.9 Source: HSER 2002, Table 5.5. aWeighted sum o f a list o fidentified services (e.g. consultations). See footnote 62, p. 66. Maputo City and the Hospital Central de Maputo are included. Excludingthemgives a similar quantitative result. Detailed data inHSER 2002. More comprehensive measures o f inequality (e.g. coefficient o fvariation, using data for all provinces) deliveredsimilar qualitative results. This improvement notwithstanding, the territorial distribution o f facilities improved only slightlybetween 1992 and2002, with the total number offunctioning facilities increasing from 1,140 to 1,22472. One speculates that the aim o f equity of access to basic services mighthavebeenreached at lower cost byincreasing further the number o fruralfacilities (with appropriate staffing andresources) inthe under-served northernprovinces; andthat one o fthe explanations for the weakness o fthe linkbetween the care effort andhealth outcomes may be that many needypeople have not yet beenreached. 135. Keyfinancial information onprograms and other important aspects is unavailable. The new hnctional classification, established in2001inaccordance with UNstandards, doesnot automatically containdetail about healthinterventionprograms. These need to be created for each country's specific situation. Since this has not yet been done, it was not possible to identifyhow much expenditure was directed at important programs: Pharmaceuticals Support, IntegratedHumanResource Development, Malaria Control, the ExtendedProgram o f Immunisation, Maternal & Child Health, Tuberculosis Control, andHIV-AIDS control. The ministryi s currently working on such a detailed classification and it is anticipated that this will be introduced at least as soon as the SISTAFE i s implemented. Another shortcoming is the tracking o fthe ministry's equity objectives: the standardbudget execution reports have no breakdown by facility level (I to N). 136. Trackingsurveys should become a standard instrument to improve expenditure quality and cut wastage. The ExpenditureTracking and Service Delivery Survey (ETSDS, Lindelow et al., 2003) inthe health service was a pilot effort to obtain survey data to track the transfer o f funds from the central government, to provinces, districts and healthposts. It attempted to link measures o f the effectiveness the funds transfer mechanisms to the effectiveness o f spending at the local level. The ultimate objective o f the survey was to examine andimprove accountability. The premise o f the ETSDS was that delays inbudget execution, as well as weak systems o f control, with consequent 72There were 1195 healthposts in 1992, o fwhich 255 hadbeen damagedbythe war, and 225 centers, o f which 15 hadbeendamaged (1992 PER, p. 105) - for a total o f 1140 fhctioning facilities. In2002 there were 1,224 facilities (Human Resource Directorate Annual Report). 76 scope for leakages anddiscretion inthe allocation o fresources, may adversely affect quality and efficiency o f service delivery. 137. Infact the ETSDS found that record-keeping was so poor that itwas impossibleto determine whether there was leakage o f funds betweenthe different levels. It did, however, turn up evidence o f leakage o f drugsbetweenthe provincial and the district level. As mentioned above (para. 120), the survey confirmed that the user charge system is inequitable, lacks transparency and creates scope for staff to pocket the moneys collected. 138. Concerning drug administration, serious failings were found. As pointed out (para. 119), 58 percent o f the clinics surveyed had suffered a stockout o f at least one essential druginthe previous six months. The average stockout time was 6-7 weeks (Lindelow et al. 2003, p. 42). Only between 66 and 76 percent o f all facilities had all essential ("EPI") vaccines instock at the time o f asking (ibid.,p. 44). 139. The ETSDS found that only 80 percent o f staff were actually inpost at any one time. Rural areaswere particularly likely to havepoorly qualified staff, andthere are highlevelso fturnover andlow levels o fsatisfaction ofstaffinrural areas. The whole question of staff incentives needs to be reexamined. It i s not only a matter o f salary, for rural staff complained o f the lack of access to public services, access to training and career advancement. 140. The survey, insum, revealed important information about the system's functioning andpointed the way to improvements. Several o f its recommendationshave been cited above. Overall the ETSDS was a successful venture and should be re eated in the health ministryand other ministries as a means of improving service quality. R D. IMPACT OF HIV/AIDS ONTHE HEALTHSYSTEM 141. Mozambique is one of theAfrican countries hit hardest by the HIV epidemic. In 2002 about 1.1millionMozambican adults and children were livingwith HIV/AIDS, with an adult prevalence o f 13.8 percent. The impact o fthe epidemic is likely to grow more serious duringthe next decade (Ministry o f Health et al., 2002), with prevalence o f HIV/AIDS among adults increasing to 16.3 percent by 2010. The number o f AIDS- related deaths i s projected to double by 2010. By 2010 life expectancy at birthis expected to drop from 43 to 36 years, rather than increasingto 50 years. Similarly, by 2010 infant mortality is expected to be at least 25 percent higher than it would have been inthe absenceofHIVIAIDS.By2010, anestimated 1.13 millionMozambican children will have lost one or bothparents to AIDS. 142. Thedemandfor health services will increase substantially as a result of the HIV/AIDSpandemic. HIV/AIDS affects the health sector bothby increasing the demand for health services and taking its toll on healthpersonnel. Some o f these increasing costs 73 Inaddition, it would be even better ifthe routine data system were improved so that detailed data on spending were collected as a matter o f course. 77 are already beingbome by the system inthe form o f treatments for opportunistic diseases; an estimated 30 percent o fhospitalbeds are occupied by HIVpatients (Haacker, 2001a). The likely cost o f treatment inthe future depends on the government's decisions on coverage andnature o fcare. To providepalliative care andtreatment of opportunistic diseases to 20-30 percent o f the population with HIV would absorb halfoftotal health expenditure; adding anti-retroviral therapy for 10 percent o f HIV/AIDS victims would cost more than another half (Table 25). The total would be 3.0 percent of GDP, compared with the 2.8 percent o f GDP spent on health currently. These costs would double by 2010. Table 25. The cost of HIV-related health services (percentage of GDP), 2000-2010 2000 2010 Total HIV-relatedhealth services, for the assumedrates o f coverage 3.0 5.9 O fwhich: Palliative care and prevention o f opportunistic infections 0.3 0.5 Clinical treatment o f opportunistic infections 1.1 1.8 Costs o f highly active antiretroviral treatments (HAARTs) 1.7 3.6 Note: 1. Assumptions: coverageof30 percent for palliative care andpreventionof opportunisticdiseases, 20 percent for clinical treatmentof opportunisticinfections, and 10percent for HAARTs. The assumedcoverageratesare not intendedas statements of desirablecoverage, and simplyreflect the fact that accessto healthservices is limited. Source:Haacker, M.(2001a), Tables6 and7. 143. The costs of the loss ofprofessionals in the health sector is also substantial. Training o f doctors andnurses will have to be expanded by at least 25 percent over the 2000-2010 periodjust to keepthe number o f doctors andnurses constant (Haacker, 2001a). At a rough estimate o f $10,000 (Pavignani 2002) for training a new professional, andassuming realistically a deathrate o f some 100per year, the incrementaltraining cost would be $1 million annually. This rough estimate does not account for the costs o f absenteeism due to HIV/AIDS. 144. Theauthorities are studying thefinancial implications of HIV as apart of the continuedpublic expenditure review process. No attempt is made here to incorporate these rough estimates o f the costs o f HIV/AIDS inthe financial projections, because (i) the study has not yet been done, and (ii) decisions on coverage andnature o f care have not been taken. However, giventhe magnitude o f the costs, it is self-evident that a comprehensive response to prevent and mitigate the impact o f HIV/AIDS needs to be operationalized as a matter o fpriority. Inthe meanwhile, attempts at prevention are clearly essential, andthe Government's strategy inthis regard should bepursued vigorously. E. POLICY PRIORITIESFORTHE HEALTHSECTOR 145. The Government is makingprogress withprioritization but there is still much to do. The Plano Estratkgico do Sector da Satide (PESS -Republic o fMozambique, 2001), despite its name, consists to a large extent of a listing o f areas where strategizing i s needed, but does not actually lay out concrete strategies to address several o fthe key problems. An example is Mozambique's very highmatemal mortality rate (1,083 per 100,000 live births, well inexcess o f regional averages): although this problem is cited several times, the PESS contains no clear policy directions to reduce the rate. Another 78 example i s the lack o f spending, output and outcome data on the major cause o f mortality, namely malaria. The Government recognizes that the PESS - a living document - shouldbe improved and adjusted to the rapidlychanging environment, and i s undertaking steps inthis direction: a revision o f the PESS is expected inthe near future; an Investment Planwas completedin2002; specific health development plans for major urbanareaswillbe developed; several studies o fgender aspects o fhealthare envisaged; under-served areas have been defined; and an attempt is being made to create a common donor fund. It is urgentthat these plans bebrought to fruition, and that there be detailed prioritization linkedto spending numbers andbased on sound analyses and monitoring, so as to pursuethe difficult objectives o f increased equity and the expansion o fthe network to include the under- and unserved. 146. Spendingand output datafor monitoring and evaluation o fthe vertical programs such as malaria control are needed for monitoring and assessment. A program-based classifier i s beingdeveloped so that the budget execution reports can report health expendituresmore effectively; inaddition, the detailed functional classifier whichwas formally adopted in2001 needs to be implementedso as to be able to track poverty- relevant expenditures. 147. Thevariouspolicy and strategy documents need to be consolidated ina consistent planninginstrumentwith clear prioritization linkedto resources. This consolidation should include the Plano Estratigico do Sector da Sazide (PESS). This can and should be done regardless o f the type o f strategy the donors adopt for conveying financial aid - Fundo Comum,budgetary support, or a continued project approach under strengthened Government leadership. 148. Inthis regard, aquote fromthe 1992PERisinstructive: "This fragmented organizational structure i s not well suitedto the acute scarcity o f qualified personnel ... Overlappingresponsibilities anda lack ofcoordination betweenunits are commonplace., No single unitwithin the Ministry has a precise overview o f all the ongoing andplanned projects for a complete picture o f extemal assistance to the health sector.. .This lack o f coordination severely limits the Ministry's ability to ensure that health expenditures are inlinewithoverallnationalpriorities." (WorldBank, 1992, p. 102). Regrettablysome of these statements are still true in2003. The underlying causes o f this apparent stasis were explored inChapter 2 (p. 33): fragmentation associated with multiple competing donors, limitedmanagerial capacity, andcivil service incentiveproblems. The solutions proposedwere an increased focus on outcomes usingthe PARPA, progressive incorporationo f expenditure analysis inthe budget cycle, the progressive replacement o f projects bybudget support, together with civil service reform, andthe pursuit o fbudget comprehensiveness. 