Report No. 40145-MW Malawi Public Expenditures Review September 2007 Poverty Reduction and Economic Management, AFTP1 AFCS2 Africa Region Document of the World Bank Republicof Malawi The WorldBank MALAWI PUBLIC EXPENDITUREREVIEW 2006 MINISTRY OF FINANCE P.O. BOX 30049 LILONGWE 3 MALAW I Tel: (265) 1 789 355 Fax: (265) 789 173 E-mail: jinance.gov.mw Website:www.finance.gov.mw CURRENCY EQUIVALENTS (Exchange Rate Effective May 25,2007) Currency Unit = MalawiKwacha (MK) MK1 = US$0.0071 US$1 = MK140.60 MEASURES Metric System FISCAL YEAR July 1to June 30 (As of July 1998) ABBREVIATIONS AND ACRONYMS ART Anti-Retroviral Therapy BHA Better Health for Africa CDSS Community Day Secondary Schools C H A M Christian Health Association o f Malawi CIDA Canadian InternationalDevelopment Agency CMS CentralMedical Stores COMESA Common market for Eastern and Southern Africa CPI Consumer Price Index css Conventional Secondary Schools DANIDA Danish International Development Assistance DFID Department for International Development DHRMD Department o f Human Resources, Management and Development EU European Union FA0 Food and Agriculture Organization o f the UnitedNations GDP Gross Domestic Product G O M Government o f Malawi HIPC Highly IndebtedPoor Countries IDA InternationalDevelopment Association IFMIS Integrated FinancialManagement Information System IHS Integrated Household Survey IMF InternationalMonetary Fund JCE Junior Certificate o f Education L D C Less Developed Countries MANEB MalawiNational ExaminationBoard MASAF Malawi Social Action Fund MDHS Malawi Demographic Health Survey MGDS Malawi Growth and Development Strategy MIE Malawi Institute o f Education MIITEP Malawi Integrated In-service Teacher EducationProgram 11 *. MOA Ministry of Agriculture MOE Ministry of Education MOF Ministry of Finance MEPD Ministry o f Economic PlanningandDevelopment MOH Ministry o fHealth MPRS Malawi Poverty ReductionStrategy MRA Malawi RevenueAuthority MSCE Malawi SecondaryCertificate of Education MTEF MediumTerm ExpenditureFramework NASFAM NationalSmallFarmersAssociation of Malawi NFRA NationalFoodReserveAgency NGO Non-GovernmentalOrganization NHP NationalHealthPlan NRA NationalRoadsAuthority NSO NationalStatisticalOffice OPC Office ofthe PresidentandCabinet PAC ParliamentaryAccounts Committee PEM PublicExpenditureManagement PIF PolicyInvestmentFramework PRSP PovertyReductionStrategyPaper PSLCE Primary School LeavingCertificate Examination RA RoadsAuthority RBM ReserveBank o fMalawi SADC SouthernAfrica DevelopmentCommunity SFFRFM Smallholder Farmers' FertilizerRevolvingFundo f Malawi SGR Strategic GrainReserve ST Secretary to the Treasury TAMA Tobacco Association of Malawi TCC Tobacco ControlCommission TIP TargetedInputsProgram TTC Teacher TrainingColleges UNESCO UnitedNations EducationSocial andCulturalOrganization USAID UnitedStatesAgency for InternationalDevelopment VLPA Village LevelParticipatoryApproach The PERwas preparedby a team o f GovernmentandWorldBankofficials ledby PerksLigoya(Director Economic Affairs, Ministry of Finance)andAntonio Nucifora(Senior Economist, AFTP1, WorldBank). The government team includesPerks Ligoya(Ministry ofFinance), ChauncySimwaka (Ministry of Finance), StanNkhata (Ministry of Finance), SteveBanda(MinistryofFinance), Saulos Nyirenda, (Ministry ofFinance), M.Makalande(Ministryo f Education),EdwardKatayika(Ministry o f Health),Mary Shaba (Officeo fthe Presidentand Cabinet), Victor Lungu(Ministryof Transport). Inaddition, the government team also includes: Chipo Msowoya, DanKhomba, Emily Zita, R. Mzonde, H.Njati andM. Kajiyanike, RJ. Nthengwe, G. Banda, A. Zubairi, K.Ng'ambi, P. Revill, D.ChilimaandMacCallum Sibande, S. Siwande, J.B.M.Phiri, S. Chirambo, M.Jalazi, A. Chinkunda, D.Nkhomba, I. Thindwa, 0. Manda, andM.Muyepa. The WorldBankteamincludesAntonio Nucifora (AFTPl), KhwimaNthara (AFTPl), Jos Verbeek (AFTPl), Pave1Lukyantsau(AFTPl), Abha Prasad(PRMED), MichaelMambo (AFTHl), Kirsten Majgaard(AFTHl), RameshGovindaraj (AFTHl), Oscar Picazo (AFTHl), Alfred Chinva (AFTHl), Rae Galloway (AFTHl), Carolina Diaz Bonilla(DECPG), Hans Lofgren(DECPG), HardwickTchale (AFTAR), DavidRohrbach(AFTAR), Tawia Addo-Ashong (AFTTR) andAntonio David(AFTHV). ... 111 PREFACE The 2006 Public Expenditure Review (PER) has been a major undertaking by Government. It represents a thorough investigation of public expenditure trends and utilization in Government, focusing on Education, Health, Nutrition and Roads. Although the original intention was to include a chapter on Agriculture in this document due to unforeseen technical delays it was ultimately resolved to include the Agriculture chapter in the next review. Many of the findings in the PER reaffirm the need to improve our expenditure planning and monitoring systems in order to reduce poverty. It provides evidence-based analysis on expenditure which serves as useful information for the formulation of future fiscal policies. Indeed, the publishing of this PER confirms Government's commitment to spur enhanced transparency in public expenditure management. Earlier drafts of this document were widely circulated for comments. Due to the late publication of the PER, some of the findings have already been taken into account in the 2006107 and 2007108 budget, and the remaining ones will be used in the subsequent budgets. This PER was coordinated by a team in the Ministry of Finance with technical work carried out jointly by Government officials from the relevant ministries and staff from the World Bank. Other Cooperating Partners including the CABS group of donors provided useful input into the process. The World Bank, GTZ and the EU Capacity Building Project provided financial assistance which is greatly appreciated. I would like to thank everybody who was involved in the writing of this Report for the high quality, depth and objectivity of the analysis. In particular, I would like to thank the coordinating team in the Ministry of Finance for their often strenuous efforts to produce the document. I would also like to thank the World Bank staff for their valuable contributions to this project which was carried in full spirit of cooperation and partnership. We believe this project has provided an excellent starting point for developing working relations between the Government of Malawi and our developing partners in carrying out analytical work. It is Government's sincere hope that the PER will help improve expenditure control and monitoring and ensure efficient utilization of resources, which is a necessary building block for poverty reduction and economic growth. R.P. Mwadiwa Secretary to the Treasury iv ACKNOWLEDGEMENTS The 2006 Public ExpenditureReview (PER) was preparedby the Governmentof Malawi with technical assistance from the World Bank. Other development partners also providedvaluableinputs. On the Government side the PER process was initially led by Mr. Ted Sitimawina and later by Dr. Perks Ligoya.Co-ordinationand day to day managementof the process was initially providedby Mr.Archangel Chinkundaand later by Mr.Saulos Nyirenda. Other staffalso involvedinthe coordinationwere Ms E. Zita, Mr D.Nkhomba,Mrs G. Banda, Ms M. Jalazi, Mr. I.Thindwa, Mr 0.Manda, and Mr. M. Muyepa. The report was preparedby Government officials from a number o f Ministries and institutions. Each group was responsiblefor a respectivechapter as shownbelow: Macroeconomic Economic Developments and theIntersectoralAllocation of Resources: Mr.SteveBanda, Mr.ChipoMsowoya,Mr.DanKhombaandMs.EmilyZita. Public Debt Sustainability and Debt Management Reforms: Mr. Stan Nkhata, Mr. Chauncy Simwaka,Mr.R.Mzonde,Mr.H.Njati andMs.M.Kajiyanike. Education Sector Review: Mr.M.Makalande,Mr.R J. Nthengwe,Mrs G. Banda andMs. A. Zubairi. Health Sector Review: Mr E.Kataika,Mr.K.Ng'ambiandMr.P. Revill. Nutrition Interventions Review: Dr.Mary Shaba, Dr. D. ChilimaandMr.MacCallum Sibande. Roads Sector Review: Mr.V. Lungu, Mr.S. Siwande, Mr.J.B.M.Phiri, Mr.S. Chirambo, andMs.M.Jalazi. Acknowledgementsalso go to Mr.M.P. Galafa(Director, DHRMD), Paul Lungu (OPC), D.E.Chunga (DHRMD), DorothyBanda (BudgetDirector, Ministry o fFinance), Richard Perekamoyo (Assistant Budget Director), Naomi Ngwira, (Director o f Debt and Aid, Ministry of Finance) and RedfordKampanje(AccountantGeneral). Specialthanks go to the Minister of Finance Hon. Goodall E. Gondwe, ( MP ) and to the Secretary to the Treasury, Mr.R.P.Mwadiwafor their support and guidance throughoutthe process. On the World Bank side the team was led by Antonio Nucifora (Senior Economist, AFTPl), and included Khwima Nthara (AFTPI), Jos Verbeek (AFTPl), Abha Prasad (PRMED), Michael Mambo (AFTHI), Kirsten Majgaard (AFTHl), Oscar Picazo (AFTHl), Alfred Chinva (AFTHl), Tawia Addo-Ashong (AFTTR), Rae Galloway (AFTHl), Carolina Diaz Bonilla (DECPG), Hans Lofgren (DECPG), Pave1Lukyantsau (AFTPl), David Rohrbach (AFTAR), Hardwick Tchale (AFTAR), Antonio David (AFTHV). The team benefitedfrom the guidance of EmmanuelAkpa (Sector Manager AFTP1, World Bank). Michael Baxter (Country Director, World Bank), and Timothy Gilbo (Country Manager, World Bank) also providedoverall directionand guidance for the analysis. The report also benefited from excellent advice from Linda Van Gelder (Lead Economist, PRMPR, World Bank), Hassan Zaman (Senior Economist, SASPR, V World Bank, and TTL o f Malawi PER 2001), Sona Varma (Senior Economist, PRMED, World Bank), Nina Todorova Budina (Senior Economist, PRMED, World Bank), who were the peer reviewers for this report. We are also indebted to Sudhir Chitale, Peter Moll, Ross Worthington, Antonio Velandia, Tadashi Endo (World Bank), who contributed insightful comments and constructive feedback. Beyond the Bank, Samira Leakey provided extensive and invaluable comments and advice in compiling and shaping the report. Alan Whitworth (DFID, Malawi) also provided invaluable contributions and guidance. We also gratefully acknowledge the many valuable comments received from Andy Berg, Thomas Baunsgaard, Thomas Dalsgaard, (IMF), Nick Dyer, Julie Kemp, Anna de Cleene (DFID), Appolenia Mbowe (AFDB), Jeremy Martins (Advisor, Ministry o f Transport), and Erik Schouten (HIV- coordinator, Ministry o f Health). The PER also benefited from the contributions o f development partners who also participated in the various Thematic Working Groups: Stella Kaendera, Levie Sat0 (IMF), Jerome Pons, Chris Indelbrecht, Raniero Let0 (EU), Jan Olav Pettersen (Norway), Leigh Stubblefield, Bernabe Sanchez, Lindsay Mangham, Alan Whitworth (DFID), Aues Scek, Joana Henseler (GTZ). Financial support was provided by the European Union, GTZ, and the World Bank. Dr. Andrew Tench o f the EU's Economic Management and Capacity Building Project deserves special mention for his tireless efforts inorganizing funding for the process. Priceless assistance in various aspects o f the management and administration o f the project, and the preparation o f the final document, was provided by Dotilda Sidibe, Rose Kumsinda (AFTP 1); Annie Jere, Maggie Mshanga, Grace Soko, Esther Lozo, Rose Ndalama, Ethel Kuniya, and Stephen Kachule (AFMMW). Zeria Banda (AFMMW) providedexcellent guidance and advice on the dissemination process. vi TABLE CONTENTS OF EXECUTIVE SUMMARY ........................................................................................................................ IX CHAPTER 1: RECENT MACROECONOMICDEVELOPMENTS AND THE INTER-SECTORAL ALLOCATION OF RESOURCES ............................................................................................................. 1 1.1 INTRODUCTION ..................................................................................................................................... 1 1.2. MAINMACROECONOMIC DEVELOPMENTS BETWEEN 1999 TO 2006 Malawi on the Verge of a Fiscal Crisis Recent developments: the turnaround ajer 2004 ........................................ 1.3 WHAT CAUSED SUCH A RAPIDINCREASE INDOMESTICDEBTBETWEEN20 Weak expenditure control ...... Volatility in donor inflows ..... ................. Inability to cope with unanticipate 1.4 INTER-SECTORAL ALLOCATIONOF EXPENDITURES: CHOOSING THE BALANCE BETWEENHUMAN DEVELOPMENT INFRASTRUCTURE AND SPENDING MMALAWI ....................................................... 13 Simulation of Alternative Expenditure Strategies using the MMS Model......................................... 15 1.5 RECOMMENDEDPOLICY PRIORITIESFOR MACROECONOMIC STABILITY AND INTER-SECTORAL ALLOCATIONOF EXPENDITURES 18 CHAPTER 2: PUBLIC DEBT SUSTAINABILITY AND DEBT MANAGEMENTREFORMS ......21 2.1 INTRODUCTION . 2.2PUBLICDEBTST 22 Evolution and characteristics of domestic debt portfolio ................................................................... Evolution and characteristics of external debt portfolio.. ................................................................... 25 2.3 PUBLIC DEBT SUSTAMABILITY ANALYSIS.. Implications of Alternative Fiscal Adjustmen Shock simulations: Potential impact of externa debt over the nextfav years ................................................................................................................ 33 34 Exploring options to restructure the domestic debt and reduce the interest bill................................. Sources offiscal risk in the medium term ........................................................................................... 35 2.4 DEBTMANAGEMENT FRAMEWORK ...................................... 39 Institutional arrangementsfor debt management 39 Coordination, information flows and reporting in debt management operations 41 The national debt management strategy ............................................................................................. 41 2.5 RECOMMENDEDPOLICY PRIORITIESTO REDUCEPUBLIC DEBT VULNERABILITY ............................... 42 CHAPTER 3: EDUCATION.... ................................................................................................................. 46 3.1 INTRODUCTION ................................................................................................................................... 46 3.2 KEY EDUCATIONSECTOR ISSUES......................................................................................................... 46 Studentflows.. .................... 46 Student repetition, dropou 47 Eflciency of resource use 5-2 Quality of education ............................................................................................................................ 52 3.3 EXPENDITURE TRENDS....................................................................................................................... 53 Recurrent expenditure............ Development expenditure Financing of higher educ Household contributions to the Primary Education sector................................................................. 64 Costs to households of Secondary and Tertiary Education ............... 3.4 POVERTYFOCUS AND INCIDENCEOF EXPENDITURESINTHE EDU OR Disparities in expenditure................................................................................................................... 65 vii 3.5 IMPACT OF HIV-AIDS ON THE EDUCATION 3.6 RECURRENTRESOURCENEEDSTO ACHIEVE Financial simulation ................................... .................................................... 3.7 IMPLEMENTINGKEY RECOMMENDATIONSFROMTHE 2001 PERON EDUCATION ................................ 3.8 RECOMMENDEDPOLICY PRIORITIES FOR MALAWI'SEDUCATION SYSTEM. .......... .. 71 72 CHAPTER4: HEALTH ............................................................................................................................ 75 4.1 INTRODUCTION.. ........................................ ......................................... .................................................................. es............................................................... Scarcity of TrainedHealth Professionals ........................................................................................... 76 Partnership with Private Service Providers. ...................... Quality of Care Delivery of Esse ................................................................................................ Preventive Beha 4.3 IMPACT OF HIV/AIDS ........................................................................................................................ 81 4.4 EXPENDITURE TRENDS................................................... ........................................................................................................... 89 4.5 POVERTY FOCUSAND INCIDENCEOF EXPENDITURE THE HEALTH IN 92 4.6 IMPLEMENTING KEY RECOMMENDATIONS FROMTHE 2001 PERON HEALTH...................................... SECTOR..................................... 94 4.7 RECOMMENDEDPOLICYPRIORITIESFOR MALAWI'S HEALTH SECTOR ... 95 CHAPTER 5: NUTRITION ...................................................................................................................... 98 5.1 INTRODUCTION ..... ............................................................. 5.2 KEYNUTRITION ISSUES Undernutrition ........................................................................... ....... ................. ........................ Micronutrient malnutrition ................................................................................................................. 99 Lack of coordination acro 5.3 EXPENDITURE NUTRITION ON , ......................................... ...................100 Targeting of expenditures 5.4 RECOMMENDEDPOLICY P CHAPTER 6: ROADS ............................................................................................................................. 107 6.1 INTRODUCTION ................... ... ....... .......... 107 6.2 KEYROADSECTORISSUES ........................... ................... .......... 108 6.3 EXPENDITURES TRENDS ......... ..................................... 6.4 POVERTY FOCUS AND INCIDENCEOF EXPENDITURES INTHE ROAD SECTOR .................................... 111 6.5 IMPLEMENTING KEY RECOMMENDATIONSFROMTHE 2001 PERONROADS .......... ...................114 115 6.6 RECOMMENDEDPOLICY PRIORITIES FORTHEROADSECTOR........... ANNEXES ................................................................................................................................................. 118 ANNEX1.1 MALAWIFISCALAGGREGATES: BUDGETED VERSUSACTUAL EXPENDITURES ANNEX1.2 NATIONAL DROUGHT INSURANCEFORMALAWI ANNEX1.3 CORECHARACTERISTICS OFTHEMAQUETTEMDGSIMULATIONS(MAMS) MODEL FOR OF THEMALAWIAN ECONOMY ........................................................................................................... 127 ANNEX2.1 BRIEFDESCRIPTIONSELECT FINANCINGINSTRUMENTS OF .................................................. 134 ANNEX3.1 BRIEFDESCRIPTIONEDUCATIONSYSTEMINMALAWI OF ANNEX4.1 BRIEFDESCRIPTIONHEALTH OF SYSTEMINMALAWI REFERENCES ......................................................................................................................................... 140 ... V l l l EXECUTIVE SUMMARY 1. The primary motivation for this Public ExpendituresReview (PER) i s to guide the government inimplementingreforms neededto reestablish fiscal sustainability, create stable macroeconomic conditions, and make better use o f scarce resources to foster more rapid and equitable growth and wealth creation, and improve the effectiveness and quality o f service delivery. Six years have passed since the Government o f Malawi carried out its last PER in Fiscal Year 2000/01 (July Ito June 30), in collaboration with the World Bank. This report builds on the analysis of the earlier work, assessing progress made in implementing its recommendations and incorporating new recommendations in light of significant developments that have occurred subsequent to the 2001report. 2. Mixed progress has been made in implementingthe main recommendations from the PER 2001. As highlighted in the relevant chapters, most o f the recommendations that have not been implemented remain valid, and are therefore included in the recommendations o f this report. Moreover, the failure to implement many o f the key recommendations from the PER 2001 suggests the need to focus more forcehlly on building political consensus for these reforms. Accordingly, the government intends to widely discuss the analysis and findings presented in this report, both within Parliament, and with the public at large, to increase public awareness and inform the policy debate. As a follow up to this report, the government could also consider preparing a prioritized PER implementation action plan, focusing first only on highly critical and easy-to-implement recommendations. 3. Since the last PER was undertaken, weak fiscal performance between 1999/00 and 2003/04 brought the country to the verge o f a financial crisis. Large fiscal deficits were financed mostly by domestic borrowing. Consequently, the domestic debt increased sharply from less than 3 percent in 1999/00 to 25 percent o f GDP in 2003/04. Coupled with high interest rates, the higher debt level meant that government's domestic interest bill shot up to a massive 9.2 percent o f GDP by 2003/04 (or 40 percent o f revenues). The government has since taken decisive action to reverse the unsustainable fiscal trends and, as a result, macroeconomic performance over the last three years has been rapidly improving. Domestic debt has decreased to 20 percent o f GDP in 2005/06, and inflation declined to 9 percent at end-March 2007. With inflation and government borrowing declining, the Central Bank was able to slash the nominal discount rate from 45 percent in September 2003 to 20 percent inNovember 2006. As a result, the domestic interest bill has decreased by almost 4 percent o f GDP over the past three years and, assuming government stays the course with its macroeconomic stabilization program, it i s expected to reduce by a further 3 percent over the next three years. 4. In light of this commendable progress, the report highlights that Malawi is now at a critical juncture, facing important choices in prioritizing public expenditures. The lower interest bill is generating much needed fiscal space that can instead be channeled to priority expenditures, such as in health, education, social safety nets, agriculture, irrigation and transport infrastructure. Thus, the government needs to decide how best to utilize these additional resources, and this report therefore addresses two ix related questions facing the government: Firstly, how best can the government take advantage o f this fiscal space to consolidate the recent gains and accelerate economic growth and poverty reduction? And secondly, as the focus o f government action moves beyond macroeconomic stabilization into policies that accelerate growth and poverty reduction, what priority reforms should the government pursue to improve the efficiency and equity o f public expenditures? The report argues that public expenditureoutcomes in Malawi can be improvedinthe next few years if: (i) government stays the course with the the macro economic program, and the reduction o f the domestic debt burden; (ii) continued efforts are made to strengthen the budget process, payroll management, and debt management, and action is taken to protect public finances from the impact o f external shocks; (iii) the efficiency and equity o f public expenditures o f intra-sectoral allocations i s improvedby implementingthe key measures recommended inthis report. 5. Furthermore, this report highlights that a strategic balance must be struck between investing interest savings into: (i) accelerating the retirement o f domestic debt, which would bring additional interest savings that can channeled to priority expenditures at a later stage; (ii)building up Malawi's external reserves and/or purchasing drought insurance, as coverage to protect against exogenous shocks; and (iii)expanding investments in infrastructure and/or social services, to boost growth rates and progress towards the MDGs. The government needs to carefully consider the costs and benefits o f each option, and these three themes underpinmuch o f the analysis inthis report. 6. The report begins by discussing the main areas for strengthening the public financial management and minimizing the impact o f external shocks on public finances. It then reviews Malawi's debt trends and outlook, and discusses measures to reduce the domestic debt burden, improve debt management, and reduce debt vulnerability. The report then turns to a brief discussion o f the inter-sectoral allocation o f resources, with a view to assessing the implications for growth, poverty reductionand progress towards the MDGs. Finally, the efficiency and equity o f intra-sectoral expenditures is assessed in three key sectors, namely education, health and roads. A chapter on nutrition has also been prepared (and i s included as part o f this report), following the analysis carried out in the 2006 Malawi Poverty and Vulnerability Assessment (MPVA) which highlights malnutrition as on o f the priority issues facing Malawi. Work has started and i s ongoing towards a chapter on agriculture, and this chapter will be publishedseparately, reflecting the critical importance o f agriculture for food security, poverty reduction, and growth in Malawi. These four sectors-agriculture, education, health and roads-constitute the four largest expenditure votes in the budget, and together account for almost 60 percent o f total voted expenditures. Further, they provide public services that are critical to accelerate growth. 7. Looking at the future, the government and the development partners have expressed the desire to see the PER analysis continue, to take advantage o f new data as it becomes available, and to extend the analysis to other areas. In this context, the government i s looking at ways to adopt a more institutionalized approach to cany out PER work on a more modular and ongoing basis. 8. This report has been carried out jointly by the Government o f Malawi and the World Bank, and with the active participation o f the group o f development partners providing budget support to Malawi (i.e., the Common Approach to Budgetary Support X group, which includes DFID, NonvaylSweden, and the European Commission; the World Bank, the IMF, and Germany, participate as observers). Preparation o f the report was overseen by a `PER taskforce' o f senior government officials (nominated by the respective Permanent Secretaries), and chaired by Director Economic Affairs in the Ministry of Finance (MOF). The World Bank also participated in the taskforce. The PER taskforce was in charge o f supervisingand coordinating progress inthe preparation o f the various chapters by sector specific `PER Thematic Working Group, chaired by the Director o f Planning inthe relevant ministry. REDUCING THE PUBLIC DEBT BURDENAND MALAWI'S VULNERABILITY DEBT 9. Reducingthe domestic debt should continue to be the overridingpriority of the government in the short term. Malawi's level o f domestic debt has decreased substantially since 2003/04. But with a small domestic capital market and the limited range of holders, it has the potential to increase again rapidly, as a resumption o f heavy borrowing could push real interest rates to high levels. Further, the interest payments on domestic debt are still significant, and continue to crowd out the amount o f discretionary budgetary resources available for expenditures on social and economic services, and investments. I O . Continued fiscal discipline is necessary for debt sustainability and domestic debt reduction. Malawi's macroeconomic situation remains fragile, and a relaxation in fiscal discipline could lead to a rapid resurgence in domestic debt. In this context the government needs to manage potential sources o f fiscal risk which could jeopardize fiscal sustainability inthe mediumterm, including:protecting the budget from the impact of exogenous shocks (notably drought), strengthening payroll management and possible wage drift, limiting the size o f the fertilizer subsidy, monitoring the possible accumulation o f arrears, and the financial performance o f parastatals. In addition, funding for the HIV/AIDS ART program and for the health sector salary top-ups has not been assured after 2009-2010, raising questions about both the sustainability o f these programs and their implications for the government budget if additional funding i s not secured urgently. 11. Strategies to reducethe domestic debt (in complement to fiscal discipline) are limited by the availability of additional financing and access to additional concessional resources. Restructuring the domestic debt portfolio is unlikely to bring about significant savings. Interest savings from any restructuring are likely to be small, and would come at the cost o f assuming increased exchange rate and inflation risks. The only possibilities open to government i s to retire domestic debt or to swap domestic debt with concessional credits, but these strategies are limited by the availability o f additional financing and access to additional concessional resources. In as far as additional resources become available (an additional USDlOO-150 million annually could become available from debt relief, IDA and M C C over the next few years), accelerating the reduction in domestic debt by using these resources to retire domestic debt or to swap domestic debt with concessional credits, would bring additional interest savings that can be channeled to priority expenditures. The greater the fiscal adjustment inthe short term, the faster the reduction in the interest bill, and therefore the greater the overall resource envelope available for priority expenditures over the medium-term. In addition, the xi shorter the adjustment period, the lower the likelihood that external shock could disrupt, or at least prolong, the period duringwhich the country i s paying a large interest burden. 12. The many risks associated with Malawi's debt portfolio emphasize the needfor management and institutional reforms, and debt market development. The characteristics o f Malawi's domestic debt portfolio, including the limited diversity o f financing instruments and their very short-term maturities, combined with the shallow debt markets, entail substantial exposure to refinancing risk, interest rate risk, and other fiscal risks. 13. The government should finalize and implement the National Debt Management Policy, to improve the legal, institutional and administrative framework, reducing both the external and domestic debt, while enhancing reserve management. As a first step, the 2003 Public Finance Management Act should be complemented by a set o f guidelines on issuing guarantees, local government borrowing, on-lending and private borrowing, degree o f concessionality on external debt, and borrowing limits (rates o f accumulation). Guidelines could clearly attribute regulatory jurisdiction over domestic debt instruments, participants, trading platforms, and clearing and settlements systems. In addition, a comprehensive National Debt Management Policy should address the following: 0 The operation of the Debtand Aid Division (DAD) should be strengthenedby establishing clear division of responsibilities across its sections (front, middle and back office), and drawing up a program to increase staff skills. DAD'S staffs need to be equipped with financial and analytical skills to produce regular debt sustainability analyses (DSAs), together with comprehensive currency-wise and maturity-wise risk analyses, interpreting loan agreements and reports sent by the creditors, and implementingthem inthe debt recording system. 0 A formal structure for institutionalized information sharing among debt managers, cash managers, budget units and the Central Bank should be established, to enable accurate and timely information flow. Rendering operational the Debt Management Committee (DMC) would enable better institutional coordination between the Ministry o f Finance (MOF) and Reserve Bank o f Malawi (RBM), and ensure timely sharing o f relevant information for debt policy formulation. In addition, the MOF should reactivate the Cash Flow Management Committee (CFMC) to ensure coordination between the agents responsible for monitoring the government's need for liquidity, and the MOF which i s responsible for issuance o f domestic debt. Also, the government should improve its cash flow forecasting capacity and link it to the RBM's liquidity management. In fact, weak debt management stems in part from lack of appropriate cash and liquidity management by government. 0 A rule-based debt management strategy with a statutory ceiling on public debt could be adopted. A direct and comprehensive approach to fiscal sustainability could entail a limit on the overall debt service as a proportion o f government revenues or GDP. 14. In order to foster the debt market development, the government should aim to lengthen the maturity of government paper and publicly Felease an issuance xii calendar. Announcing a calendar would only be possible in the mediumterm, once the macroeconomic situation allows the government to issue long term papers in a sizable manner. The calendar would set out in advance tentative dates and amounts o f government borrowings, thereby facilitating better planning by market participants and smoothing out supply-demand matching for government's issuances. It would also enable better liquidity management by financial institutions. 15. In the medium term, as the range of investors increases, the government could consider introducing the primary dealer system; this should not be rushed, however, as the systempresentsrisks. Introducinga primary dealer system ina market with a small range o f investors, could give rise to a potential for primary dealers to form cartels and end up restricting competition in the primary market, pushingup the interest rates. 16. In order to foster the secondary market for government securities, the government can aim to consolidate Treasury bills and elongate maturities (to increase fungibility) and reduce the high withholding taxes on government securities trading, which discourage discount houses from trading ingovernment securities inthe secondary market. Inaddition, suitable clearing and settlement mechanisms need to be developed. action plan in early 2008; implement actionplan in2008-2010 includingplan to strengthenthe operationof DAD, and formal coordination structure between MoF and RBM on matters of cashmanagementanddebt. Fiscal implications: Savings from improveddebt management. Foster developmentofmarketfor Sequencing: Discuss with RBM and forum of commercial governmentsecurities banks chief executives, with a view to reducing withholding taxes on government securities trading starting in 2008, and developing suitable clearing and settlement mechanisms 2008- 2009. Assess risks o f introducing a primary dealer system, with a view to adoptingthe system inthe mediumterm. Fiscal implications: Expectedsavings inthe mediumterm from reductionin the risks of debt portfolio, and resulting reduction ininterestrates. STRENGTHENINGTHE BUDGET AND PAYROLLMANAGEMENT 17. Reforms that strengthen the budget process and payroll management should remain at the center of the reform agenda. There is a need to strengthenthe policy context o f budget preparation, improve the technical infrastructure for budget management and rationalize the budget structure and budget documentation. Secondly, besides the continued phase-in o f the IFMIS, there is a needto improve internal controls and reconciliation in budget execution. Thirdly, there i s a need to strengthen audit capacity by: (i) further developing the capacity ofthe Central InternalAudit Unit;and (ii) developing the capacity o f the NAO to shorten the delay inthe presentation o f the audited accounts and report to the legislature. These reforms have not been discussed in detail in X l l l ..* this report, as they are already part of the authorities PFM action plan and have been examined inprevious studies. Similarly, the government i s already planning to conduct a review o f the HRMIS inorder to establish its robustness, and to carry out a payroll audit o f all civil servants in 2007 to verify and update the civil service database; these measures could entail substantial savings for the government, due to the reduction in the number o f ghost workers. 