149. The "Fundo Comum"is an attempt to correct budgetaryfragmentation, but is not the only option. Similar to the PRO-AGRIinagriculture, the hope is that the Fundo Comum will enable a common platform for the donor community to engage the authorities inpolicy debate, to provide fundsjointly, to ensure common procurement rules, and to monitor effectively. The aims are laudable. Reducing the total number o f projects probably does reduce the Government's bureaucratic burden. However, as the experience inroads and inagriculture showed, the negotiationinvolved insetting up and runningsector-wide instrumentsis extremely time-intensive for bothGovernment and 79 donors, and inneither case didthis result ineffective monitoring o fthe impact o f expenditures on outputs and outcomes. Hence it should not be assumed that this i s the only solution. There are two alternatives. One is for the donor community to move an increasing share o f its fundinginto budgetsupport and to rely on the Government's procurement setup, which inturnwould require more focus on the overall fiduciary framework; given that project funds are inany case fungible, the gains incontrol over a small segment o f Government spendingare lost ifthe overall fiduciary quality i s low. Another solution -a short term stopgap expedient only -i s that the Government adopt stronger leadership, take on fewer andlarger projects, and seek to upgrade its policy capability, its output and outcome monitoring, and its financial information. 150. Extension of the health service to under- and unserved areas should be apriority. Only about halfofthe populationis reasonably served bythe health system (videhospital births o f45 percent, immunization 57-61 percent). An extensionis needed o fthe primary health system to cover the remainder who are often inareas distant from large towns. Travel times to medical assistance should come down, qualified staff shouldbe placed in sufficient numbers, and drugs shouldbe available consistently. Equality o f access to healthservices i s already a key objective o fthe Government's strategy andthe health system is already seeking to attain these objectives. This expansioncould a key aspect o f the Government's pro-poor orientation. Inaddition -based on evidence inother countries about roads andhealth- extended road rehabilitation inareas difficult of access couldbe an effective health intervention. 151. Pursuit of basic health will require reallocations and cutting waste. Improvements are needed inthe use o f drugs, as explained above. Efficiency gains in hospital spendingneed to be made, startingwith the low bedoccupancy rates. There should be full cost recovery for services beyond the basic health service package, at least at the level I11and IV institutions. 152. Consideration should be given to extending the roles of for-profits, non-profits, community health workers and of traditional healers, so as to cover the under-sewed. As privatesector health activity increases (especially inthe cities) the NationalHealth Service could increasingly target its efforts at those unable to pay. The role o f the NGOs has increased inrecent years and could be encouraged further. The PESS suggests, but does not develop a clear strategy for, extended collaborationwith community health agents (e.g. midwives). The PESS notes that current healthpolicyis infavor o f collaborating with the traditional medicine sector, but that inpractice little has been done. Since a significant number o f Mozambicans use these services, either exclusively or inconjunction with the National Health service, collaboration could be cost-effective. 153. The next round of civil service reform in health should learnfrom the lessons of thepast. An increase inoutput per staff o f 2 percent per annum for three years, in exchange for an increase inremuneration o f 87 percent cannot be pronounced a success. Effective methods o f linkingpay andpromotion to performance, and o fproperly remuneratinghealth care professionals, must be found so that the next round o f civil service reform is fruitful. 154. Theuserfee system needs a complete overhaul. The user fee schedules lack transparency and vary hugelyby facility. Overcharging i s common and exemptions are 80 not consistently granted, discouragingthe poorest from seeking care. As the Expenditure Tracking and Service Delivery Survey found, it is often the case that the amounts charged are not duly remittedthrough the official channels.74 Among the elements inthe needed reform are: civil service reform so that all staff are appropriatelyrewarded; a requirement that all units advertise the user fee schedules and exemptions inuser-friendly ways; unannounced inspections to penalize non-compliance; andregular beneficiary surveys and expenditure tracking surveys to assess the extent o fnon-compliance. The Government could also consider abolishing fees at Level Ifacilities altogether, because the contribution the fees make to the district budgets i s quite small, and little is retained bythe facilities themselves. Finally, all userfees should be reflected in the budget. 155. Theauthorities could consider contracting out the management ofparts of the health sewice, e.g. certain o fthe hospitals. This may be a way to attract highly competent management, permit greater freedom inthe selection o f staff, and lightenthe administrative burden for the Government. l4 The Government has already considered such a reform, and has done a costing study o f the Maputo Central Hospital with this in mind. 81 CHAPTER 5. ROADS A. INTRODUCTION 156. While Mozambique still has one o f the least developedroad networks inthe southern African region, it can boast o fremarkable achievements innetwork rehabilitation. The classified roadnetwork consists o f 25,000 kmo froads, o fwhich 4,275 kmare primary, 7,880 km secondary and the remainder tertiary. Some 5,000 km are paved. The country i s extremely sparsely served, with 32 kmo froadper 1000 square kilometers, lowest inthe southern Africa regionwhose average is 135 km and median i s 90 km.75Only 57 percent o fthe classified network is maintainable, i.e. ingood or fair condition, again one ofthe least developed inthe region, whose average was 71 in 1997percent. Rural roads comprise 53 percent o f the network, not far o f fthe median in the southern African region o f 59 percent. 157. In1992, Mozambique was emerging from aprolonged conflict that leftthe road network inextremelypoor condition androad management institutions lacking in capacity to planor execute the civil works to rehabilitate andmaintain the network. At the same time, it was acknowledged that transport policies hadbeenbiased toward the major east-west, port-rail transit corridors, andhad ledto a neglect o f road infrastructure to serve the domestic and export economy. 158. The mainobjectives were to restore serviceability to the 1973 level by 1998 through roadrehabilitation, bridge replacement and recurrent and routine maintenance. This was to be achieved through: (a) policy reforms, including private sector involvement incivilworks, cost recoveryinfinancing roadmaintenance, anddecentralized decision making; (b) institutional development, through training engineers andtechnicians to plan andimplementcivil works; and (c) rationalizing investments, bybalancing maintenance andrehabilitation, meetingthe transport needs o fthe agriculture sector, and alleviating poverty. The donor community contributed to the effort invarious ways, including the Roads and Coastal Shipping (ROCS) Programs 1and2 (1994-2002), and Roads I11which began in2002. 159. The overall performance o fthe sector against the objectives andbenchmarks established in 1992 was fair, barringone serious reservation, namelyperiodic maintenance: . Institutions. Substantial progress was madewith institutional capacity building, developing the execution capacity o f local roadnetworks, with the training o f technicians and engineers, andwith private sector involvementinroad rehabilitation andmaintenance. ''Similarly, roaddensity per million inhabitants i s 1,440 km, again the lowest inthe region, whose average i s 6,393. . 82 Rehabilitation. Roads ingood or fair condition rose from 10 percent to 57 percent o fthe classified network underthe two donor- andBank-supportedroads . programs, the Roads and Coastal Shipping (ROCS) 1and 2. Impassable roads decreased from 50 percent to only 8 percent. Regional roads: Between 1992 and2000 some 6,630 km o f tertiary and . unclassifiedroads were rehabilitated, most o f it inManica, Zambezia, Cab0 Delgado and Nampula. Routine maintenance made considerable progress. Less than4000 kmwere . givenroutinemaintenance in1994; this rose to 15,000 kmannuallyby 1998, achieving 80 percent o f the target. Periodic maintenance fell well short o f expectation. The planwas for 3000 Myr,butonly anaverage of434M y rwas done, or 18percentofplan. The cumulative maintenance backlog o f over 16,000 km exceeds the size o f the maintainable road network. Considering that a maintenance backlog already existed, it i s clear that the failure to perform the required level ofperiodic maintenance undermined some o f the positive impacts on road condition from the program o froadrehabilitation. The reasons for this massive failure are examined below. B. KEYROADSSECTORISSUES 160. Autonomy of roads management: much done, much to be done. Autonomy i s called for inorder to maximize efficiency inallocation o f investment resources for the national road network. An important step was the creation in 1999 o f the National Roads Administration (ANE),to replace the National Directorate o f Roads andBridges. The ANE is responsible for the administrationofnational andregionalroads, and for road financing through the Road Fund. ANE is an autonomous public institution subordinate to the Minister o fPublic Works and Housing. The management o f ANE reports to a Board o f Directors which includes bothprivate andpublic sector representation. 161. With the objective o f achieving objectivity inresource allocation and a concomitant separation o f local political influence from national road allocation decisions, a second phase o f road sector reforms calls for the separation o f the RoadFund from ANE, each with its own board comprising bothpublic- andprivate-sector representation, the creation o f a National Roads Council with responsibility for road policy, and more effective decentralization o froadplanning management responsibilities. 162. Little decentralization has been achieved. Very little planning, policy or management capacity exists at the provincial level. The desired devolution o f decision- makingandmanagement to provincial and local authorities who are best able to assess local needs will not succeed until such time as such capacities are inplace. The Roads I11 program contains substantial resources to this end. 163. Upgrading the analytical capacity o fthe road administration system is an urgent task. The statistical and monitoring capacity o fthe ANE i s deficient, and inparticular there i s no regular monitoring o f donor-funded expenditures. The upgrading should be 83 sufficient to permit expost analysis, to guide planning, to permit effective monitoring, to monitor donor-funded expenditures and to permit further improvements inefficiency. 164. Moreprogress with developingprivate contractors is needed. Routine maintenance shifted from being entirely state-run on force account to being split among the parastatal roadmaintenance companies (Empresas de Construqiio e de Manutenqiio de Estradas e Pontes) or ECMEPs (40 percent), labor-intensive brigades76(20 percent) attached to the latter, three large private contractors (25 percent) and several small companies (15 percent). The ECMEPs were originally formed inthe ten provinces to execute roadworks for the Government. Suffering from severe bureaucratic inefficiencies, they were reorganizedinto three regional companies which were to operate on a commercial basis, and eventually to be privatized. They obtained financial autonomy in2000. However, they are still incharge, without bidding, o f a significant part o fmaintenance services inthe provinces. Repeated efforts to privatize the ECMEPs have so far failed andthis remains as one o fthe main tasks o f the Roads 111. As a step thereto, there shouldbe a formal requirementthat all contractsbe subjectedto competitivetender,whichwould induce the ECMEPs to restructure andpossibly even provide them with an incentive to seek privatization. Also, a timetable for privatization should be established. c. EXPENDITURE TRENDS 165. Theroads sector is the highest-spending sector on the investment account and third highest in total, after education and health. All roads expenditures are recorded, for historical reasons, as being on the investment account, even ifthey take the form o f recurrent expenditures such as routine maintenance. Inthe 2002 Budget, for example, roads were allocated Mt 700 bn(23 percent) on the internal investment account andMt 1,476 bn(19 percent) on the external (donor-funded) investment account, totalling Mt 2,176 bn, or 20 percent o f the total investment account. This exceeds plannedinvestment account spending inhealth o f Mt 2,074 bn. Summingrecurrent and investment accounts, education was the largest planned spender at Mt4,448 bn, health the second largest at Mt 3,573 bnand roads the third at Mt 2,176 bn. The roads sector accounts for about 14 percent o f plannedspendingo fthe PARPA "priority sectors"77 and 9 percent o f total spending. Roads spending i s around 2.3 percent o f GDP, similar to that o f Malawi o f 2.4 percent in2001. 166. Data on actual donor-funded expenditures are being collected from the donors at the time o fwriting (May 2003). These will be reportedwhen they become available, and will be compared with budgeteddonor-hnded expenditures. 76Termed (Brigadas de) Us0 Intensivo de M9o de Obra (UIMOs). From 1992to 1998, the UIMOs rehabilitated 4000 kmof tertiary roads. There are over 40 brigades operative inall ten provinces. 77Priority sectors: see footnote 3, p. v. 84 167. Rehabilitation and maintenance are moving towards a better balance.78 As may be seen from a summary o fbudget allocations inTable 26, rehabilitation (including district and rural roads) used to account for as much as three times the spendingon maintenance inthe mid-1990s when the post-war rehabilitation program hadjust begun. Since then the ratio has settled at a level closer to 2 or 2% to 1. Table 26: Road sector budgets, 1996-2002 Item 1996 1997 1998 1999 2000 2001 2002 Current Mt lo9 Average Rehabilitation 590 551 456 390 332 716 870 558 Maintenance 146 310 459 405 246 332 536 348 District-Rural Roads 68 558 283 387 305 465 333 343 Institutional Costs 191 249 129 78 131 168 436 198 Total 995 1,668 1,327 1,260 1,014 1,681 2,176 1,446 Constant 2000 M t 1,233 1,942 1,536 1,414 1,014 1,570 1,861 1,510 US$ (millions) 88.1 144.5 112.0 99.3 64.6 81.6 91.2 97.3 Share Of expenditures 14.7% 17.5% 13.1% 9.8% 6.1% 11.2% Rehab+Dist-Rur. I 4,5 Maintenance 3.6 1.6 1.9 2.6 3.6 2.2 2.8 Institutional costs as 19% 15% 10% 6% 13% 10% 20% 14% % of total - 168. Maintenance expenditures have leveled out in real terms since 1996despite growing maintenance needs and a growing economy. As Table 27 shows, maintenance spending was $24 millionin 1996 and despite some fluctuations, stood at $25 millionin 2001-notwithstanding the fact that the amount o fmaintainable roads increased from 2,500 kmin 1993 to about 13,000 kmin2002. Also, the Mozambican economy grew 48 percent inreal terms between 1996 and 2001. The upshot o f the failure to expand maintenance expenditures is thatperiodic maintenance has been increasinglyneglected (whereas routine maintenance has fulfilled targets to the extent of 80 percent, as mentioned above). The best performance against target inperiodic maintenance was in 1994 when 42 percent o f what was requiredwas done, and the second best was in 1996 with 20 percent. In1998 the amount was 5 percent andin1999 12percent. 78There is virtually no new roadconstruction as the costs ofrehabilitation ofexisting roads are much lower. 85 Table 27. Expenditureon roads,by maintenance/ investment, 1995to 2001 1995 1996 1997 1998 1999 2000 2001 Billions ofMeticais Total n.a. n.a. n.a. n.a. a a a Maintenance (reported by RoadFund) 125 266 334 347 345 541 513 Investment (viz. rehabilitation) n.a. n.a. n.a. n.a. a a a As reported byRoad Fund 546 653 284 536 302 371 421 Other donor-funded expenditures n.a. n.a. n.a. n.a. a a a Theequivalent in US$ millions Total n.a. n.a. n.a. n.a. a a a Maintenance (reported by Road Fund)d 13.9 23.6 28.9 29.3 27.2 34.5 24.8 Investment (viz. rehabilitation) n.a. n.a. n.a. n.a. a a a As reportedbyRoadFund 60.5 57.8 24.6 45.2 23.8 23.6 20.4 Other donor-funded exDenditures n.a. n.a. n.a. n.a. a a a Memo: M W S $ 9,022 11,294 11,546 11,850 12,689 15,689 20,670 Source: RoadFunddata; see Herman (Jan.2003, Table 38). Note: a Donor-fundedexpendituresnot recordedby the RoadFundhave to do with rehabilitation. The datawere not yet availableat the time o f goingto press. bNoimportanceshouldbe attachedto the apparentdecline inrehabilitation spending. There is probablyaproblemwith coverageby the RoadFunddata sets. This i s beingcorrectedby obtaining datafrom MPF anddonors directly. These datawere not yet availableat the time ofgoing to press. Most maintenanceis intemally funded - see Table 28. 169. Donors contribute to all areas of roads expenditure, including maintenance. The objective o f the Road Fundis to finance recurrent expenditures, using chiefly the proceeds o f the fuel tax. Nevertheless donors have contributedinall areas, as i s shown in Table 29. Donors are responsible for nearly all the funding o f rehabilitation (91 percent), as this i s a capital item, andmost o f district andrural roads (70 percent) andperiodic maintenance (76 percent). c Table 28: Fundingsources (internal andexternal) for the roadsector budget,1996-2002 Budget Item Internal Extemal Total Rehabilitation 9% 91% 100% Routine Maintenance 98% 100% Periodic Maintenance 24% 76% 100% District-Rural Roads 30% 70% 100% InstitutionalCosts 16% 84% 100% Total 25% 75% 100% ISource: Orgamentodo Estado. 170. Budgeted and actual spending show wide variation. The figures inTable 29 shouldbe regarded with the utmost cautionbecause insome cases the annualbudgets were changed duringthe year but the changes were not published, andbecause until2000 budgets were done inreal terms, andhighintra-year inflation would result inan apparent budget executionrate inexcess o f 100percent (e.g. as muchas 164percent inthe mid- 1990s). Inaddition, the data are from different sources (MPF, Road Fund, anddonors). Withthese caveats inmind, it is clear from Table 29 that plannedand actuals are often far 86 apart. In2000 and2001,execution o f the government-fundedcomponent o fthe budget was only fairly low, at 87 percent and 72 percent respectively. Table29. Budgetedvs. actualspending,by funding source, 1995to 2001 1996 1997 1998 1999 2000 2001 Billions of Meticais Domesticallv generated (fuel tax and govt. budget) Budgeted 148 181 252 263 389 581 Actual 93 244 294 323 339 417 Actual as % of budgeted 63% 135% 117% 123% 87% 72% Externallv funded (donors) IBudgeted 847 1,487 1,075 996 625 1,100 Actual (total) a a a a a As reDortedbv the RoadFund 826 374 589 324 573 517 Othe; donor &ding a a a a a a Source: Orpmento do Estado, RoadFunddata. aDonor-fundedexpendituresnot recordedby the RoadFundhaveto do with rehabilitation. The datawere being assembledat the time of goingto press. 171. Execution rates will need to rise in order tojustifi the appealsfor further transfers to the Road Fund. Road Fundrevenues from roadtaxes andtolls was inexcess o fMt 555 billion in2001, close to the budget o f Mt 581 billion. Yet spending o fthese revenues was only Mt 417 billion, an execution rate o f 72 percent. As pointed out above, maintenance activity needs to be increased and more budget will be required, but a necessary step to get there i s that the roads authorities succeed inexecuting a significantly higher share o fthe budget than 72 percent inorder to justify larger allocations. 172. Inadequatefunding contributed to the massivefailure inperiodic maintenance. In1994the Government decidedto earmark the petroleumtax for the RoadFundinorder to provide a predictable level o f funding for routine andperiodic maintenance to support the rehabilitation efforts o f ROCS 1and2. However, only 57 percent o f the planned Government financing for the RoadFundbetween 1994 and 1998 was actually provided". This was part o fthe reason for the massive under-performance o f periodic maintenance cited above. 173. Fundingfor maintenance was inadequate becausethe real value ofthe petroleum tax fell 40 percent between 1994 and2000. The levy on regular fuel was kept at Mt 671Aitre from 1995 to 2001, resultingina decline from $0.078 to $0.036. Furthermore, the authorities failed to transfer to the Road Fundeven the portion o fthe tax that had beenpromised. As Table 30 shows, the total fuel tax in2000 was $36 million, o fwhich $22 million was transferred to the Road Fund, but this was well below the Government commitment (as o f 1998) o f $42 million, and also below the estimated maintenance requirements o f $45 million. The persistent failure o fthe authorities to provision properly for maintenance obliged the donors, inRoads 111,to foot a substantialportion o f the periodic maintenance bill, on a decreasing basis, over a 10-year period, at the end o f 79The data for the period 1996 to 1998 are presented inTable 30 onpage 110. 87 which time the authorities would be wholly responsible for it. This is an unsatisfactory situation since it flies inthe face o f the principle o f sustainability, viz. the principle that the authorities shouldnot seek investments that they are not preparedto maintain. Table30. Transmissionof RoadFundFuel-BasedRevenues Item 1996 1997 1998 1999 2000 Total Fuel Tax $22.9 $36.9 $39.3 $38.7 $35.8 Transferred to Road Fund $16.4 $19.5 $21.6 $19.6 $22.4 Percent Transferred 72% 53% 55% 51% 63% Government Commitment $34.0 $40.0 $42.0 n.a. n.a. Routine Maintenance Requirements $8.1 $8.5 $8.8 $9.2 $9.6 Periodic Maintenance Requirements $19.4 $23.5 $27.6 $31.7 $35.8 Total Est. Maintenance Requirements $27.4 $31.9 $36.4 $40.9 $45.4 Source: RoadFund. 174. Poor organization also contributed to the massivefailure inperiodic maintenance. Periodic maintenance is more difficult than routine maintenance, as it requires a more complex procurement process and design, whereas routine maintenance was readily assigned to the parastatal maintenance companies. Inaddition, periodic maintenance lacked an institutional advocate within the Direcqgo Nacional de Estradas e Pontes and later ANE, whereas other areas o f work had their champions: routine maintenance fell under the Department o fMaintenance, rural and feeder roads fell under the Directorate o fRural Roads, and rehabilitation programs enjoyed donor attention. Consequently when allocations to the Road Fundwere below budgeted amounts, periodic maintenance was cut back. Itwill be essential to reorganizethe periodicmaintenance arrangements. 175. Looking forward, the Roads I11program will succeed ifthe maintenance program i s funded, as the Government intends, by restoringthe fuel tax to its real 1997 level or committing to alternative funding sources providing the same funds. Everyeffort should be made to ensurefullfundingof maintenance. 176. Rural road efforts are evidence of the authorities 'drivefor poverty reduction. Rehabilitation efforts between 1992 and 2002 focused more on the tertiary andruralroad network (6,630 km) thanon primaryand secondary roads. The Feeder Roads Program (FRP), h d e dbythe ILOand avariety o fdonors includingASDI, DFID, andUSAID, operates inall provinces, but particularly inZambezia andNampula. Ithas been effective inpromoting labor-based techniques, thereby reducing unit costs and expanding employment. These techniques should be considered for applicationto civil works on the national network. 177. Roads expenditure allocation choices have stressedpoverty reduction but have not gonefar enough. Spending has beenreasonably equally distributedbyprovince, giving access to rural people andpeople insmall towns (whereas, for promotion purelyo f industry, or middle class interests, efforts mighthave been directed at the two main cities and their links with the exterior). As mentioned above, district andrural roads have formed a large (24 percent) part of total spending. But at the district level there i s a 88 negative correlation o fpoverty incidence and road density (viz. good-to-fair roads per unitarea)", implyingthat poor people haveto go furtherto get to a good-to-fair road. This is a result o fthe poor beingconcentrated insparsely populatedrural areas, which raises the unit costs o froad services. It mightnot be efficient to seek to compensate fully for their isolationby seeking to provide equal access to roads to all, irrespective o f relative cost. But some compensation i s surely appropriate. Poverty indices should be explicitly be taken into account and significantly weighted when decisions on rehabilitation andmaintenance are taken. A greater effort could be made with rural roads. It was stressed inthe Mozambique Agricultural Sector Memorandum (World Bank, 1997) that one ofthe bestgovernment interventionsonbehalfo fsmallholders would be to strengthen the road system andimprove district roads and even rural tracks to market locations. 