18. Reformsto insulate budgetimplementationfrom the impact of exogenous factors should be pursued. Malawi's economy is highly vulnerable to adverse weather shocks. This stems both from the direct impact o f shocks on agricultural production (and GDP), but i s also due to the indirect impact on government finances resultingfrom the unanticipated need for emergency interventions (which may translate into increased domestic borrowing), and increased pressure on the current account because o f the need for exceptional food imports. In spite o f the frequency o f (weather) shocks, only a very small allowance i s made in the budget for such contingencies in Malawi. As a result, shocks usually translate into higher domestic borrowing, and have a substantial impact on fiscal performance and macroeconomic stability. Possible remedial actions consist o f some kind o f reserve funds which can be used in case o f shocks. Given the substantial difficulties with managing a Contingency Reserve Fund (see Chapter I), and the high sterilization costs o f rapidly increasing forex reserves, purchase o f weather-based insurance i s recommended. In addition, the government should continue to work with CABS to improve the predictability o f aid, which has also been a source o f volatility duringthe past decade. (i)Purchasenationalweather insurance:The government canbuyrainfall-based insurance contracts to transfer the financial risk o f severe and catastrophic national drought that adversely impacts the government's budget to the international risk markets. Such a contract would secure timely and reliable funds for the government if a contractually specified severe and catastrophic shortfall in precipitation occurs during the agricultural season, as measured by weather stations throughout the country. Access to such contingency funds in a time o f crisis would generate a supplemental source o f rapidly available emergency financing to complement existing budget resources, giving the government more flexibility in its drought planning and enhancing the government's ability to launch an efficient and cost- effective drought response. Since development partners often provide additional financial resources inresponse to a humanitarian crisis, it would be important to seek their support to pay at least part o f the premium. (i)Increase forex reserves: An alternative to purchasing insurance would be to increase forex reserves to around 5 months o f imports. This would allow the government to run down reserves as a first line o f defense to buffer against most shocks without reserves disappearing. Increasing reserves would entail substantial sterilization costs, and therefore can only take place gradually. (iii)Liaise with donors to improve the predictability of aid disbursements: Several options are possible to improve aid predictability and aid coordination within the framework o f the CABS, by looking at the experience o f neighboring countries. xiv PROPOSEDREFORMSTO STRENGTHENTHE BUDGET AND PAYROLL MANAGEMENT Priority Policy recommendations Sequencingand Fkcal Implications Adopt nationalweather insurance Sequencing: Discuss with development partners and World Bank inJune-August2007, for a pilot to start inOctober 2007. Pilot with smallpremiumfor at least 2 years beforeextending. Fiscal implications: Pilot costs 0.02 percent o f GDP (USD0.5 million). Subsequently, national insurance premium around 0.05 percent o f GDP (USD10 million annually for a payout of USD70 millionin case o f large drought). Gradually increase forex reserves to Sequencing: Discuss with IMF during September mission, and around5 months of imports implement over the next4 years Fiscal implications: Substantial `cost' at around 8-10 percent o f GDP, to be gradually accumulatedand sterilized depending on additional inflows; but expected to lead to net savings over the longterm, by facilitatingmacroeconomicstability. Liaise with developmentpartnersto Sequencing: Discuss with CABS in June-August 2007, to improve the predictabilityof aid improve predictability o f aid disbursements starting with 2008 disbursements review. Fiscal implications: No costs; expected to lead to savings over the longterm, by facilitatingmacroeconomicstability. Strengthenpayrollmanagementby (i) Sequencing: Carry out review o f HRMIS, and payroll audit in conduct a review of the HRMISinorder 2007/08. Implementrecommendations o f HRMIS in2008109. to establishitsrobustness, and(ii) carry out a payrollaudit of all civil servants to Fiscal implications: No costs; possible savings related to verify andupdate the civil service eliminationo f ghost workers, estimated at 0.4 percent of GDP, database and improvedmanagement o fpayroll. CHOOSINGTHE INTER-SECTORAL ALLOCATIONOF RESOURCES 19. Looking ahead, there i s a need to consider how to invest the savings f r o m the reduction in domestic interest payments to accelerate growth and poverty reduction. The lower interest bill will generate much needed fiscal space to increase other (recurrent and development) expenditures, such as in health, education, social safety nets, agriculture, irrigation, and transport infrastructure. The allocation o f expenditures will affect the extent o f progress towards the Millennium Development Goals (MDGs). The government has recently adopted the Malawi Growth and Development Strategy (MGDS) as its overarching strategy for economic policy, and the MGDS aims to allocate savings from the domestic interest bill to increase spending on `economic sectors' and infrastructure investments. The MGDS recognizes that spending towards social services i s an investment in human development. Human development indicators measure important dimensions o f poverty, which are not captured by improvements in income. Further, a healthy and educated population i s itself necessary to achieve sustainable economic growth and development. However, the government believes that in previous strategies, resource allocations were tilted towards general administration and social services. Hence, the government wants to channel the funds released from the reductionindomestic bill to adjust this balance. xv 20. A strategy focused on increasing expenditures on infrastructure and agriculture would likely lead to relatively faster economic growth (and poverty reduction) but comparatively slower progress towards other human development indicators. The MAMS (Maquette for MDG Simulations) model has been used to analyze the implications o f alternative expenditure strategies. The model simulates the evolution o f major social and economic indicators in Malawi until 2015, including monitoring the effects o f government policies on poverty, education, health, and water and sanitation coverage. The results highlight that a moderate increase in infrastructure spending leads to faster GDP growth (and reductions in monetary poverty), but at the expense o f slower improvements in human development indicators (and vice versa; a moderate increase in spending on social sectors leads to faster progress in health and education, and other human development indicators, but at the cost o f a slower growth rate and monetary-poverty reduction). The results also highlight that maintaining sound macroeconomic policies i s critical for improvements in both growth and human development indicators. 21. Ensuring the `quality' of public expenditures (both in social sectors and infrastructure sectors) is critical to accelerate growth and progress towards the MDGs. The results of the MAMS model assume that the `quality' o f expenditures and investments (within each sector) does not change substantially. However, ensuring that resources within each sector are used efficiently i s also critical to realize the impact o f increased expenditures. The rest o f this report therefore examines the allocation and use o f resources within major sectors. IMPROVINGTHE EFFICIENCY AND EQUITYOF INTRA-SECTORAL EXPENDITURES 22. This report reviews expenditures in four key sectodareas: education, health, nutrition, and roads. The criteria by which public expenditures are assessed in these sectors revolve around two main themes. The first relates to the extent that public spending improves economic efficiency by intervening in areas where there are significant market failures. The second broad theme i s the extent to which public expenditures promote equity. The poor will require essential services that they may not be able to access if left to the market and hence there is a legitimate role o f the government inmeeting its poverty alleviation objectives. The nature o f the efficiency and equity arguments for public spending will also determine the type o f government activities inthese sectors interms o f whether public provision, financing or regulation are the appropriate means to achieve desired outcomes. For each sector, a summary o f critical issues i s presented below as background to introduce key policy recommendations. The MAMSmodel is an economy-wide simulation model created to analyze development strategies. It integrates a relatively standard dynamic recursive computable general equilibrium (CGE) model with an additional module that links specific public expenditures to poverty and other MDG achievements. The relatively detailed treatment o f government activities in MAMS makes this link possible. The model allows for the fact that growth inthe stock o f human capital (education and health) andor public infrastructure capital (including roads, energy and irrigation) contribute to overall growth by adding to the productivity of other production activities. The model also allows for complementarities between expenditures in different sectors. xvi Efficiencyand EquityinEducationExpenditures The efJiciency of education expenditures is low, and the allocations within the sector have not been rationalized 23. Significant additional financial resources are required to achieve the MDGs. Current trends indicate that Malawiwill fall short of achieving universalprimary education by the year 2015, unless both adequate policies are introduced and additional resources identified. The recurrent costs o f primary education would need to grow by 40 percent, from the current USD46 million to USD64 million by the year 2015, in order to achieve universal primary education. The MGDS does not envisage real increases in hnding to education, however. Assumingthe total recurrent educationbudget (including tertiary) remains at 5 percent o f GDP, the recurrent costs o f primary education would needto receive an increasing share o f the total budget. This is due to the growing salary envelope from the higher number o f teachers, and to an increasing share o f the teachers beingqualified (therefore earning higherwages). 24. Teacher salaries in primary and secondary education are relatively high. The Education For All (EFA) benchmark i s that teacher salaries at primary level are around 3.5 times GDP per capita, and at secondary level teachers are 6-8 times GDP per capita (and no more than 100 percent o f primary salaries). However, teachers' salaries in Malawi are comparatively high with average teacher remuneration in 2004 at about 5.2 times GDP per capita in primary and 6.2 times GDP per capita in secondary.2 Such relatively high levels o f teacher salaries in Malawi may not be financially sustainable, given the substantial increase in the number o f qualified teachers necessary to reach the education MDGs. 25. Around 50 percent of the resources spent on primary education are effectivelywasted. The cost o f producing a completer (Standard 8 student) would cost eight years o f schooling, if there were no repetition and no dropout. Infact, the cost was about 22 school years in 2004. Furthermore, over 70 percent o f drop outs exit the system inthe first four standards, before achieving lifelong literacy and numeracy, so at least 40 percent o f the `pupil years' provided by primary education are effectively wasted, and another 8 percent i s wasted on repetition. 26. Such high drop outs and repetition are due to a combination of the high schooling costs faced by poor households and the poor learning environment. Though there i s no fee for attending public primary, around 80 percent o f parents o f primary school students in public schools reported paying various expenses (DHS 2002; IHS2 2005). In addition, parents also face an opportunity cost o f sending children to school, because o f the need for their labor at home. High drop outs are also due to the low perceived benefits from education, since many primary schools lack sufficient teachers (the pupil-per-qualified-teacher ratio, PQTR, was 95 in 2004), adequate classrooms, safe drinking water, toilet facilities, and furniture (desk, chairs), such that learning outcomes are low. As discussedinChapter One, these figures have increasedby a further 50 percenton averagein2005 and 2006. However, recentlyreleasedrevisedGDP figures suggestthat GDP is around38 percenthigherthan previously estimated. Onbalance, therefore, the ratioo fwages to GDPwould onlybe slightly higher. xvii 27. Improving the PQTR is critical to raise both quality and efficiency of education. The pre-service teacher educationprogramrecently put inplace is a good step forward, but additional efforts are necessary to increase the number o f trained teachers; around 6000 new teachers are needed every year, against the current output o f less than 3000. Part o fthe problemhas beencaused by a tendency for primary teachers to move to teach in secondary schools (where salaries are higher). The pupil per teacher ratio in secondary education i s currently around 20: 1,which i s well below the government target o f 40:l. Hence, it would be feasible to raise the pupil per teacher ratio in primary by moving the excess o f trained primary teachers from secondary schools back to primary schools. In addition, teacher absenteeism i s a major problem contributing to a poor learning environment, as almost one fifth o f teachers are absent on any given day in Malawi (partially due to the impact o f HIV/AIDS). 28. Regional inequities in the distribution of education resources accentuate these problems. The great variation in the amount o f funding available to different schools does not follow any apparent criteria. Further, there i s a wide disparity in the number of teachers available to schools with similar number of pupils and even greater randomness inthe distribution o f textbooks and o f classrooms. As a result, some schools are starved o f resources and cannot function properly. Ideally the allocation o f staff and finances to the various districts should depend on predetermined criteria such as the number o f schools, enrollment levels, number o f teachers, and geographical distances for the school-age populationto attend school. 29. The distinction between relatively well-funded Conventional Secondary Schools (CSSs) and very low funded Community Day Secondary Schools (CDSSs) results in a highly(inequitableand) inefficient system. CDSSs account for almost 70 percent o f secondary enrollment in government schools. However, CDSSs have almost no facilities, lack qualified teachers, and receive a much smaller share o f funding, such that CDSSs facilities are inmuch worse condition and their PQTR i s substantially higher, compared to CSSs. On the other hand, the government i s spending on boarding facilities in a few selected CSSs, and it still finances a range o f operating costs for boarding facilities in CSSs. Although G o M officially stopped subsidizing boarding fees in 2001, the current boarding fee i s too low to cover costs, and in practice boarding i s still subsidized from the school's funds. The highly imbalanced distribution o f resources within the secondary system plays a major role in the low efficiency o f the secondary education system. More than 50 percent o f the secondary students still fail in the MSCE exam, mainly inCDSSs, indicatingthat a large share o f the resources spent on secondary education are also wasted. While there are many reasons for this performance, better funding for CDSS would contribute to substantially improve the efficiency o f the secondary education system. In order to reduce the inequalities o f resources allocation and improve secondary education quality, the status o f CDSSs should be improved to match CSSs and the formula developed to ensure a proportionate share o f funding goes towards CDSSs. 30. The HIV/AIDS pandemic currently consumes roughly 4.4 percent of government spending on education, and the cost could increase three-fold. The current total costs o f HIV/AIDS to the education system in terms o f lost teachers and absenteeism are around 4.4 percent (USD4 million) o f total government education xviii spending. This is a conservative estimate because it excludes the impacts on higher education, and on the MOE administrative staff. In 5-10 years, if teacher attrition due to HIV/AIDSrises to 4 percent, and if stipends were given to orphans, the incremental cost may reach around USD11.5million per year. Further, these estimates do not include the amount o f resources already spent on funerals, which i s estimated at around 3-5 percent (above USD200,OOO per year) o f the education ORT budget on average, and likely to be much larger share o f higher inthe primary sub-sector, where the available ORT budget i s small and the number o f teachers large. 31. Several measures are proposed to address these challenges, by rationalizingexpenditures and increasingefficiencyineducation spending: Unless the government increases the allocation to education, there is a need to reallocateresources from within education and/or implementcost savings, in order to reachthe MDGs. As discussedbelow, substantial resources could be saved in tertiary education and reallocated towards (lower) primary education. Efforts should also be made to control the level o f (primary and secondary) teacher salaries inreal terms. Within the primary cycle, lower primary education (Standard 1-4) needs to receive its fair share of resources through internal reallocation of, particularly,classrooms and teachers. The current practice inMalawi assigns one teacher per class, independento f the number o f students. However, the lower standards have a much higher enrollment; consequently, they have much higher pupil per teacher ratios. Moreover, the best qualified teachers are often assigned to higher standards. Given that 70 percent o f dropouts occur in the first four standards, it i s crucial to increase attention to quality o f education at lower levels. Improving the PQTR is critical to raise both quality and efficiency of education; this can be achieved by increasing the output of teachers, rationalizing the allocation of teachers across sub-sectors, fighting absenteeism, and intensifyingthe fight against HIV/AIDS. Additional efforts are necessary to double the annual output o f trained teachers, to approximately 6000 new teachers every year. To complement these efforts, the excess number o f under-qualified teachers in secondary could be transferred back to primary schools. The alarmingly high rate o f teacher absenteeism needs to be addressed through increased monitoring, and by raising the quality o f school management, and also possibly by introducing legislation to give communities the option o f managing their local school and the right to hire/fire teachers (Le. the State would continues paying for a certain number o f teachers depending on school size, but cannot influence the choice o f teacher). In addition, it is critical that MOE prioritize funding implementation o f the Ministry's HIV/AIDS policy, to help curb the spread o f HIV/AIDS and to mitigate its impact on the education sector. The geographicaldistributionof funding must follow an equitable financing formula, and not perpetuatepast inequities.The government education budget needs to be rebalanced between regions. The significant geographical imbalance in the provision of public education services needs to be redressed, in order to give all Malawians equal access to education. xix Teacher allocation and incentives should be rationalizedto improve equity and quality in education services. Also, the current trend of teacher migration away from rural posts must be reversed through teacher incentives (notably, improve teacher accommodation in rural posts) and a financing mechanism (Le., the money should follow the school, not the teacher). In secondary education, the emphasis of reformsshould be to reduce the gap in education quality offered at CDSSs versus CSSs. All resources, including qualified secondary teachers, should be allocated equitably between the different types o f public secondary schools to level out quality. As a priority, the physical structures o f CDSSs should be upgraded to an acceptable standard. Government should not pay for boarding facilities. The boarding school system i s inefficient (and inequitable-see below). Construction o f boarding facilities should be limited to schools where boardingi s necessary for girls (where alternative safe accommodation i s not available). In any case, the government should refrain from subsidizing boarding services, which should be outsourced andpaid for by students (see below about financial support to poor students). Education sector resources should not be used for funeral assistance. The amount o f resources spent on hnerals i s substantial and growing, and i s eating into scarce resources that are meant for education. Such support should not be taken from the Ministry's budget. At the very least the policy should be modified by replacing the current practice o f covering the hllcost with a fixed monetary subsidy. The efficiency of the education system could be also improvedby reducing student repetition, for example through automatic promotion between certain grades (inas far as improvements inthe quality o f education allows). The higher repetition and dropout rates for girls must also be addressed (see below-under improving equity). The university system in Malawi is highly inefficient. 32. University students are now 167 times more expensive than primary students, compared to only about 64 times in Zambia. While recurrent unit expenditureper primary pupil in 2003/04 was about USD17 which is slightly below the average for African countries, secondary and tertiary educations in Malawi are much more costly than average for African countries. Further, subventions to universities more than doubled in real terms from 2001/02 to 2004/05, despite the fact that university enrollment has hardly grown, increasing the annual student unit cost from about USD1,895 to almost USD4,OOO. As a result, the gap in per student spending across education levels has been growing rapidly inthe past few years. 33. Considerable efficiency gains could be made by increasing the studenthtaffratio and outsourcing boardingfacilities. Such highcosts are due inpart to the extraordinarily low studentktaff ratios at the various colleges o f UNIMA. In particular, the College o f Medicine stands out with less than 2 students per lecturer. Colleges also have an abundance o f support staff, which contribute to the high average PE/student o f MK530,OOO per year (corresponding to around 32 times GDP/capita). The xx breakdowno f UNIMA's income and expenditure shows that the bulk o f expenditure goes to administration, boarding, maintenance and other non-academic activities. 34. Wastage due to drop outs in universitiesis considerable. Further driving up the cost o f tertiary education are low survival ratios. Since 1993/94, enrollment has been around 3,500 or more; with most programs being 4-year degrees, so we would expect around 850 graduates every year. Instead, EMIS reports only about 360 graduates per year between 1999 and 2002. That brings the total cost to government o f producing one university graduate to around USD25,OOO-30,000. 35. Given the large inefficiency of the University system, the government should consider introducingthe followingreforms: Universities need to be made more accountable for the use o f public funds, and for progress in implementing needed reform, through setting up a reporting mechanism (such as semi-annual or annual reports). Considerable efficiency gains could be made by increasing the studentlstaff ratio, and reducing administrative overheads. As inthe case o f secondary education, the government should not be running or subsidizing hostels, and outsourcing boarding facilities could lead to substantial savings. There are substantial inequities in the distribution of benejhfrom public spending in education. 36. Education expenditure on primary disproportionately benefits the childrenfrom non-poor households. Despite the highprimary GER, at any given time 20 percent o f school-age children are not going to school. Further, around 40 percent of children from the poorest quintile o f households do not ever start primary school. Moreover, very few children actually complete primary, largely because their families cannot afford it, and primary schools close to poor communities appear to have substantially higher pupil-per-teacher (PTR) ratios (88 students per teacher) than non- poor communities (68 students per teacher), implying access to lower quality o f primary education for the poor. 37. The benefits from expenditure on secondary education are heavily skewed toward the non-poor.Enrollment in secondary education is low (GER is only 17 percent), and very inequitable. Three times as many non-poor students as poor students are enrolled in secondary education, and boys and girls from the richest decile are 10 times as likely to attend secondary school compared to those inthe poorest decile. Thus, the richest 50 percent o f households reap 70 percent o f the benefit from government expenditure on secondary education. Further, within the secondary sub- sector, richer students generally attend boarding schools, while poorer students tend to attend CDSS (which are under finded). However, government i s still paying for the costs o f boarding, which therefore benefitmostly students from the richer income quintile. 38. The substantial expenditures on tertiary education exclusively benefitthe richest households. Enrolment in tertiary education is only 55 per 100,000 inhabitants (i.e. 0.05 percent o f the population), and i s associated almost exclusively with households from the richest decile. Starting from 2001/02 UNIMA increased tuition fees from xxi MK1,500 to MK25,000 per academic year. However, the fee increase coincided with the establishment o f the student loan scheme covering tuition fees, benefitingpractically all students enrolled through the normal intake; to date no students have started repayment o f their loans, which students regard as a grant. Thus, inspite o f the fee hike, government i s still bearing almost the full cost o f public higher education. Considering that tertiary education i s many times more costly than secondary and primary education, it i s clear that the current system i s highly inequitable. Students from the richest quintile are receiving free tertiary education, while 40 percent o f children from the poorest households do not even start primary school largely due to lack o f money. In their university reform study (2001) the Malawi Institute o f Management (MIM) recommended that maintenance and student living services be outsourced, and that students pay for food andtransport. 39. The gender gap in primary enrollment is closing but gender parity has not yet been attained in secondary and tertiary education. The number o f hostel spaces i s larger for boys than for girls, limiting equal access to elite secondary schools. At the Universityo f Malawi, women made up only 34 percent o f students in2003 (which i s an improvement fromjust 26 percent in2000, however). 40. The financing of higher education needs urgent reform, to reduce government subsidies and considerably increase cost sharing with students who belong to the richest income quintiles. Possible measures include: The bursary scheme for secondary needs to be adequately targeted to reach pupils from the lower income quintiles, to increase their access to secondary. The intake o f girls to government secondary schools should be increased to 50 percent, by increasing the number o f places in boarding facilities reserved for girls, implementing the recently approved `pregnancy policy', and increasing the number o f female teachers. The loan collection system for tertiary education needs to be enforced, or scrapped, and the money targeted towards bursaries for the needy across the education sub-sectors. Improve efficier Rebalance educationbudget: Sequencing: Phased implementation over the next 3-4 years - Rebalancebudgetallocations from startingwith 2008/09 budget; definenew formula for allocating tertiary to primary. resources to districts in 2007/08 for implementation starting with 2008/09 budget. - Rebalancebudgetandresources (teachers and classrooms) within primary Fiscal implications: Improvedefficiencyinuse o f resources sub-sector from upperto lower primary. - Reducegap betweenCDSS versus csss. - Introduceequitablefinancing formula, for decentralizationofbudget. Improveavailabilityandmanagement of Sequencing: Phased implementation over the next 3-4 years. xxii teachers: Identify strategy to further increase output o f trained teachers - Devised strategy to further increase in2007108, with timetable for implementation. Develop plan in output o f trained teachers 2007108 to transfer surplus untrained secondary teachers from secondary back to primary over the next 3-4 years. Agree on - Transfer excess under-qualified sustainable system for `rural incentives' starting in 2008109. teachers in secondary to primary schools. Agree on action plan to fight teacher absenteeism in 2007108, - Stop teacher migration away from rural and implement monitoring system in 2008109, possibly posts through (i) teacher incentives, and entailing the introduction o f legislation to give communities (ii)adoptanewfinancingmechanism. right to hireifire teachers. Approve raise in qualifications for school management in 2007108. Start funding and - Reduce teacher absenteeism by implementation o f the HIV/AIDS strategy immediately with increased monitoring, raising the quality 2007108 budget. o f school management, and/or by giving communities the right to hireifire Fiscal implications: Substantial costs may be needed in teachers. developing additional teacher training capacity and introducing rural incentives (depending on strategies selected). All policies - Prioritize implementation o f HIV1AIDS expected to improve quality o f education and efficiency in use policy o f resources. Revise funeral assistance policy such Sequencing: Revise funeral policy in 2007108 and implement that resources are not taken from the starting in2008109. Ministry's budget, and a re paid directly from DHRMD.Also, limit funeral Fiscal implications: Substantial cost savings (estimated at 3-5 benefits to a fixed monetary subsidy. percent o f education ORT). Improve efficiency o funiversities: Sequencing: Phased implementation over the next 3-4 years. - Increase studenvstaff ratio, and take Agree on reporting format and mechanism in 2007108. measures to reduce administrative Implement in2008109. overheads inuniversities Fiscal implications: No costs. Improved efficiency in use o f - Set up annual reporting mechanism for resources. universities on their use o f funds. Improve eq ?yin the distribution of resources Increase boarding fees in secondary Sequencing: Review fees in 2007108 and implementation education. starting in2008109. Fiscal implications: Improved equity in use of resources as funds from cost recovery canbe channeledto bursary fund. Review bursary scheme for secondary to Sequencing: Review system o f allocation o f bursaries to adequately target poorer pupils. introduce means testing in 2007/08. Fiscal implications: Improved equity in use o f resources as funds from cost recovery can be channeled to bursary fund. Increase fees intertiary and enforce loan Sequencing: Review fees in 2007108 and implementation collection system. starting in 2008109. Carry out rapid review o f loan administration system to improve transparency and loan collection in 2007108, and implement recommendations in 2008109. Review system o f allocation o f loans 1grants based on means testing in 2007108. Fiscal implications: Improved equity in use o f resources as funds from cost recovery canbe channeledto bursary fund. Increase intake o f girls to government Sequencing: Phased implementation over the next 3-4 years secondary schools to 50 percent. Fiscal implications: Improved gender equity in use of resources. xxiii ... Limit secondary boardingto girls and Sequencing: Phased implementation over the next 3-4 years vulnerablegroups. Fiscal implications: Improvedequity inuse o f resources. Efficiency and Equity in Health Expenditures Supply stock-outs and a lack of trained health workers, are major sources of inefficiencies in theprovision of health services. 41. Most of Malawi's health facilities are not capable of delivering essential health services. Access to health services in Malawi i s distributed with considerable equity across socioeconomic groups. However, the quality o f services i s extremely low, reflecting both lack o f resources and severe inefficiencies. Just 9 percent (54 out o f 585) government and missionhealth facilities are capable o f providing the essential package o f health services (EHP) on-site (only 1or 2 facilities ineach district). 42. The first major constraint is severe lack of trained health workers; but action is being taken. The scarcity o f skilled staff such as physicians, nurses and midwives, i s manifest in high vacancy rates (percent o f unfilled posts) ranging between 30 to 80 percent o f skilled posts in government (and to a lower extent also in CHAM) health facilities. These alarmingly high vacancy rates impede the delivery o f even basic health services. An aggressive recruitment policy for new skilled health staff has been adopted in 2005: the `Six Year Emergency Training Plan' (ETP). As part o f the ETP, remuneration packages for skilled staff have been increasedby introducing salary top-ups o f 52 percent on average for skilled health staff. In addition, the ETP also aims to increase the number o f training places, to triple the number o f doctors and double the number o f nurses intraining. The adoption o f the ETP has already resulted inincreased hirings and a slow down in the out-migration o f nurses from Malawi. This program is addressing a critical bottleneck inhealth services provision, but its financial sustainability after the initial six-years period depends critically on continued donor support. Although the ETP has already received significant government and donor support, funding for the salary top-ups beyondthe six-year period remains to be secured. 43. The lack of skilled health workers and overloading of the available staff results in low quality of services; it is not uncommon for less-skilled workers to do technical and clinical work for which they are not trained. Due to the shortage o f skilled medicalworkers, the workload has skyrocketed. For example, inantenatal care, a typical health worker faces as many as 100 patients per 8-hour day, implying an average encounter o f 4.8 minutes per patient, compared to the standard 30 minutes. It should be noted, that while health facilities suffer from severe shortages o f skilled staff, there are significant surpluses o f unskilled staff. Inresponse to the heavy workloads, trained staff are `down-loading' clinical tasks, with weak and infrequent supervision, which also has serious effects on quality o f care. For instance, a study inone health facility found that 20 out o f 34 deliveries (or 59 percent) were attended by ward attendants. 44. The second major constraint is unavailability of even basic drugs, mainly due to slow progress in reforming the Central Medical Stores (CMS). Malawi drugs procurement and distribution system remains almost solely reliant on the CMS, which has been plagued for many years by institutional, management, and financing problems. xxiv CMS suffers from a lack o f qualified staff, poor planning, poor logistics, and weak support systems. Poor stock management and storage often results in health centers receiving medicines not ordered, or outright non-delivery o f ordered medicines. Organizationally, CMS i s under the MOH and lacks the autonomy to operate as a commercial enterprise. An action plan has been developed outlining the short-term steps to change CMS management, including international recruitment o f key positions. In 2006, the government awarded a contract to hire an external management team for Central Medical Stores. 45. Tertiary hospitals are hobbled by lack of autonomy. The granting o f autonomy to central hospitals has been in discussion since the mid-1990s. Inaddition to insufficient financial and human resources, the limited control that tertiary hospitals have over their operations i s hampering their performance, notably in the key functions o f strategic management, procurement, financial management, human resources management, and administration. Further, the lack o f autonomy has resulted in stalled fee reforms, such that current fees are grossly inadequate to cover the cost o f service provision(and correspond to less than 20 percent o f the actual cost). 46. With a national prevalence rate of 14.7 percent in 2005, Malawi is the eighth worst affected country in the world from HIV/AIDS, and the overall burden of disease i s very substantial. HIV-related conditions now account for around 40 percent o f all inpatient admissions and about 70 percent o f admissions to medical wards. HIV co-morbidity (tuberculosis, other opportunistic infections) is also increasing. HIV/AIDS has substantial adverse effects on the government's fiscal situation. The epidemic i s a large drain on scarce domestic resources, compounding the constraints to development that the country faces. Further, HIV/AIDS reduces the availability o f health workers. 47. The medium-term sustainability of HIV/AIDS programs is not assured, even using rough estimates that arguably understate the resource requirements. There i s an inadequate funding stream to meet the projected cost o f Malawi's current goals and commitments towards HIV/AIDS programs. Financial simulations indicate that international funding and domestic availability for HIV/AIDS related expenditures will be adequate to meet the needs as estimated until2008. However, from around 2009- 2010 Malawi would need to have sourced around USDlOOmillion annually in additional funding (either from donors or domestic resources) in order to maintain existing programs; the cost o f expanding ART coverage in line with government targets i s substantially higher, rapidly rising above USD200 million annually even using very conservative estimates. 