178. The Government's newprogram (Roads III)will incorporate equity considerationsprominently. A positive aspect o fthe Government's road sector policy and strategy is the recognition o f the tradeoffs, inroad investment, between the competing objectives o f alleviating poverty andpromoting economic growth. Although the PARPAprioritizes roadinvestmentsinthose areas o fthe country that havethe greatest economic potential and largest populations as well as the highest levels o f poverty, the Road Sector Strategy recognizes that these objectives may not always be in accord. The Road Sector Strategy and Roads I11include a provision for allocating the investment and periodic maintenance budgets through the use o f multi-criteria analysis which incorporates both economic feasibility anda measure o f equity. This process should be strengthened by utilizingdistrict-based poverty measures inthe ex ante social impact analysis and by incorporating the expost socio-economic impact assessments into the process; andbyimproving the economic analysis o fprojects through improved applicationo f A N E ' s HighwayNetwork Management System (HNMS). D. POLICY PRIORITIESFOR THE ROADSSECTOR 179. Continued strengthening of institutional reforms is key toprogress in the roads sector. The management o fANE should become more goal-based andfocus more effectively on achieving instituional efficiencies and meeting its targets. The performance indicators that are part o f the Roads I11program should be used as a management tool to that end. 180. Roads administration needs to be autonomous. The RoadFundshould be separated from ANE, as planned, and the RoadFundshould be strengthened to better forecast fimdingneeds, managing revenues, coordinating and auditing road sector budgets and expenditures, and ensuring transparency and accountability inits operations. The simple correlation coefficient (r) i s -0.18. There is no correlation betweenpoverty incidence and the rehabilitation and maintenance effort, or road density, at the provincial level. But this means only that the provinces are so large that they incorporate districts that are vastly different, so that n o correlation would be expected. 89 The restructuredANE and RoadFundboards should include private sector representation capable o f requiringaccountability ofmanagement. 181. Fundingfor maintenance must be adequate. Ifsufficient maintenance finding cannot be assured, the size o frehabilitationprojects shouldbe reduced. The Roads I11 program, supported by the World Bank throughthe Roads and BridgesManagement and Maintenance Project (RBMMP)will be reviewed at mid-term and cut ifadequate funding for maintenance has not beenprovided. 182. Restructuring and reorganization is needed in order to ensure adequateperiodic maintenance. Periodic maintenance needs an institutional champion. Increases o f fundingalone will not be sufficient. 183. Thelevel of thepetroleum tax should be raised substantially. Raising it to the real level o f 1997 would be sufficient to ensure full funding o fthe pluri-annual maintenance program for 2003. The conclusion o fthe Poverty and Social Impact Analysis (Nicholson, 2002) was that raising the tax would have a minor impact on the welfare o f the poor. 184. In order to reduce maintenance costs, theparastatal maintenance companies (ECMEPs) should be subjected tofull competition andprivatized. All contracts should be subjected to competitive tender, which would compel the ECMEPs to restructure and possibly evenprovide them with an incentive to seek privatization. 185. The equity orientation of road rehabilitation should be improved through incorporating district-based poverty measures into the multi-criteria analysis called for in the Roads Sector Strategy. 186. Roads expenditure data need a thorough overhaul. Analysis o f actual expenditures i s complicated bythe fact that neither the Ministryo f Finance nor ANE maintain systematic data on actual externally funded expenditures on roads. The data available from the Conta Geral do Estado do not reflect the detailed budget lines inthe Budget, do not distinguish consistentlybetween external and internal financing for road sector expenditures, and are not always consistent with data providedbythe RoadFund. Road Funddata on revenues distinguishamong sources o f finds, permittingcalculation o f internally financed expenditures, but they do not report the vast majority o f externally financed expenditures. A unifiedaccounting system for all road sector expenditures should be put inplace. It i s hoped that A N E ' s new Financial Management System will solve this problem. Furthermore, reliable data are needed on unit costs o frehabilitation and maintenance so as to permit policy-relevant comparisons over time and over space. 90 CHAPTER 6. WATER SUPPLY AND SANITATION A. INTRODUCTION 187. Water supply and sanitation are o fprime importance inthe PERbecause o ftheir complementaryrole" inpreventing disease. Access to improved water supply and sanitation, accompanied by effective hygiene educationg2,canbe expected to improve bothhealth andproductivity, reduce drudgery, improve quality o f life andprovide increased opportunities for education and employment, especially for women and girls. 188. As of 1992, around ten percent ofrural dwellers hadregular access to safe water. In1997about 12percent ofruralpeoplehadsafewater accordingtothehousehold surveyg3.Although the PERo f 1992 enunciated the ambitious goal of raising access to 35 percentg4by 1995, this target was infact reached only in2001, when there were 12,490 waterpoints, o fwhich 8,098 were functioning, serving a ruralpopulationo f 12.6 million. Mozambique's coverage still lags behind Malawi (44 percent), Zambia (48 percent), Tanzania (42 percent) andUganda (46 percent). 189. Access to household piped water inurban areasg5was 29 percent in 1992. The goal cited inthe PERo f 1992 was 50 percent by 1995, but the subsequent decade saw no progress, andwith urbangrowth, coverage had fallen to 25 percent by 200286. Inthis respect Mozambique lags behind its neighbors -Madagascar (35 percent), Uganda (35- 40 percent)87. Adding inpeople with access to cleanwater via standpipes and vendors (legal kiosks and illegal resellers), coverage is estimated to be some 70 percent inurban areasg8,rising to some 90 percent inMaputo City. Onthe other hand, World "e.g.Esrey's(1996) surveyofAfricanandothercountriesfoundthatbenefitsfromimprovedwater occurred only when sanitation was improved. See also the summary by Wagstaff et al. (2002). 82For example, Jalan and Ravallion (2003) study inIndia shows that the health gains o fpipedwater largely by-pass children inpoor families, particularly when the mother is poorly educated. See also Varley and Tarvid et al. (1998). 83MinistryofPlanning andFinance et al. (1998), Table 2.47. Access to safewater inthe Poverty Assessment was defined as access to pipedwater (inside ow outside dwelling) or a public tap. 84Note that the 35% coverage figure includes people with access to Small PipedSystems (towns), which account for four percentage points. 85Urbanareas are here defined as the 21 largest cities, whose total populationis 4.3 million. (The population o f Mozambique i s 16.9 million.) Smaller cities/towns fall inthe category o f Small Piped Systems which for convenience are groupedunder "rural" water access. 86In1997 about 27% ofurbanpeoplehadpipedwater either in-house or ouside the house (Ministry of Planning andFinance et al., 1998, table 2.42), but the true figure might be lower because some o fthe outside-of-house access may reflect purchases from neighbors. "WorldBankstaffestimates. In1997about 54% ofurbanpeople hadsafe water accordingto the householdsurvey (Ministry of Planning andFinance et al., 1998, table 2.47). Also in 1997, 92% o f dwellers inMaputo City had access to 91 Development Indicators reports that access to cleanwater (including standpipes and vendors) i s 85 percent, which i s still behindMalawi (95 percent) and Zambia (88 percent) but higher than Tanzania (80 percent). 190. Total urbanand peri-urban sanitation coverage i s probably over 35 percent. This represents a considerable improvement over 1993 when the figure was closer to 10 percent. 191. Water resource management i s a crucially important function for the Government to play, particularly inthe light o fMozambique's vulnerability to floods and droughts. For reasons o f cost, it was decided not to include water resources a major focus ofthis PER,because a thorough examination ofMozambique's water resource management issues would have requireda large separate study. Nevertheless the key aspects are cited this report and the importance o f appropriate investments andbudgetary allocations i s emphasized. 192. This chapter starts byprovidingthe context behindMozambique's mix o f (modest) progress andregress. It then discusses the allocation and effectiveness o fpublic expenditure inthe water supply and sanitation sector. It thenproposes a shift ofpriorities inseveral keyrespects. B. KEY SECTORISSUES 193. The most urgent sector issue is the inadequacy of rural and urban water supplies, interms ofbothcoverage andquality ofservice, despitethe decade ofeconomic growth since the end o f the war in 1992. The lack o f regularsupplies o f safe water is particularly acute inurbanareas, on account o f the health risks involved. 194. Poorphysical sustainability of rural waterpoints. Ofthe country's 12,490 rural waterpoints (mainly boreholes with handpumps), some 35 percent are broken, though in some parts the percentage is higher.89This i s due to the lack o f spare parts and the lack o f human capacity for maintenance, inturn stemming from a lack o f community involvement andcommitment. At least 10donors are involved infundingrural waterpoint installation, andvarying degrees o f attention have been givento sustainability, community involvement and "ownership". Since the authorities' adoption o f the Demand Driven Approach in 1999, requiringa payment by the community o fbetween 2 percent and 5 percent o fthe installation cost, it i s expected that the breakdownrate will fall. In addition, training inplanningand monitoring shouldbe given to the provincial and district directorates responsible for water supply, and community training programs inthe operation and management o fwaterpoints should be extended. 195. Rural waterpoint installation is expensive. Sinking a borehole costs $10,000 in Mozambique, versus $5-7000 inMalawi, Tanzania, Zambia and Ghana. Some o fthe piped water (inhouse, outside house, or public tap), whereas the average for Beira, Nampula and Matola was 61%, and other cities' access was 41% (ibid.,Table 4.15, p. 251). 89InCab0 Delgadoprovince, where reliable data does notexist, at least 60% ofruralwaterpoints were not working as o f 2002 (African Development Bank, 2002, iii). 92 differential has been attributedto reasons beyondthe control o fthe authorities, such as aquifer conditions, the small size o fthe Mozambique market which limits the gains from economies o f scale, poor roads, large distances, inadequate infrastructure and services, and a highlevel o fbusiness risk associated with highinterest rates. However, since several countries inthe region also face these problems. Hence a deeper investigationo f this very specific issue may be called for inorder to see whether there are additional factors that explain Mozambique's higher cost structure, and to identify opportunities for reducingthe cost. 196. Slow implementation in urbanpiped water access. Unlikeruralwater supply, urbanpipedwater supply requires major policy, regulatory and procurement interventions. Urban pipedwater access registered little progress inthe decade after the 1992 peace agreement because the National Water Policy was formulated only in 1995. The policy o f delegated management for urbanwater supplywas innovative and one o f the most forward looking inthe region. The implementationof the policybegan inlate 1998 when FIPAG, a newly created asset holding body, signed a lease contract and a management contract with a private sector consortium, Aguas de Moqambique (AdM), to manage the water supply o fthe five largest cities (Maputo, Beira, Nampula, Quelimane and Pemba). These five cities make up 70 percent o f the total urbanpopulation. The delegated management process has been supported by the National Water Development I1project (NWDP11). Five donors are involved inthe sub-sector, the African Development Bank, the World Bank, the EuropeanUnion, Netherlands andthe French Agency for Development. Since then, with the setback o f the floods o f 2000, difficulties were experienced with the contract, necessitating a change o foperators, explaining why little progress has beenmade to date. It is hopedthat as the NWDP I1picks up, coverage o fhouseholdconnections will rise, inthe five major cities, from 33 percent in2002 to 40 percent by 2005, though it is doubtful that the installation rate can be increased as quickly as thisgo.Nevertheless, progress hadbeen made inpolicy formulation and institutional development, including with the regulatory function. The Conselho de Regulaqgo do Abastecimento de Agua (CRA, Regulatory Board for Water Supply) has been established as an independent regulatorybody responsible for regulatingprivate operator water supply and setting tariffs for municipalwater supply entities. 