48. To address these challenges, several measures could be considered to increase efficiency of expendituresin health: Additional action is required to increase the level of skilled staffs. Fundingfor the outer years of the Emergency Training Plan needs to be identified, and government should request more donors to support the emergency training and recruitment plans. In addition, the government should negotiate with the Global Fund to include a major human resources component in its Round IV proposal. Some resources can also be saved by gradually reducing the above-establishment xxv numberso f surplus unskilled staff. Inaddition, there is a needto develop a system for tracking trainees and recruiting them at the completion o f their training (only 2 percent o f the 309 initial set o f trainees under the ETP joined the government health services, while the rest went to C H A M and NGOs). Inthis context, formal bonding arrangements should be introduced as a requirement for those whose training has been paid (or subsidized) by the government. Further, in order to promote staffing in hard-to-fill posts inrural settings, there i s a needto consider a system o f staff incentives (notably, improve accommodation in rural posts) and a financing mechanism (Le., the money should follow the health post not the staff). The potential for public-private partnerships should be better exploited, notably with CHAM, given the scarcity of human resources in government facilities. A more unified and comprehensive contract betweenthe government as financier and the C H A M facility (or other NGO) as service provider, covering staff salary subventions from MOF, provision o f drugs, and service agreements for other recurrent transactions (ORT), would help to ensure greater cohesion in the financing o fgovernment and C H A M facilities. Community-based health services should be expanded and strengthened. In addition, the government should pursue expansion o f community-based programs for essential health services (e.g., community-based `directly observed treatment, short course' DOTS for tuberculosis, family planning services relying on community-based distributors, distribution o f bednets); however, success will depend on the availability o f health commodities, which hinges critically on CMS reforms. Reformsof the CMS should be accelerated, and inparallelthere is an urgent need to improve the procurement and delivery systems for essential drugs. CMS should undergo comprehensive institutional reform Donors have supported GOM to convert CMS into a non-profit trust with an independent board of trustees, but government is yet to make a decision on the status o f CMS. Procurement and drug tracking systems should be rationalized to improve drug stream and realize the cost saving benefits o f bulk procurement. Switching to a demand-driven (requisition) system embodying principles o f a `cash-and-carry' financing which define the Tanzanian and Zimbabwean reformed drug systems would achieve this. Drug registration and quality assurance and control mechanisms should be clarifiedand revised. Guidelines for essential drugs should be improved inorder to decrease use o f inappropriate therapies, and wastage. Tertiary hospitals should be reformed to improve quality of care and cost recovery. The hospital autonomy bill, which envisions making tertiary hospitals legal corporate entities under the direction and control o f a hospital board, should be enacted and its provisions implemented.Cost recovery at tertiary care facilities should be improved through design o f a sliding scale fee structure that keeps up with current prices, but accounts for the variety o f patients that tertiary hospitals have. A Poverty and Social Impact Analysis should be carried out to inform the new fee structure, to ensure continued access to the poorest. xxvi Medium-term financing to pay for costs of meeting the GOM's target in HIV/AIDS programs needsto be urgentlyidentified,as availablefunding will run out within the next 2-3 years. By 2009-2010 at the latest Malawi would need to have identified additional funding, either from international donors or from alternativedomestic sources (domestic revenue or the private sector). Inthis context, the government should apply for the new Rolling Continuity Channel of the GlobalFundinNovember2007. Improve efjcie v and rationalization of resource use Increaselevelsof skilledstaffs: Sequencing: Approach donors and Global Fund in 2007108 to - Proceedwith implementationo f seek funding for salary top-ups after 2011, andor devise emergencytrainingplan strategy to deal with the facing out o f top-ups. Commission study to explore possible mechanisms for bonding - Seek funding for salarytop-ups after arrangementsin2007108 for introductionin2008109. Agree on 2011. sustainable system for `rural incentives' startingin2008109. - Introduceconditions for those being Fiscal implications: Urgent need to resolve fiscal risks subsidizedby trainingprograms, such as associated with current salary top-up program after 2011. formal bondingarrangements. Savings as costs o f training staff are repaid in kind though - Facilitate staffing inhard-to-fillposts bondingarrangements. inruralsettings through (i) staff incentives(accommodation)and (ii) adopt a new financing mechanism. Removebottlenecksindrugs Sequencing: Request management team to develop options for availability: future restructuring of CMS during 2008. Rationalize drug - Improveprocurementanddelivery stream by switching to a demand-driven (requisition) system systems for essentialdrugs during 2008. - CMS shouldundergo comprehensive Fiscal implications: Substantialreductionin inefficiencyand institutionalreform. leakage at CMS. Substantialcost savings frombulk procurement. Tertiary hospitalsshouldbe reformedto Sequencing: Enact hospital autonomy bill in 2007 and improve quality o fcare and cost implementprovisionsin 2008. recovery. Fiscal implications: increased cost recovery from higher fess for hospitaltreatment Strengthen partnerships:Developa more Sequencing: Evaluate government's experience with unifiedandcomprehensive contract establishingservice agreements with CHAM hospitalsin 2008; betweenthe governmentandthe CHAM design moreunifiedand comprehensivecontract in2008. facility (or other NGO)as service provider. Fiscal implications: None Urgentlysecure additionalfinancing for Sequencing: Develop financing strategy to raise required HIV/AIDS. financing, either fiom international donors or from alternative domestic sources (domestic revenue or the private sector) in 2007108. (NB: This strategy should include, but not be limited to, applying to the new Rolling Continuity Channel of the GlobalFundinNovember2007). Fiscal implications: Urgent need to clarify fiscal implications associatedsustainability of currentART programs. xxvii Efficiencyand EquityinNutritionExpenditures 49. Malnutritionis the most severe challenge facing Malawians,irrespective of their income status and level of calorie consumption. Nationwide, a staggering 48 percent o f preschoolers are stunted (of which 18 percent are severely stunted), largely irrespective o f household income and level o f caloric intake. Worse, these figures have remained more or less static for the last 15 years. Such exceptionally high levels of malnutrition have persistent long-term impacts, as malnutrition diminishes future productivity, thus perpetuating their vulnerability to poverty traps in the future. Child malnutrition appears not to be highly correlated with poverty levels (and with caloric intake) at the household level, suggestingthat other factors needto be tackled to eradicate it. Evidence from the 2006 Poverty and Vulnerability Assessment (PVA, based on analysis o f the 2004/05 Integrated Household Survey) shows that, in addition to income, interventions need to be geared towards improvements in nutritional and feeding practices, as well as water and sanitation infrastructure. 50. The cross-sectoral nature of the problem makes nutrition interventions difficult to coordinate and difficult to measure. Many ministries, donors and NGOs are involved in nutrition-related activities in Malawi, which may be part o f health, agriculture or other sectoral programs. This has made coordination difficult, and has also diminished the impact o f the interventions. The appointment in 2006 o f a Principal Secretary for Nutrition, HIV and AIDS, directly under the Office o f the President and Cabinet, has yet to produce new programs and more government hnding to the sector, butcouldundertakethis role. 51. Very low resources are spent to prevent malnutrition,when considering the severity of the problem. Annual expenditureson direct nutrition interventions are about 1.5 percent o f GDP, almost entirely funded by donors. Government spends only about 0.05 percent o f GDP (USD300,OOO) on nutrition programs, in spite o f the fact that one in two children under-5 are malnourished. For comparison, government spends annually approximately USDl.5 million on feeding students inboarding facilities at the two public universitiesand inelite secondary boarding schools. This i s in spite o f the fact that secondary and, in particular, tertiary students are known to belong to the highest socio-economic quintiles inMalawi. 52. The current portfolio of nutrition interventions in Malawi does not prioritize cost-effective interventions. In spite o f the fact that supplementary food is not a cost-effective strategy to prevent and thus reduce maln~trition,~more than one third o f all expenditures are on therapeutic and school feeding programs. When funding to `feeding programs' and other curative interventions are subtracted, the resources spent to prevent malnutrition from occurring are even more limited. Almost half o f nutrition expenditures (44 percent) in Malawi are curative in nature, and can thus not expect to have any lasting effects on improving feeding practices, which i s a major cause o f malnutrition. This compares to expenditure o f just 1 percent on the highly cost-effective Other benefitsare derivedfrom school feeding, namely, reducedrates of dropout, andimprovedlearning. Neverthelessthe cost o f suchprogramsis very high. xxviii measure o f micronutrient supplementation (of vitamin A to children under 5 and iron to pregnant women). 53. Interventions are not targeted at the most vulnerable. Young children, particularly children less than 2 years o f age, and pregnant and breast feeding women have the highest nutritional requirements o f any age group and are generally more vulnerable to malnutrition than the rest o f the population. In Malawi, no interventions specifically target the under-2s, which i s the period when most malnutrition occurs. About a third o f nutrition expenditures target young children, pregnant women and mothers o f young children, but half o f all expenditures are not targeted to any group. 54. To address these challenges, several measures could be considered to increaseefficiency and equity of expendituresinnutrition: The considerable share of nutrition expenditure invested in programs that are costly and not expected to have a lastingimpact should be scaled backin place of more cost effective strategies. Inparticular, feeding programs that are untargeted (food aid distribution, school feeding programs, etc.) have been shown not to be cost effective in reducing malnutrition in school-age children and certainly do not reach the primary target group o f childrenunder-2 years o f age. A national community nutrition program would target the priority age groups in a low cost and effective manner. Improving the nutritional status of the under-5s (and particularly under-2s) inMalawi should be a key priority inthe coming years. Such a program, modeled along the lines o f the Honduras Community-Based Integrated Child Care Program (see Chapter Five) could channel more resources into the prevention o f malnutrition among under-5s, and would be possible to sustain due to its relatively low cost (USD6-7 million per year). This would not all be additional costs, however, as it would replace some existingservices. Micronutrient supplementation is highly cost-effective and should be boosted. However, it may not be sufficient to deliver the micronutrient supplementation to young children through the primary health care system, as not all children under 5 come into regular contact with clinics and other health care providers. Moreover, any national fortification program must take into account the fact that many Malawians buy only a limited number o f food stuffs and most maize meal i s not centrally processed. Diet diversification and education about good nutrition practices must be part of the response to combat malnutrition. There is a need to disseminate knowledge about the benefits o f consuming new foods that are rich in micronutrients (and how to prepare them), and small livestock. In addition, nutrition education programs are needed in order to improve the distribution o f food among household members and give information to caregivers and other family members about good practices for feeding young children. Improvedcoordination among government and donors, and across sectors, is needed to tackle the daunting challenge of malnutrition in Malawi. The government should take the lead to rationalize and coordinate the many existing xxix programs addressing nutrition within a prioritized set o f interventions, in order to provide an effective response to the malnutrition crisis. A research agenda on the determinants of malnutrition in Malawi needs to be pursued in order to design evidence-based policies and programs. Evidence from the 2005 IHS2 highlights that child malnutrition is not highly correlated with poverty levels (and with caloric intake), suggesting that other factors need to be tackled to eradicate it. However, there i s a need to deepen the understanding o f non-income factors, and what policies can be introduced to mitigate them. POLICYPRIOR^ Improve efficiel y and rationalization of resource use Scale back nutrition expenditure invested Sequencing: Discuss with development partners what type o f infeeding programsthat are untargeted nutrition interventions ought to be prioritized, and agree on a (food aid distribution, school feeding refocusing plan in 2007/08, for implementation in a phased programs, etc.). manner over the next 3-4 years. Fiscal implications: No additional costs; improved efficiency inuse o fresources. Introduce a national community nutrition Sequencing: Develop program in 2007/08 and identify program focused on under-5s (modeled funding, to start implementation in2008/09. along the lines o f the Honduras Program or other community-based nutrition Fiscal implications: Less than U S D10 million, to be sourced as programs), and including nutrition part o f the refocusing exercise. education programs to teach mothers about good practices for feeding young children. Expand program o f dissemination o f Sequencing: Prepare and carry out nationwide nutrition nutritionknowledge (new foods and how education campaign in2007/08 and 2008109. to prepare them). Fiscal implications: Less than U S D l million, to be sourced as part o f the refocusing exercise. Undertake a program to increase Sequencing: Develop national fortification program in 2007108 micronutrient supplementationto young and begin implementation in 2008/09. children. Fiscal implications: Depend on extent o f program activities, butto be sourced as part o fthe refocusing exercise. EfficiencyandEquityinRoads Expenditures 55. The road network is heavily imbalanced: a small network of main roads i s mostly paved and in good condition, and the rest of the network i s unpaved and mostlyin poor condition. The paved network is almost entirely ingood or fair condition (98 percent). However, the greater part o f the network i s unpaved (74 percent), and about 40 percent o f the unpaved roads remain in poor condition. In fact, the majority o f the roads in poor condition are the rural roads. Further, the condition o f the network, whilst not deteriorating, has not improvedover the last six years. xxx 56. Funds spent on routine maintenance, have been inadequate to meet the maintenance needs of the network, and to clear the sizeable backlog of maintenance. The official policy in the road sector i s to first maintain the existing road network followed by rehabilitation, upgrading and construction o f new roads. In practice, however, full maintenance o f the network has never been funded, and there continues to be a bias towards financing rehabilitation, upgrading and construction o f new roads at the expense o f systematic under funding o f routine maintenance. To a lesser extent, the problem is also due to the fact that the revenues from the Road Fund (almost entirely originating from fuel levies), which i s the main source o f funding for maintenance activities, are limited inscope. 57. There is inadequate planning and prioritizationin road maintenance and construction activities. Decisions on road maintenance and construction have traditionally beenmade on an ad hoc basis, and project selection and planning have been overshadowed by political interference. A computer-based model has recently been developed (the `Road Data Manager') to calculate the priority interventions in terms o f maintenance and road construction, in order to allow decisions on road maintenance and construction on thorough project analysis and appraisal. 58. There has not been a poverty focus in the road sector expenditure, and access to roads is heavily skewed in favor of the richest households. The poor tend to live in remote areas with limitedroads and/or means o f transportation, which constrains their access to markets and trading centers, and limits their economic opportunities. Further, roads inrural areas are often impassable, on average for up to four months inthe year, such that sizeable portions o f the population remain isolated from the rest of the country for substantial amounts o f time. This emphasizes the need to be addressed to increase the condition in the overall network and ensure rural accessibility and mobility. As discussed, more funds are beingspent onrehabilitation, upgrading and construction o f new roads compared to routine maintenance. There is thus an imbalance between funding for the paved and unpaved roads. More expenditure i s focused on paved road in order to preserve the high value o f the asset much to the detriment o f the unpaved road network (and poor households). 59. To address these challenges, several measures could be considered to increaseefficiency and equity of expenditures in roads: Allow for regular review and adjustments in fuel levy to increase funding in Road Fund. The fuel levy i s the most consistent source for maintenance funding. The Road Sector Programindicates that a fuel levy equivalent to at least 16.5 US$ per liter (an average o f 14 US$ for diesel and 19 US$ for petrol) will ensure adequate funding for routine and periodic maintenance for the entire network. The increase should be carried out gradually, however, and poverty and social impacts o f raising funding from the Fuel Levy should be assessed in comparison to alternatives. To ensure adequate funding for the future, it i s important that the FuelLevy rates are set inUSD, and the Malawi Kwacha amount is updated on an annual basis. It i s also important that funds levied from the road users through the fuel levy are utilized for the sole purpose o f maintaining the roads, which has beena probleminthe past. xxxi Adopt the recently developed Road Data Manager to ensure planning and prioritization in road maintenance and construction activities. This is especially important as government i s planning to expand expenditures on roads. The introduction o f the Road Data Manager system should assist inprioritization and allocation o f funds more efficiently, by basing decisions on roadmaintenance and construction on thorough project analysis and appraisal. It will be critical that the Road Authority i s able to operate autonomously o f political interference, in the selection o fpriorities inmaintenance and new construction. Funding for routine and periodic maintenance is the top priority, and must be adequate and consistent, includingif neededfrom donor inputs. There is a need to fully fund routine maintenance o f at least the core network and carry out systematic periodic and backlog maintenance on the entire network, to bring the whole network to maintainable condition. If there is not enough funding available, decisions will need to be taken as to what size o f network can realistically be managed based on available resources, and on the amount o f rehabilitation to be undertaken. The sustainability o f the sector i s in question when what is being rehabilitated cannot be or is not maintained. Given that internally generated funds for the sector are insufficient, it will be necessary to prioritize any external funding towards covering backlog and periodic maintenance before reconstruction. Currently most donor funding i s negotiated bilaterally, with emphasis on particular roads rather than the improvement o f the network as a whole. A sector wide approach will create opportunities for donors to have a global picture o f the sector and focus funding on priority programs. M o r e attention and finances need to be spent on the rural roads. A more equitable balance o f finance needs to be established between rural and main roads. MOTPW and the NRA need to work with MLGRD to identify an appropriate mechanism o f sourcing and allocating funds for road maintenance at district level, increasing the share o f funding from the Road Fund devoted to district roads, and tertiary roads. Increase the capacity of domestic construction industry to meet demands. The NRA 5yr strategic plan estimates that about USDSOO million will need to be spent over the next five years to bring the network up to a standard where 70 percent o f the network i s in good condition. This requires about 11,OOOkm o f roads to be rehabilitated. The local contracting capacity will need to be strengthened and opportunities for their participation in the works need to be created. Possible actions include, increasing training activities with the National Construction Association, privilege foreign companies who enter into joint ventures with local companies in bidding selection process for contracts, and consider expanding the `term maintenance contracts' (which are mainly for routine maintenance) to cover larger areas and longer time periods, in order to allow smaller companies to build up their capacity. xxxii Improve efficiei y and rationalization of resource use 4llow for regular review for adjustments Sequencing: Increase fuel levy gradually, starting with 8 US$ nfuel levy to increase fundinginRoad and 10 US$ in 2007108; then carry out PSIA before further 7und. increases. Fiscal implications: Increase level o f revenues from around USD14 to around USD30 million initially, and possibly to USDSO million insecond phase, after PSIA. 4dopt a system to ensure planning and Sequencing: Introduce use o f Road Data manager as aprt o f the xioritization inroadmaintenance and procedures envisaged by the law, to be prepared and approved :onstmction activities. in2007108. Fiscal implications: No costs; but improved efficiency in the use o f resources. Funding for routine and periodic Sequencing: Prioritize finding for all maintenance starting in naintenance i s the top priority, and must 2008/09 budget, including from donor resources if necessary. 3e adequate and consistent, including if needed from donor inputs. Fiscal implications: Cost o f maintenance o f existing network is estimated at around USDSO million, which can b e identified from a refocusing o f sector expenditures. Equity I the distributionof resources More attention and finances need to be Sequencing: Revise formula to allocate maintenance spent on the rural roads. expenditures between paved and unpaved network in favor of unpaved network in 2007108 (jointly with increasing overall share o f sector resources towards maintenance). Fiscal implications: No costs; but improved equity o f benefits distribution o f public investments inroads. Increase the capacity o f domestic Sequencing: Adopt clause to facilitate involvement o f local Gonstmction industry to meet demands industry in future rehabilitation and construction contracts, starting in 2007108. Revise system o f contracts for road maintenance to increase length and amount o f contracts starting in2007/08. Fiscal implications: No costs; but increase share o f public resources benefiting local companies. xxxiii xxxiv CHAPTER1: RECENT MACROECONOMIC DEVELOPMENTSTHEINTER-SECTORAL AND ALLOCATION OF RESOURCES 1.1Introduction 60. Malawi's macroeconomic performance since 1999 has been characterized by a rapid increase in the level o f domestic debt between 1999 and 2004, and a remarkable turnaround since then. Increased indebtedness during 1999 and 2004 has resulted in a high interest burden, thus compromising the government's ability to allocate resources for critical poverty alleviating expenditures. Since 2004 a number o f reforms have been implemented, domestic debt has declined, and the outlook is promising. The reductiono f interest payments on domestic debt remains a key priority inthe fight against poverty. It will generate much needed fiscal space to increase other recurrent and development expenditures, such as in health, education, social safety nets, agriculture, irrigation, and transport infrastructure. The government needs to prioritize how best to utilize these resources. 61. This chapter briefly reviews the main macroeconomic developments since the late-1990s, with a focus on the rapid increase indomestic debt during 2000-2004. It then examines the causes o f the increase in domestic debt, with a view to identify reforms to prevent a recurrence o f the crisis. The chapter then discusses the composition o f public expenditures, in the context o f the recently adopted Malawi Growth and Development strategy (MGDS). The chapter examines the implications o f alternative expenditure strategies for growth and poverty reduction, and in particular, it explores the impact o f changing the balance o f expendituresbetween economic and social sectors. The chapter concludes by providing some lessons and recommendations, and issues for hrther analysis inthe PER. 1.2. Main Macroeconomic Developmentsbetween 1999 to 2006 Malawi on the Verge of a Fiscal Crisis 62. Macroeconomic performance between fiscal years 1999/00 and 2003/04 was not satisfactory, but it has improved markedly after June 2004. Sustained growth has been elusive, reaching no more than 2 percent average during the period between 1999 and 2005,4 with significant volatility from year to year (Table l.l).5 Much o f the poor performance has been the result o f the recurrent weather shocks on smallholder agricultural production, notably maize production. Malawi's agriculture NationalAccounts estimates havebeenrevisedin2007. However,revisedestimates were notyet availableat the time ofwriting, andtherefore all NationalAccounts figures inthis report are based on the oldnationalaccount series. It shouldbepointedout, however, that the preliminaryestimates from the revisednational accounts indicatethat the levelof GDP has beenrevisedupwardswith an average o f around38 percent. Giventhat populationgrowth is around2 percentper annum, this implies no growthinper capitaterms (Le.that poverty has stagnated). Around 6 percentgrowth is requiredto make a meaningfulimpacton povertyreduction(WorldBank2003). 1 Figure 1.1: Maize production and GDP growth in Malawi, 1985-2005 = 20 200 2 & 15 150 e 10 100 g 2 =2 o5 50 sgm 0 -5 -50 s2.22e n a -10 -100 N a, 0 0 -15 -150 I I I G D P Growth Rate +Maize Production Growth Rate Source: NSO and IMF Statistics constitutes about 40 percent o f the economy (and more considering that most manufacturing i s agro-industry), largely smallholder farming and mostly rainfed agriculture. As a result, weather patterns deeply affect agricultural production and Malawi's GDP (Figure 1.1). Most recently, the adverse weather in2001, 2002 and 2005 brought dramatic swings inagricultural output over the last five years.6 Inaddition to the lackluster performance o f agriculture, the poor economic growth has been exacerbated by an even worse performance in manufacturing, partly due to exceptionally high government borrowing that `crowded out' financing to the private sector. 63. Weak fiscal performance between 1999100 and 2003104 brought the country to the verge of a financial crisis. The government ran large fiscal deficits o f more than 7 percent o f GDP in each fiscal year. Given the substantial reduction in external budgetary aid,7 the deficits were financed largely by domestic borrowing. Consequently, the domestic debt increased sharply from less than 3 percent in 1999/00 to 25 percent o f GDP in 2003/04.8 Further, the increase in government borrowing (from local banks) pushedup interest rates. The combination o f the increased amount o f debt and the high interest rates meant that government's domestic interest bill shot up to a massive 9.2 percent in2003/04 (Table I.Iand Figure 1.2). Malawi's economy is highlyvulnerable to adverse weather shocks not only because o f the direct impact on agricultural production (and GDP), but also due to the indirect impact on government finances resulting from the unanticipated need for emergency interventions (which may translate into increased domestic borrowing), and increased pressure on the current account because o f the need for exceptional food imports. Further, the long-term impact o f the volatility is substantial, as economic uncertainty hampers the country's ability to generate and/or attract productive investments, to effectively compete ininternational markets, and, most important, to translate economic growth into employment and income generation that benefit those who need it most. Following the weak fiscal performance in 2000/01, and in light o f the authorities' decision to halt the privatization program, the IMF suspended its PRGF program inNovember 2001. Other donors followed suit by stopping their budget support grants and concessional lending. * Malawi's fiscal year run from July 1st to June 30fh. 2 Figure 1.2: Malawi Budget and Fiscal 64. Nominal and real interest Performance at a Glance, 2000-2006 rates have been extremely high, hurting private sector growth, Malawi's fiscal performance, 2000-2006 during 2000-2004. The high levels o f domestic borrowing pushed up interest rates, such that the real interest rate has been above 20 percent between 2000 and 2004, peaking at almost 40 percent in 2003 (Figure 1.3). Such real interest rates were amongst the highest in the world. High interest rates in Malawi have generally been attributed to a combination o f the high levels o f Fiscal years government borrowing, high inflationary expectations, and the Source of deficit financing (as %GDP) highly oligopolistic structure o f the 14 financial sector (World Bank 2004; b '* :100 Mlachila and Chinva, 2002). The high m - interest rates have effectively made it $ 6 almost impossible for farmers and 6r 4 business to borrow, with deleterious 2 0" 2 effects on economic growth. In fact, I 0 government borrowing effectively -21 'crowded out' the funds for private investment in Malawi which dropped from 4.4 percent o f GDP in 1999 to 1.5 percent in2003 (Figure 1.4). Domestic debt (as %of GDP) 65. Malawi experienced fairly high inflation with substantial volatility over the past decade. The relaxation o f monetary policy in the run up to the second multiparty general elections in 1999 resulted inan increase in inflation from 9 percent in 1997 to 45 percent by 1999. Inflation was subsequently brought under control with tight monetary policy Source: Ministry o f Finance and RBM, and IMFand during 1999-2002, reaching 10 percent World Bank staff calculations by 2003. The escalation o f the debt crisis, however, led to increasedmonetization o f the budget deficit during2003 and 2004. Inturn, the expansion inmoney supply fuelled the resurgenceof inflation to a peak of 17 percent inFebruary 2006. The recent increase ininflation was partly compounded by the low maize harvest due to drought in2005, and the soaring world energy prices. 66. The current account deficit (excluding official transfers) also worsened over the period, from an average of 14 percent of GDP in 1999-2002 to an average of 24 percent in2003-2006. The deterioration i s partly due to the higher oil prices inrecent 3 Figure 1.3: Movements inInterest Rate, 1992-2005 years, and has been compounded by the need Nominal Interest Rates (annual average) for exceptional food imports in 2005 as a result o f the drought. The current account deficit including grants, however, improved from 10 percent in 2004 to 7 percent in 2006, as a result o f increasing grants. 67. The exchange Note: The Reserve Bank of Malawi (RBM) sets the discount rat rate vis-&vis the US which guides commercialbanks and is also reflected inthe interest dollar has depreciated by rates for Treasury Bills. Interest rates in Malawi are market- around 100 percent in determined, with commercial banks largely free to set their own lending and deposit rates. The Prime Lending Rate is the interest nominal terms during the rate charged by commercial banks to their most creditworthy five year period. Until customers (usually the most prominent and stable business 2003 the RBM intervened customers). The rate is almost always the same amongst major solely to manage the extent banks. o f exchange rate fluctuations. In more recent years, however, the government has pursued an exchange rate stabilization policy, apparently to prevent the rate fiom depreciating during the `import ~eason'.~The result has been a substantial loss o f reserves, and more recently forex rationing and an increase inblack market trading, which have accompanied periodic one-offadjustments to realign the exchange rate. The foreign reserves position has deteriorated throughout the period, with the level remaining below two months of import cover since 2002. The level o f reserves was below 1.5 months at the end o f 2005, down from almost 4 Figure 1.4 Private Investments as percent months o f import cover at the start o f the period. of GDP, 1999-2006 68. In addition to poor fiscal performance and external shocks, economic growth has also been significantly hampered by the 4 0 impact of HIV/AIDS. The adult a prevalence rate o f HIV/AIDS in Gs Malawi at the end o f 2004 was 2 0 estimated at 14.2 percent, corresponding to around 900,000 infections. The total number o f 0 0 orphans i s currently estimated at I 1999 2000 2001 2002 2003 2004 2005 2006 I Source: RBMand IMFdata I Malawi'strade balanceis usuallyundermostpressureduring Septemberto January whenmost of the agriculturalinputs and food importstake place. Onthe other hand, the trade balance is usually strongest duringthe April to Septemberwhen the tobacco exports,whichaccount for almost 60 percentof Malawi's exports, are concentrated. 4 around 1 million, o f which about half are thought to be directly due to HIV/AIDS. Studies on the evolution o f the epidemics estimate that by 2010, the number o f people with HIV/AIDS may reach 2 million. In addition, life expectancy is expected to drop further from 38 to 35 duringthe nextten years. Most studies estimate that GDP growth is adversely impacted in the region o f 1-2 percent per year, through lower labor productivity, lower public and private savings, and reduced capital formation, as public expenditures are diverted to meet growing health demands (see Chapter Five). While the magnitude o f the effect o f the epidemic on GDP growth and per capita income differs in various studies depending on the model assumptions, there i s strong consensus that the epidemic i s having a significant adverse impact on growth and poverty. Recent developments:the turnaroundafter 2004 69. The Governmentand the RBMhave taken rapid steps since June 2004 to stabilize the fiscal situation and reign in liquidity growth and inflation. The administration has enforced strict fiscal discipline and, as a result, macroeconomic performance over the last three years has been rapidly improving.loInflation declined from 17 percent at end-February 2006 to 9 percent at end-March 2007. Improved fiscal performance combined with donor budget support provided for a gradual reduction in domestic debt stock to 20 percent o f GDP in2005/06. The domestic interest bill has also decreased to 5.5 percent o f GDP in 2005/06 (or 23 percent o f revenues). This was achieved, despite the severe food crisis in 2005, through fiscal restraint, lower interest rates, as well as higher budget support from donors. With inflation and government borrowing declining, the Reserve Bank has gradually reduced the nominal discount rate from 45 percent in September 2003 to 20 percent inNovember 2006, and i s assessing the conditions for further reductions (Figure 1.3). l1Private investment has since increased back to 3.7 percent of GDP in 2006, as interest rates have been declining in recent years (Figure 1.4). The nominal exchange rate has remained fairly stable vis-a-vis the U S dollar (at around MK 138 = USD I), as a result o f RBMinterventions in2005/06 initially (see above), and more recently reflecting a stronger real exchange rate (from HIPC completion point and MDRI debt relief). Gross foreign exchange reserves at the Central Bank remain very low, with import cover at 1.6 months o f imports as o f December 2006. 70. Malawi reached the HIPC Completion Point in August 2006 and subsequently qualified for the MDRI. This resulted in a decline o f Malawi's debt-to- exports ratio from 229 percent to 32 percent. Malawi's Fitch Credit Rating has also been upgraded from CCC to B-, raisingprospects for private capital inflows. 71. The situationremains fragile, however,and the governmentneeds to stay the course on its program of macroeconomic stabilization. Malawi's level o f domestic debt at the end o f fiscal year 2005/06 i s not as large as the corresponding ratios in some other countries. But with a small domestic capital market and the limited range o f holders, it has the potential to increase rapidly, as a resumption o f heavy borrowing IoThe IMFandthe Governmentagreedon a StaffMonitoredProgram(SMP) startingfromJuly 1,2004. Followingstrongperformanceinthe implementationofthe SMP, a new PRGFwas approvedinAugust 2005. Malawihas completedthe 31dreview of its PRGFprograminFebruary2007. liFurther reductions inthe interestrate are limitedbythe needto roll over the stock o f domestic debt at cheaper cost, while avoiding excessive moneycreation. 