197. Access to water through standpipes in urban areas is one of several ways that water utilities willprovide water to unserved customers. The provision o fthe service will depend on affordability, water supply availability andthe costs o fprovidingservice andwhat users are willing to pay. Rolling out o findividual connections will continue as densification increases andis the most cost effective way to provide water services. However, 11-29 percent o furban dwellers rely on standposts for lack o f money for an individual connection or because o f land tenure issues. Since standpost services are cross subsidized bythe utility from other consumers, a strongutility is a key to providingthese services on a sustainable basis. Standpost service providedbythe public utilities has beenpoor andusually a reflection o fthe financially viability o fthe utility. Data shows Extrapolating this rate of expansion, coverage of household connections inthe five major cities would rise to 76 percent by 2015. 93 that only 19percent o f standposts are functional. The cost o fwater purchased at publicly operated standposts is low at about $0.15/m3, butpressure i s often low and water stoppages common. A successful standpost or kiosk model i s one that sells water at a price that reflects the market andprovides a sustainable service." An example is that o f Angoche, which sells water to private individual managers, who inturnon-sell at prices determined by competition. The price i s muchhigher ($0.77), but quality i s better and people havebeen willing to pay for a more reliable service. At least 25 percent o furban dwellers rely on resale o fwater from those with connections. Officially, resale by individuals with connections i s legally prohibited, but this law should be repealed, inpart because the law i s not enforceable, but primarily because allowing the sale ofwater to neighbors encourages more individual connections to the system, increasing competition, and bringingdown the cost o f water for those without their own connection. 198. TheSmall Piped Systems (SPS) are a major development challenge because o f the large number involved (270), their poor state o frepair (38 percent fully operational), andtheir generally poor financial situation. The number o fdomestic connections inthe SPSs varies from 5 to 170, and inaddition most SPSs have three or four standpipes. The number o fpeople served by the SPSsvaries from 2,500 to 19,400. Dueto a lack o f economies o f scale, unitcosts o f service are high,but currently cost recovery i s low. The complexity o f SPSs i s also a challenge for operationby community organizations. Currently the only donor involved with SPSs i s the World Bank. Models o f financially sustainable SPSs are presentlybeingdeveloped. It maybe that the Government will need, inthe meantime, to shoulder some o fthe burdeno f the capital and rehabilitation costs o f SPSs, while bringingthe standard o f SPS service up to acceptable levels, rationalizing and strengtheningmanagement, andimproving cost recovery. 199. ThePARPA goals and the Millennium Development Goals (MDGs)in rural water are achievable. The PARPA goal o f40 percent rural coverage by 2005 i s attainable on present projections, from a coverage level of 35 percent in2002 and on the assumption o f continuation o f the rate o f 1,300 new connections per year, even without reducing the breakdown rate. To achieve the MDGo f 67 percent coverage by 2015 can be achieved with the plannedrate o finstallation o f 1,400 per year.92193Hencethe MDGis likelyto be attained. 91Inaddition, amodelofstandpipe management involvingcommunities, localauthorities, NGOsandthe water utilityi s being tested. A contract for water supply through standpipes will be set upbetweenthe water utilityand the vendors that will pay for water based on the approved tariff and resell water at a price agreed with the communities. The O&M costs o fthe standpipe will be paidby the standpipe users. 92Assumptions o fthe model: 35% breakdownrate; total population growth o f2.3%; urban population growth of 5%; implicly, growth o fthe ruralpopulation o f 1.1%. This assumes that the problemo f low execution of already allocated government funds (see Table 33) canbe overcome. 93Furtherexamination ofthe definitions ofcoverage is called for. InMozambique the coverage estimate uses 500 people per water point. More conventional estimates inother countries are between 250 and 280 persons per water point. The water directorate had a study done o fusage in2003, and found that some waterpoints were usedby 800 to 1200 people per time period. It was not possible to review this paper thoroughly intime for the issuance o f this PER. The matter will be further examinedby a national study to 94 200. ThePARPA and MDGgoals in urban water may not be achievable. The PARPA goal o f 50 percent coverage o f household ipedwater by 2005 may not be achievable on the present projections o f the NWDP IL94,95The MDGrefers topiped household water excluding standpipe^.^^ The target is to reduce the unserved population by halfby 2015, viz. to raise coverage from 25 to 62.5 percent. This i s possible only ifthe NWDP I1 expands service continually at the ambitious rate planned for the period 2002-200597, and the projects fundedby the ADB and others inthe remaining cities expand at similar rates. . As matters stand at present it is unclear whether the MDGwill be attained. 201. Otherfactors influencing water supply. Other factors involved inthe prospects for water supply are the limitedpublic sector institutional capacity, coupledwith the inadequate water sector informationbase and data collection and monitoring system. As mentioned above, there i s an urgent need for reinforcing the training programs for provincial anddistrict-level ministerial staff, as well as training inmanagement for local communities. 202. Sanitation has received less attention than water supply. There i s a need for improved sanitation, especially inurban andperi-urban areas, as is demonstrated by frequent outbreaks o f faecal-borne diseases. To date, public investment for sanitationhas been less than 10percent o f that for water supply. The under-funding i s partly due to the lack o f a strategy for the sub-sector, as well as to unclear responsibilities o f the various actors involved. According to the National Water Policy, DNA is responsible for sanitation, but it has only recently formed a department for sanitation. InMaputo and Beira, the only cities with sewerage systems, the municipalities are responsible for operating the sewerage systems, not FPAGand the private sector operator. At a minimum, coordinated planningis needed for water supply,hygienepromotionand sanitation, as noted inthe 1999-2003 Rural andPeri-Urban Sanitation Strategy. At the Johannesburg 2002 Summit it was agreed to adopt the objective o freducing the proportion o fthe population without access to sanitationbyhalfby 2015. 203. Decisions need to be taken on the appropriate mix of sanitation instruments. The forthcoming Strategic SanitationPlanning Study for seven major cities will indicate the most appropriate mix o f sanitation `technologies' and the likely costs involved. Maputo andBeira are the only cities withpiped sewerage systems andthese serve only a small proportion o f the urbanpopulation inthe more developed city centers. Other residents be undertaken by the Centro EstratCgico e de Desenvolvimento do Sector de Aguas (CEDESA) and the Water Sanitation Programo f the World Bank. 94 The target o f 50 percent may, however, be feasible by 2007, given the expansion o f the network, new construction and construction o f standpipes. 95 The PARPA goal i s stated as raising urban coverage from 44 percent in2001to 50 percent by 2005 (see the matrixinthe operational annex). Note that the definition o f coverage here includes not only piped water supply but also standpipe usage. 96 Note that the MDGdoes not distinguish between rural and urban water supply. Yet for convenience here it is split mallurban, enabling amore precise focus which is usefulbecause expectations for expansionin the two sub-sectors are radically different. 97 See footnote 90. 95 rely on septic tanks and improved latrines. A relatively successful Low-Cost Sanitation Program (LCSP) has beeninoperation since 1980, focusing on improved latrines. Some 33% o fthe target population o f 3.61 million inthe urban centers included inthe LCSP currently have access to improved latrines at a cost o f about $4 per capita. The ADB's Four District Centers (South) studyproposeda mix o f septic tanks intown centers and improved latrines inthe peripheral areas, combined with urbandrainage, at a cost o f $15 per head. An additional policy suggestion is to allow house connections to the piped water system only for households that are connected to the sewerage systemor have installed a septic tank (Others could have yard taps). Such a policy would limit the amount o fwastewater generated andreduce adverse health impacts that can result from increasingwater supply without corresponding wastewater removal. 204. Expensiveseweragesystems are unlikely to be the solutionfor theforeseeable future. Piped sewerage is muchmore expensive than LCSP, risingto over $100 per capita. There are numerous examples inAfrican countries o f expensive sewerage systems that standidle due to lack o f funds for operation andmaintenance. Insome cases, sewerage is collected, but not properly treated before beingdischarged to waterways. While the cost o f collecting andtreating sewerage is higher than the cost o f supplying water, customers are typically muchless willing to pay connection fees and monthly charges. Indeveloped countries, governments often subsidize the cost o f investment and cross-subsidize runningcosts from water tariffs or from tax revenues, arguingthat the environmental benefitsjustify the subsidies. InMozambique, the Government i s unlikely to be able to afford such subsidies for the foreseeable future. The strategy shouldthenbe the continuation ofthe LCSP for peripheral areas, combined with the requirement that households with in-house connections inthe central areas o f cities connect to the existingsystem or, where that option is not available, install a septic tank at their own expense. Regulation o f septic tank haulers will also be necessary to ensure proper disposal. Inthis regard, DNA has adopted a strategy o f decentralization o f responsibility for sanitation to the municipal level, and o f introducing a sanitation tariff. 205. In rural communities thefocus should be on hygiene education, sanitation promotion andfacilitation of the entry of small scaleprivate contractors to meet the demand for latrine construction. The Government has developed materials for such education andpromotion inrural communities and has trained staff insome districts and provinces intheir use. Education andpromotion about hygieneand sanitation should become an integralpart o f community preparation inall rural water supply programs. Furtherinvestments will beneededto achieve these objectives, including investmentsin research for appropriate technologies. 206. Water resource management issues need urgent attention and adequate budgetary allocutions. Concerning strategy development, the Ministryo fPublic Works and Housing is formulating an IntegratedWater Resource Management (IWRM) Strategy with NWDP Ifunding. Mozambique needs to be well equippedto negotiate agreements concerningits river basins with adjoining countries. Provision for the costs o fnegotiation (e.g. over the Incomati-Maputo basin) should be included infuture DNA budgets, and consideration should be givento obtaining contributions to the costs from the donor community. Finally, inview o f the severe flooding experienced in2000 and 2001, there i s an urgentneed for the government to develop a practical flood forecasting, warning 96 and management policy, strategy and program. A flood risk analysis study i s due to be carried out during 2003. e. TRENDSINPUBLICEXPENDITUREINWATER SUPPLY AND SANITATION 207. Water supply expenditures are mostly donor-funded. Inthe last "normal" (viz. pre-flood) year for which data were available, 1999, total expenditures inthe water sector, including bothwater resource management andwater supply, were some $15 million, o f which the Government funded $4.9 million anddonors the remainder (Table 31). There was a large increase indonor-funded expenditure (though not o f Government- funded expenditure) in2000 and 2001 on account o f the floods: in2001 total expenditure on water rose to $24.5 million, o fwhich donors contributed $19.7 million. Nearly 90 percent o f water sector expenditures inMozambique go to water supply and sanitation, andthe remainder to water resourcemanagement. 