5 couldpushreal interest rates at highlevels. Also, the country's vulnerabilityto external shocks can easily disrupt, or at least prolong, the adjustment period during which the country is payingan enormous interestburden. Table 1.1 Malawi: Basic Macroeconomic Indicators, 1999-2006 Key Macroeconomic Indicators 1999 2000 2001 2002 2003 2004 2005 2006" 19,-,$-v~oo6 National Income and prices' GDP (USD million) 1,799 1,727 1,731 1,935 1,765 1,903 2,076 2,239 1,897 GDP Growth (%) 4.0 1.1 -4.2 2.1 3.9 5.1 2.1 8.5 2.8 Inflation (eop) 28.1 35.4 22.1 11.5 9.8 13.7 16.5 9.9 18.4 Fiscal sector indicators' Revenues and grants 24.1 27.4 23.5 26.7 34.7 37.5 42.5 42.1 32.3 Revenues 17.2 18.3 16.8 20.0 22.6 25.1 24.4 24.2 21.1 Grants 6.9 9.1 6.7 6.7 12.2 12.4 18.1 17.9 11.3 Expenditures 29.7 33.2 31.2 38.1 42.5 42.9 43.5 43.5 38.1 Current expenditures 19.2 22.6 24.0 30.9 31.4 31.6 33.9 27.9 31.1 Wages and salaries 4.7 5.2 6.8 6.8 6.5 7.5 7.3 7.3 6.5 Interest payments 3.7 4.6 5.0 6.9 10.6 8.7 6.6 4.5 6.3 Domestic interest payments 2.3 3.0 3.9 5.5 9.1 7.4 5.5 4.1 5.1 External interest payments 1.4 1.6 1.2 1.3 1.5 1.3 1.2 0.4 1.2 Other current expenditure 10.8 12.7 12.2 17.2 14.3 15.4 20.0 16.2 14.9 Development expenditure 10.4 10.1 7.2 7.4 11.1 11.0 9.5 15.4 10.3 Overall balance (excluding grants) -12.5 -14.9 -14.4 -18.3 -19.9 -17.8 -19.0 -19.3 -17.0 Overall balance (including grants) -5.6 -5.8 -7.7 -11.6 -7.8 -5.4 -1.0 -1.4 -5.8 Domestic balance -2.7 -4.7 -7.7 -11.0 -9.2 -6.5 -9.7 -6.2 -7.2 Primary Balance incl. grants (% o f GDP)4 -0.4 -1.7 -3.8 -5.5 -0.1 0.9 -4.2 -2.1 -2.1 Underlyingbalance' -1.6 -1.7 -2.7 -1.7 -0.9 1.8 0.9 0.6 -0.7 Net domestic borrowing 2.1 1.2 6.9 11.8 7.7 3.0 0.3 -1.9 3.9 Domestic Debt 2.8 8.9 10.2 20.4 24.8 23.8 19.9 14.8 15.7 Monetary sector indicators ' Growth inM 2 (%) 33.6 42.4 12.1 25.2 29.3 29.8 14.3 22.0 26.1 Grossreserves' 4.2 4.4 3.3 1.4 1.4 1.1 1.4 1.6 2.4 Interest rate (3 months Tbill, average) 40.1 39.5 42.4 41.7 39.3 28.6 24.5 23.2 33.3 Externalsector indicators Current account balance (percent o f GDP) Excluding official transfers -16.9 -11.7 -12.5 -29.3 -17.2 -21.3 -29.8 -21.8 -20.1 Including official transfers -8.2 -3.0 -6.8 -17.2 -7.9 -10.1 -16.2 -7.2 -9.6 Exchange rate M W K N S D (period average) 44.1 59.5 72.2 76.7 97.4 108.9 118.4 136.3 89.2 *Estimate Source: Government of Malawi and I M F 1. Calendar year basis 2. Percent of GDP, fiscal year basis (Malawi fiscal years run from July lst to June 30th) 3. Definition: Overall balance excluding grants, foreign-financeddevelopment expenditure, and foreign interest payments 4. Definition: Domestic balance, excluding domestic interest payments. 5. A measure o f domestic adjustment effort (Le. domestic primary balance excluding the health SWAP). Definition: Overall balance plus statistical discrepancy, excluding grants, revenue and expenditure from maize, interest, foreign financed development expenditures and the health SWAp. 6. Months o f current year's imports o f goods and non-factor services, calendar year basis. Source: Ministrv ofFinance and RBM and IMF and World Bank staffcalculations 6 1.3 What Caused such a Rapid Increase in Domestic Debt between 2000 and 2004? 72. The rapid growth in domestic debt and the resulting increase in the interest bill were pushing the country towards a fiscal crisis. The fiscal deficits o f the early 2000s led to an increase in domestic interest costs. Adverse weather shocks then exacerbated the impact o f fiscal indiscipline. In fact, the 2002/03 maize imports compounded the already rising interest costs, leading to a rapid increase in domestic expenditures and the crowding out o f discretionary (i.e. non-statutory) expenditures. Although government started borrowing on a large scale in 2001/02, the full effect o f increased interest costs was not felt until2003/04, when the domestic interest bill reached 9.1 percent o f GDP, corresponding to a massive 40 percent o f domestic revenues. The breakdown between interest (plus the 2002/03 and 2004/05 maize operations) and all other expenditure on conventional government services, as a share o f GDP, also shows that other expenditure has been relatively stable over the period (Annex 1.1). In other words, the increase intotal expenditure since 1999/00 i s largely attributable to growth in domestic debt and the ensuinginterest bill. 73. The `underlying balance' highlights the role of weak fiscal discipline and external shocks in explaining the fiscal crisis of the past few years. An exam o f movements in the `underlying balance' can facilitate an assessment o f the relative importance o f weak fiscal discipline versus external shocks, e.g. weather-related shocks and aid flow volatility, in explaining the fiscal crisis o f the past few years.13 The `underlying balance' was negative by around 2-3 percent o f GDP during fiscal years 1998/99 and 2001/02, reflecting a weak fiscal stance (Table 1.1). It improved slightly in 2002/03 to a 1 percent deficit, but the overall balance suffered substantially from the impact o f external shocks (i.e. drought-related expenditures). The underlying balance turned positive is 2003/04, suggesting an improved fiscal effort, but not sufficient to offset the amount o f interest payments due on the accumulated stock o f domestic debt. It i s only after 2004/05 that the combined effect o f improved fiscal stance and increased aid flows resulted inthe stabilization o f domestic debt. 74. Weak budget preparation and execution, volatility of donor inflows, and exposure to external shocks have been the main sources of the surge in domestic debt between 1999/00 and 2003/04. As discussed in detail below, the increase in the level o f domestic debt between 1999/00 and 2003/04 was due to a combination of: (i) weak expenditure control, including both an inability to keep ORT expenditures in line with the budget, as well as weak forecasting o f the wage bill and o f the interest bill; (ii) differences between the anticipated and actual donor budget support; (iii)lack o f mechanisms to cope with the occurrence o f shocks, including both financial losses by domestic parastatals and/or the impact o f external weather shocks. Hence, while the fundamental cause o f the fiscal crisis was the government's inability to control The totalinterestbillincludingforeign debt reacheda staggering10.7 percentof GDP, or around32 percentoftotaldomestic expenditures. l3 The `underlying balance' providesa measure of the domestic adjustment effort, which separatesthe issues of aidflow volatility fromthe issues o f expenditureoverruns(this indicatoralso excludes expendituresrelatedto maizesales andpurchases, andto the health SWAp). It is calculatedas the overall balanceplus statisticaldiscrepancy,excluding grants, revenue and expenditurefrom maize, interest, foreign financeddevelopment expendituresandthe healthSWAp. 7 expenditure before mid-2004 and to live within its means, the impact o f fiscal indiscipline was made worse by the volatility in donor disbursement o f funds, and the impact o fthe adverse weather shocks. Weak expenditure control 75. Weak public expenditure management is reflected in the significant variations in the allocations made to most individual agencies, as compared to the approved budget. Between 1999/00 and 2003/04, actual expendituresusually differed significantly from the original budget estimates, mainly as a result o f overshooting o f the recurrent budget (Table 1.2). Variations in discretionary recurrent expenditures were quite large, both in the aggregate (plus or minus 10 percent on average) and in terms of changes to the allocations made to individual agencie~.'~Further, about half o f public institutions regularly received either 10 percent more or 10 percent less than originally programmed and it was not uncommon for institutions to receive 30 percent more or less than originally anticipated. As budget outturns bore little resemblance to approved estimates, the budgetprocess lost credibility (Fozzard et al., 2002; DFID, 2004). 76. Significant overspending in Other Recurrent Transactions (ORT) has been a major cause of the overruns in the recurrent budget, reflecting weak commitment to fiscal discipline. For instance, ORT were above budgetby 5.2 percent of GDP in 1999/00, by 2.8 percent o f GDP in 2000/01, 7.8 percent in 2002/03, and 5.1 percent in 2003/04 (Annex 1.1). An examination o f the areas in which unbudgeted expenditure took place reveals a consistent pattern o f over-expenditure on activities and institutions which bring little direct benefit to the poor, e.g. travel, state residences, foreign affairs, defense, National Intelligence Bureau, the Police, the Office o f the President, and Special Activities (World Bank, 2003). Hence it would be difficult to argue that the path o f fiscal consolidation was not chosen because o f the need to protect pro-poor spending priorities. 77. Poor debt management and payrollmanagement have also been a major source of weak fiscal performance. A large portion of the discrepancy inthe recurrent budget was associated with higher than anticipated interest payments on debt and repeatedunderestimation o fpersonal emoluments. For instance, the domestic interest bill was underestimated by 1.7 percent o f GDP in 1999/00, by 1.5 percent o f GDP in 2000/01, 2.1 percent in 2001/02, 2.8 percent in 2002/03, and 4.9 percent in 2003/04 (Annex 1.1). Similarly, the wage bill was underestimated by 0.7 percent o f GDP in 1999/00, by 0.5 percent o f GDP in 2001/02, 1 percent in 2002/03, and 0.5 percent in 2003/04. 78. Measures have been introduced recently to strengthen budget preparation and execution, including payroll management, but further work is necessary. Progress has been made in improving the coverage of donor support, thus improving the comprehensiveness o f the budget starting in 2006. Significant progress has also been made to reform and streamline the execution o f the budget, notably by rolling out the Integrated Financial Management Information System (IFMIS), l4Discretionaryrecurrentspendingis definedas totalrecurrentminus'statutoryexpenditures' (Le. interest on debt, pensions and gratuities and Office of the President). 8 rationalizing all central government bank accounts into a Single Treasury Account, and implementing a centralized payment system in 2005. Areas o f weakness remain, however. Firstly, a recent IMF assessment o f the budget system highlights opportunities for: (i)improving the technical infrastructure for budget management; and (ii) rationalizing the budget structure and budget documentation, so as to strengthen the policy implementation through the budget (IMF, 2007a). Secondly, besides the continued phase-in o f the IFMIS, there i s a need to improve internal controls and reconciliation in budget execution. Thirdly, there i s a 2-3 year time lag between the end o f the fiscal year and the presentation o f the audited accounts and report to the legislature, implying that the official accounting records were not very useful for fiscal management purposes. Fourthly, internal audit reports are not being acted upon as well. These reports are a major source o f identifyingcorruption, maladministration, inappropriate expenditure, etc. A t the moment all internal audit reports are sent to the Accountant General (AG) for action and follow-up, but little follow up takes place due to lack o f capacity inAG office. 79. Significant steps have been taken by government to strengthen control of the payroll inrecentyears, and additional actionsare planned.The civil service wage reform carried out in 2004, entailed a rationalization o f grades and salary levels, and a consolidation o f most non-discretionary allowances into the wage (Box 1.1). The Government also established a Public Sector Remuneration Board (PSRB) which has begun the exercise o f further implementation o f the Medium Term Pay Policy through the consolidation into base salaries o f all remunerative allowances not already consolidated, in all public sector entities. The PSRB i s also examining the development o f a single national salary and allowance structure for all public sector entities including the civil service and parastatals. The government also introduced a new system in September 2006 for managing personnel and pay roll records, the Human Resource Management Information System (HRMIS). Unlike the earlier system it replaced, the HRMIS ensures that payroll data is automatically updated each time there is a change to personnel data. A database o f all government employees has also been compiled, which-once hlly operational-will help payroll management and mitigate problems o f ghost workers, inaccurate wage projections, unauthorized allowances, hirings, promotions and wage increases. As part o f this exercise, the government has also reconciled payroll records with the backlog o f personnel data as o f December 2006. Additional work i s requiredto strengthen the HRMIS and its data base. There are plans to carry out an independent review o f the HRMIS in 2007 in order to establish its robustness. Further,the government plans to undertake a payroll audit o f all civil servants in2007 to verify andupdate the payroll database. These measures could entail substantial savings for the government. l5 l5The mostrecentpersonnelaudit, in 1998,revealedthat, due to weaknesses inthe payrollmanagement system, around 5 percento f the civil service complementwas composedo f ghost workers. This suggests that government has beenincurringcosts in excess of MWK 100(USD2.5 million) annually due to ghost workers. The audit also revealedthat about 20 percent of civil servants were occupyingunauthorizedposts, costingapproximatelyMWK 277 million (USD7.5million) annually. 9 Volatility in donor inflows 80. Donors' budget support has been erratic over the past decade, further exacerbatingthe impactofweak fiscal discipline. Malawi receives substantial external assistance, amounting to around 17 percent o f GDP on average over the past decade. In fact, the amount o f assistance has varied substantially, with a low o f 10 percent o f GDP infiscal year 2001/02, and a highof 25 percent in2005/06. An important factor behind Box 1.1Civil Service CompensationReformand RelatedAdjustments In 2004 the government embarked on a broad civil service reform process to consolidate and streamline wages and salaries, increase transparency, and to improve management o f government employees. The reform process covers the main civil service (including teachers and health workers), the army and the police. It excludes parastatal enterprises, the judiciary, parliament and independent state institutions. The initial 2004 wage reform was followed by a PAYE-Pay as Y o u Earn-tax reform in July 2005 (continued inJuly 2006), a second round o fwage adjustment inFebruary 2006 and pension reform inJuly 2006. The October 2004 wage reform was a substantial step forward in terms o f transparency as it consolidated all fringe benefits and allowances (for housing, fuel, transportation, education, loans etc.) into a basic, taxable salary. The reform included a 25 percent general pay raise, a 50 percent increase in the minimum wage (implying a substantial salary raise for the lower grades o f the civil service) and a narrowing o f the gap between contractuals and regular middle-to-upper grade levels. Although this gap was again addressed in the February 2006 wage adjustment, a significant gap still exists between contractuals and regular civil servants in the upper grades. The February 2006 wage adjustment consolidated the professional allowances o f teachers, lawyers and health workers into the basic salary. It also provided top grades with a 50-70 percent nominal salary increase, while the lower grades received a much smaller increase o f 13-15 percent. In addition, the July 2005 tax reform raised the annual tax exempt PAYE threshold from MK36,OOO to MK60,000, changed the rate structure and lowered the top rate from 40 to 35 percent. The threshold was increased to MK72,OOO and the top rate further reduced to 30 percent in July 2006. Due to inaccurate payroll information, the 2004 wage adjustment turned out substantially more costly than anticipated, resulting in an annual overrun o f about 1% percent o f GDP. Another unintended side effect o f the 2004 reform was a very large expansion o f the base on which pensions for government employees are calculated, which would have implied a tripling o f the government pension bill. A short-term measure in July 2005 required the use o f the salary inthe past five years o f employment, rather than inthe final year, in calculating the pension benefits. A more comprehensive revision to the pension formula was introduced in July 2006 to ensure the long-term viability o f the government pension plan. K e y elements o f the revision were reducing the replacement rate and discounting lump-sum payments. Increasesin Civil Servants Take HomePay since October 2004. 10 the increase indomestic debt was the irregular pattern o f external fundingbetween 1999 and 2004. The actual amount o f budget support received varied between 5.5 percent o f GDP in 2000/01 and 0.8 percent o f GDP in 2002/03. This volatility appears not to have been anticipated at budget preparation, such that the program grants were always expected to be around 5.5 to 6 percent o f GDP. As a result, program grants were overestimated by an average 2.6 percent o f GDP over the period, and as much as 3.8 percent in2001/02 and 4.9 percent in2002/03 (Annex 1.1). Such systematic lower-than- forecasted donors' disbursements reflect an optimistic budgetingo f external inflows. The impact on fiscal outcomes was severe because o f the inability to adapt the budget to changes inavailable financing. l6 81. The mechanisms for disbursement of donors' budget support were not conducive to good public expenditure management. Undoubtedly, the volatility in budget support was the direct result o f weak Government performance in implementing agreed macroeconomic and reform programs, underpinning the disbursement o f budget support. However, until 2004 the mechanisms and timing o f the performance evaluation by donors, which determine disbursements, were not aligned with the budget process and did not help remove the uncertainty in the level o f available resources by the time o f budget preparation. Such uncertainty contributed to poor fiscal performance, especially for a country as heavily dependent on donor funding as Malawi. 82. Substantialprogress has already been made by several donors with the establishment of CABS, which could be further refined. Since 2004 the bilateral donors providing budget support to the government have strengthened their coordination under the Common Approach to Budgetary Support (CABS). One of the goals o f CABS i s to facilitate the predictability o f aid flows, by providing ex-ante commitments on levels of funding (for the following fiscal year) at the time o f budget preparation. A few steps could be undertaken by CABS to further facilitate predictability o f budget su~port:'~ First, at the completion o f the annual joint review in FebruaryMarch o f each year the CABS could provide government a firm commitment on the level o f aid to be disbursed inthe following fiscal year. These commitments could be changedonly inthe event of a violation o f the "fundamental principles" o f the agreement (respect for human rights, democratic principles, sound macroeconomic management, good governance, including sound public financial management, accountability and effective anti-corruption programs, and the rule o f law). Second, the CABS could commit to disbursethe amounts within a given number o f months following completion o f the review, according to a timetable to be coordinated with the authorities. Third, as envisaged in the Joint Framework, the CABS partners could provide government indicative commitments on budget support for the two years following the coming fiscal year. Inability to cope with unanticipated shocks 83. The impact of unanticipated shocks has translated into increased domestic borrowing. In 2000/01 unbudgeted expenditures o f 2 percent o f GDP l6 Budgetsupport was includedinthe budget eventhoughthe CABS grouphadclearly indicatedthat their budgetsupportwas dependent upon government staying `on-track' with the IMFPRGF. Mozambique,Tanzania andZambiaprovidegoodpracticeexamples for budgetsupport harmonization by development partners. 11 originated from unauthorized debt accumulated by ADMARC and NFRA. ADMARC and NFRA accounted for an additional 0.6 percent o f GDP inunbudgeted expendituresin 2001/02. Further, in preparing the 2002/03 budget, the government did not set-aside resources to respond to the emerging food crisis (which resulted from the drought and floods in2001/02), and eventually spent an unbudgeted5 percent o f GDP because o f the need to provide food relief.'* In fact, every few years there is an unexpected crisis, usually because o f the impact o f adverse weather on agricultural production (and GDP and public finances). In spite of the frequency of (weather) shocks, only a small allowance i s made in the budget for such contingen~ies.'~As a result, shocks usually translate into higher domestic borrowing, and have a substantial impact on fiscal performance and macroeconomic stability. The significant fiscal incidence o f shocks, notably the recent drought in 2002/03, highlights the need to plan for contingencies. Evenin2005/06, when the government carefully planned its humanitarian response to the 2005 drought and included funding in its 2005/06 budget, the actual cost o f the interventions turned out to be much higher than initially envisaged, resulting in a larger fiscal deficit (including grants) by 0.8 percent o f GDP. 84. Remedial actions should focus on having access to financial resources which can be used in case of shocks. Possible remedial actions includes establishing contingency funds, and/or increasing forex reserves for external shocks, and/or piloting the use o f national weather insurance.20 85. Establishing a contingent reserve fund poses several risks. The government could consider establishing a Contingency Reserve Fund (CRF), linked to the potential scale o f shocks (at around 8-10 percent o f GDP). These fimds are normal practice in developed economies, which are much less vulnerable to shocks. CRFs are important to lower vulnerability to shocks, as they can smooth out peak financing requirements,and are especially important for countries with limitedor no capital market access. However, there are also a number o f issues related with them, and notably (i) difficulty in properly accounting for CRFs; (ii) challenges related to asset management and investment strategy o f the CRFs; and most important, (iii)CRFs may create corruption and governance issues, especially inweak institutional environment (such that the CRF becomes a slash fund for unauthorized expenditure). 86. An alternative to purchasing insurance would be to accumulate foreign exchange reserves to around 5 months of imports. This would allow the government to run down reserves as a first line o f defense to buffer against most shocks without l8 This was a combination o f emergency imports o f maize (amounting to 3.8 percent o f GDP), and transfers to the ADMARC and NFRA parastatals (1.2 percent o f GDP). l 9A budget line for 'unforeseen expenditures' allows for such contingencies, amounting to about 1percent o f GDP. Malawi has traditionally relied on international assistance following climatic disasters. 2o One may ask whether for a l o w income country like Malawi it makes sense to set aside funding up front which might not need to be spent on shocks, rather than spending it on essential immediate needs. M a n y OECD countries regard fiscal contingency reserves as essential to efficient budgeting: a case that is allthe stronger for low-income countries which are highly vulnerable to shocks. Highlevel o f risk has been identified as an important explanation for slow growth in Africa in general (Collier and Gunning, 1999).In fact, insufficient anti-shock action and finance has been a recipe for magnifying economic instability and other distortions, ending up costing far more inthe long-term (Gelb and Eifert, 2006). A s discussed above, such costs have been extremely highin the case o f Malawi. 12 reserves disappearing. While this strategy removes most o f the risks inherent in establishing a CRF, increasing reserves would also entail substantial sterilization costs, and therefore could only take place gradually. 87. A more innovativeapproach wouldbe to pilot the use of nationalweather insurance. Recent developments ininternational risk markets imply that the government can buy an index-based weather derivative contract to transfer the financial risk o f severe and catastrophic national drought that adversely impacts the Government's budget to the international risk markets (Hess and Syroka, 2005; G o M and World Bank 2006; see Annex 1.2 for details). The aim o f such a contract would be to secure timely and reliable funds for the Government if a contractually specified severe and catastrophic shortfall in precipitation occurs during the agricultural season, as measured by weather stations throughout the country. Access to such contingency funds in a time o f crisis would generate a supplemental source o f emergency financing in May to complement existing budget resources, giving the Government more flexibility in its drought planning and enhancing the Government's ability to launch an efficient and cost-effective drought response.21 Such a new approach to financing a government response to drought would offer greater autonomy for the Government, as Malawi shifts away from an appeals- based model that relies on unpredictable and often untimely donor emergency relief funds, to an ex-ante risk management and financing 1.4 Inter-sectoralAllocationof Expenditures:Choosing the Balance between HumanDevelopment and InfrastructureSpending in Malawi 88. Following recent improvements in the macroeconomic situation, the government needs to prioritize how best to utilize the resources released from the reduction in domestic interest payments. In 2004/05 about 28 percent o f the budget was spend on administration, 30 percent in social services (including education, health, social security and community development), and around 18 percent on economic services (including agriculture, transport, energy and mining, and industry and commerce), with the balance o f 23 percent taken up by debt interest payments (Figure 1.5). The lower interest bill will generate much needed fiscal space to increase other (recurrent and development) expenditures, such as inhealth, education, social safety nets, agriculture, irrigation and transport infrastructure. Operating with limited resources, the government needs to decide how best to utilize these resources. 2'This instrument effectively serves to smooth out expenditures o n drought-related crisis evenly across years, and it would obviate the need for fund-raising inresponse to crises, thereby making the response more efficient and timely. It would therefore provide a first step towards ex-ante risk management by the government and donors, by selling the drought risk to private-sector international risk markets. Establishing event-specific, contractually guaranteed contingency finding also creates fiscal stability for the government and has the added advantage o f providing risk price information to assist government in its investment decisions with respect to managing and mitigating drought risk. 22Donors often provide additional financial resources inresponse to a humanitarian crisis; hence, there may be a financial loss to Malawi if it takes on the cost o f the insurance premium. Such a perverse incentive against an ex-ante approach to risk management could be discussed with development partners to seek support inthe payment o f at least part o f the premium. 13 89. The emphasis of the recently finalized Malawi Growth and Development Strategy (MGDS) is to rebalance government expenditure from social to economic sectors. The government has recently adopted the MGDS as its overarching strategy for economic policy. It differs from its predecessor (the Malawi Poverty Reduction Strategy, MPRS) in that the primary focus o f the MGDS i s on measures to spur productive activities in economic sectors. The MGDS highlights six focus areas where government will concentrate its efforts: agriculture and food security; infrastructure development; irrigation and water development; energy generation and supply; integrated rural development; and prevention and management o f HIV and AIDS. As stated in the MGDS, the government believes that its main contribution should be to create an enabling environment for private sector development and improve economic infrastructure such as road networks, energy, water systems and telecommunication. The MGDSrecognizes that spendingtowards social services ought not to be seen as an end in itself.23 However, the government believes that in previous strategies, resource allocations were tilted towards general administration and social services. Hence, the government wants to channel the funds released from the reduction in domestic bill to adjust this balance. Figure 1.5 Allocationof PublicExpenditureby Function,2004/05 GeneralAdministration 0Education 0Health Social Security and Welfare Services 0CommunityandSocialDevelopment* Agricultureand Natural Resources 0Transport Other Economic Services Unallocable Servlces 0Debtarnmortization Source: GOM Economic Report, various years. Note: `General Administration' includes a large number o f smaller votes: State Residencies, National Audit Office, National Assembly, Office o f the President and Cabinet, Human Resources Management and Development, Directorate of Public Procurement, Defense, Malawi Defense Force, National Statistical Office, ForeignAffairs, Malawi Revenue Authority, Home Affairs, Police, Prisons, Immigration, Justice and Constitutional Affairs, Directorate o f Public Prosecution and State Advocate, Registrar General, Human Rights Commission, Electoral Commission, Anticormption Bureau, L a w Commission, etc. 23Human development indicators measure important dimensions o f poverty, which are not captured by improvements inincome. Further, a healthy and educated population i s necessary to achieve sustainable economic growth and development. Notably, in countries like Malawi where shortage o f skilled manpower is a real constraint to growth, spending on social sectors would also support growth. Further, for growth to b e sustained, it will also need to be shared, requiring the improved provision o f social services and the provision o f social protection programs for the most vulnerable. 14 1 Box 1.2 The MAMS (Maquette for MDG Simulations)model The MAMS is an economy-wide simulation model created to analyze development strategies. The modelintegratesa relativelystandard dynamic recursivecomputable general equilibrium(CGE) model with an additionalmodule that links specific MDG or poverty-relatedinterventions to poverty and other MDG achievements. The relatively detailed treatment o f government activities in MAMS (including three educationsub-sectors, various types of health services, water-sanitation, infrastructure, and other government) makes this link possible. Provision o f education, health, and water-sanitation services contribute directly to the MDGs. Growthinthe stock o fboth humancapital (educationand health) and public infrastructure capital (including roads, energy and irrigation) contribute to overall growth by adding to the productivity of other productionactivities. The model also allows for complementarities between expenditures in different sectors, and allows for key interactions betweenachievingthe MDGs and the rest of the economy. It monitors what happens in terms o fpoverty reduction(MDG 1) and the evolution o f key human development indicators. To keep it relatively simple, it does not cover all MDGs. It focuses on the ones with the greatest cost and the greatest interaction with the rest o f the economy: universal primary school completion (MDG 2), reduced under-five and maternal mortality rates (MDGs 4 and 5), and increasedaccess to safe drinkingwater and basic sanitation (MDGs 7a and 7b). Additionaldetails on the modelare providedinAnnex 1.3. Simulation of Alternative Expenditure Strategies using the MAMS Model 90. The MAMS (Maquette for MDG Simulations) model has been used to analyze the implications of alternative expenditure strategies (Box 1.2 and Annex 1.3). The composition o f public expenditures is important for growth and poverty reduction. While there are no universal rules on what constitute the optimum allocation, it is possible to explore the consequences o f changing the balance of expenditures between economic and social sectors. The MAMS model has been used to simulate the evolution o f major social and economic indicators duringthe period up to 2015, including monitoring the effects on poverty, education, health, and water and sanitation coverage. Several simulations are conducted to assess the impact o f alternative expenditure allocations. Key findings and results are presented in Table 1.2. Details o f the assumptions underpinningeach simulation and the full results are presentedinAnnex 1.3. I t should be clarified upfront that the model assumes that the quality o f public expenditures does not change substantially. In fact, in addition to the level o f resource allocation, it i s the quality o f those expenditures (Le. how efficiently those resources are spent) that determines the total level o f goods and services provided by the government. Further,it should be made clear that the results o f the model provide an indication o fthe broad direction o f changes inthe economy, and the likely trade-offs which exist between the different strategies. However, the model and the results o f the simulations do not aim to provide precise forecast o f the path and level o f individual variables. 91. The base simulation(BASE-PRGF) simulates the implementationof the economic programpursuedby the government since mid-2004,inlinewith the IMF PRGF. The simulation envisages that Malawi can grow by 6 percentper year on average throughout the period to 2015, under the implicit assumption that adequate reforms are adopted and that Malawi i s not hit by external shocks.24 The baseline assumes that the 24 The PRGFprogramassumes that Malawi can grow by 6 percent peryear on average inthe next few years (as a result of a short-termcatchingup effect from years o fpoor economic management), andthat growthwill drop to 4.5 percentover the mediumterm (fromyear 2011) convergingto the regionalaverage. Inthe modelwe assumea constant growthrate at 6 percent throughout the period. To assess the impact of 15 fiscal envelope available in any Box 1.3 I s there more to povertythan just income? one year i s in line with the assumptions under the IMF PRGF Poverty is a multidimensional concept encompassing program. An increase in primary numerous aspects o f well-being. In Malawi this fiscal deficit would be recorded as multifaceted nature o f poverty has been highlighted by the reductions in the domestic interest findings o f the recent Malawi Poverty and Vulnerability Assessment, M P V A (GOM and World Bank, 2006). bill (starting in mid-2005) are According to a measure o f income-poverty (as proxied reallocated towards other using consumption data from the IHS2), the rural Southern government expenditures, leaving region i s the poorest, the rural Northern region is second the overall balance roughly poorest, and the rural Central region is the richest. unchanged. The baseline scenario However, according to people's subjective self-assessment o f poverty (also collected by the IHS2), the results show assumes no changes in the current that while the rural Southern region i s confirmed as the composition o f expenditures, and poorest, the rural Northern region, which `objectively' i s the shares across sectors are second poorest, has lower perceptions o f poverty than the maintained constant (as in 2003/04 Central region. In other words, people inthe North region fiscal year).25 feel richer than people in the Central region, even though the opposite is true when w e measure actual consumption. 92. Poverty i s reduced A s explained in the (MPVA), this finding reflects the substantially, and all human higher availability o f social services in the Northern region, and notably the higher levels o f education and indicators improve under the health indicators. Hence, this finding confirms that there is baseline simulation. The results o f more to poverty than just income. People's o w n perception the simulation highlight that by o f poverty reflects other important dimensions o f well- 2015 the poverty rate decreases being, and notjust monetary-poverty. from 52.4 percent in 2004 to 31.5 Source: Malawi Poverty and Vulnerability Assessment, percent, directly due to the M P V A (GOM and World Bank, 2006). Tncrease inhousehold per capita consumption. Monitoring all other education, health, and water-sanitation measures also shows improvements by 2015. The net completion rate for primary school doubles, although remaining at a very low level, while the maternal mortality rate i s cut by almost half from 960 (per 100,000 live births) in 2004 to 490 by 2015. Under-5 mortality rate also decreases from 133 (per 1000 live births) in 2004 to 108by 2015. Although access to safe drinkingwater and to safe sanitation improves, the increases are quite small. 