208. Thedistinction between recurrent and capital account is not meaningful in the water and sanitation sector. The recurrent account i s devoted almost entirely to salaries (98 percent). Salaries o fproject staff are paid out o f the investment budget, as are expenses for goods and services. 97 Table31. Water and sanitationsector expenditures(actual) 1999 2000 2001 Mtbil. US$mil. Mtbil. US$ mil. Mtbil. US$ mil. 3DP (IMF2002 data) 4,090 3,750 3,610 rota1Water and SanitationSector Expenditure (includingwater resources) rota1(Govt. + donor-funded) n.a. 15.3 n.a. 28.3 n.a. 24.5 Government-funded expenditure 62.1 4.88 68.5 4.37 100.0 4.84 Recurrent account 2.4 0.19 2.6 0.17 3.8 0.19 Capital account 59.7 4.70 65.9 4.20 96.2 4.65 "Recurrent" costs 24.2 1.91 27.8 1.77 39.6 1.91 Salaries 9.0 0.71 11.6 0.74 16.9 0.82 Materials 15.2 1.20 16.2 1.03 22.7 1.10 "Capital" costs 35.5 2.80 38.2 2.43 56.6 2.73 Donor-funded expenditure n.a. 10.4 n.a. 23.9 n.a. 19.7 Wem. item: Total exp. as % o f GDP 0.37 0.75 0.67 Water Supply and SanitationSub-sector Expenditure rota1(Govt.+ donor-funded) n.a. 13.7 n.a. 27.3 n.a. 20.3 Government-funded expenditures 57.8 4.56 63.9 4.1 92.6 4.5 Recurrent account 2.2 0.18 2.5 0.16 3.6 0.17 Capital account 55.6 4.38 61.4 3.91 89.0 4.30 "Recurrent" costs 20.6 1.62 23.5 1.50 33.0 1.59 Salaries 8.1 0.64 10.1 0. 65 13.6 0.66 Materials 12.5 0.99 13.4 0.86 19.4 0.94 "Capital" costs 35.0 2.76 37.8 2.41 56.0 2.71 Urban water supply & sanit. 13.3 1,05 18.9 1.20 25.8 1.24 Rural water supply & sanit. 21.7 1.71 18.9 1.21 30.3 1.46 Donor-funded expenditures n.a. 9.13 n.a. 23.2 n.a. 15.8 Wem. item: Total exDend. as % of GDP 0.33 0.72 0.56 Sources: Conta Geral do Estado, Ministryo f Finance, and DNA. Votes: (1) 98% o fthe recurrent budget goes o n salaries. Hence no breakdowni s given. (2) Exchange rate mumptions: $1 = Mt 12,691 (1999), Mt 15,689 (2000), 20,707 (2001). (3) UWSS = urbanwater supply and sanitation (WSS) and RWSS =rural WSS. (4) With the data available it i s not possible to separate donor :xpenditure into U W S S expenditure and RWSS expenditure. 209. Mozambique's expenditure levels are similar to those of comparable countries. Mozambique's proportionof public hnding allocated to water supply and sanitation i s not dissimilar to that of comparable African countries, as is seen inTable 32. Government-funded spending inMozambiquei s 2.4 percent o fbudget, close to that o f Tanzania (2.5 percent) and Zambia (2.6 percent). Table 32. Water supply and sanitationspendingin relationto the budget Percentage o f Government budget Mozambique (1999-01) 2.4 Tanzania 2.5 Uganda 5.0 Zambia 2.6 98 210. Budget execution in the water supply and sanitation sector is low. Table 33 shows that in2001 the National Directorate for Water (DNA) succeeded inspending only 54 percent o f the investment budget allocated to it. The average for the period 1999-2001 was 63 percent. Year Total spent (Mt Equivalent inUS$ % ofGOM billion) million Investment Budget 1999 52,936 4.19 77.2 2000 57,699 3.73 56.4 2001 84,607 4.13 54.1 The under-spending is due partly to late arrival o fDNA's first duodkcimo allocation (as much as 8 months insome years) andpartly to the complexity o f the bureaucratic process involved inthe replenishment system, as noted inan earlier chapter. 211. The rate o f execution indonor-fundedprojects is not high. Actual donor-funded water sector expenditures in 1999 to 2001 as percentages o fthe pledged amounts ineach year averaged 70 percent. One o f the largest projects, NWDP I, i s particularly behind schedule. Among the causes i s the slow release o f the GOM counterpart funds from the InvestmentBudget, as well as the concentrationo f institutional effort on the complex procurement and legislativeprocess for the privatization of the management o f the five largest cities.98 212. Budgeting and goal-setting need to be improved. Budgetproposals by DNA for future years are unrealistically optimistic. Projections submitted in2002 for FY 2003 anticipate an increase inspending from the $24 million actually spent in2001 to no less than$88 million. Ofthis, the Government-funded component rises from $4.5 millionto $8.7m, which i s vastly inexcess o fwhat could realistically be executed. The donor- fundedcomponent o fthe projections envisages arise from previous levels o f $10m to $23m to an improbable $79.8m. Although increases can be anticipatedin2003 and 2004, as the NWDP IandI1programs take hold, the increases suggested here are not feasible. Preparation o f a new, realistic program for 2003-06 is called for. 213. The quality of urban water service ispoor, though not unlikethat o f countries at a similar level o f development. While the costs o f operation and maintenance, at $0.10- 0.15/m3, are within the rangeo f comparable countries, the continuity o f supply (11 hours/day as o f2002) i s below the average for developing countries (20 hourdday). Furthermore, unaccounted-for-water percentages are high(54 % for Maputo andBeira) though similar to those incomparable African countries. Staffing ratios are high (24/1000 connections, vs. 17/1000 for East and Central Africa and 19/1000 for South Asia). Collection rations inMaputo and Beira are 76 percent but average about 50 percent inthe smaller cities. Incontrast, inTete, where the management capacity o fthe 98Another factor i s the degree o f management decentralization granted to their Mozambique offices by the donors. The July 2002 NWDP IBi-Annual Report quotes the slow approval process o f boththe Government and the World Bank as a cause o f the NWDP Idelays. 99 local water authority has beenstrengthened, the collection rate is 90 percent. The overall picture is one o f low expenditure efficiency. Thus far, it appears that the introductiono f private sector management inthe five major cities three years ago has not yet solved the long-standing problems o f service delivery. Inpart this i s the result o fpoor infrastructure coupledwith poor accountability and an absence o f appropriate incentives. (Inaddition, at least ayearwas lost inimplementationofprivatemanagementbecauseof contract difficulties at the start.) Efficiency improvements are a major objective o f the NWDP II'fivecities program and other urbanwater supplyprojects. 214. Connections coverage is afunction of thepoverty level. In 1997, according to the household survey, 10 percent o f urbanhouseholds had in-house piped water (Table 34, line 1); 1.9 percent o f the ultra-poor, 4.2 percent o f the poor and21percent o f the non-poor hadin-housepipedwater. Inaddition, some households hadpipedwater from sources outside the house (Table 34, line 2) but the data are ambiguous because some o f these may include purchases oftap water from neighbors. But even addinginout-of- house pipedwater, the contrast i s marked: the ultra-poor at 16 percent, the poor at 19 percent andthe non-poor at 42 percent. Table34. Water sources of urbanpeople, 1997 (percent) Source of water Ultra-poor Poor Non-poor All 1. Piped (inhouse) 1.9 4.2 21 10 2. Piped (outside ofhouse) 14 15 21 17 3. Piped (public tap) 28 28 23 26 4. Own well 11 13 9.0 11 5. Public well 27 24 11 19 6. River or lake 8.3 5.3 1.5 3.9 7. Other sources 10 11 13 12 Source: Ministryof Planning andFinance et al., 1998, Table 2.42. Inturnusing data of the Household Survey of 1997. Note: Some households have more than one water source, so that the totals sumto more than 100. 215. Full cost recovery isfeasible, over time, in urban areas. The Government's policy is to phase infull cost recovery tariffs inurbanareas. This will help ensure sustainability o furbannetwork water supply systems. The full cost o fwater inMaputo i s $0.60/m3. The average charge for household connections inMaputo in2001 was $0.26/m3, so that full cost recovery would imply a real increase o f 131percent. Other cities' charges are mostly lower but fall inthe range o f $0.15-0.37/m3. As prices rise, it i s likely that consumption will drop so the key to expand individual connections to encourage an increased volume o f sales. The planned series o ftariffincreases will pay for O&M by 2005 and for all costs by 2008. 216. Thepercentage of income paidfor water is key to understanding expected sustainability. It i s typical that urbanconsumers with connections pay significantly less per cubic meter than those who rely on private vendors or private standpipes, as emerges from Table 35. Consumers o f water from venders, stand posts, etc. use considerably less water than those on individualconnections though the share oftheir income going to water may be the same. The regional average is 4 to 6 percent o f income. The Beneficiary Survey inthe major cities in2001 established that the majority o f 100 respondents would be willing to pay for a household or a yard connection giventhat water from other sources i s more expensive. Table 35. Prices paid for water in urban areas, by location and source U S cents per m3 Household connections: Average charge, Maputo, 2001 (Finney 2003, p.38, from FIPAG data) 0.26 Average charge, Beira, 2001 (Beneficiary Assessment Summary) 0.30 Standpipes: Maputo, 2001 (Beneficiary Assessment Summary, Sustem) 0.53 Beira, 2001 (Beneficiary Assessment Summary, Sustem) 0.53 3 Purchasedfiom neighborshendors 0.53 The norm applied is that a household should generally not pay more than 4 percent o ftwo minimumsalaries for water. Currently this would implyapayment o fjust under $3 per monthper connection, which i s well above the current minimumtariff charge o f $1.80 per month. To cover the fillsupplycost, however, the minimumcharge would need to be raised to just over $4 permonth or 5.5 percent o ftwo minimumsalaries. However, a survey in2001 showed that 59 percent o fhouseholds inBeira andMaputo were paying more than the 4 percent o ftwo minimumsalaries for water, because o f the highprice o f unofficial water purchases. This suggests that the limit o f 4 percent o f two minimum wages is too conservative. It appears that fillcost recovery i s attainable andwould be socially acceptable. Inthe case o f the five biggest cities, the present planis that tariffs will be raised steadily untilfull costing is attained in2008. D. POLICY PRIORITIES 217. The previous discussion leads to a thoroughgoing restructuring of the water supply sector. A key requirement i sfull-cost pricingfor urbanpiped water, coupled with a major improvement in thepresent low service standards which should arise from improved management inthe context o f the contracting-out procedures presently under way. 218. Urbanstandposts will continue to bepart of the landscapefor theforeseeable future, given the capital cost o finstallation o fhouseholdconnections for resource stretched utilities. Serious consideration should be given to licensing privately operated kiosks as i s common inmany African cities because ofthe reliably o f such services andimproved access. Additionally, community management o f standpipes should be tested (see footnote 91). Removal of theprohibition on neighborhood vending would also promote competition and also increase access as the pipednetwork expands; currently the prohibition i s not being enforced inany case. 219. Theauthorities shouldforge ahead with the Demand-responsive Approach to rural water so as to reduce the 35 percent waterpoint breakdownrate. 220. Theprovincial allocation of rural water supply efforts should take into account poverty and assessed needs. The provinces that will benefit most from the 2002-03 rural 101 water supply (RWS) programme are Inhambane, Zambezia and, to a lesser degree, the three other northernprovinces. This is appropriate, since these are generally the poorest. However, Nampula, despite being densely populated andhaving the lowest RWS coverage (17 percent), is programmed to receive only 317 new waterpoints in2002/03, as compared with 671 inZambkzia (whose coverage i s 23 percent), 444 inInhambane (whose coverage i s 29 percent), and 235-240 each inthe less populous Niassa and Cab0 Delgado. Poverty reduction would bebetter servedby allocating more to Nampula. The new AfDB project i s expected to be based inNiassa and Nampula, leading to significant increases ininvestment inthese provinces. 