93. Two simulations SIM-SOCIAL and SIM-INFRASTR analyze the trade- off between allocating a larger share of resources to human development than infrastructure, or vice versa. The base simulation has been used as the benchmark against which two additional simulations are compared. The BASE-PRGF simulates a scenario inwhich additional resources are being allocated across human development and lower assumptions regarding GDP growth rates, however, we also run an alternative simulation assuming a 4 percent GDP growth rate (throughout the period). The results indicate that the assumption on the level o f GDP growth has a substantial effect on the level o f the outcomes o f the simulation, but does not changes the direction o f the tradeoffs between the different strategies. The base simulation also assumes that Malawi is not hit by external shocks. The M A M S is not suited to examine the impact o f shocks; an external shock, such as a drought, can be simulated as a combination o f a one-off reduction in GDP growth inany one year andan increase inthe fiscal deficit. The net result would beto slow downprogress towards all the MDGs, both directly by reducing expenditures available for relevant government activities, and indirectly through the one-off reduction inGDP growth. 25Additional resources are allocated proportionally across social and infrastructure sectors. 16 infrastructure in line with past trends. Two additional simulations have been run to simulate the possible trade-offs faced by the government under alternative development strategies, while still maintaining the same assumptions about the expenditure envelope available in any one year (in line with the IMF PRGF program). In particular, these simulations analyze the trade-offs between allocating more public resources to "economic services and infrastructure"26, on the one hand, or to "human de~eloprnent"~~on the other, given the expenditure envelope available in any one year (see details in Annex 1.3).28 For both simulations, spending on "other government services" remains at the same growth rate as under the baseline. 94. Shifting more resources to infrastructure leads to faster growth rate, exports, investments and (monetary)povertyreduction,but slower progress in other human development indicators. The increased expenditure focus on infrastructure leads to an increase in the growth rate of real GDP at factor cost from 6 percent to 6.3 percent per year. The poverty headcount i s estimated to fall slightly faster than under the baseline decreasing to 29.6 percent by 2015, reflectingthe faster increase inaverage GDP per capita (Table 1.2). The improvements in human development indicators, however, are more modest. Primary completion rate only improves from 8 percent to 11 percent, Table 1.2 SummaryResults of the MAMS Simulations:BaselinePRGF simulationand Simulationsof Alternative ExpendituresAllocations,2004-2015 sim-infrastr base-prgf sim-social Values Units Groth rate of GDP at factor cost 167.3 bn Kwacha 6.3 6.0 5.6 MDG I:headcount poverty rate 52.4 % 29.6 31.4 33.9 MDG 2: primary (1st cycle) net completion rate** 8.0 % 10.8 15.4 18.5 MDG 4: under-5 mortality rate (share of live births) 133 per 1000 113 108 106 MDG 5: maternal mortality rate (share of live births) 960 per100000 587 499 473 MDG 7a: acess to safe drinking water 66.1 % 67.4 67.9 68.1 MDG 7b:acess to safe sanitation 63.7 % 64.7 64.9 64.8 Simulations base-prgf base Poverty Reduction Growth Facility scenario sim-infrastr Spending on educ, health, and water-sanitation decreases by 1.5 times the growth rate in base-prgf simulation, while spending on "irrigation", "Agriculture Gov", and "Public Infrastructure" adjusts as required. sim-social Spending on educ, health, and water-sanitation increases by 1.5 times the growth rate in base-prgf simulation while spending on "Irrigation", "Agriculture Gov". and "Public Infrastructure" adjusts as required. For both sim-infrastr and sim-social simulations, spending on "Other Government Services" remains at the same growth rate, `*`MDG 2 here refers to the rate of primary school completion for students of the correct age cohort AND within the 8 year time span (a very stringent definition of MDGP). Source: WorldBankstaff 26Includesthe government agriculture, irrigation,and other public infrastructuresectors. 27Includesthe sectors for education(allthree cycles), health, andwater andsanitation. 28SIM-INFRASTR exploresthe impact ofdecreasing spendingon "human development" by 1.5 times the growthrate inBASE-PRGF. SIM-SOCIAL,on the other hand, simulates an increase inspendingon "human development"by 1.5 times the growth rate inBASE-PRGF.Inbothsimulations, spendingon "economic services and infrastructure" adjusts as required, given the expenditureenvelopeavailable (and after accountingfor the growthrate of spending on "other government services" whichremains at the same growthrate as under the baseline).The resultingdivergenceinresource allocationbetweenthe two alternatives is quite marked(see Annex 1.3 for details). 17 instead o f reaching 15 percent as under the ba~eline.'~Maternal mortality decreases to 587 (compared to 499 under the baseline) and under-5 mortality decreases to 113 (compared to 108). 95. Alternatively, focusing expenditures on social sectors leads to a slower economic growth rate and slower poverty reduction, but more rapid progress on human development indicators. GDP growth is projected at 5.6 percent per year which i s lower than under the baseline and the alternative scenario that focuses on infrastructure Ospending (at 6 percent and 6.3 percent, respectively). Poverty reduction i s also slower than under the baseline and alternative scenario that focuses on infrastructure spending. The poverty headcount reaches 34 percent by 2015 (as comparedto 31.5 percent and 29.6 percent under the BASE-PRGF and SIM-INFRASTR, respectively). On the other hand, the results for the primary net completion rate, the mortality rates, and access to safe drinking water and basic sanitation show that a focus on HD spending is beneficial. The primary completion rate increases from 8 percent to 18.5 percent (compared to 15 percent in BASE-PRGF and 11percent in SIM-INFRASTR). Under-5 mortality and maternal mortality rates also decrease faster than under the BASE-PRGF, and much more than under the infrastructure simulations (Table 1.2). Similar results can be seen for the improvements inaccess to safe drinkingwater and basic sanitation. 1.5 RecommendedPolicyPrioritiesfor Macroeconomic Stability and Inter-sectoral Allocationof Expenditures 96. The government needs to stay the course with its macroeconomic program. After June 2004 the government has taken rapid steps to reverse the unsustainable fiscal trends pursuedbetween 2000 to 2004. As a result, macroeconomic performance over the last two years has been rapidly improving. The situation remains fragile, however, and the government needs to stay the course on its program o f macroeconomic stabilization. The results o f a simulation using a MAMS model o f the Malawian economy confirm that economic policies pursued by the government after 2004 are conducive to good macroeconomic performance andpoverty reduction. 97. Reducing the domestic debt should continue to be the overridingpriority of the government in the short term. Malawi's large domestic debt stock not only threatens macroeconomic and financial stability, but interest payments are also crowding out the amount o f discretionary budgetary resources available for expenditures on social and economic services, and investment. 98. Reforms to strengthen the budget process and payroll management remain at the center of the reform agenda. In order for the budget to reflect government policy there i s a need to improve budget management and rationalize the budget structure and documentation. Secondly, besides the continued phase-in o f the IFMIS, there is a need to improve internal controls and reconciliation in budget execution. Thirdly, there i s a need to strengthen audit capacity by: (i) developing further the capacity of the Central Internal Audit Unit; and (ii) developing the capacity o f the Accountant General (AG) and the National Audit Office (NAO) to shorten the delay in 29 Primary Completion Rate here refers to the rate of Standard-5 completion for students o f the correct age cohort and within the 8 year time span (Le. a very stringent definition). 18 the presentation o f the audited accounts and report to the legislature. Fourthly, lack o f follow up on internal audit findings needs to be addressed. These reforms have not been dealt with here indetail, as they are already part o f the government's Public Financialand Economic Management (PFEM) action plan. Similarly, the government i s already planning to conduct a review o f the HRMIS in order to establish its robustness, and to carry out a payroll audit o f all civil servants in2007 to verify and update the civil service database. 99. Reformsto insulatebudgetimplementationfrom the impactof exogenous factors should be pursued to prevent a resurgence of the debt levels. Given the substantial difficulties with managing a Contingency Reserve Fund, and the high sterilization costs o f rapidly increasing forex reserves, piloting o f weather-based insurance i s recommended. In addition, the government should continue to work with CABS to improve the predictability o f aid, which has also been a source o f volatility duringthe past decade. (i) Purchase national weather insurance: The government can buy rainfall-based insurance contracts from an international provider to transfer the financial risk o f severe and catastrophic national drought that adversely impacts the government's budget to the international risk markets. The weather insurance contract would trigger a contingent lump sum for the government in the event o f food emergencies. Such a new approach to financing a government response to drought implies a shift away from an appeals-based model that relies on unpredictable and often untimely donor emergency relief funds, to an ex-ante risk management and financing model. A few donors have expressed an interest in supporting an initial pilot in Malawi by paying the insurance premium on behalf o f the government. If the program moves beyond the pilot phase, it i s suggested that the government allocates some premium funds within the budgetto protect its own fiscal exposure to drought infuture years.30 (ii) forexreserves:Analternativetopurchasinginsurancewouldbeto Increase accumulate foreign exchange reserves to around 5 months o f imports. This would allow the government to rundown reserves as a first line o f defense to buffer against most shocks without reserves disappearing. Increasing reserves would also entail substantial sterilization costs, however, and therefore an increase inreserves can only take place gradually. (iii) Liaise with donors to improve the predictability of aid disbursements: Buildingon the progress made by since 2004, the CABS partners could revise their Joint Framework for budget support cooperation with government such that a binding commitment on the level o f aid to be disbursed inthe following fiscal year i s made at the completion o f the annualjoint review inFebruarymarch o f each year. Following this commitment, only a violation of the "fundamental principles" o f the agreement (respect for human rights, democratic principles, sound macroeconomic management, good governance, including sound public financial management, accountability and 30Donors often provide additional financial resources inresponse to a humanitarian crisis; hence, there may be a financial loss to Malawi if it takes on the cost o f the insurance premium. Such a perverse incentive against an ex-ante approach to risk management could be discussed with development partners to seek support inthe payment of at least part of the premium. 19 effective anti-corruption programs, and the rule o f law) would providejustification to reduce or halt aid disbursements. Second, the CABS could commit to disburse the amounts within a given number o f months following completion o f the review, according to a timetable to be coordinated with the authorities. Third, as envisaged in the Joint Framework, the CABS partners could provide government indicative commitments on budget support for the two years following the coming fiscal year. 100. Lookingahead, there is a needto consider how to investthe savings from the reduction in domestic interest payment: the MGDS elects faster growth at the expense of slower progress in human development indicators. The MAMS model highlightsthat maintaining sound macroeconomic policies is critical for improvements in both growth and human development indicators. The reduction of interest payments on domestic debt i s making available substantial budgetary resources for other recurrent and development expenditures. The government needs to decide how best to use these resources to accelerate growth and poverty reduction. The MGDS aims to allocate savings from the domestic interest bill to increase spending on economic sectors and infrastructure investments. The results o f the MAMS model highlight a tradeoff between a strategy focused on increasing infrastructure spending versus one that focuses on human development. A moderate increase in infrastructure spending leads to a faster GDP growth rate and poverty reduction, but at the expense o f improvements in human development indicators. Conversely a moderate increase in spending on social sectors leads to faster progress inhealth and education, and other human development indicators, but at the cost o f a slower growth rate (and poverty reduction). 101. Improvingthe `quality' of public expenditures bothin social sectors and infrastructure sectors is critical to accelerate growth and progress towards the MDGs. The results of the simulations highlight how changes in the allocation of expendituresaffect progresstowards the various MDGs; however, the results also suggest that the differences from alternative expenditure strategies are not dramatically different. Ensuringthe efficient use o fresources within each sector, i.e. maximizing the `quality' o f expenditures and investments, is also critical to accelerate progress.31 The next few chapters will examine the allocation and use o f resources within major sectors. ''The MAMSmodelassumes that the `quality' ofexpendituresandinvestmentsremainsconstant. 20 CHAPTER2: PUBLICDEBTSUSTAINABILITYAND DEBT MANAGEMENT REFORMS 2.1 Introduction 102. Malawi has benefited from substantial debt stock cancellation from the HIPC Initiative and the Multilateral Debt Relief Initiative (MDRI) for which Malawi qualified inAugust 2006. Nevertheless, Malawi's debt exposure worsened substantially between 1999 and 2005, due to an increase in the level o f domestic debt and to a series o f exogenous factors which negatively affected its external debt sustainability. While a number of recent reforms have been implemented, Malawi remains vulnerable to a moderate risk o f debt distress. 103. This Chapter examines Malawi's public debt sustainability and debt management framework, with a view to recommend options for reducing Malawi's debt vulnerability. The chapter reviews the composition and recent trends o f Malawi's public debt and reports the results o f the recent Debt Sustainability Analysis (DSA). It then explores the implications o f alternative fiscal paths for reducing the domestic debt burden, and highlights a few potential risks to fiscal stability in the medium term. The chapter also discusses the institutional and legal framework for public debt management. Recommendations focus on measures to strengthen the institutional framework for debt management and foster the development o f domestic debt markets. 2.2 PublicDebtStructure,RecentTrendsand Sustainability 104. Malawi's publicdebt burdenhas beenreduced substantially followingthe HIPC completion point and MDRI debt relief in 2006. Total public debt has been reduced from USD3.4 billion (equivalent to 165 percent o f GDP in nominal terms) as o f end-2005 to about USD0.8 billion (equivalent to 37 percent o f GDP innominal terms) as o f end-2006 (Table 2.1 and Figure 2.1).32 Most o f this remaining public debt stock i s domestic debt, accounting for 20 percent o f GDP, while external debt now accounts for around 17 percent o f GDP. These solvency ratios are not very worrisome by international standards. 105. Interestpaymentsremainvery highbecausethey mainly service domestic debt. Most of Malawi's high debt interest payments since the late 1990s have been on domestic debt. The interest burden o f the external debt stock was always very small because external debt was contracted on very concessional terms. For instance, interest payments on external debt were barely 1.5 percent o f GDP in 2004, while payments on domestic debt were 8.5 percent o f GDP (Figure 2.1).33 Therefore, the cancellation o f external debt has not reduced significantly Malawi's total interest payments. Liquidity ratios remain fairly high, with the ratio o f total interest payments to GDP estimated at 5.3 32This estimate i s based on data collected in2006 for HIPC Completion Point. A full loan-by-loan reconciliation based on information provided by the authorities and the creditors was conducted as part o f the preparationfor the HIPC Completion Point Debt Sustainability Analysis. 33The external interest payments are even lower if measured on a cash basis rather than on a due basis. 21 Figure 2.1: Total Public Debt and Interest Pa ments (as percentage of GDP), 1999 - 2006 - Total Government Debt 160 9.0 v i Total Interest Payments 140 8.0 I 2 0 7.0 8 8 6.0 t!s100 5.0 0 80 60 50 4.0 3.0 40 2.0 20 1.0 1999 2000 2001 2002 2003 2004 2005 2006 1999 2000 2001 2002 2003 2004 2005 2006 Domestic External 0Domestic External Source: Ministry of Finance and RBM Table 2.1 Key Public Debt Indicators before and after HIPC completion point and MDRI debt relief (nominal values in 2005 and 2006, respectively) 2005 2006" Solvency ratios Public Debt to GDP 158.7 37.4 o/w External Debt to GDP 136.8 17.3 o/w Domestic Debt to GDP 21.9 18.9 Liquidity Ratios Interest payments to GDP 7.7 5.3 o/w Domestic Interestpayments to GDP 6.4 4.8 o/w External Interestpayments to GDP 1.3 0.5 Interest Payments to Government Revenues (Including Grants) 30.2 21.9 o/w Domestic Interest Payments to Government Revenues (Including Grants) 25.9 19.8 oiw External Interest Payments to Government Revenues (Including Grants) 5.3 2.1 External Debt Service to exports (percent of exports) 21.5 6.4 Memo items GDP per capita (USD) 175 175 Nominal GDP (USD, million) 2,076 2,239 Government revenues (as percentage o f GDP) 24.4 24.2 Long-TermForeign Currency Sovereign Credit Rating (Fitch rating) ccc B- Source: WorldBank, IMF, Fitch-IBCA,andReserve Bank of Malawi. Note: *Liquidity ratios are basedon estimates of externaldebt service payments in2007, after HIPC andMDRIdebt relief. The Fitchratingwas upgradedinFebruary2007 percent in 2006 and the ratio o f total interest payments to government revenues (including grants) at 21.9 percent in2006. Evolution and characteristicsof external debtportfolio 106. The external debt burden has dropped substantially after Malawi qualified for HIPC completion point and the MDFU debt relief in August 2006. Upon reaching HIPC Completion Point, Malawi has also benefited from the MDRI whereby Malawi received 100 percent debt cancellation from IDA, IMF and the African 22 Development Fund,34its three largest multilateral creditors who collectively held 75 percent o f Malawi's debt outstanding after HIPC relief.35 The outstanding stock o f external debt (after accounting for HIPC, topping up, and MDRIdebt relief) i s estimated at approximately USD380 million in nominal terms (as o f end-2005), equivalent to 17 percent o f GDP (and/or 8.2 percent o f GDP inN P V terms). Following HIPC completion point, Malawi's external debt burden indicators have decreased to levels significantly lower than the average o f low-income countries. The ratio o f interest payments to exports i s around 6.4 percent, and the ratio o f interest payments to government revenues (including grants) i s around 2.1 percent (Table 2.1). 107. In the case of Malawi, reaching HIPC completion point has meant additionaldebt service savings of about USD45 millioneach year (about 2 percent of GDP). 36 This amount i s additional to the interimdebt relief which Malawi has already been receiving since HIPC decision point in 2000, amounting to about USD55 million per year. However, it should be noted that, inthe case o f IDA, the MDRIdebt relief is subtracted from IDA'Sallocation to Malawi. The net increase in financial resources to Malawi from HIPC completion point i s estimated at around USD25 million. 108. The currency and interest rate composition of Malawi's external public debt do not constitute a significant source of external vulnerability. Of the total remaining external debt stock, 78 percent i s owed to multilateral creditors, with IDA being Malawi's largest official creditor accounting for 27 percent o f the debt. About 22 percent o f the debt i s owed to bilateral creditors, and almost nothing to commercial creditors (Table 2.2). Hence most o f Malawi's remaining external debt i s at highly ~~~ ~~~ Table 2.2 Compositionof Malawi's public debt stock following HIPC completion Doint and MDRIas of end-2006 Nominalterms NPV terms USD YOof Share of USD % of Share of million GDP total debt million GDP total debt Total publicdebt 820.7 37.4 100 621.2 28.3 100 Externaldebt 381.8 17.4 46.5 180.1 8.2 29.0 Multilateral 291.9 13.6 36.3 110.1 5.0 17.7 olw IDA 103.6 4.7 12.6 35.8 1.6 5.8 olw IMF 3.8 0.2 0.5 2.9 0.1 0.5 Bilateral 83.9 3.8 10.2 10.7 0.5 1.7 Commercial 0.6 0.0 0.1 0.6 0.0 0.1 Domesticdebt 441.1 20.1 53.7 441.1 20.1 71.0 GDP 2194.5 2194.5 Source: Malawi Joint World BanklIMF Debt Sustainability Analysis Basedon Low-Income Country Framework, August 2006 34For IDA, eligible debt covers debt disbursedand outstanding as of end-2003. For the IMF and AfDF, credits on debt outstanding and disbursedas of end-2004. 3sMalawi's external debt at HIPC decision point (as of end-1999) amountedto USD2.6 billion innominal terms, equivalent to 144percentof GDP. Innet presentvalue (NPV) the stock of external debt was estimated at USD1.5 billion, equivalent to 269 percent of exports of goods and services. 36Over the periodbetween FY07 to FY15, the savings in debt service costs from IDA, IMF andthe AfDF are estimatedto amount to USD433 million innominal terms and USD334 million inNPV terms. These resources are additional to the interim debt reliefthat Malawi has alreadybeenreceiving since the decision point (inDecember 2000), and are mainly a result of MDRIdebt relief. 23 concessional rates. The foreign currency composition o f Malawi's external public debt tracks its main creditors. Most o f external public debt i s denominated in SDR with the shares o f dollar and Euro denominated debt amounting to 11 percent and 7 percent respectively. In light o f Malawi's main trading partners and the pattern o f its capital inflows, the overall currency mix appears prudent and does not raise concerns about hndamental foreign exchange mismatchproblems. Similarly, as there i s no external debt that has been contracted with floating rate, the real cost o f external debt will be unaffected by sharp interest rate fluctuation. 109. The sharp Figure2.2 Weighted averageYield and deterioration in debt Maturity of T-bills sustainability between 1999 Figure 3: Weighted Average Yield and Maturity o f T-Bills and 2005, highlights Malawi's vulnerability to external borrowing. At the time o f decision point (based on data for - - - end-1999) it was projected that - - 2 5 even after accounting for HIPC .- 20 debt relief the NPV o f debt-to- exports ratio would remain -Ave. Msrurltv- - above the HIPC debt AW. W e l d sustainability threshold (of 150 percent o f GDP) till 2010, Figure2.3 CurrentT-bills MaturingIssues peaking at 181 percent in 2002, January 2004 to February2007 and decrease to reach 169 percent o f exports by end 2005. Current T-bills MaturingIsues In fact, the actual outturn in 16,000 2005 was 229 percent. It 14,000 therefore exceeded decision 12.000 g 10,000 point projections by 60 j 8,000 percentage points. As discussed 3 6.000 in the HIPC Completion Point 4,000 document, the substantial 2,000 deterioration in Malawi's debt 0 g ; s a s a a a s s s s s burden indicators between a 2 ~ j ~ $ ~ dg a $decision point and completion $ ~ point i s primarily attributable to exogenous factors, notably lower than expected exports growth and changes in discount rates (World Bank 2006). Virtually none o f the increase inthe debt-to-exports ratio is due to higher than anticipated external borrowing.37 Malawi's vulnerability to debt distress i s also confirmed by the results o f the public debt sustainability analysis carried out in2006 (see below). 37Limits on non-concessionalexternal borrowingunder the IMF's PRGF program also helped to prevent an increase in external non-concessional debt. 24 Evolution and characteristics of domestic debtportfolio 110. Malawi's domestic debt i s heavily concentratedin a few instrumentswith very short-term maturities. The debt is comprised o f Treasury Bills (T-bills) with maturities o f 91, 182 and 273 days, Reserve Bank Bills (RBMbills) with maturities o f 63 and 91 days, Treasury Notes (maturities o f over one year up to 10 years), and Treasury Bonds (maturity o f more than 10 years).38 The most common financing instruments are T-bills which accounted for about 70 percent o f the total outstanding stock during 1999- 2005. Given the predominance o f T-bills inthe market the weighted average maturity o f domestic debt as o f end-2005 was below 250 days, with more than 40 percent o f domestic debt having maturity o f less than 91 days (Figure 2.2). 111. The maturity schedules show the existence of a "bunching problem", with a highvolume of obligations maturingon the same dates. Figure2.3 reveals the repayment humps in debt payment schedules that characterized end-2004 and continued in 2005. There has been an improvement since April 2006 inparallel with the reduction Figure 2.4 Domestic debt holdersin 2005 in the volumes o f maturing Treasury bills. Figure Holdings o f treasury Bills September Often debt markets in 2005 developing countries present similar bunching problems due to the lack o f effective and coordinated cash management planning 11 and an ad hoc approach Reserve Bank rn Commercial Banks-major 0 Commercial Banks-other 0 Discount Houses in meeting government's borrowing requirements. ' Other Financial Institutions Cor orate sector Institutional investors 0 Brofers 112. The domestic Figure2.5 Domesticdebt by types of holder,1999-2006 debt market is also heavily concentrated Domestic Debtby Holder among a few participants.As in other 60% countries which are in 1 -- 5nx" similar stages o f 40% development, Malawi's 30% I debt market is at an 20% incipient stage. The main 10% participants in the 1999 2000 2001 2002 2003 2004 2005 2006 domestic market are El RBM W Commercial Banks 0Nonbank Sector commercial banks, non- banking institutions Note: Data for 2006 i s as o f end-September (financial institutions, 38Treasury notes and bonds replaced Local Registered Stock (LRS). The Reserve Bank o f Malawi issued the first Malawi government LocalRegisteredStock inAugust, 1965, under Section 45 o fthe Finance and Audit Act, 1963. 25 discount houses, and institutional investors figure) and the RBM (Figure 2.4).39Notably, the non-bank financial sector represents the major source o f government financing, accounting for almost 50 percent o f the government deficit financing as o f September 2006. The discount houses, in the non-bank sector, have to hold 70 percent in liquid assets, and preferholding government and RBMsecurities. 113. The structure of debt holdings has been evolving in recent years with the RBMshare of debt holdings reducing slightly infavor of non-banking institutions. The Reserve Bank of Malawi has maintained its exposure into Treasury bills since 2001 at around 30 percent, which i s lower than it was in 2001-2002. The non-bank financial sector has been steadily increasing its share o f debt holdings and from Table 2.3 selectedCommercial Banking 2003 it has been holding above 40 indicators, end-2005 percent o f the debt, thus overtaking Malawi SSA the RBM to become the largest Bank concentration index 100 88.9 holder o f domestic debt (Figure M 2 (% o f GDP) 21.2 31.4 2.5). Assets (% o f GDP) 11.8 23.8 Deposits (% o f GDP) 16.2 24.3 114. The highly Bank credit to Private (% of GDP) 6.5 17.0 concentrated debt holdings Real deposit rate (%) -3.9 -0.9 reflect the weakness of the Lending deposit rate (%) 22.2 12.4 Malawi financial system. The low Source: IMF and World Bank level o f financial intermediation in the economy, measured as the M2/GDP ratio, shows the relatively low financial depth even as compared with other countries in the region (Table 2.3). The financial market i s highly concentrated, with commercial banks holding about 76 percent o f the total financial sector assets. Further, within the banking sector, the two largest banks hold roughly 75 percent o f Figure 2.6 Inter-bank trading activity commercial banks loans and MKBill Figure 7. Inter-bank Trading Activity 73 percent o f deposits. 18 115. The secondary 16 market for Treasury bills is 14 illiquid and volatile, also 12 reflecting the limitations of I O the financial system. 8 Despite a recent increase in 6 inter-bank trading, the 4 secondary market for 2 Treasury bills remains very thin (Figure 2.6). Such illiquidity can be explained I -Volume traded +Inter-bank interest rate I by several factors: (i) inadequate liquidity 39The commercial banking sector comprises 8 commercial banks; in addition there are 23 non-bank institutions, including 5 development finance institutions, 3 leasing operations, 12 insurancecompanies, 2 discount houses andthe New Building Society. 26 management that causes the typical "one-way" market, where banks face similar liquidity constraints at the same time; (ii)lack o f alternative investment opportunities that can compete with Treasury billdbonds as profitable and risk free investments; (iii)the absence o f instruments for hedging; and, last but not least, (iv) inadequate incentive structure in the financial markets. The latter factor i s mainly the result o f the high withholding taxes on government securities trading (that discourages discount house to trade government paper inthe secondary market and make it more profitable to hold until maturity) and also the high liquidityreserve requirementsfor banks, which together with restrictions to operate repurchase agreements, encourage banks to rely heavily on Treasury bills. 116. There is limited competition in the domestic debt market. The average nominal T-bill rate has remained above 30 percent until mid-2004 (Figure 2.7). More important, real yields during the period 2001-2004 were above 20 percent, one o f the highestinthe world. Such highrates of return on T-bills reflect among other factors, the highly concentrated investor base and the absence o f a liquid secondary market. These issues have complicated Figure2.7 MonthlyInterestRates andInflation, debt management and June 1999-December2005 monetary policy I arrangements, and raised 80.0 , I questions about 70.0 the 60.0 competitiveness and 50.0 accuracy o f primary market 40.0 valuations. 30.0 4 & * 117. Domestic yields 20.0 10.0 1l-Y have beensteadily coming 0.0 t down in nominal and real "" "L ' ' " " " " " ' """"" """"""""" " " " ' -16 @,& terms since 2003 (Figure -20 o J I 2.7). The reduction initially 1-Bank rate (eop) +Treasury Bill Rate ( a 4 ) followed the tight monetary Deposit Rate ( a 4 ) Lending Rate ( a q ) policy by the RBM during -12 months inflation (CPI) +Treasury bill rate (in real terms) 2000-2002, which lowered Source: MoF, RBMand IMF data inflationary expectations. More recently, the government's fiscal consolidation has reduced the demand for domestic borrowing in 2004-2006. This trend i s expected to continue inthe near term. 118. The characteristics of Malawi's domestic debt portfolio make it vulnerable to a high degree of risk. The domestic debt portfolio is held in domestic currency and therefore does not have an exposure to foreign exchange risk. However, the discussion above has highlightedthat the domestic debt stock i s characterized by limited diversity o f financing instruments and very short-term maturities which, combined with the shallow debt markets, entail substantial exposure to refinancingrisk, interest rate risk, and other fiscal risks: Refinancing(or rollover) risk: The heavy concentration ina few instrumentswith very short-term maturities o f less than one year entails substantial exposure to refinancing or rollover risk. The data for Q1 2005 reveal that 72 percent o f the T- 27 bill issuance was towards rollover. The exposure to refinancing risk is exacerbated by the existence o f the "bunching problem", with a high volume o f obligations maturing on the same dates. Further, the heavy concentration o f the domestic debt market among a few participants also raises additional concerns about the exposure to refinancing risks. Interest rate (and fiscal) risk: The short maturity structure and constant rollover o f the domestic debt stock increases the exposure to interest rates risk, and therefore poses an important fiscal risk to the government. Infact, sudden movements in T- bill rates would rapidly have a large impact on total interest expenditures. The volatility o f market interest rates makes it difficult to accurately budget for interest expenditures, and repeatedly resultedinunbudgeted expenditures between 2000 and 2004. Financial risk: Fiscal risks are exacerbated by Malawi's shallow domestic financial system, which implies a large risk exposure o f the few financial institutions. Given the absence o f large domestic institutional investors, banks hold a highly concentrated portfolio dominated by T-bills, accounting for 20 percent o f all Treasury bills and notes as o f end-2005. These investments represented around 24 percent o f the banking system's total assets and around 28 percent o f the Kwacha deposits respectively at end-2005, resulting in a concentration o f fiscal risks. Further, investment holdings within the banking system are concentrated in the hands o fthe two largestbanks, resulting ina further concentrationo frisk. Inflation risk: During the last few years, the RBM has been financing an increasingproportion o f the government's fiscal deficit. The increasing RBMshare o f deficit financing in recent years could translate in a rise in money supply and result in future inflation. The government finances its deficit by resorting to a combination o f Ways and Means advances from RBM and through issuing money market instruments. According to Malawi's law, the government can rely on financing up to 20 percent o f anticipated revenues in a specific year through ways and means advances. When government exceeds this limit, it instructs the RBMto convert the excess into T-bills inthe primary market 2.3 Public Debt SustainabilityAnalysis 119. A Debt Sustainability Analysis (DSA) was prepared jointly by the World Bank and the IMF in 2006, based on the common standard framework for low-income countries (LIC).40 The analysis uses the updated (as o f end-2005) debt database reconciled for the HIPC completion point and incorporates macro-economic projections as o f June 2006. The analysis accounts for the fact that Malawi reached HIPC completion point inmid-2006 thereby benefitingfrom HIPC topping up and MDRIdebt cancellation. Full results o f the DSA have been presented in the Malawi HIPC completion point 40The public debt sustainabilityframework consists of an excel-basedmodelwhichprojects debt levels basedonprojectionso f key macroeconomicvariables andthe structure o fthe economy (such as GDP growth, exchangerates, interestrates, tax rates, andfiscal variables), developedby the WorldBank andthe IMF in2004. For additionaldetails on the dgbt sustainabilityframework, pleasesee: http:Nsiteresources.worldbank,orglINTDEBTDEPT/Resources/DSAGUIDE~EXT2006 1O.pdf 28 document (World Bank/IMF 2006) and only the main conclusions are discussed below. The main assumptions and additional details are provided inBox 2.1. 120. The projectedpublic debt burdenis vulnerable to deteriorationin either the externalor domestic debt indicators. The results o f the DSA highlight that even assuming good policy performance accompanied by strong GDP and export growth and substantial inflows o f external financing in the form o f grants, all debt burdenindicators are expected to deteriorate through the projection period. If the government does not implement its reform strategy to enhance growth, diversify exports and improve governance, Malawi's debt burden indicators would increase steeply and the external debt stock indicators would breach their thresholds within a decade. Similarly, the debt burden would breach its external-debt-to-GDP ratio threshold under stress tests. Moreover, Malawi's highlevel o f domestic debt further increases Malawi's probability o f debt distress. Insufficient donor financing inthe form o f grants could lead to pressures to accumulate domestic debt, and endanger debt sustainability. Even with the level o f external support assumed in the baseline, a reversion to the historical record o f poor macroeconomic policy during 1999/00-2003/04 would retard the reduction in domestic debt and cause serious deterioration inthe public debt indicators. 121. Even after full delivery of HIPC debt relief and implementation of the MDRI, Malawi remains at a moderate risk of debt distress. For Malawi's debt to remain sustainable in the future it i s important that authorities make prudent use o f the resources freed-up through debt relief and strengthen public expenditure management. Moreover, contracting o f new loans on less concessional terms or shifting from grants to loans would significantly deteriorate Malawi's debt sustainability outlook. Implications of Alternative Fiscal Adjustment Pathsfor Domestic Debt Reduction4' 122. Building on the analysis above, several scenarios on Malawi's fiscal sustainability outlook have beenexamined here usingthe low-income country public debt sustainability framework. The framework has been used to explore the impact o f alternative fiscal paths on reducing the domestic debt burden, as well as the several scenarios to assess the vulnerability o f the stabilization program to various types o f shocks. Specifically, the framework has been used to run four scenarios: (i)The historical scenario projects historicaltrends between 1999-2004 forward to 2015; (ii) the 41 Iti s technically possibleto link the MAMS model(used in Chapter One) with the DSA LIC framework. This wouldallow an examinationof: (i) the implicationso f differentcompositions o fexpenditureson growth(using the MAMS)and on debt sustainability (usingthe LIC DSA), and (ii) the implications o f differentfiscal paths (using the LIC DSA) on growth andthe MDGs(usingMAMS). However, such analysis has notbeenpursuedinthis PERbecauseof two reasons. Firstly,as discussedbelow, the results ofthe fiscal scenarios indicatethat the debt sustainabilityanalysisis not very sensitive to changes inGDP growthrates, andthereforethere is little value inexploring(i) above. Secondly, the MAMSresults in Chapter Onehighlightthat the exact progress towards the MDGsdepends critically onthe expenditure envelope, and therefore it is obviousthat faster fiscal adjustmentleads to progress towards the MDGs. 29 Box 2.1 MalawiDebt SustainabilityAnalysis 2006 The macro-framework underlying the DSA i s characterized by implementation of structural reforms to support private sector development, export diversification and growth. Annual output growth is projected to be about 6 percent over the near term, recovering from a prolonged period o f poor performance. It is assumed to drop to 4.5 percent over the longer term converging to the regional average. Private sector investment and FDI inflows are projected to boost export growth inthe long term, leading to an increase in the export share o f the economy from 27 percent in 2005 to 30 percent by the end o f the projection period. Donor support, as measured by net official assistance is assumed to increase gradually and modestly by about 2 percent o f GDP to about 23 percent o f GDP in 2015. It is assumed that even after HIPC completion point and implementation o f MDRI about 84 percent o f total aid, excluding HIPC debt relief grants, i s providedinthe form o f grants. The share o f loans remains relatively l o w amounting to around 3.8 percent. Although the volume o f new loans increases relative to the past five years, the grant element underlying the new borrowing assumptions is in line with past experience. All external debt burden indicators are projected to remain below their thresholds under the baseline simulation. Malawi's N P V o f debt ratio to GDP is projected to increase from 11 percent in 2006 to 24 percent by the end o f the projectionperiod. Similarly, the N P V o f debt to exports ratio climbs from 39 percent in 2006 to 81 percent by 2026. Malawi's debt service to exports ratio i s projected to remain on average at 4 percent throughout the projection period. Stress tests (carried out as part of the standard LIC DSA analysis) indicate that failure to implementreforms, lack of donor financing in the form of grants and exogenous shocks would lead to a substantial deterioration of Malawi's debt burden indicators. Ifkey macroeconomic variables, such as GDP growth, the current account deficit and FDI inflows, would assume their historical average, Malawi's debt stock indicators would breach their thresholds within a decade. Debt stock indicators would also approach their thresholds toward the end o f the projection period if scaling up o f aid would be financed through concessional debt instead o f grants. Finally, under the most extreme stress test, Malawi's N P V o f debt to GDP ratio would be lifted above its indicative threshold by 2011 for a sustained period o f time. Reducingthe large domestic debt burdenis expected to remainthe cornerstoneof the government's fiscal strategy through the medium term. Continued fiscal restraint, further reductions ininterest rates intandem with a sustained reduction ininflation, and debt reliefunder the MDRIare assumed to enable the government to reduce the domestic debt below 10 percent and interest payments to about 2 percent o f GDP. Over the longer term, the baseline scenario assumes that the government will target a stable domestic debt burden o f about 2 percent o f GDP at a cost o f less than 0.5 percent o f G D P in interest payments. External debt relief combined with the government's commitment to reduce domestic debt would lead to a substantialdecline in Malawi's public debt stock. Malawi's public debt stock is expected to drop from 159 percent o f GDP in 2005 to 37 percent o f GDP in 2006 as a result o f HIPC debt relief and the implementation o f MDRI. It i s projected to decline and bottom out around 32 percent at the turn o f the decade. Similarly, the N P V o f public debt would decline from 74 percent o f GDP in 2005 to around 18 percent o f GDP at the end o f the decade. Source: World B a d d I M F (2006), "Malawi: Debt Relief at the Heavily Indebted Poor Countries (HIPC) Initiative Completion Point and Under the Multilateral Debt Relief Initiative (MDRI)", Prepared by the Staffs o f the International Development Association and the International Monetary Fund, Washington DC. Note: The World Bank's Country Policy and Institutional Assessment (CPIA) rates Malawi as a medium performer. Under the joint World Bank/ IMF debt sustainability framework for LICs the corresponding thresholds are 40 percent for the NPV o f external PPG debt to GDP ratio, 150 percent for the N P V o f external PPG debt to exports ratio, 250 percent for the N P V o f external PPG debt to revenue ratio, 20 percent for the external PPG debt service to exports ratio, and 30 percent for the external PPG debt service to revenue ratio. ("Operational Framework for Debt Sustainability Assessments in Low-Income Countries -Further Considevations," SMIO5IlO9, 3129105). 30 baseline scenario (PRGFprogram) reflects the path already chosen by the government, with support by the IMF; (iii)the moderate reform scenario suggests a slower implementation o f reforms; and (iv) the rapid adjustment scenario highlights a faster adjustment path. The baseline scenario is the same as in the LIC DSA, i.e. is underpinned by the same macroeconomic framework and assumptions. The other scenarios explore the impact o f different assumptions regarding the path o f GDP growth and fiscal variables (Le. they are not underpinned by a new and consistent macroeconomic framework for each scenario), and the overall balance i s then assumed to close with domestic financing. The specific assumptions under each scenario are discussed below. 123. The historical scenario simulates the resumption after 2006 o f the historical fiscal trends experienced during 2000-2004, i.e. a return to the unsustainable fiscal policy pursued between 2000 and 2004. The results highlight that continuation o f the fiscal policies pursued during that period would have led to an explosive increase in the domestic debt and its interest bill (Figure2.8). By 2013 the interest payment would have reached 25 percent o f GDP, such that the debt could not be serviced without paralyzing all government activities. Hence, the government would have been forced to default on interest payments, plungingthe country ina financial and economic crisis. 124. The baseline scenario (PRGFprogranz) is the same as under the LIC DSA. I t simulates the macroeconomic stabilization program undertaken by the government since May 2004. Given the gravity o f the fiscal situation, the government, in agreement with the IMF, undertook an ambitious (but realistic in hindsight) macroeconomic stabilization program aimed at restoring a sustainable fiscal balance and reigning in the growth indomestic debt. Annual output growth i s projected to be around 6 percent over the near term, recovering from a prolonged period o f poor performance. It is assumed to drop to 4.5 percent over the longer term (after year 2011) converging to the regional average. The implementation o f structural reforms to support private sector development, export diversification and growth, while maintaining macroeconomic stability characterize the macro-framework underlying this scenario.42 The government i s assumed to continue the tight fiscal discipline followed in recent years, allowing it to reduce the stock o f domestic debt (inreal terms, as a percentage o f GDP), and its interest costs. The results highlight a steady decrease in the stock o f domestic debt reaching 7 percent o f GDP by 2011 and an interest bill rapidly decreasing to around 2-3 percent o f GDP. 125. The moderate reform scenario simulates a slower macroeconomic stabilization program than under the baseline scenario (PRGF program). The baseline scenario (above) assumes an annual GDP growth rates o f 6 percent on average in the 42 Private sector investment and FDIinflows are projected to boost export growth inthe long term, leading to an increase inthe export share o f the economy from 27 percent in 2005 to 30 percent by the end o f the projectionperiod. Donor support, as measuredby net official assistance is assumed to increase gradually and modestly by about 2 percent o f GDP to about 23 percent o f GDP in 2015. It i s assumed that even after HIPC completion point and implementation o f MDRIabout 84 percent o f total aid, excluding HIPC debt relief grants, i s provided inthe form o f grants. The share o f loans remains relatively l o w amounting to around 3.8 percent. Although the volume o f new loans increases relative to the past five years, the grant element underlying the new borrowing assumptions is in line with past experience. 31 medium-term,which is double the growth rate experienced during the past decade. The moderate reform assumes instead that the pace o f reforms is not as sustained as would be required to meet the PRGF targets. Hence, this scenario simulates the impact o f slower GDP growth, at 4 percent, and assumes that revenues grow 2 percentage points less than under the baseline PRGF scenario, and that expendituresgrow 2 percent faster than under the baseline PRGF scenario. The results o f the simulations show that a slower adjustment path as compared to the baseline scenario would not be sufficient to stabilize the domestic debt. After slowing down for a few years, the domestic debt would increase rapidly such that the interest bill would reach 29 percent by 2015, forcing a default and a financial and economic crisis (as under the historical scenario). It should be noted that these (quite dramatic) results result from the assumptions made for the period after 2011. Nevertheless, even with more accommodating assumptions, the simulation highlight the vulnerability o f the current macroeconomic pathto a relaxation inthe fiscal effort. 126. The rapid reform scenario simulates a faster macroeconomic stabilization program than under the baseline scenario (PRGF program). The assumption i s that, following the reaching of HIPC completion and MDRI, additional resources would become available from debt relief and scaling-up o f aid, and could be used to accelerate the reduction in domestic debt (see details below). Specifically the rapid adjustment scenario assumes that about USDlOO million in additional grant resources are channeled towards domestic debt reduction during 2006-2009, resultinginthe full retiremento f the domestic debt stock. The results o f the simulations show a very rapid adjustment path with domestic debt and the interest bill decreasing to zero by 2009-2010. In fact, the faster reduction in interest bill corresponds to savings amounting to 7 percent o f GDP between 2006-2015, compared to the baseline scenario. It should be pointed out, however, that eliminating all domestic debt i s not advisable from a financial development point of vie^.^^ Figure2.8: AlternativeFiscalAdjustmentPathsfor DomesticDebt Reduction goO Domesticdebt (percentof GDP) Interest on domestic debt (percentof GDP) Rapid reform S C L ~ M O(grantshigher by 25%, Ie approx USSIOO) 1I I 70 0 -sModeratereformscenario(exphigher by 2%, revenueslower by 2%, GDP I O 0 growthat 4% pa.) I 40 0 1 100 100 *NO ReformScenariolFiscalbalanccas m history) 20 0 10 0 0 0 1999 2OW 1WI >WZ 1W3 tW4 ZWI 1W6 1Wl 2WB 2W9 2010 1011 2011 2013 2014 101 0 0 +Baseline reform Sccnmo (IMF PRGF baseline) -100 Source: World Bank staff simulations 43Further, the speed o f an ambitious reform scenario that would rapidly reduce the availability o f Treasury bills may need to be moderated in consideration of the limited alternative investment opportunities with an acceptable risk profile available to financial institutions inMalawi. 32 Shock simulations: Potential impact of external shocks (e.g. drought) on adjustment pathfor public debt over the nextfew years 127. The modelhas also been used to explore the vulnerability of the domestic debt stabilization program to various types of shocks, which are plausible given Malawi's recent economic history. The main difference as compared to the shocks carried out under the LIC DSA i s that here we focus on the fiscal impact o f possible shocks, and we also explore the implications o f shocks that have a direct fiscal impact (which i s not part o f the standard LIC DSA). Specifically, five different shocks have been simulated to assess how they would affect performance compared to the baseline scenario. Shock 1 simulates that government expenditures are higher than programmed by 5 percent and that grants and revenues are below program by 5 percent. Shock 2 simulates higher interest rates by 10 percent and a stronger exchange rate by 10 percent. Shock 3 simulates a slower GDP growth rate o f only 2 percent per annum. Shock 4 simulates the impact o f a drought in 2007, such that the expenditures would be higher than programmed by 15 percent (in 2007 only) and the GDP growth rate lower by 5 percent (in2007 only).44 Finally, Shock 5 simulates thejoint impact o f Shocks 1and 3. 128. The results of the sensitivity analysis highlight the vulnerability of the stabilization program to policy slippages and externalshocks. A higher fiscal deficit than programmed (Shock 1, Shock 5) and/or a higher interest rates and exchange rate than programmed (Shock 2), would lead to an explosive increase in domestic debt and force a financial crisis within just a few years (Figure 2.9). Similarly, the impact o f a severe drought (Shock 4) could offset the stabilization program and lead to a resumption in the domestic debt spiral. On the other hand, a slower growth rate than anticipated (Shock 3), by itself does not pose a serious threat to the stabilization program provided all other fiscal objectives are maintained. Figure2.9: PotentialImpact of ExternalShocks on AdjustmentPathfor Domestic Debt Domesticdebt (percent of GDP) Interest on domestic debt (percent of GDP) - - 10 0 Shack I:expendirureshigherby 5%. and revenueand grants lower by 5% 80 0 4-Shack2:interesthigherby10% 70 0 and ERsnongerby 10% 20 0 600 40 0 100 i+Shack 4 GDP shack (draughtlea 100 1 IO-5% GDP growth rate and 15% 20 0 100 0 0 99 ?OW 2WI ZW? 2Wl 2W1 2WS 2W6 2W7 2W8 2 w 9 2010 2011 2012 2013 2014 201 -100 Source: World Bankstaffsimulations 44Inpractice, inthe event ofa drought(or other naturaldisaster)it is likely that developmentpartners wouldprovide additional funding to pay for at leastpart of the humanitarianresponse. 33 Sources ofjiscal risk in the medium term 129. The results of the analysis highlight the importance to monitor and control possible sources of fiscal slippage. Areas requiring special attention include management o f the payroll management and control o f wage drift, resisting pressure to increase expenditures on fertilizer subsidies, accumulation o f new arrears, and monitoring borrowing by parastatals and the risk o f bailouts. Each o f these areas i s described inmore detail below. 130. Continuingto strengthen management of the payroll and controllingthe wage drift. Weak payroll management has underminedbudget implementation inrecent years and remains a major concern. The government has started a program o f reforms to increase civil servants wages and improve payroll management. As a share o f GDP, civil service compensation in Malawi i s now in line with other countries' in the region. In designing future wage adjustments, the challenge will be to contain the overall payroll at current levels relative to GDP. However, the government will need to make difficult decisions about priorities within this overall envelope. Future wage adjustments will also need to ensure sufficient compensation at the lowest employee grades to cover the cost o f living. The government will have to balance the need to contain growth inthe wage bill while at the same time providing adequate incentives for the civil service and strengtheningeconomic management capacity. Pension outlays could also turn out to be higher if the demographic assumptions underlying the current reforms turn out to be incorrect. To reduce these risks, the authorities need to continue implementingreforms in payroll management, inparticular, ensuring that the civil service database remains up-to- date, and improve the compilation o f data related to current and hture retirees and deaths. 131. Controlling the costs of fertilizer subsidy and adopting a sustainable policy for agricultural productivity enhancements. The cost of the fertilizer subsidy has been increasing rapidly in recent years, and in 2006/07 its cost was over 7 percent o f the budget,and around 3 percent o f GDP, and amounting to around 70-80percent o ftotal fertilizer consumption (with negative impacts on the private sector operations in fertilizer markets). Current plans for government interventions in fertilizer markets envisage a long term price subsidy, which i s neither sustainable nor constitutes a comprehensive approach towards agricultural productivity enhancements. In light o f the approaching elections season, it i s likely that pressures will build to increase fertilizer subsidies even further. It i s important, therefore, that the government develops a sustainable medium term policy for agricultural productivity enhancements, which places the role o f fertilizer subsidies in a broader context, and within a fiscally sustainable (and market-friendly) strategy for promoting productivity growth and food security. 132. Avoiding the accumulationof new arrears remainsa source of fiscal risk. The government has been making substantial progress in repaying the large stock o f arrears accumulated in past years. In addition, the government has discussed with creditors a plan to securitize a portion o f the stock o f domestic arrears into government bonds (with a maturity o f Ito 3 years). Weak accounting systems pose a risk that new expenditure arrears may be accumulating without being reported. Notably difficulties continue to be experienced with the delayed payment o f utility bills by government 34 ministries and agencies. The introduction o f the IFMIS should help prevent new utility arrears. Administrative measures have also beendeveloped to preventnew utility arrears, such that the government will centralize the payment o f utility bills o f government entities which are not included inIFMIS to prevent hture utility arrears. It i s crucial that government continues to monitor the efficacy o f these measures, and act promptly at any signo f accumulation o fnew arrears. 133. Monitoring parastatals' contingent liabilities and reducing risks of bailouts: There is a significant risk that the unbudgeted bailout o f parastatals could increase pressure on the fiscal reform program over the coming years. This risk has significantly increased with the new domestic borrowing by ADMARC and the Smallholder Farmers Fertilizer Revolving Fundand the recourse by ESCOM to external commercial borrowing in 2006. Potential contingent liabilities to the budget need to be managed by ensuringproper governance inthe administration o f parastatals. Inaddition, the government could introduce measures to strengthen its oversight o f parastatals' performance and reduce the risk o f future parastatalbailouts.45 Exploring options to restructure the domestic debt and reduce the interest bill 134. Continued fiscal discipline is the key pre-conditionin any credible debt reduction strategy. Given the (decreasing but still) high cost o f the domestic debt burden, it is usehl to consider whether there are additional strategies that can be pursued to reduce the debt costs to the economy. Several countries have implemented debt reduction strategies, ranging from strengthening macro environment to reducing debt stock, swapping with external debt and restructuring domestic debt. Common to all strategies aimed at debt reductionor restructuring i s the firm resolve and commitment of authorities to strengthen fiscal and monetary policies and ensure macro stability and debt sustainability. The recent unsuccessful experiences o f Cape Verde, Lebanon, and Kenya all highlight the vulnerability o f debt reduction strategies if they are not underpinned by serious commitment to fiscal discipline. Other reforms, aimed at strengthening the macro environment, would reinforce credibility in domestic debt markets, thus allowing the government to float longer maturity debt and borrow at lower interest rates. 135. In fact, there is limited scope for alternative measures to reduce the domestic debt burden. Given the state of development of Malawi's domestic debt market, the viability o f options for restructuring domestic debt are limited. We briefly discuss a few possible strategies: (a) debt retirement through exceptional resources, (b) debt swap with low-cost external funds, (c) diversify financing instruments, and (d) set- up a Trust Fund. Debt retirement through exceptional resources 136. A balance must be struck between investing any additionalresources (on infrastructure and social services), increasing Malawi's low external reserve coverage to protect against exogenous shocks, and retiring domestic debt. A 45The Public Enterprises Monitoring Unit (PERMU) inMinistry o f Finance i s supposed to prepare quarterly monitoring reports that, inter alia, include at least liability estimates o fparastatals. Such report, however, have not been readily available in recent years. 35 preliminary review suggests that exceptional financing resources o f up to [USDlSO million] annually could become available (from debt relief and additional foreign grantd~redits).~~Reaching the HIPC completion point and accessing MDRIrelief will have a relatively modest impact on net aid flows to Malawi, at around USD25 million annually starting in2007/08. While substantial scaling up o f donor aid inthe short term i s unlikely, an increase of up to USDIOOannually mainly from the World Bank, MCC and DFID looks possible over the next few years.47 Little additional resources are envisaged from privatizationproceeds. 137. The government could retirea portionof the higher cost domestic debt by usingexceptional financingresources.This strategy was pursuedby Ghana, Nicaragua, Lebanon and Kenya.48For example, a rough calculation suggests that if the government uses around USDIOO million o f the additional resources to retire the domestic debt during 2006/07 to 2009/10, this would allow the government to retire an additional M K 1 4 billion in domestic debt annually, and reduce the high cost o f servicing the domestic debt by 2009/10. This strategy would result in savings on the interest bill amounting to around 7 percent o f GDP (or USD200 million) in total over the 5-year period 2006/07 to 201l/12.49 To implement this strategy, the government would needto: (i) ascertain the amount o f resources that might become available for a debt retirement; (ii) examine the detailed terms o f each loan to calculate the maturity structure and interest costs o f each loan, in order to decide which loans should be retired first; (iii) into take account the monetary impact o f the operation, which would have to be sterilized. Debt swap with low-cost externalfunds 138. The high-interest debt instruments could be swapped with low-cost funding. Inrecent years, the interest rate differential on Malawi's external and domestic debt has been significantly large, averaging nearly 35 percentage points on an annual basis between2000 and 2005. While the Kwacha depreciated over this period by around 15 percent every year on average, the significantly large interest differentials have well outstripped the Kwacha depreciation by around 20 percentage points on average. Improved economic performance since 2004 has slowed down the annual rate o f depreciation and domestic interest rates have also been reduced substantially. Hence the average interest differential net o f depreciation has also reduced. Nevertheless, ifMalawi had access to external concessional loans it might benefit from interest costs savings by substitutingdomestic debt with external concessional debt. The actual amount o f interest savings would depend on the amounts considered for the swap and the actual 46For comparison,net official aidhas increasedfrom about 16percentof GDPbetween2001/02-2003104 (or about $3 10 million) to about 19 percent o f GDP in2004/05-2006107 (or about $440 million). 47The WorldBankapprovedinFebruary2007 a new four-year Country Assistance Strategy which could provide Malawi some $20 million more annuallythaninitially anticipated. Malawimay also become eligible for funding from the MillenniumChallengeAccount from the UnitedStates that couldamount to aroundUSD250million over four years. A modest steady increase inDFID supportis anticipatedinthe shortto mediumterm. Other donors havenotyet committedhigheraid flows. 48For instance, inGhana20 percent of the proceeds fromHIPCreliefwere usedfor domestic debt reduction. Nicaraguain2002, supportedby a PRGF, decidedto reducedomestic debt by using privatizationreceipts, asset recoveryproceeds, and grants. 49Malawiwould spendonly $138million ininterestpaymentsover the period2006107 to 2011/12, comparedto $337 million without the additionaldebt retirement. 36 concessional rates on the new debt. As the fiscal front stabilizes, a shift o f debt from domestic to foreign concessional sources also presents the advantage to reduce the overall risk of the government's debt portfolio (by reducing interest rate and refinancingrisks). A key risk to consider when substituting domestic debt for external concessional debt, however, i s the risk that a substantial devaluation may wipe out the gains from lower interest payments on foreign debt, making this debt just as costly as high interest domestic debt. 139. While several countries have successfullyapproached donors for this type of operation, it i s unclear that additional funding would be available to Malawi." For Malawi, the additional IDA resources mentioned above could be used for this type o f operation. I t i s unlikely, however, that substantial additional concessional resources would become available from other donors. As in the case o f debt retirement, the first step would be to ascertain the amount o fresources that mightbecome available for a debt swap. Once the hnding has been secured, the government would have to examine the detailed terms o f each loan to decide which loans should be restructured first, and the monetary impact o f the operation that would have to be sterilized. Diversibjnancing instruments 140. Diversify financing instruments to increase the liquidity and stabilize the demand for government securities. The diversification o f instruments would not reduce the debt stock in the short run, but would reduce the interest costs and the refinancing risk. Options to diversify and float instruments in Malawi are presently limited given the shallow financial markets, and diversification will only be feasible in the mediumterm. Possible instrumentsthat could be considered inthe mediumterm, to pursue financial deepening and lengthen the maturity o f government debt, could include floating rate bonds, inflation indexed bonds and bonds with embedded calVput options (see Annex 2.1 for a description). Such instruments, however, entail a reduction o f the interest bill at the cost o f assuming increased exchange rate and inflation risks inthe debt portfolio, which needs to be carehlly assessed. Before introducing any o f these instrumentsin the market, the RBMwould need to consult with market participants and investors to ascertain their understanding and ability to participate in such issuances.51 An agreement could be reached to issue such instruments on an experimental basis to explore the market appetite and reaction o f investors. In general, a track record in fiscal discipline and macro stabilization are pre-conditions for all these instruments. 141. Consolidating Treasury bills and re-issuing them into relatively longer dated paper would reduce refinancing risks. Issuing long-dated paper would entail a higher cost, but would reduce refinancing risk. In order not to increase the coupon the RBM could consider revising the liquidity reserve requirements (LRR) to allow 50Nicaragua, for instance approached multilateral creditors to request support to reduce domestic debt burden. Kenya obtained partial substitution o f domestic debt by foreign concessional loans. InLebanon a different variant was followed. At the Paris I1Conference, a number o f donors agreed to lend Lebanon on favorable terms, which the government then used to repay both domestic and external debt, either as it fell due or through repurchase inthe market. Notably, the RBMwould need to consult with the quarterly forum o f commercial banks' chief executives to ascertain their views and appetite for such instruments. All chief executives o f commercial banks meet with the RBMevery quarter to discuss issues relating to the markets. 37 commercial banks to hold longer term bonds as part o f LRR. At present the LRR for banks i s 17.5 percent o f the total demand and time liabilities, and includes the holding o f vault cash and deposits in discount houses. In order to elongate the maturity o f treasury bills, the RBMcould issue longer term bonds (e.g. 2 years), which could be permittedas part o f bank's holdings o f LRR.52Prior to issuing longer paper, the RBM needs to consult the Forum o f chief executives to keep them informed and to assess the market liquidity. Setting-upa TrustFund 142. A Trust Fundcould also be usedto re-profilethe interestand maturity of the existing debt. This operation would entail converting existing debt into new longer maturity bonds to be issued and serviced by a Trust Fund. The Trust Fund would be financed by earmarking foreign currency receipts and grants. The objectives would be to: (i) theinterestcost inthe shortterm; (ii)bringdowndomestic debttoprudent reduce levels over the medium term; (iii) minimize the operation's liquidity impact; and, (iv) mobilize foreign resources. Clearly this type o f operation would first require agreement by the current holders o f domestic debt. In this context, a key aspect o f the strategy would be the ring-fencing the funds such that they can be utilized only towards debt reduction. This operation i s expected to reduce interest cost by increasing creditors' confidence, and to allow the government to redeem domestic debt (Le. buy back the trust fund) over the medium-term, without increasinginflationarypressures.53 143. A Trust Fund would allow substantial benefits to the government. As compared to strategy o f debt retirement and/or debt swap discussed above, a Trust Fund offers the advantage to minimize the monetary impact o f the debt re-profiling. As the monetary and fiscal conditions permit, the government would buy-back the securities from the Trust Fundover a mediumto long term time horizon (around 20 years), thereby permitting retirement o f the debt overhang without injecting inflationary pressures into the economy. The risks associated with this strategy are minimal. Exposure to foreign exchange risk would have to be assessed, but following HIPC completion point and the MDRI, risks can be expected to be low. This strategy has been tried inCape Verde and Lebanon, but the success has beenmixed. 144. Setting up a Trust Fund would entail mobilizing the required funding and carrying out a loan by loan analysis to identify a restructuringplan. Again the first step would be to discuss with donors ascertain the amount o f resources that might become available for a debt swap. Further the creditors have to agree to the terms o f the debt re-profiling. Apart from credible steps towards fiscal consolidation, implementation o f such a strategy would require meetings with donors to persuade them o f the government commitment towards fiscal prudence and reduction o f its debt burden. 52It shouldbe noted,however, that the preferentialtreatment given to discounthouses, allowing commercialbanksto count their deposits with discount housesagainst the LRR,inan attemptto stimulate the secondarymarket has notworkedwell. It is also regardedas anundesirablemarket distortionby many financialsector playersinMalawi. 53Dependingonthe country conditions (inflationlevels, exchangerates, absorptive capacity of the banking system), countries adaptthis scheme by either issuingforeign currency bonds or domestic bonds, or setting the Trust Fundabroador domestically. 38 2.4 DebtManagementFramework 145. The assessment of existing legal and institutional arrangements to manage public debt highlights several areas for improvement. An assessment o f Malawi's institutional arrangements for debt management was carried out in 2000 by the World Bank, the Macroeconomic & Financial Management Institute o f Eastern & Southern Africa (MEFMI), and the South African Debt Office (MEFMI et al, 2000).54 The report highlighted weaknesses in: (i)the institutional arrangements for debt management, notably in the operation and staffing o f the Debt and Aid Division o f the MOF; and, (ii) the coordination among the various institutions that participate in debt in management operations, both within the MOF, and with the RBM. As discussed below, progress has been made by the Government since 2000 in establishing effective debt management practices, setting and operationalizing the CS-DRMS software system and inpreparing the draft NationalDebt and Aid Policy. Inseveralother areas there has been limited progress, however. Institutional arrangementsfor debt management 146. Malawi's legal and institutionalframework for debt management clearly defines roles and responsibilities; but operationalguidelines for implementationare needed. Under the 2003 Public Finance Management Act, the Minister o f Finance is the only authority to contract public debt, including the authority to raise domestic debt through the issue o f Treasury bills, bonds and notes. The Debt and Aid Division (DAD) o f the MOF i s responsible for managing and recording external debt. Being the fiscal agent o f the government, the RBM i s responsible for making all payments to external creditors upon receipt o f payment instructions issued by the MOF. The RBM i s also responsible for contracting and managing domestic debt. Hence, in practice, debt management functions are shared between the MOF and RBM. Implementation o f the existing legal framework would be facilitated by specific guidelines on issuing guarantees, local government borrowing, on-lending and private borrowing, degree o f concessionality on external debt, and borrowing limits. Also guidelines could clearly attribute regulatory jurisdiction over domestic debt instruments, participants, trading platforms, and clearing and settlements systems. 147. The organizationofDADinto a front office, middleoffice andback office, i s not fully operational in practice. DAD i s organized in three divisions, with an additional division being set up to deal with aid issues (Figure 3.9).55 The Resource Mobilization Division (RM) i s responsible for the functions o f the Front Office, the Planning and Information Division (PI) i s in charge o f performing the middle office 54 The study evaluated the main aspects of Malawi's debt management, cash management and domestic debt marketdevelopment. 55 The typicalhnctionof a front office is to implementthe government's debt managementstrategyby identifyingnew sources of funding, negotiatingthe terms ofthe new loans, andgenerallymaintaining and improvinginvestorrelations.The typicaltask o f a middle office is to prepare andmonitor the implementationo fthe government's debt strategy. The responsibility o f a back office is the settlementof the daily transactionstakenby the front office, ensuringthe validation andverificationofthe information. It is also incharge o f debt recordingandmaintainingthe debt databases.Havinga sound debt recording systemis fundamentalto independentlydeterminedebt servicingpayments andto pay on a timely basis. 39 functions, and the Disbursement and Debt Servicing Division (DDS) i s in charge o f performing the Back Office tasks. The current organization provides for duality between staff working with preparation o f loan agreements, and those settling and accounting the transactions. However, there i s no separation between staff entering and checking data entries in the debt recording system; this duality i s a necessary organizational preconditionto ensure data quality. 