221. Water sector statistics need drastic improvement. These should include an improved water sector information base, including the status andperformance o furban water supply systems including the SPSs, and a regular rural waterpoint inventory system. FIPAGhas already startedthe process o f establishing a data base for the five biggest cities. Inaddition there i s a need for regular information about donor disbursements, and regular information about the outputs o f donor-funded programs. 222. More attention needs to be given to urban sanitation. The present approach by DNA,based onthe Low-Cost SanitationProgramintheperi-urban andurbanareasand healthpromotion and other `soft' measures inboth urbanandrural areas, i s sound and should be continued. The on-going Strategic Sanitation Plans Study is expected to provide recommendations for improving sanitation inurbanareas through a combination o f affordable options. At a minimum, institutional arrangements need to be clarified and more coordinatedplanningneeds to be done so that improvements to sanitation and hygiene promotion are carried out inconjunction with water supply improvement. 102 ANNEX I. ACTION PLAN, AND MILLENNIUMDEVELOPMENT GOALS 223. This suggestedactionplanis a summaryofwhat is already containedinthe sections "Policy Priorities" ineach chapter of the maintext. It introduces no new informationbeyond what is contained inthe maintext. Itmay serve as aconvenient checklist to monitorprogress. Table 36. Action plan Area/Objective ReformMeasures Timetable Fiscal manage- Disseminatethis PER(Public Expenditure Review) inGovernment and 2003 4 3 ment: increase civil society with workshops transparency Integrate public expenditure review ina systematic way into the budget cycle and the allocation decision-making process: Identifyanarrow set of targets for specific monitoringinthe 2003 PARPA process Strengthen the statisticalcapacityof staff inthe PARPA priority 2003-2005 ministries so that there are real-time data available on spending and on targets Develop systems for reportingsystematically and regularly on 2003-2004 achievement o f output and outcome objectives, and on all funds including donor-funded expenditures Clean-up o f Governmentaccounts: complete inventory o f Government accounts at centralbanks and 2003 Q1 commercial banks close all accounts not linkedwith the 2003 financial year 2003 4 2 Address the problem o f chronicunder-spendingby increasing the amount Continuous o f training inaccounting and inthe budgetary process Improve internal auditingby substantial new hiring o f qualified staff 2003-2005 (originally planned for 2002) Introduce the integrated financial management information system (SISTAFE)by: rolling out to Education and Finance ministries 2003 rolling out to the remaining ministries 2004 Cross-cutting areas Civilservice reform: follow the Public Sector Reformprogram, withthe sequence (i) 2003-2006 restructuring, (ii)linking pay with performance, (iii) reviewing pay scales [note: this will be a combination of raisingpay and moderatingpay, depending on the sector and occupation] investigate the alleged 20 percent "ghost employees" inprimary education, and also investigate other sectors for HIV/AIDS: do the planned "sector expenditure review" study o f the 2003 42-43 implications o f HIV/AIDS for Government expenditures Education Raise completion rates inlower primary (EP1) from 30% (2001) to 60% (2008) by: curriculumreform includingnear-automatic promotionwithin 2004 cycles reducing school fees at primarylevel 2004 103 increasingrecruitment o fwomen teachers, especially inprovinces Continuous where there are few women teachers, untilbalanced(50%) Eliminate "ghost teachers" by: investigating the 20% "ghost teachers" inlower primary 2003 eliminating "ghost teachers" from the payroll 2004 Merge the EP1 and EP2 cycles into a single primary educationcycle 2005 Increase funding for CFPP teacher training and revise curriculumto one- 2005 year basic training to increase annual output o fprimary teachers Introduce ceiling for cost o f classroom construction (US$lO,OOO, including 2004 servicing and furnishing) and decentralize school construction programto local authorities Health Consolidate the strategy andpolicy documents ina single planning system 2003-2004 withprioritizationlinkedto resource allocation. The user fee system: 2003 Thoroughly reviewthe system, so as to reduce over-charging and grant exemptions consistently On the basis o fthe review, introduce reforms which may include: 2004 posting readable informationabout fees and exemptions at all clinics andhealthposts; improving remuneration levels inthe context o fthe civil service reformprogram, linkingperformance with compensation; conducting regular beneficiary assessments; introducing more unannounced inspections Reflect all user fees and other funds on budget 2003 Roads Separate Road Fundfrom Administraqlo Nacional das Estradas 2003 Restore fuel tax to its real 1997 level 2003 Provide full funding for routine andperiodic maintenance 2003-2005 Create a suitable institutional "home" for periodic maintenance Start in2003 Subject all to competitive biddingall routine andperiodic mainenance Start in2003 contracts for parastatal maintenance companies Privatize the parastatal maintenance companies (ECMEPs), raising the share 2003-2008 o fprivately executed maintenance from 40% (2002) to -% by-[target to be set by authorities] Reorganize the periodic maintenance arrangements to raise maintenance 2003-2005 from 434 krdyr (1995-2001) to -Wyr by -[target to be set by authorities] Water supply Urban water supply: increase rate o f household connectioninstallation from 2,50O/yr 2003-2015 (1999-2001) to 21,8OO/yr (NWDP I1plan for 2002-2005) 0 apply full-cost tariffs by 2008 2003 - 2008 improve urban standpipe service quality, by adopting the private 2003-2005 standpipe operator approach scrap rules against resale o fwater 2003 Rural water supply: raise waterpoint development rate from 900-1300/yr (1999-2002) to Start in2003 1400/yr 0 reduce the % o f non-functioning waterpoints (currently 35%) and continue application o f the Demand DrivenApproach 104 Table 37. Mozambique'sprogresstowards the MillenniumDevelopmentGoals99 MillenniumDevelopmentGoal Will the goal Comments be met? Extremepoverty: halve the proportiono f unknown Results o f the 2ndhousehold survey - people living inextreme poverty between 1990- expected inlate 2003 -will will enable an 2015, vu. from about 69% to 33.5% assessmento f the prospects HIV/AIDS: Halt and reverse the spreadof unknown Impact o f the Government's new strategy, HIV/AIDS by 2015 backed up by a the WB-financed HIV/AIDS ResponseProject, could be examinedin2004 which would permit an assessment Hunger: Halve the proportion o f people who unknown Results o fthe 2"d household survey - suffer from hunger by 2015 expected inlate 2003 - will will enable an assessmentofthe prospects Access to safe water:Halve the proportion o f Rural: yes See chapter on water supply people without safe drinkingwater by 2015 Urban: (vu. rural: raise access from 35% to 67%, uncertain urban: from 25% to 62.5%) Education:Universal completion o f primary Unlikely Unlikely to be met despite EducationFor All . education by 2015 due to constraints on recurrent expenditure Gender equality: Eliminate gender disparity in: primary educationby2005 Yes Access: Aclueved w.r.t. primary by 2002. .. (Completion: likelydate unknown; depends on efforts to recruit female teachers.) secondary educationby 2005 unknown Access: likely date unknown. (Completion: likely date unknown.) all levels o f educationby 2015 unknown Access at tertiary: likely date unknown. (Completion: likely date unknown). Child mortality: reduce under-5 mortality by Unlikely Optimistically projecting recent 2/3 by 2015, vu. from 200/1000 to 67/1000 improvements, the rate is 180by 2015. Maternal health: reduce maternal mortality by Unlikely Ministrylacks ananalysis ofthe causes of % b y 2015, vu. from 1083/105live births to matemal mortality, or a specific plan for 271/105 reducing it. Optimisticallyprojecting recent improvements, the rate is 644 by 2015. Malaria:Have haltedby 2015, andbegunto Unlikely No evidence that the actions taken hitherto reverse, the incidence of malaria have reduced incidence Environment: Integrate the principles o f Potentially n.a. sustainable development into country policies andprograms and reduce the loss o f environmental resources 99For general information about the MDGs, see see http://www.undp.org/mdg/countryreports.html. 105 ANNEX 11. DATA Table 38. Functionalclassificationof expenditure(percentage of GDP) 1998 1999 2000 2001 2002 -actual actual actual actual est. - - - romans indicate possible exclusion of foreign financing - - Bolded italics indicate the inclusion of all foreign financing - - Total expenditures 21.0 23.8 27.5 28.3 28.2 GeneralAhnistration 3.0 4.1 3.9 2.2 2.2 Education a 2.5 2.9 5.0 4.6 3.1 Health 2.4 2.8 3.5 3.8 3.7 Agriculture 0.4 0.4 3.2 0.3 0.4 Roads 1.9 1.2 1.6 1.3 Water 0.4 0.8 0.7 Other including residual 10.7 12.0 9.5 15.4 18.8 Total recurrent expenditures 11.2 12.2 13.8 14.2 14.4 GeneralAdministration 2.3 2.6 2.2 1.8 1.8 Education a 2.0 2.5 3.2 3.0 2.8 Health 2.1 2.4 2.8 3.5 3.5 Agriculture 0.2 0.3 0.3 0.2 0.2 Roads 0.7 0.7 I.0 0.7 1.4 Water 0.I 0.I 0.I Other including residual 3.9 3.7 4.2 5.0 6.1 Total investment expenditures 9.8 11.6 13.7 14.1 13.7 GeneralAdministration 0.7 1.5 1.7 0.4 0.3 Educationa 0.5 0.3 1.8 1.7 0.2 Health 0.4 0.5 0.6 0.3 0.3 Agriculture 0.2 0.1 2.9 0.1 0.2 Roads 1.1 0.6 0.7 0.6 Water 0.3 0.7 0.6 Other including residual e 6.9 8.2 5.3 10.4 12.7 Memo item: Total expenditures inContas Geral do Estado 13.8 15.6 28.0 15.6 16.4 and Relat6rios de Execuggo e Sources: Conta Geral do Estudo for 1998,1999 and 2000; Relatdrios de Excecupio for 2000,2002 and2001(to September, but annualizedinthe table); water from Finney(2002); roads fromHerman(2002); healthfromHSER (2002), education from e-mailsent by MafaldaDuarteto Alex Valerio on 03/24/03, FileAnnex "Exec 2001 ". Note: Datafor GeneralAdministration,EducationandAgriculture are from the Contas Geruis do Estudo (CGE) andRelatdrios de ExecuFZo (RE). Datafor Educationin2001includeextemalfinancing fromMafaldaDuarte'se-mail. Datafor Roads, HealthandWater are fromthe correspondent sectoral PERpapers. Datafrom CGE excludes most extemallyfinancedexpenditures, except for the year 2000, which includesit. The Relatdrios de ExecuGZoalso exclude most extemallyfinanced expenditures. Datafor total expenditures, total recurrentexpendituresandtotal investmentexpenditures correspondto data inthe consolidatedCentralGovemment accountsas presentedinTable 2, "Govemment finance, 1997-2003 (percentageof GDP)" onpage 20. These data includes most extemallyfinanced expenditures. a Includesprimary, secondaryandtertiary education. Health andWater are the only sectors for which the whole series includesextemallyfinanced expenditures. Between 1998and2001, roads includesa smallproportionof extemallyfinancedexpenditure.. This category is the residualbetweenthe total andthe sectoral information. It includesmissingsectors andnon-accounted-forextemal financing. e The differencebetweenthe information in this memo itemandthe total expendituresi s the externalfinancing that i s not includedinthe Contas Geraisdo Estado andRelatdrios de ExecuqZo. 106 Table 39. Health sector recurrent expenditures by level of care (percentage) Level 1982 1989 2001 IandI1(healthposts/centers) 64 44 42 I11and IV(rural / general hospitals) 36 56 58 Total 100 100 100 Sources: WorldBank (1992), Table IV.9, p. 105; andMinistCrioda Sahde(2001b), table 6. Figure 7. Inequity in budget allocations across regions"' " I I . loo Source: Serviqo Nacionalde Saude (1999) andHSRP Indicators (2002). 107 REFERENCES African Development Bank, 2002. Mozambique: Water and Sanitation Sector Review. Volume 1. MainReport. May 2002. Preparedby SEED, Lda. Agencia Suiqapara o Desenvolvimento e a Cooperaqgo (SDC), 2001. Provedores Privados Ngo-Lucrativos de Serviqos de Saude emMopmbique: Situaqgo presentee suas perspectivasde desenvolvimento. Vol. 1:Relatdrio de Sintese. Vol. 2: Resultadosda FASE I do Estudo. Vol. 3: Resultados da FASE I do Estudo. Study on non-profits. Amdt, Charming, 2003. HIV/AIDS, Human Capital,and Economic GrowthProspectsfor Mozambique. Africa Region Working Paper Series, May 2003. Austral Consultoria and Projectos Lda, (2002). Estudo de Custo e Eficacia Externa do Ensino Tkcnico eProfissional em Mozambique: Relatdrio Final (Draft). Maputo. Austral Consultoria e Projectos, Lda. e MinistCrio do Plan0 e Finanqas, 1999. Identijkariio dos "Off-Budgets e mecanismospara a sua integraqiio no " orramento do Estado. 