148. Scarcity of qualified staff is the major problemwithin DAD. The work of DAD is hampered by the weak capacity o f the DAD staff in debt assessment and monitoring skills. The actual number o f staff i s sufficient if properly trained. The high turnover o f trained staff has resulted in the remaining staff being inadequately trained. The Front Office (FW Division) staff i s not trained to assess the degree o f concessionality o f new borrowing and to assess the risks from different financing sources and terms o f new loans. At presentthis i s not a concern since most o f the borrowing i s at set terms from multilateral creditors (e.g., IDA, AfDB). However, it expected to assume greater importance as the profile o f Malawi's creditors evolves. The Middle Office (PI Division) has not articulated a formal debt management strategy due to lack o f technically qualified staff, and lack o f a clear mandate for the role o f the middle office. The Back Office (DDS division) i s responsible for the accuracy o f debt service payments, but is hamperedby weak staff capacity and the lack o f an administrator. There is no staff dedicated for risk monitoring and compliance. There are no operational risk management plans, business continuity or disaster recovery plans. 149. Data recordingand reportingneed to be strengthened. Debt records as o f end-2005 are complete and have been validated against creditor advice and bank statements. This i s the result o f several years o f work to verify the information in the database, and preparations for HIPC completion point accelerated the process o f data Figure2.9 InstitutionalStructurefor DebtManagement I * Ministry of Finance I I Debt and A i d ManagementDivision (DAD) I______________. Resource Planningand Disbursement and ; II I DomesticDebt I IAid Coordination I Mobilization Information Debt Servicing Section Unit I Division Division Division II(Not operational) I (Not operational) I Front office Middle office Back Uj'ice I DomesticDebt I IUnit is expectedto I -Implements -Prepares and Performs operations Division in Reserve I I become operational I government's debt monitors IIBankofMalawi i s in2007108. I management strategy implementation of - monitoring II I -identifies new government's debt I I I responsiblefor I I issuing and I Staff: 6recmitedin -sourcesof funding strategy -disbursementsexternal servicing o f IImanagingdomestic I I 2006107 I negotiatesterms o f ; debt 1 1 new loans Staff: 4 out of - debt recording and 1 1 II - maintains and recommended 10 maintaining the debt I I 1 I I I I I improves investor databases I 1 1 I I ` I I relations. I 1 1 I Staff: 12 out of I 1 1 I I 1 1 I Staff: 6 out of recommended 14 I 1 1 I I 1 1 I recommended 19 I 1 1 II 40 reconciliation. There are, however, no documented policies and procedures for accessing the database. In the past, external debt data and HIPC debt relief provided by creditors was reconciled on an ad-hoc basis, and this resulted in inaccurate forecasts o f debt servicing obligations and discrepancies betweenbudget estimates and actual debt service payments due. The recording is conducted by various institutions; DAD for external debt and guarantees, RBMfor domestic debt and statutory bodies for their debt. The RBMand MoF have the same debt recording system (Commonwealth Secretariat DRMS). There are no electronic file interfaces and the compilation i s subject to problems and delays. 150. The domestic debt management is carried out by the RBM. For the present, given the accumulated knowledge and synergies that the RBMhas with respect to the financial markets and monetary policy, it may be more efficient for the RBM to continue to manage the operations related to domestic debt. However, over the medium term there i s a need for MOF to assume control o f the management o f domestic debt and limit the role of the RBM to that o f executing agent.56 This would provide greater flexibility and autonomy to the RBM in managing inflation and interest rates, and it would enhance accountability o f both monetary and fiscal policy. Coordination, information flows and reporting in debt management operations 151. Channels of communication between the MOF and the RBM have improved,but need to be strengthened further. Although the Budget Division meets regularly with the DAD and the RBM to discuss macroeconomic policy, including debt related issues, there i s little formal coordination between the government's cash management, debt management, budgeting, and monetary policy. There are currently two technical committees operating: the Monetary Policy Committee (MPC), which meets every month to discuss recent developments in the economy and to recommend appropriate actions; and a Committee which meets every week to monitor progress towards PRGF targets, which includes the MoF, including DAD, and the RBM, but the meetings seem to focus more on budget and monetary issues; debt issues are residual ifat all discussed. Proper communication between the various divisions involved in debt management inMOF and RBMi s desirable, and information sharing would result inless bunching o f repayments and enable lowering o f interest rates. The national debt management strategy 152. Malawi lacks a comprehensive debt management strategy. Although the government has increased the share o f domestic debt in total debt to meet its financing needs duringthe past years, it does not analyze the costs and/or risks associated with the different financing options, and does not have a forward-looking medium-term debt management strategy. Minimizing borrowing costs i s important to ensure that Malawi remains on a fiscally sustainable path after having reached its HIPC completion point. 56Debt management arrangements vary across countries: (a) central bank exclusively undertakes all activities concerned with debt management; (b) external debt management is handled by Government and internal debt management by central bank; (c) Government manages both external and internal debt, and central bank functions as agent to central government and does book keeping; and, (d) separate autonomous organization undertakes both external and internal debt management. 41 As a stop-gap measure inthe absence of a comprehensive debt management strategy, the government plans in 2007 to develop and issue guidelines for the entire public sector to ensure that new external borrowing i s sufficiently concessional and accumulated within sustainable levels. At the same time, the government has started preparation of a debt strategy, which it plans to finalize in 2008. The strategy should clarify the legal, institutional and administrative framework for public debt management, including addressing key weaknesses in current arrangements, notably the limited information sharing between DAD, RBM and budget-making departments, and the inadequate preparedness for loan negotiations. More broadly, the strategy should lay down the framework, policy objectives and strategies for public debt management, and should be based on a risk management framework that formulates alternative assessments o f expected costs and risks for the borrowing options available to the Government. 2.5 RecommendedPolicyPrioritiesto ReducePublicDebtVulnerability 153. Fiscal discipline i s necessary for debt sustainability and domestic debt reduction. Malawi macroeconomic situation remains fragile, and a relaxation in fiscal discipline could lead to a rapid resurgence in domestic debt. In this context the government needs to take steps to control potential sources o f fiscal risk which could jeopardize fiscal sustainability in the medium term, including: the payroll management and possible wage drift, the size o f the fertilizer subsidy, the possible accumulation o f new arrears, and the financial performance o f parastatal. Complementary strategies to reduce the domestic debt are limitedby the availability o f additional financing and access to additional concessional resources. Similarly, the interest savings from restructuring the debt portfolio are likely to be small, and come at the cost o f assuming increased exchange rate and inflation risks. 154. Accelerating the reduction in domestic debt by allocating new resources to retire the domestic debt stock in the short term, would bring additional savings that can be channeled to priority expenditures. There is a tradeoff: the greater fiscal adjustment in the short term, the faster the reduction inthe interest bill, and therefore the greater the overall resources available for priority expenditures over the medium-term;on the other hand, the higher the government expenditures inthe short term, the longer the adjustment period during which the country pays a large domestic interest bill. In addition, the longer the adjustment period, the higher the likelihood that external shocks could disrupt, or at least prolong the period during which the country i s paying a large interest burden. The opportunity o f more ambitious domestic debt reduction, however, needs to be examined in the context o f the limited level o f development o f financial markets. A more rapid debt reduction strategy would need to be balanced by the consideration that Government debt currently plays a big role for both banks and non- bank financial institutions (pensions, insurance funds), as it provides the core o f their low-risk investmentportfolio. 155. Reduced financing from the RBM and developing the domestic debt market to diversify the investor base and lengthen the maturity curve are also important. In recent years, efforts by the RBM to extend the domestic debt maturity profile have been constrained by the lack o f macroeconomic stability, a liberalized interest rate environment and underdeveloped domestic markets for government bonds. 42 Allowing for favorable market conditions, it i s expected that the authorities will issue more Treasury notes including longer-dated ones, e.g., with a maturity o f 2-5 years. 156. The many risks associated with Malawi's debt portfolio emphasize the need for management and institutionalreforms, and debt market development. The characteristics o f Malawi's domestic debt portfolio, including the limited diversity o f financing instruments and their very short-term maturities, combined with the shallow debt markets, entail substantial exposure to refinancing risk, interest rate risk, and other fiscal risks. Reforms of the regulatory framework and institutional arrangementsfor debt management 157. The 2003 Public Finance Management Act should be complemented by a set of guidelines on issuing guarantees, local government borrowing, on-lending and private borrowing, degree o f concessionality on external debt, and borrowing limits (rates o f accumulation). Also guidelines could clearly attribute regulatory jurisdiction over domestic debt instruments, participants, trading platforms, and clearing and settlements systems. 158. The government should finalize and implement a medium term debt management strategy, to improve the institutional and administrative framework, reduce and restructure domestic debt, while enhancing reserve management. Several issues need to be considered: The operation of DAD should be strengthened by establishing clear division of responsibilities across its sections (front, middle and back office), and increasing staff skills. DAD'Sstaffs need to be equipped with financial and analytical skills to produce regular debt sustainability analyses (DSAs), together with comprehensive currency-wise and maturity-wise risk analyses. Capacity building is also required in interpretingloan agreements and reports sent by the creditors and implementthem in the debt recording system. Also, additional staff with skills in financial management, accountancy or IT should be hired particularly for the middle office. The M o F may recruit new staff, train both old and new staff in debt management skills, and/or staff could also be transferred from RBMunder a secondment scheme. Improve the coordination between the Ministry of Finance and the RBM. A formal structure for institutionalized information sharing among debt managers, cash managers, budgetunits and the central bank should be established, to enable accurate and timely information flow. The budget making departments could share expenditure-revenue or outflow-inflow information with the debt managers ina coordinatedmanner, to alleviate sudden market issuances. Strengtheningsuch coordination between the DAD, the budget-making departments and the RBM would improve both cash and liquidity management. This could be easily facilitated through a formal committee structure. Rendering operational the Debt Management Committee (DMC) would enable better institutional coordination between the MOF and RBM, and ensure timely sharing o f relevant information for debt policy formulation. In addition, the MOF should reactivate the Cash Flow Management Committee (CFMC) to ensure coordination between the agents 43 responsible for monitoring the government's need for liquidity, and the MoF which i s responsible for issuance o f domestic debt. A rule based debt management strategy with a statutory ceiling on public debt could be adopted. The draft debt management policy proposes that new external loans should not be contracted that increase future debt service to export ratio beyond 15 percent. Though somewhat ad-hoc, in a long-term perspective this strategy could improve debt sustainability and reduce fiscal imprudence. A more direct and comprehensive approach to fiscal sustainability would entail a limit on the overalldebt service as aproportiono f government revenues or GDP. 159. Weak debt management stems in part from lack of appropriate cash and liquidity management by government. A proper cash management strategy would imply that government's periodic expenditure needs are managed in tandem with its revenue inflows. The lack o f such a strategy leads to mismatches between inflows and outflows, leading to large borrowings, often at high interest rates. The recent implementation o f IFMIS, and the centralization o f all central government bank accounts into a Single Treasury Account, has reformed and streamlined budget execution in Malawi. The government now needs to improve its cash flow forecasting capacity and link it to the central bank's liquidity management.57 A recent IMF report in 2007 provides an in-depth analysis and suggests measures to strengthen cash management systems in Malawi, and the government has started implementation o f the key recommendations (IMF, 2007b). Reforms tofoster debt market development 160. To enhance the functioning of the primary market, in the medium term the DAD should consider publiclyreleasing an issuance calendar,setting out in advance tentative dates and amounts of government borrowings. The Government, through the central bank, conducts weekly auctions o f Treasury Bills. The mandate given to the central bank i s mainly to roll-over outstanding debt. The announced volumes for the respective maturities are indicative and allotted amounts often deviate from the announced amounts. In the medium term, once the macroeconomic situation allows the government to issue long term papers in a sizable manner, announcing, a calendar would facilitate better planning by market participants and smooth out supply-demand matching o f for government's issuance^.^^ It would also enable better liquidity management by financial institutions. 161. Plans to introduce a primary dealer system need to be carefully assessed, as the system presents substantial risks in the case of Malawi. The RBM has developed guidelines which were discussed with the forum of chief executives o f commercial banks for their views to introduce a primary dealer system. This would prevent the need for RBM to be the first buyer o f government debt and enable lengthening the maturity structure. Regular contact with non-bank financial institutions, such as pension funds could be maintained with a view to enabling issuance o f longer term securities. However, there have been mixed experiences with the primary dealer 57The RBMhas developed a liquidity model with assistance from the IMF, but the inadequate policy coordination between RBMand M o F has hampered the accuracy o f the model. 58Infact, for short-term papers, an issuance calendar is unlikelyto work without sound cashmanagement. 44 system. It generally works best incountries inwhich there is an active range o f potential investors, which would not participate in auction but could be reached by financial institutions (i.e. the primary dealers). InMalawi, however, this may not be the case, and introducing a primary dealer system could instead have the effect to restrict competition in the primary market, pushingup the interest rates. For instance, the experience in Tanzania shows that there i s a potential for primary dealers to form cartels and perhaps raise prices evenmore. 162. Foster the secondary market for government securities. An active and liquid secondary market for government securities would mitigate high financing cost of the government debt in the medium- and long-term. It would help extend the government's securities yield curve, which would help the government reduce its rollover risks, and serve as first step for introducing new debt instruments.Consolidating Treasury bills and elongating maturities would increase fungibility and enhance prospects for secondary market development. Further, the secondary market can be fostered by reducing the high withholding taxes on government securities trading, which discourage discount house from trading ingovernment securities inthe secondary market. 163. Establishsuitable clearing and settlement mechanisms. InMalawi trading i s on equities, treasury bills and local registered stocks and clearing i s done by transaction. Settlement i s on equities (thus trading plus seven days). There are no custody systems in Malawi as such securities are physically held by institutions. Brokers are the intermediaries e.g. Stockbrokers MW Limited. 45 CHAPTER EDUCATION 3: 3.1 Introduction 164. This public expenditure review concerns the activities o f the Ministry o f Education (MOE) and the two public universities (for a detailed description o f the education sector inMalawi, see Annex 3.1). The education sector i s an integrated system that needs all three levels, but government funding o f the primary sub-sector i s particularly important due to the significant public returns to investment o f basic education, and its demonstrated correlation with reduced levels o f poverty. 165. This chapter analyzes the expenditures o f the primary, secondary, teacher education and higher education sub-sectors, and proposes a series o f actions to improve efficiency o f spending in the sector. It also describes the impact o f the HIV/AIDS pandemic on the education sector, and attempts to quantify its impact on the costs o f providing education. Progress in implementing the recommendations in the 2000 PER i s reviewedand new policy recommendations are prescribed. 3.2 Key education sector issues Studentjlows 166. Primary enrollment is still considerably higher than school-age population,but the GrossEnrollment Ratio (GER) may be overestimated. According to the annual school census (known as EMIS), enrollment in primary education stood at 3.2 million in 2004, while the school-age populationinMalawi is estimated to be around 2.4 million. This gives a GER5' for primary o f 132 percent in2004. Moreover, there are wide differences between the GER for standards 1-4 (I 86 percent) and for standards 5-8 (74 percent). However, household surveys have consistently found the GER to be lower, which i s more consistent with what we would expect 10 years after Free Primary Education was introduced (by which time most o f the over-age students who enrolled in 1994 should have exitedthe system). 167. The number of new entrants to primary education seems to be overestimated. The high GER is often explained by a large number o f repeaters, but according to the 2004 school census, 666,000 out o f 873,000 pupils enrolled instandard 1 were non-repeaters. However, there were only about 321,000 6-year olds according to NSO population estimates, so the Net Intake Rate6' (NIR) to primary was twice the number o f children inthe corresponding age group. Infact, the NIRhas been around 200 percent for the past 10 years, and such a sustained high NIR clearly suggests a data problem. 59 The GER is calculated as the total number o f pupils at a certain level o f education relative to the number o f children inthe relevant age group inthe population. For primary education, the relevant age group for Malawi is defined as 6-13 years, and for secondary 14-17 years. 60The Net Intake Rate i s the number o f non-repeaters in the first grade divided by the number o f children o f relevant age inthe population. For primary education, the relevant age group for Malawi is the 6-year olds, for secondary the 14 -year olds. 46 168. Despite the high primary GER, at any given time 15 percent of school- age children are not going to school. The IHS2 survey shows that school age children (6-13 year olds) may be out o f school, because they have not yet started school, have dropped out of school, or will never go to school. School attendance and literacy rates are highest inthe Northern region. 169. The GER in secondary education is 17 percent. In 2004, the school census reported a total secondary enrollment o f 180,000, translating to a GER o f about 17 percent o f the corresponding age group (14-17 year-olds). A much larger share (23 percent) o f the student body is enrolled in private secondary schools than in primary. However, enrollment data for secondary education has been o f variable quality up to 2004, due to frequent under-reporting by schools. Regional differences in access to education appear to be evenlarger at the level o f secondary educationthan for primary. 170. Tertiary enrollment is 55 per 100,000 inhabitants. Higher education is mainly provided at the two public universities, University o f Malawi and University of Mzuzu. Universityenrollment was around 5,900 at the two public universitiesin2004, up from 3,600 in 2000, bringing tertiary enrollment per 100,000 inhabitants up to 55 in 2004. 171. The gender gap in primary enrollment is closing but gender parity has not yet been attained. In2004, girls made up 50 percent o f primary school pupils as opposed to 47 percent in 1994/95. However, in standard 8 only 43 percent are female, and among those who pass the PSLC exam at the end o f the primary cycle, just 39 percent are female. 172. There are noticeableimprovements in gender equity at secondary and tertiary levels, but hostel capacity i s a limiting factor. The selection to secondary schools gives preference to admission o f girls, ensuring a 45 percent share o f girls in Form 1 in 2004 (compared to 36 percent in 1994/95), even though girls made up only 39 percent o f the PSLC passers. However, the number of hostel spaces i s larger for boys than for girls, limiting equal access to elite secondary schools. At the University o f Malawi, women made up 34 percent o f students in2003 compared with just 26 percent in 2000? Student repetition, dropout, and completion rates 173. There is still a way to go along the road to Universal Primary Completion. The most reliable indicator o f the proportion o f children that complete primary school i s the Primary Completion Ratio.62For completion o f the whole cycle, the PCR was only about 34 percent in 2004 (Table 3.1). The Proxy PCR63for standard 5 i s 61 The sources o f datafor this paragraphare EMIS andthe NationalExaminationBoardMANEB. 62 The PrimaryCompletionRatio is calculatedas the numberof students completingthe primarycycle (PSLC passers) dividedby the populationof 13 years old. Another indicator o f systemperformancei s the survival rate.However, this indicator depends on standard 1new enrollment, whichappears to be inflated inMalawi(see paragraphonNetIntakeRate). 63 The proxyPCR is an indicator of the levelo fcompletiono f a certainnumber o fyears of schooling, and i s usefulinMalawiwhere the primarycycle i s relativelylong. The proxy PCRfor standard 5 i s the new enrollmentinstandard5 (enrollment minusrepeaters)dividedby the populationof 10 year olds. 47 2000 2001 2002 2003 2004 PCR (PSLC passers) 42% 37% 38% 36% 34% Proxy PCR (reached standard 8) 51% 53% 52% 47% 46% Proxy PCR (reached standard 6) 71% 74% 74% 67% 69% Proxy PCR (reached standard 5) 98% 102% 93% 84% 86% 175. Secondary education is producing more graduates, but more than 50 percent of students still fail inthe MSCE exam. More students sit for the MSCE exam at the end o f secondary, and the MSCE pass rate has also improved (45 percent in 2004 compared to 20 percent in2000). However, 50 percent o f those who sit for the exam fail, indicatingthat a large share o f the resources spent on secondary education are wasted. 176. Repetitionand dropout rates are not improving. High rates o f repetition and dropout are endemic to the primary system, and neither repetition (on average 17 percent per grade) nor dropout (about 16 percent per grade) rates have been improving in the past couple o fyears. Incontrast, repetition insecondary education is the fairly low, at 3.7 percent on average per grade. Dropout in secondary i s 12 percent on average per grade. Inboth primary and secondary, the dropout and repetition rates for girls are higher than for boys, and there are fewer dropouts in the Northern region compared to the Central and Southern regions. Across regions, the proportion o f repeaters in primary school was comparable. 177. On the demand side, direct and indirect costs, poverty and malnutrition contributeto dropout, repetitionand absenteeism. According to IHS2 respondents (in 2004/05), the cost o f schooling i s the major cause for the high rates o f drop out in primary education (GOM and World Bank, 2006). As many as 49 percent o f students report lack o f money for fees and uniforms as the major cause o f primary school drop In addition, parents also face an opportunity cost of sending children to school, 64The fact that direct costs o f schooling play a role indropout from primary education, particularly in rural areas, is also evidenced by the surge in enrollment in 1994, when fees were eliminated. Though there is no fee for attendingpublic primary, as many as 83 percent o f parents o fprimary school students inpublic schools reported paying for one or more school supplies duringthe year 2001 (DHS 2002 Survey, NSO). Parents are also required to contribute to school development funds, inmany cases the only source of funds for school maintenance or development activities. 48 because o f the need for their labor at home. About 6 percent o f students report that parents forced the child to stop and that their help was neededat home. Both direct and opportunity costs increase as the child progresses through the grades, and also increase with high levels of repetition. In addition, hunger and micronutrient malnutrition are highly prevalent in Malawi, with almost half o f children under 5 years o f age being stunted. Stunted children are less likely to enroll in school at the appropriate age, and have lower test scores. Especially during the hunger period from November to April coinciding with the beginning o f the school year in January, families often keep children out o f school. Another factor contributing to dropout i s the large proportion o f underage and overage children: these are more likely to drop out o f the formal education system than peers who enroll at the correct age. 178. On the supply side, a poor learning environment factors for higher rates of dropout, repetition and absenteeism, particularly for lower grades and for girls. Many primary schools lack adequate classrooms, safe drinking water, toilet facilities, and furniture (desk and chairs), particularly in the lower standards. In addition, the school calendar and perceived long hours o f instruction may not match cultural and economic activities in the community. Moreover, teacher absenteeism i s a major problem contributing to a poor learning environment, as almost one fifth o f teachers are absent on any given day in Malawi.65 The lack o f female teachers in rural schools in Malawi i s often cited as a main reason parents do not send their daughters to InMalawi, female teachers make up 38 percent o f the primary teacher establishment, but because disproportionately more female teachers work at urban schools, only 31 percent o f teachers inrural schools are female. 179. The higher dropout rates for girls are also due to a number of social factors. Other reasons for higher dropout o f girls include factors external to the school environment such as the increasing prevalence o f orphanage, as girls look after orphans in most households, the need to care for sick parents and sick relatives, which increasingly common as a result o f the HIV/AIDS pandemic, early marriage and teenage pregnancies, and initiation practices for girls. The analysis o f IHS2 data indicates that early marriage or pregnancy account for 9 percent o f drop outs, and also confirms that girls also do disproportionately more house chores and look for food during the lean season (GoM and World Bank, 2006). The government has put in place a `re-entry pregnancy policy' to enable girls who dropout o f school due to early pregnancies to come back and finish their secondary education 180. Primary Teachers lost to attrition are not being replaced, leading to a rising pupil-per-teacher ratio. Table 3.2 shows that the concerted effort to upgrade under-qualified teachers hired in the 1990s has led to a reduction in PQTR from 123 in 2000 to 95 in 2004. However, overall there are more primary pupils per teacher now 65Unpublishedfindings o f the pilot Public ExpenditureTracking Survey(PETS) in education. A pilot PETS was carriedout by the NSO in2004, in collaborationfrom DFID andthe World Bank.However, due to problemswith the executiono fthe survey the expendituretracking analysis couldnotbe carried out on the datacollected, anda surveyreporthas not beenprepared. 66A highshare of female teachers increases completionrates o f girls (Kengue andMingat,2002). 49 (72:l in 2004, compared to 63:l in 2000).67The Policy Investment Framework (PIF) o f 2000 stipulates that the pupil-per-qualified-teacher ratio in primary education should be 60:1, and the government is attempting to meet this benchmark by 2015, intime to meet the Millennium Development Goals (see below for a discussion on the resource and financial implications involvedinmeetingthese goals). Table 3.2 Numberof primaryteachers and pupil-teacherratios, 2000-2004 2000 2002 2003 2004 Number o f primaryteachers 47,840 46,032 45,100 43,952 Qualified teachers (Grade PT1-8) 24,566 23,632 32,019 33,018 Unqual. teachers (TT & month-to-month) 23,274 22,400 13,081 10,934 % unqualifiedteacher 49% 49% 29% 25% Percentage o f Female Primary Teachers 38% 38% 38% 38% Rural, % female 32% 32% 31% 31% Urban, % female 84% 84% 81% 82% Primary PTR 63 69 69 72 PTQR 123 134 97 95 Rural PTR 65 70 73 77 Urban PTR 46 58 47 44 Source: EMIS. Note: For this table, the salary grade i s used as the indicator o f the teacher's qualification. A large number o f teachers were upgradedfrom temporary teacher to primary status (PT1-PT8) around 2002-03, so, there is a large shift inthe share o f trainedhntrained teachers around that time. 181. Annual enrolment in Teacher Training Colleges falls short of required numberof teachers. Reachingthe pupil-per-qualified-teacher ratio inprimary education o f 60:l by the year 2015 would require increasing the number o f trained teachers from around 33,000 in 2004 to around 57,000 in 2015. Given an annual attrition o f about 6 percent per year, this would require yearly enrolment in Teacher Training Colleges (TTCs) o f around 6,000 students annually. At present, however, yearly enrollment remains below 3000 students.68An expansion program o f the existing five TTCs i s underway, with financial assistancefrom donors, and is expected to be complete in2008, after which enrolment will increase from the current level o f around 3,000 to 3,810. A new college is also beingbuilt and is expected to become operational in2009, after which total enrolment in all the colleges will increase to around 4,300. In 2006 the MOE adopted a new teacher training program, the Initial Primary Teacher Education (IPTE), under which students will be in college for one year and then gain teaching experience for another year (the "one-plus-one" teacher-training program). However, the new system does not increase annual enrolment in TTCs, and the number o f teachers deployed in primary schools will be the same as under the old two-year residential training program. 67This is the result of a defacto `hiringfreeze' inrecentyears due to a combination of too few teachers being trained and the Ministry's wish not to hire additional untrained teachers. Infact, for a number o f years, there has not been a pre-service teacher training programinplace inMalawi; the only teacher training programwas the Malawi Integrated In-service Teacher Education Program (MIITEP), which was an up-grading program for untrained teachers already inthe teaching force. In2006 the Ministry phased out MIITEP and put inplace the InitialPrimary Teacher Education (IPTE), which is a more conventional pre- service teacher training program. Yearly enrolment climbed to around 5,500 in2005 by enrolling two cohorts ina one-year crash course. However, the crash-course approach was discontinued in 2006 due to concerns with the quality o f trained teachers. 50 182. Moreover, the allocation between urban and rural posts is becoming more inequitable. Teachers are supposed to be allocated to District Assemblies (DAs) based on established criteria. In practice, teachers avoid placement in rural schools because o f the shortage in teacher housing, and female teachers inparticular desire urban posts. Thus from 2000 to 2004, the overall number o f primary teachers declined, but while rural schools lost 4,700 teachers, urban schools actually added 850, due in part to post migration away from remote rural schools. Indeed, EMIS reported that 4,400 primary teachers, or 10 percent o f all primary teachers, had transferred to another post within a year. The Country Status Report (World Bank, 2004) found a wide disparity in the number ofteachers available to schools with similar numberofpupils in2000, and 34 percent o f the variation in teacher deployment could not be explained simply by the number ofpupils. 183. There is even greater randomnessin the distribution of textbooks and of classrooms. Data collected in the 2004 school census also show considerable unexplained variation (49 percent) in the distribution o f textbooks, and even more variation in classrooms distribution (71 percent). Inpractice, as with teacher allocation, there i s no coordination between Physical Facilities, whose mandate requires it to carry out construction works for the MOE, and the zonal Primary Education Advisors/ Divisional Planners. Construction i s done fairly haphazardly, with decisions beingbased more uponpolitical will than greatest need. Clearly, donors and government need to work together to prioritize where new classrooms should be built.69 184. Many small schools need more classrooms to reach the national average pupil: classroom ratio. Figure 3.1 shows that schools with only one or two classrooms have extremely highpupil: classroom ratios on average (almost 250 pupils per classroom in 1-classroom schools, and more than 150 pupils per classroom for 2-classroom schools), while the schools with five or more classrooms schools have average pupil: classroom ratios o f around 80:l. This indicates that there are many schools in Malawi with a high enrollment but only Figure 3.1 Pupils: classroomand pupils: one or two classrooms. The more i teacher, bv school size. 2004 classrooms a school has, the lower the pupil-per-teacher ratio, I " 250 ______ 250 but the variation is far smaller (Figure 3.1). 200 200 185. There is a shortfall 150 of qualified teachers in Secondary Education, 100 i.100 especially in rural areas. The pupil-per-teacher ratio in 50 secondary education i s 0 deceptively low, at one teacher 1 2 3 4 5 8 7 8 9 10 11 1 2 > 1 2 per 20 students. However, the Number of classrwms in the school pupil-per-qualified-teacher ratio 69 Itshould be noted that while teachers are funded by government, classroom construction i s offen funded usingdonor funds. Hence, it appears that donors are not being successful indistributing resources where they are needed most. 51 i s much higher, at 73:1. The PIF stipulates that the pupil-per-qualified-teacher ratio in secondary education should be 40: 1, implyinga current shortfall o f 2,073 trained teachers (EMIS 2005). Efficiency of resource use 1997 2000 2004 Averagenumber o fyears a primary pupil spends inthe system 4.4 5.2 5.3 Averagenumber o fyears ofprimary educationcompleted 3.7 4.4 4.3 Averagenumber o fyears a completerhas spent inprimary 11.8 10.8 11.1 Averagenumber o fyears a dropouthas spent inprimary 2.9 3.6 3.9 Averagenumber o fyears financedto producea completer 19.4 18.6 21.7 Inputloutput ratio 2.4 2.3 2.7 Quality of education 187. A new curriculum and primary cycle is under implementation. The implementation o f the new primary curriculum in Malawi, which will begin in 2007, i s expected to improve quality significantly, provided it i s accompanied by sufficient teachers and learning materials. The reduction o f subjects from the previous 7 to 4 should have a significant impact on learning outcomes and efficiency. Government should also consider reducing the offering o f subjects at secondary from the current 11 to something within the available resources and trained teachers. 