2" versgo. Vol. 1. Beattie, Allison andDaniel Kraushaar, 2000. HealthExpenditure inMozambique: A Review ofMajor Policy Issues. Volume 2 of 3 volumes. Management Sciences for Health(MSH). FundingbyUSAID. Beneficiary Assessment Summary. Fumishedby Mr.D.Barraclough (FIPAG). Powerpoint presentation about water inMaputo and Beira. Data from 2001. Bonnel, R., 2001. Costs of scaling HIVprogram activities to a national level in Sub- Saharan Africa. The World Bank,Washington, D.C. Brushett, Stephen, 2000. Republic of Malawi: Public Expenditure Review: Roads Sector Review. March20,2000. Backgroundpaper for the Malawi PER. Cabral, Zaida, 1999. A study of access to basic education and health in Mozambique. Finalreport: Communitylevel. Maputo, February 1999. Oxfam inMozambique. CGE, Conta Geraldo Estado, see under Republica de Mopmbique, 2000-2001. Chichava, Ana, Arlindo Lopes, Femando Pacini and Jorge Graqa, 2001. Relat6rio tCcnico sobre o processo de gestgo de politicas publicasem MoCambique. "Estudo realizado no h b i t o do processo de implementaqgo dareforma do sector publico". Setembro, 2001. EMCO -Empresade Consultoria Organizacional. Coelho, Isaias, Paul0 dos Santos, Ant6nio Teixeira, Jog0 Amaral Tomas, andRicardo Varsano, 2001. Para uma Reforma do Sistema Tributbrio e sua AdministrapTo. IMF, DepartmentofPublic Finance, March2001. Colclough, C., Al-Samarrai, S. (2000). Achieving Schoolfor All: Budgetay Expenditures on Education in Sub-SaharanAfrica and South Asia, World Development. Community Agency for Social Enquiry (CASE), and Centre for HealthPolicy (CHP), 2000. Beneficiary Assessment of Health Users in Mozambique. June 2000. 108 DNPO (Direcqgo Nacional do Plano e Orpmento, Ministerio do Plano e Finangas) and DNEA(DirecqZo Nacionalde EconomiaAgriria, Ministkrio daAgriculturae Desenvolvimento Rural), 2003. Anhlise da DespesaSectorial (Ministkrio de Agricultura e Desenvolvimento Rural): Relat6rio da 1"Fase. September2003. Economist, 2002. Special report. Health care inpoor countries: For 80 cents more. August 17. ESER, 2002-See Ministbriodo Plano e Finangas e Ministerio daEducaqZo(2002). Esrey, S. A., 1996. "Water, waste, andwell-being: amulticountry study." American Journal of Epidemiology 143 (6): 608-623. ESSP Supervision Mission, 2002. Estimating theImpact of HIV/AIDSfor the Education System of Mozambique and the Financials Implications to 2015. Annex 4: HIV/AIDSCost and Impact. ESSP: EducationSector StrategicPlan. Filmer,D.,J. Hammer and L.Pritchett, 2000. Weak links inthe chain: adiagnosis of healthpolicy inpoor countries. The World Bank Research Observer 1512. Finney, Chris E., 2003. Mozambique Public Expenditure Review: Background Paper on Agriculture. May 2003. Finney, Chris E., andE.Kleemeier, 2003. Mozambique Public Expenditure Review: Background Paper on the Water Sector. May 2003. Fiscal Affairs Departmento fthe IMFandthe PREMPublic Sector Group o fthe World Bank, 2002. Public expendituremanagement: country assessment and action plan. September 2002. Fozzard, A., 2002. How, When and why does Poverty get Budget Priority. Poverty Reduction Strategy and Public Expenditure in Mozambique. Working Paper 167. ODI. GeneralInspectorateo fFinance (IGF), Mozambique, and SwissAgency for Development Cooperation (SDC), 2002. Performance audit of the pharmaceutical sector pilot study: Preliminaryfindingsfor the stock management sub-system. Haacker, M.,2001a. Providing health care to HIVpatients in Southern Africa. TMF Policy Discussion Working Paper, PDP/O1/3, Washington D.C. Haacker, M.,2001b. The economic consequencesof HIV/AIDS in SouthernAfrica. IMF Policy Discussion Working Paper, PDP/02/38, Washington D.C. Handa, Sudanshu, FarizanaOmar, and Maimuna Ibraimo, 1998. Humancapital and social well-being inMozambique. In:Ministryo fPlannning andFinance, 1998. UnderstandingPoverty and Well-Being in Mozambique: TheFirst National Assessment. Maputo. Pages 265-301. Handa, Sudanshu, and KennethR. Simler, 2000. Quality or quantity? Thesupply side determinants ofprimary schooling in rural Mozambique. InternationalFood Policy ResearchInstitute DiscussionPaperNo. 83. 109 Harding, Alan, 2003. Review of Budget-related issues. Reportto G11Donors and Government of Mozambique: Joint Donor Review 2003. Maputo, May 2003. Heltberg, Rasmus, Kenneth Simler and FinnTarp, 2001. Public spending andpoverty in Mozambique. UNU/WIDER project onNew Fiscal Policies for Growth and Poverty Reduction. HEM, viz. HealthResearchofAction, 1999. Health CareFinancing in Tanzania: Costing Study of Health Services. FinalReport. Ministry o fHealth, Tanzania. Herman, Larry, 2003. MozambiquePER: Background Paper on Roads. January 2003. HSER, 2002 -See MinistCrio da Sadde e MinistCrio do Plano e Finanqas (2002). HSRPIndicators, 2002, viz. Health SectorRecovery Programindicators. HSRP is a World Bank-fundedproject. INE-DHS1998. See InstitutoNacional de Estatisticae Demographic andHealth Surveys, 1998. Instituto Nacionalde Estatisticae Demographic and Health Surveys (INE-DHS), 1998. MoCambique: InqukritoDemogrdflco e de Sazide 1997. Macro Intemational Inc. Manuel da Costa Gaspar, Humberto A. Cossa, et al. September 1998. InstitutoNacionalde Estatistica, 2001. Questionbrio deIndicadores Bdsicos de Bem- Estar (QIBB). Survey o fwell-being. Jalan, J. andM.Ravallion, 2003. "Does Piped Water Reduce Diarrhea for Childrenin Rural India?"Journal of Econometrics 112 (1): 153-73. Kengue, Valkse Mapto, andAlain Mingat, 2002. Analyse comparative internationale de lafkminisation du corps enseignant et de I'impact du genre de 1'enseignant sur la performance des systdmes kducatifs primaires en Afrique. Communication Colloque AFEC, Caen, May 23-25,2002. Kostermans, Kees, 2000. Memo for preparation o f Poverty Reduction StrategyPaper: Mozambique. July 2000. Health, Nutrition andPoverty Sector. Lindelow, Magnus, Patrick Ward andNathalie Zorzi, 2003. Expenditure Tracking and Service Delivery Survey (ETSDS): the health sector inMozambique. Finalreport -forcomments. February2003. Funding:DfID. AssistancefromMISAUis acknowledged. Lonstrup, Esther, 2002. Status ofpublicfinancial management in Mozambique: Fiduciary risk assessment. Consultant to DfID. Maputo, August 2002. Mingat, A. (2002). Teacher Salary Issues in African Countries. Internal document. The World Bank, Washington D.C., March2002. MinistCrio da Educaqzo de Mopmbique, 1998. Plano Estratkgico de EducaqEo, 1999- 2003: Combater a exclusEo,renovar a escola. Maputo. MinistCrio da Sadde, 2001a. Cendrio deDespesas e Financiamento do Sector Saude de Mkdio Prazo 2001-2005. 110 Ministerio da Safide, 2001b. Quadro de evoluqio da despesapor gastos correntes de salarios, bens e sewiqos e capital. Seconddraft, Feb.20,2001. Preparatory paper for the strategy (PESS). Ministerio da Safide e SDC, 2001. Analise dofinanciamento externopara o sector de Saude, 1997-2000, com estimativaspara 2001-2003. DPC-MISAU/SDC. Authors Pavignani andRusso. MinistCrio da Safide, 2002. Estudo sobre o impacto do SIDA nos serviqos de sa~deem Moqambique. Financiado pela CooperaqTio Italiana. Vol. 1:Sumario Executivo. Vol. 2: Relatdrio Final. June 2002. Ministkrio da Safide e MinistCrio do Plano e Finangas (2002). Health Sector Expenditure Review (HSER): Draft Reportfor Discussion. Maputo, 31October 2002. MinistCrio do Plano e Finangas, 2000. RelatBrio deExecuqio do Orqamento do Estado. Periodo: Janeiro aDezembro. MinistCrio do Plano e Finanqase Ministbrio da EducaqTio, 2002. Education Sector Expenditure Review (ESER). Maputo, November 1,2002. Ministry of Education. (2002). Plano Estratkgico de Educaqio, 4a ReuniTio Anual de RevisTio, Maputo, September2002. MinistryofHealth, MinistryofPlanningandFinance,National Instituteof Statistics and NationalAIDS Council, 2002. Demographic impact of HIV/AIDS in Mozambique: update 2000. Maputo. MinistryofPlanning andFinance, Eduardo Mondlane University, andInternational Food PolicyResearchInstitute, 1998. Understanding Poverty and Well-Being in Mozambique: TheFirst NationalAssessment (1996-1997). Ministry o fHealthand Management Sciences for Health, 1999. Health Expenditure in Mozambique: An Analysis of Major Policy Issues. 3 volumes. Moon, Alistair, 2003. Public expenditurereviews andthe budget process inUgandaand Tanzania. Informalmanuscript, World Bank. NationalDirectorate o f Water, 2002. Preparationofthe water chapter for the PER (Public Expenditure Analysis). NationalConsultant FinalReport. September 2002. Nicholson, Kit, 2002. Mozambique Fuel Tax: Poverty and Social Impact Analysis. DfID. SecondDraftReport. Oxford Policy Management (2002). Unit Costs of Education in Zambia, Final Report. Oxford. PARPA, see Republic o fMozambique (2001). Pavignani, E. (2002). 0 impacto do SIDA sobre o sector de Saude.Unpublisheddraft. Pavignani, Enrico, Stefan Sjolander andDagAarnes, 2002. Moving on-budget in the health sector in Mozambique: requirements,features and implications of future aid management andfinancing mechanisms. October 2002. Draft for discussion. 111 PEMR, 2001. See WorldBank, 2001a. Pesquisa Salarial. Comparaqiio entre afun@o ptiblica e o sector empresarial nas cidades deMaputo, Beira e Quelimane. Maputo, September2002. Unofficial paper from staff o fMAE. PESS, 2001. Plano Estrate'gico do Sector da Satide. See Republic o fMozambique, Council o fMinisters, 2001. Pfeiffer, James, 2003. "International NGOs andprimaryhealth care inMozambique: the need for a newmodel o f collaboration." Social Science & Medicine 56 (2003): 725-738. Republic o fMozambique, Council ofMinisters, 2001. Health Sector Strategic Plan 2001-2005-2010. Approved April 24,2001. PESS -Plano EstratCgicodo Sector da Safide. Republicof Mozambique, 2001. Action Planfor Reduction of Absolute Poverty 2001- 2005 (PARPA; 4thdraft March 19,2001). Unofficialtranslation by the British High Commission inMozambique, May 2001. Republica de Moqambique, 2000-2001. Conta Geral do Estado 1998and 1999and 2000. Maputo. Referred to here as CGE. RepublicofMozambique, 2003. PARPA Implementation Evaluation Report: 1'' half 2002 (Sectoral Performance), 3rdquarter 2003 (Public expenditure). Postedon the external Web site ofthe World Bank. ServiqoNacional de Safide, 1999. Informaqiio Estatistica Sumdria, Janeiro a Dezembro 1999. NationalHealthService. SDC, see Swiss Agency for Development and Cooperation, andAgencia Suiqapara o Desenvolvimento e a Cooperaqzo. Simler, Kenand SarahHarrower, 2003. Estimating poverty indices using light monitoring surveys. PowerPoint presentation, May 29,2003. IFPRI. Sulemane, JosC A. and SteveKayizzi-Mugenva, 2001. The Mozambican civil service: incentives, reforms andperformance. Draftpaper, 40 pages. WIDER project in May 2001. SwissAgency for Development andCooperation (SDC), 2000. Beneficiary Assessment o fHealth Users inMozambique. June 2000. Fundedby SDC. UNESCO (2000). WorldEducation Indicators, 2000. New York. UTRAFE(Unidade TCcnicapara aReformada Lei daAdministraqgo Financeira do Estado), 2002. SISTAFE:Plano deAcqiio e Orqamento. October 17,2002. MinistCrio do Plano e Finanqas, Gabinete da Ministra. UTRESP (Unidade TCcnica da Reforma do Sector Pfiblico), 2002. Reforma salarial: opq6es depolitica de me'dioprazo. Maputo, Janeiro de 2002. Varley, R. C., J. Tarvid, et al., 1998. "A reassessment o fthe cost-effectiveness o fwater and sanitation interventions inprogrammes for controllingchildhood diarrhoea." Bulletin of the WorldHealth Organization 76 (6): 617-31. 112 Wagstaff, Adam, Jennifer Bryce, FlaviaBustreo, Mariam Claeson, Niklas Danielsson, and the WHO-World Bank Child Health andPoverty Working Group, 2002. ChildHealth: Reaching the Poor. Draft. World Bank, 1992. Mozambique -SecondPublic Expenditure Review. December 31, 1992. Report No. 11524 - MOZ. Country Operations Division, SouthemAfrica Department. World Bank, 1994. Better Health in Africa: Experience and LessonsLearned. Washington, D.C. World Bank, 1997. Mozambique: Agricultural Sector Memorandum. Report No. 16529 MOZ. April 30. Africa Region: Agriculture andEnvironment. World Bank (2000). Cost and Financing of Education: Opportunities and Obstaclesfor Expanding and Improving Education in Mozambique. Intemal Draft, Washington D.C., January 2000. World Bank, 2001a. Mozambique Public Expenditure Management Review (PEMR). Report No. 22985-MOZ. December 2001. Africa Region, Macroeconomics 1. WorldBank, 2001b. Moqambique. Andise da Gestio daDespesaPtiblica. Dezembro de 2001. Regigode Africa, Macroeconomia 1. Translation ofPEMR. World Bank, 2001c. Malawi. Public Expenditures: Issues and Options. September 2001. Report No. 22440 MAI. Africa Region, Macroeconomics 1. Yates, Robert, andNathalie Zorzi, 2000. HealthExpenditure Review inMozambique. Volume 3. HealthFinancing Studies. Ministryo fHealth and Management Sciences for Health.