188. Teacher absenteeism is rampant, and is highest in rural, and in poorly managed, schools. More than one in five teachers are absent on any given day. When the head-teacher held an MSCE, absenteeism is lower, at 16 percent, than in schools where the headmaster only helda JCE (23 percent). Teacher absenteeism is also higher in rural than inurban schools, and inpoorly rated schools (PETS pilot analysis). 52 189. Teacher absenteeismis affectingthe actual hours of instructionreceived. The number o f hours o f instruction pupils receive has been shown to be a very important determinant o f student learning. HIV/AIDS has exacerbated the problem o f absenteeism (teachers take time off for illness, to care for sick relatives and to attend funerals). The new curriculum will demand an increased number o f hours o f instruction. Reducing teacher absenteeism and raisingthe effective number o f hours o f instruction are important and highly cost-effective measures to improve quality. In a number of countries (e.g. Nepal), the Government has passed landmark legislation that gave communities the option o f managing their local school and the right to hire/fire teachers. The State continues its funding obligation of paying for a certain number o f teachers depending on school size, but cannot influence the choice o f teacher. This i s an option that could also be considered inMalawi. 190. The lower standards of primary do not receive their fair share of resources. The current practice in Malawi assigns one teacher per class, independento f the number o f students. However, the lower standards have a much higher enrollment; consequently, they have much higher pupil-per-teacher ratios.70 Moreover, the best qualified teachers are often assigned to higher standards. Given that 70 percent o f dropouts occur in the first four standards, it is crucial to direct particular attention to quality o f education at lower levels. 3.3 ExpenditureTrends 191. Total education expenditure by government and donors was USD127 million in 2003/04. Recurrent expenditure was USD91 million (all government) and development expenditure USD36 million (98 percent donors, 2 percent government; Table 3.4). The donors' share o f total education expenditure was 28 percent in 2003/04; excluding external assistance on the government recurrent budget (budget support). Spending on education was up more than USD4 million compared to the previous year, 2002103; due to increased government expenditure. 192. Households contribute another USD30 million to education, mostly at about USD4 per child inprimary to pay for learning materials, school maintenance, et^.^^ primary and secondary levels. According to IHS2 respondents, households contribute Total household expenditure for primary education can thus be estimated to be around USD13 million for 2003/04. Households reported spending on average USD77 per student insecondary education; or USD14 million intotal. Higher education has the least cost sharing with parents, as tuition for regular students i s fully sponsored by the government.72 A small, but growing share o f university students in "parallel" programs do pay fees and organize own accommodation, however. 70 Further, classes often tend to take place outside due to the shortage o f classroom, which contributes to highdrop out rate and absenteeism especially inadverse weather. 7 1The DHS-Ed 2002 survey gave a higher estimate o f parent expenditure at USDlO per primary pupil. 72 Parents still make a significant contribution interms o f boarding fees, clothing, transport, learning and teaching materials etc. 53 Other Adminis Prim- Secon- Teach. Univers subv. tration aryl daryl edu. ities' ~ r g . ~ Total Percent Recurrentexpenditure Government 8.0 46.1 14.7 2.1 16.2 4.1 91.2 100 Percentage 8.8 50.5 16.5 2 2.3 4.5-- 100 Development expenditure Government 0.1 0.6 0.7 2 Donors, incl. bilateral 2.7 21.4 4.0 1.4 35.4 98 Total 2.7 27.4 4.6 1.4 36.1 100 Percentage 8 16 13 4 0 0 100 Total expenditure Government 8.1 46.1 15.3 2.1 16.2 4.1 91.9 72 Donors, incl. bilateral 2.7 27.4 4.0 1.4 35.4 28 Total 10.7 73.5 19.3 3.5 16.2 4.1 127.3 100 Percentage 8 58 15 4 12 2 100 193. Government expenditure on education is higher now than 10 years ago. Table 3.5 shows that as a share o f GDP, government expenditure on education increased from 3.8 percent in 1993/94 to 5.3 percent in2003/04, a faster rate than GDP growth over the same period. Government education expenditure per capita (over the whole population) has increased by 74 percent in real terms, reaching USD8.5 per capita in 2003/04. Government education expenditure per student has grown by 31percent inreal terms (in spite o f the fact that the number o f students has more than doubled over the period), reaching USD29 per student enrolled in education in 2003/04 (average o f all 3 levels). 194. The level of resources allocated to Teaching and Learning material has beenincreasing inreal terms over the past few years, whilst government subventionto secondary school boarding has decreased. 195. University subventions grew faster than MOE budget, despite a slower growth in enrollment at tertiary level. The MOE recurrent budget has grown at a rate o f about 8.8 percent per year in real terms in the past 10 years. The government subventions to the autonomous universities grew slightly faster over the same period, at 9.1 percent per year (constant Malawi kwacha), despite the fact that university enrollment has grown the least. 196. About two-thirds of donor spending on education is off-budget. Donors account for the bulk o f development spending, particularly when their off-budget spending (about two-thirds o f donor spending) i s included. Donor development spending has been increasing in real terms at an average rate o f about 2.4 percent per year since 1993/94. 54 Table 3.5 TotalEducationExpenditures,by Recurrentand Development Classification,and by source of funding, 1999/00to 2004/05 1999/00 2000101 2001102 2002103 2003104 2004105 Nominal Values (MK Million) Actual Actual Actual Actual Actual Actual Total government expenditure 27,220.8 37,266.4 42.490.3 61,260.3 80.536.3 96.625.0 Total Government recurrent expenditures 17,637.6 25,736.0 32,674.6 49,473.0 59.537.1 71.656.0 Total Development expenditures 9,583.1 11,530.4 9,815.7 11,787.3 20,999.2 24,969.0 Total Voted Expenditure on Education 3,694.6 3,742.9 7.080.9 8,707.9 10,178.6 Total Voted RecurrentExpenditureson education 2,929.9 3,206.7 5.893.6 7,266.4 9,284.9 11,569.5 Ministry of EducationVote 2,416.9 2,398.0 4,947.3 5,965 I 7,235.3 8.631.9 Universities(subventedorganizationsVote) 411.4 664.2 678.6 1,139.4 1.638.6 2,399 5 Other support organizations(subventedorganizationsVote) 101.7 144.4 267.8 161.9 411.0 538.2 Total Voted DevelopmentExpenditures on education 764.7 536.2 1,187.2 1,441.4 893.7 GovernmentofMalawi (Ministry of EducationVote) 189.1 59.7 133.9 351.7 70.8 Donors(Development Pamers) 575.5 476.6 1,053.3 1,089.7 822.9 Total Expenditureon Education(on-budget and off-budget) 4,760.0 4,974.0 8,459.7 10,807.2 12,936.6 Total domestically-fimdedexpenditure on education 3,119.1 3,266.3 6.027.5 7,618.2 9.355.7 11,569.5 Total externally-fundedexpenditure on education 1,640.9 1,707.7 2,432.1 3.189.0 3.581.0 GDP (Nominal) 91,809 113,871 136.141 160,137 189,563 226,272 Exchangerate(Malawi KwachaAJS$) 51.8 65.9 69.9 85.2 106.6 111.1 1999100 2000101 2001102 2002/03 2003/04 2004105 (USD million) Actual Actual Actual Actual Actual Actual Total government expenditure 525.3 565.8 607.9 719.0 755.5 869.7 Total Government recurrent expenditures 340.4 390.7 467.4 580.7 558.5 645.0 Total Development expenditures 184.9 175.0 140.4 138.3 197.0 224.7 Total Voted Expenditure on Education 71.3 56.8 101.3 102.2 95.5 Total Voted RecurrentExpenditureson education 56.5 48.7 84.3 85.3 87.1 104.1 Ministry of Education Vote 46.6 36.4 70.8 70.0 67.9 77.7 Universitics (subventedorganizationsVote) 7.9 10.1 9.7 13.4 15.4 21.6 Other support organizations(subventedorganizationsVote) 2.0 2.2 3.8 I.9 3.9 4.8 Total Voted DevelopmentExpenditures on education 14.8 8.I 17.0 16.9 8.4 Governmentof Malawi (Ministry of EducationVote) 3.6 0.9 I.9 4.1 0.7 Donors(Development Pamers) 11.1 7.2 15.1 12.8 7.7 Total Expenditureon Education (on-budget andoff-budget) 91.9 75.5 121.0 126.8 121.4 Total domestically-fundedexpenditure on education 60.2 49.6 86.2 89.4 87.8 104.1 Total externally-fundedexpenditure on education 31.7 25.9 34.8 37.4 33.6 Percent Shares Recurrentexpenditureson educationas % of total votedrecurrent spending 16.6% 12.5% 18.0% 14.7% 15.6% 16.1% Total Voted expenditureson Education as % oftotal votedexpenditures 4.0% 3.3% 5.2% 5.4% 5.4% Total expenditureon educationas % of GDP 4.0% 3.3% 5.2% 5.4% 5.4% Total votedexpenditureon education as % of GDP 4.0% 3.3% 5.2% 5.4% 5.4% Total votedEducation recurrentexpendituresas % of GDP 3.2% 2.8% 4.3% 4.5% 4.9% 5.1% Total votededucationdevelopmentexpendituresas % of GDP 0.8% 0.5% 0.9% 0.9% 0.5% Domesticallyfunded spendingon education as % of GDP 3.4% 2.9% 4.4% 4.8% 4.9% Recurrcnt spendingon educationin total Governmentrecurrent spending(9 16.6% 12.5% 18.0% 14.7% 15.6% 16.I% Total expenditureon educationas % o f GDP 5.2% 4.4% 6.2% 6.7% 6.8% 0,0% % Governmentto Total Voted DevelopmentExpenditures 24.7% ll,l% 11.3% 24.4% 7.9% %Donors to Total Voted Development Expenditures 75.3% 88.9% 88.7% 75.6% 92.1% % Donorsto Total educationExpenditures 34.5% 34.3% 28.7% 29.5% 27.7% Total Voted expenditureon cducationper capita 7.I 5.6 9 8 9.7 8.9 Total recurrent spendingper capita 5.7 4.8 8.1 8.1 8.1 Per capitadevelopmentexpenditures 1.5 0.8 1.6 1.6 0.8 Governmentexpenditureon educationper capita 6.0 4.9 8.3 8.5 8.2 Total expenditureon educationper capita 9.2 7.4 11.7 12.0 11.3 Total student enrollment 3,269.m 3,444.058 3,296,640 3,250,888 3,177,488 Governmentexpenditureper student 4.5 2.4 5.2 5.2 2.6 Population 10.0 10.2 10.4 10.6 10.8 GDP (nominal) 1771.7 1728.7 1947.7 1879.5 1778.3 2036.7 Source: Ministryof Finance,Annual Appropriations Accounts, andDFID donor disbursement survey 2004. 55 Notes: 1. The 2003104 government recurrent expenditure is a revised budget estimate. 2. The university tuition loan scheme is included under Universities, even though it appears under Secondary education in the government accounts. Program Admini- Primary Secondary Teacher edu. Total Cost center stration growth HQ 24% 56% 118% 24% Northern -34% 46% 49% --40% 100% 42% Central Western 308% -11% -37% -100% -10% Central Eastern 219% -3% 169% -100% 7% South Western 134% 9% -20% -100% 4% South Eastern -13% -23% 37% -15% Shire Highlands 1644% -25% 69% -100% 23% Malawi College o f 13% 13% Distance Edu. D O M A S I -3% -3% Total growth 105% 5% 37% -46% 11% 198. The 2003/04 budget allocations are fundamentally different from the previous fiscal year, raising concerns about the data quality and about prioritization. Table 3.6 provides a detailed analysis o f how the increase in expenditure was distributed across education levels and cost centers. The budget allocations differ widely, raising a concern about the quality o f the accounting data, as it seems inconceivable that a budget would be distributed so differently in two consecutive years. Insome instances, expendituresappear to have been accountedunderthe wrong program: e.g., a large part o f primary education expenditure in Shire Highlands was accounted under Administration, hence the 1644 percent increase in the normally small administrative budget and the 25 percent decrease in the normally large primary budget. Moreover, if the accounting data do reflect actual spending, they raise concerns about consistency, as each division seems to have prioritizedvery differently across the various education levels. Similarly, across geographic divisions, some budgets were cut, while other divisions received substantial additional funding. 199. The recent budget increases have been heavily skewed away from primary. MOE's budget increased by 11 percent in real terms from fiscal year 2002/03 to 2003/04, but spending on primary education only grew 5 percent, compared to 37 percent for secondary, and 105 percent for administration and support services. Spending on teacher education was cut by 40 percent. Overall, about half o f the increase went to wages (PE), and half to other expenditure (ORT). PE grew most in the Northern 56 Division, where wages for primary education and secondary education expanded as much as 50 percent in real terms, likely due to training o f existing teachers. PE spending also grew inthe Shire Highlands and Central Eastern divisions, while HQand the other three divisions actually spent less on salaries. The growth in the ORT budget was more evenly distributed across divisions, though the bulk o f it was spent in HQ on teaching and learning materials for primary and secondary. The large increase in HQadministration costs reflects an expenditure o f USD1.l million in 2003/04 on Formation and Maintenance o f Capital Assets. Table3.7 Educationindicatorsand expenditure,by region North Central South Average primary school enrollment, 2004 (# pupils per schools) 420 641 727 Primary Pupil: teacher ratio, 2004 64 70 78 Primary Pupil: classroom ratio, 2004 84 116 110 Primary Unit expenditure, 2002103 (current MK) 1,403 1,334 1,189 Primary Unit expenditure, 2003104 (current MK) 2,277 1,332 1,088 Average secondary school enrollment ((# pupils per public school) 145 207 195 Secondary Pupil: teacher ratio (public schools) 19 19 19 Secondary Pupil: classroom ratio (public schools) 34 43 41 Secondary Unit expenditure, 2002103 (current MK) 5,158 2,218 5,883 Secondary Unit expenditure, 2003104 (current MK) 8,422 2,752 7,403 200. The Northernregion receives a disproportionatelylarge share of primary and secondary education expenditures, and the Central Region the least. Moreover, in 2003/04, the Northern division received the largest budget increase73.Part of the explanation i s that more children attend school in the Northern region, but per student expenditure is also higher. For primary education, for example, the unit expenditure in the Northern region in 2002/03 was 5 percent higher than in Central and 18 percent higher than in the Southern region (Table 3.7). In 2002/03, the government spent only half the resources per secondary student in the Central region than they did in the other two regions and the difference was even bigger in2003/04. Recurrent expenditure 201. The allocationof recurrentexpenditureover time across education levels has fluctuated from changes in accounting practices to render comparisons meaningless. Spending on administration was 13 percent o fthe MOErecurrent budget in 2001/02 but only 5 percent the following year, while allocation to universities was 21 percent in2000/01 but only 12 percent the year after. The magnitude o f such fluctuations cannot be explained by periodic adjustments o f staff salaries, or changes inpriorities, but can only reflect different accounting practices in different years. For planning purposes, 73However, it is not known what the budget increase was used for, as there was no noticeable increase in number of teachers according to EMIS-thus, this might be an accounting error. 57 and to successfully implement specific education policies, recurrent budgets allocations should not generally fluctuate much. 202. Primary education's share of recurrent cost is barely above the 50 percent mark. Allocation to primary education was around 50-51 percent o f recurrent expenditure^^^ inboth 1993/94 and 2003/04 (though inthe intermediateyears it has been higher). This barely exceeds the EFNFTIbenchmark o f 50 percent for countries striving to achieve the MDG o f universal primary completion. The financial simulation included inthis Chapter will discuss ifthe allocation o frecurrent expenditureto primary needs to behigher. 203. Education recurrent expenditure relative to GDP i s above the regional average. Recurrent spending on education as a share o f GDP, at 5.3 percent, is above the average for African countries o f about 3.0 percent o f GDP (based on data for 1997-2000 presentedinWorld Bank 2003). However, Malawi has one o f the lowest GDP per capita of the region, while maintaining the highest gross enrollment rates in primary. As such, simulations carried out in preparation o f the ESSUP indicated that the current financing level would need to be sustained over the next 10 years to achieve the government's policy targets o f a pupil-per-qualified-teacher ratio o f 60:l in primary and 40:l in secondary by 2015, unless enrollment or teacher salaries (the two main cost determinants) are lowered (World Bank 2004b). 1993194 2001102 2002103 2003104 Primary 15.9 17.4 17.9 17.3 Secondary 99 92.4 69 126 Teacher education N a 757 1,235 424 University 2,414 1,895 2,445 2,884 Note: Includesadministrativeoverhead 204. Primary per student expenditure is USD17.3, slightly up from 1993194, but the gap in per student spending across education levelsi s growing. Recurrent unit expenditure was about USD17.3 per primary pupil in 2003/04, including administrative overhead, compared to the USD15.9 spent per pupil in 1993/94 (Table 3.8). This i s slightly below the average for African countries (11 percent o f GDP per capita, compared to 13 percent). Secondary and tertiary education in Malawi are much more costly than average for African countries, however: university students are now 167 times more expensive than primary students, compared to only about 64 times that o f primary in Zambia). 205. The unit expenditure for primary is USD11.7 on wages and USD3.0 on ORT per student in 2003/04. Figure 3.2 shows how both PE/student and ORT/student are consistently higher at higher levels of education. Part o f the explanation i s the lower pupil: teacher ratios and higher teacher salaries. The ORT/student was unusually highin primary in2003/04 due to extraordinary spending on teaching and learning materials; per student ORT was USD3.0 compared to USD2.2 in 2002/03. PE/student in 2003/04 was 74When the wage bill for the CDSS teachers is takenout o fthe primarybudget. 58 USD11.7 compared to almost USD13 the year before. The remainder o f the unit cost i s the administrative overhead. The unit cost for the TTCs (Teacher Training Colleges) and DOMASI have an unusually high proportion ORT spending, because students' allowances are paid out o f the ORT budget. Donor-supplied primary school textbooks are pre-packagedand directly suppliedto schools Figure3.2 Governmentunit recurrent expenditureby inputin2003104 (USD) 80% 60% 40% 20% 11 0% Primaw 1Secondaiv I Trcolleaes 1 DOMASI Universitv moverhead 2.6 . 19.1 . 60.0 8 2 L , 140.8 ~ ~ OOtherORT 0.8 ii 19.7 246.1 353.5 ----I 617.2 'MT&Lfrom H Q ~ 2.2 18.7 ~ ~ ! - 206. The gap in pupil-per-teacher ratio between Conventional Secondary Schools (CSSs) and Community Day Secondary Schools (CDSSs) has closed, but CSSs still have a higher proportionof qualified teachers. The Country Status Report found that government schools and grant-aided schools had a lower pupil-per-teacher ratio, and a much higher percentage o f qualified teachers than CDSSs-and, as a result, a much higher unit cost (World Bank 2004). The latest data (Table 3.9) show that the pupil-per-teacher gap has since closed, but CSSs schools still have more qualified teachers. CSSs have more non-teaching staff, also contributing to a higherunit cost. 207. The allocation of ORT funding is skewed in favor of CSS schools over CDSSs. When preparing allocations for the secondary education sector, MOE came up with an objective formula such that schools with the same characteristics receive the same allocation o f funds, based on: (i) basic facilities, such as class rooms, boarding facilities etc.; (ii) school curriculum; and (iii) auxiliary factors, such as ownership o f a car, a borehole, remoteness, etc. The current formula concentrates on CSSs only, however. This i s because originally CSSs were established by the government, whereas CDSSs schools were established by the community and financed from a specific budget line (separate from government secondary schools). Nowadays, however, some CDSSs 59 are built/ expanded by government.75To reduce the inequalities o f resources allocation and improve efficiency and quality in secondary education, the status o f CDSSs should be improved to match CSSs and the formula developed to ensure a proportionate share goes towards C D S S S . ~Unless additional resources can be identified to increase the ~ education budget, this measure would entail a reduction infkding for the CSSs, in favor o f CDSSs. Table 3.9 Pupilteacher ratio by type of government secondary school, 2004 I CDSS csss ODen Private or Total Gov. Boarding Gov. Day Schools Sciools grant-aided Number o f schools I 561 35 59 16 296 967 Enrollment 83,492 16,223 19,828 1,957 58,657 180,157 Teachers 4,724 643 878 155 2,644 9,044 PuDils*Der teacher II1 - 18 25 23 13 22 20 Pupils per 159 32 31 29 76 72 qualified teacher Pupils per non- 101 30 43 115 34 50 teaching staff Average # o f non- 1.5 15.2 7.8 1.1 5.9 3.7 teaching staffper school Average 149 464 336 122 198 186 ~ enrollment Source: Author's calculation based on EMIS 2004. 208. The pupil-per-teacher ratio at secondary is low, but there are not enough qualified secondaryteachers.As mentioned above, the current pupil-per-teacher ratio in secondary education i s well below the government's target o f 40:l so the teacher department (DTED) believes that it would be feasible to raise the pupil-per-teacher ratio to 30:l. However, staff reduction i s a sensitive issue. Given the deficit o f trained teachers in primary, the obvious solution i s to move (trained primary) teachers from secondary schools back to primary schools. 209. Many secondary schools are too small to realize economies of scale, partly due to under-enrollment due to a poor learning environment. Experience from other countries has shown that economies o f scale are usually substantially higher in secondary than in primary education, due to the higher degree o f specialization o f teachers in secondary (World Bank, 2005). However, secondary schools in Malawi enroll on average 186 students (Table 3.9), while primary schools on average enroll more than 3 times this number. Some secondary schools, in particular the CDSS schools and the government day schools are under-enrolled (in contrast, the elite government boarding schools are over-enrolled). The MOE has called for a considerable expansion o f boarding facilities at secondary schools. However, improving the quality and learning environment o f the poorest schools (i.e. the CDSS), might be a more cost-effective way to providebetter quality secondary education to local populationcenters. 75 The African Development Bank continues to spend resources inthe rehabilitation and expansion o f CDSSs and their contribution inthis areas probably exceeds government's efforts. 76 Inthe short term, the ministry should also consider introducing a `shortfall index' inthe funds allocation formula in order to jack-up allocations and improve services inCDSSs. 60 210. Cost recovery of secondary boarding has improved, but boarding fees are still too low. The current boarding fee o f MK1,500 (USD12) per student per term i s insufficient to cover even the cost o f purchasing food for the students (indeed, the cost o f feeding the students i s greater than the total amount o f fees paid by parents). Boarding schools therefore use their general-purpose funds (also paid by parents) to supplement the cost o f food, thereby reducing their ability to fund school maintenance and development activities. There are other recurrent costs o f boarding currently being met by government, including utilities, fuel, maintenance, etc. In addition, boarding schools have more non- teaching staff than any other category o f secondary school (Table 3.10). These staff assist in running hostels, food preparation, cleaning, security, etc. Divisions may be helping cover the cost o f boarding facilities (there is no other source o f fimding and the schools are still running). 211. Personal Emoluments to teachers inMalawi are comparatively high, and, moreover, may not be sustainable. When total recurrent expenditure i s divided by the number o f teachers77,average teacher remuneration is found to be about 5.2 times GDP per capita in primary and 6.2 times GDP per capita in secondary (Table 3.10). These figures correspond to salary comparisons to GDP per capita based on payroll department data (Table 3.11). Both primary and secondary salaries are comparatively high: the Education For All (EFA) benchmark for primary i s 3.5 times GDP per capita, and at secondary level teachers are usually paid 6-8 times GDP per capita and no more than 100 percent o f primary salaries. Current qualified primary teacher salaries inMalawi may not be financially sustainable: as the primary sub-sector has strongly increased its share of qualified teachers, the total wage bill has increased, as the total number o f teachers has declined. Table 3.10 PE expenditure per teacher, 2003/04 Total PE expenditure # o f PEiteacher PEiteacher in (adjusted for CDSS) Teachers multiples o f GDPicapita Primary sub-sector (25% o f MK3,7 12,62 1,000 43,952 MK84,470 or 5.2 teachers are under-qualified) USD835 Secondary sub-sector (most MK954,729,000 9,044 MK105,565 or 6.2 are trained primary teachers) U S D1,043 Source: Expenditure i s from annual appropriation accounts, adjusted for CDSS teachers inorder to be comparable with no. of teachers according to EMIS. Table 3.11 Annual teacher salary and teacher salary/GDP per capita, 2005 Gross salary Gross salary (minimumo f grade, incl. allowance) inmultiples o f GDPicapita Qual. Primary teacher MK100,920 or USD926 5.6 Qual. secondary teacher (diploma) MK148,224 or USD1,360 8.2 Qual. secondary teacher (degree) MK256,368 or USD2,352 14.3 Source: Salaries are from MOE; teacher salaries were last adjusted inDecember 2004. 77EMIS data on the number o f teachers i s reported by headmasters; and not independently verified inthe field. 61 212. Budget allocationtowards training is very low and training activities are haphazard. Training activities are haphazard and driven more by individual needs than sector needs, resulting inan accumulation o f debts. Sector needs for all types o f training over a mediumterm period should be identified ina training plan which has been costed, with an annual reviewo f the trainingplan. Development expenditure 213. Over two thirds of donor-financed capital expenditure remained off budget as of 2004/05. Donor expenditure is shown inTable 3.5 above. The proportion o f off budget funding was around 70 percent until 2004/05, though anecdotal evidence suggests that the situation may have improvedsince 2005/06. 214. Primary education received the most development expenditure. The MOE development budget mainly financed construction and rehabilitation (about 75 percent); but also some technical assistance, training, educational supplies, among other categories. 215. Unit construction costs for primary school construction do not vary greatly but there is considerable variation in the cost of building teacher housing. Most donors build a single classroom (half a classroom block) for about USD7,OOO- 8,000, which increases to USD8,500-10,000 with furnishings (Table 3.12). The difference in cost between donors i s fairly moderate. Only MASAF has provided a significantly lower unit classroom cost at USD5,OOO per unhmished and USD7,OOO per furnishedclassroom78,and it would be usefulto determine ifthe MASAFmodel could be more widely applicable. Construction o f an entire school often includes additional site works (e.g., an administration block, a borehole, latrines, etc.) which bring the total "per classroom unit" cost up considerably. There i s much more variation, at 84 percent, inthe cost o f a 3-bedroom teacher's house. Again, MASAF has the lowest unit construction cost. Table 3.12 Unit constructioncosts inprimary school construction(USD) Unit Costs: Donor/Program Per classroom (rural) 3-bedroom teacher's Furnishingo f 1 house classroom DFID 7.1 15 10.000 1,625 EUNicroprojects 7,600 161800 1,600 Plan International 8,600 14,400 1,850 UNICEF 7,749 na na W B M S A F 5,035 9,143 1,92 1 % difference highest to lowest 71% 84% 20% Financing of higher education: 216. 18 percent of government recurrent spending on education went to universities in 2003/04. According to the Annual Appropriations Account, the 78The difference in costs i s likely due to a greater community input inthe M A S A F school construction and also the perceived lower standards o f buildingin MASAF-administered projects. 62 subvention to universities more than doubled in real terms from 2001/02 to 2004/05, despite the fact that universityenrollment has hardly grown, increasing the annual student unitcost from about USD1,895 to almost USD4,OOO. 217. The cost of public higher education is almost fully met by government. Table 3.13 shows the breakdown o f UNIMA's income and expenditure, with the bulk o f expenditure going to administration, boarding, maintenance and other non-academic activities. The apparent increase in fees collected by UNIMA starting from 2001/02 i s due to an increase in tuition fees from MK1,500 to MK25,OOO per academic year. However, the fee increase coincided with the establishment o f the student loan scheme, benefitingpractically all students enrolled through the normal intake. With the exception o f a student book allowance (MK5,000), the loans are not disbursed to students but directly to the universities to cover tuition fees, and to date, no students have started repayment o f their loans (which many students undoubtedly regard as a grant).79 Thus, in spite o f the fee hike, government i s still bearing almost the full cost o f public higher education. In their university reform study (2001) the Malawi Institute o f Management (MIM)recommendedthat maintenance and student living services should be outsourced, and that students pay for food and transport. Table 3.13 UNIMA, percentbreakdownof income and expenditure 2000/01 2001/02 2002103 Income Subvention 91% 79% 79% Fees 2% 14% 15% Other income 7% 7% 6% Total income 100% 100% 100% Expenditure Teaching and research 28% 32% 37% Student living 20% 14% 14% Administration 16% 17% 16% Maintenance 16% 16% 16% Transport & traveling 10% 9% 8% Other charges 11% 11% 11% Total expenditure 100% 100% 100% Source: UNIMA Financial Statements. 218. Universities have introduced new parallel full fee-paying programs. The Education Act prescribes that students must live in university accommodation, but universities enroll a small, but growing number o f students in a parallel full fee-paying program anyway. These students pay significantly higher fees" than normal-intake students and are usually non-boarders. These programs have allowed the universities to increase enrollment without expanding hostels and boarding facilities in the past couple o f years. 79A trust has recently been establishedto disburse and collect funding from graduates but its effectiveness still has to be felt. At Mzuzufee-paying students pay MK 257,000 annually for a degree program and MK181,OOO for a diploma course (Whitworth, J., 2005). 63 Table 3.14 Efficiency Ratios for University of Malawi, 2003/04 Chancellor College o f Bunda College Kamuzu College College Medicine o f Agriculture o f Nursing Polytechnic Students 1,812 169 570 400 2,549 Lecturers 301 95 80 51 169 Non-teaching college staff 525 75 466 66 166 Students per lecturer 6.0 1.8 7.1 7.8 15.1 Students per non-teaching staff 3.5 2.3 1.2 6.1 15.4 PE per staff (USDiyear) 4,527 8,619 2,627 7,498 8,748 Inmultiples of GDP/capita 28 54 16 47 55 P E per student (USDiyear) 2,065 8,675 2,5 19 2,193 1,146 ORT per student (USDiyear) 287 4,5 15 2,223 1,097 10 Source: Table is adapted from Whitworth, J., 2005. Notes: Expenditure data are from UniversityUnit, MOF for 2003104 financial year, and staff and student data from UNIMA.Excludes Central UniversityAdministration. 219. Considerable efficiency gains could be made by increasing the studenthtaff ratio. Table 3.14 demonstrates the extraordinarily low studenthtaff ratios at the various colleges o f UNIMA. Inparticular, the College o f Medicine stands out with less than 2 students per lecturer. Colleges also have an abundance o f support staff, which contribute to the average PE/student o f MK530,OOO per year, or around 32 times GDP/capita. 220. Wastage due to dropout from universities may be considerable. Further driving up the cost o f tertiary education are low survival ratios. Since 1993194, enrollment has been around 3,500 or more; with most programs being 4-year degrees, so we would expect around 850 graduates every year. Instead, EMIS reports only about 360 graduates per year between 1999 and 2002. As a result, the total cost to government o f producing one university graduate reaches around USD25,OOO-30,000. Household contributions to the Primary Education sector 221. In spite of the `Free Primary Education' policy adopted in 1994, significant costs of primary schooling are still faced by households, and there are regional disparities. Seven out o f ten pupils' households spend money on uniforms, clothing and shoes to be worn to school. Almost six in ten pupils' households spend money on the buildingor development fund, whilst one inthree pupils' households spent money on food. Parents also contribute heavily towards the administration o f terminal examination fees, utility bills (mainly contributing towards water and electricity bills), wages for security guards (inurban areas) and teaching and learning materials. There are regional discrepancies in the costs to households, with students in the Southern region likely to contribute most often and those in the Northern region the least likely. In addition, households often support the construction or maintenance o f school buildings and teachers' houses, sanitation, or other school projects. Households may also provide materials for the school, such as roofing, stone and sand, and donate their labor. As many as 74 percent o f parent/ guardian households interviewed for the IHS2 in2004/05, report making one or more contributions (money, materials, or labor) to primary schools. A 64 smaller proportion o f parent/ guardian households (8 percent) have contributed money, food or labor to primary school teachers. Not surprisingly, wealthier households are more likely to contribute money, while rural and poorer households are more likely to contribute indirectly (labor and materials). 222. Better procurement in the Teaching and Learning Materials Supplies Unit of the MOE would significantly reduce the costs to parents for primary education. The SuppliesUnitfaces a backlogo f arrears owed to various suppliers,which have consumed their budget inrecent financial years and severely hampered the delivery o f teaching and learning materials to schools. This i s a significant cost to parents, so addressing this issue would do much to reduce the cost to households o f primary education. Costs to households of Secondary and Tertiary Education 223. Householdscontributeto the RevolvingFund,the GeneralPurposeFund, School Development Fund, and Boarding Fees but these fees are inadequate. The government stopped subsidizing the purchase o f secondary textbooks in 2001, resulting in acute shortages of textbooks within secondary schools because the student contributions towards the Textbook Revolving Fund (TRF) only provide for a minimal number o f textbooks. The GoM also stopped subsidizing Boarding Fees in 2001, but the current boarding fee i s too low to cover costs and in practice i s subsidized from the School Development Fund. 224. Revenues collected by Tertiary Educationinstitutionsremaininsufficient. Funds collected by the universities are often inadequate to cater for their ever growing needs in terms o f size and technological advancements. Salaries for university staff, board and lodging deplete most o f the funds that universities receive, so text books and other reading materials are in short supply. Contracting out accommodation and other auxiliary services so that universities concentrate on their core functions could free some resources to priority areas. However, modalities need to be worked out. 3.4 Povertyfocus and incidence of expenditures inthe educationsector Disparities in expenditure 225. Education expenditure on primary disproportionately benefits the children from non-poor households. Figure 3.3 shows that primary education expenditure per household appears to be distributed fairly equitably across quintiles8'. However, because the number o f children in poor households i s almost double the number in non-poor households, per child, expenditures in fact favor richer quintiles. Indeed, while one in five children do not attend primary school, this doubles to two in five children for the poorest 20 percent of the population. Moreover, very few low income quintile children actually complete primary, largely because their families cannot afford it, and primary schools close to poor communities appear to have substantially ~~~ The Lorenz Curve for primary lies just above the line o f perfect equality, implyingslightly more resources are spent on lower income deciles. 65 higher PTR ratios (88 students per teacher) than non-poor communities (68 students per teacher), implying access to lower quality o f primary education for the poor. 226. The benefits from expenditure on secondary and tertiary education are even more heavily skewed toward the richest households. Three times as many non- poor students as poor students are enrolled in secondary education, and boys and girls from the richest decile are 10 times as likely to attend secondary school compared to those inthe poorest decile. Thus, the richest 50 percent of households reap 70 percent o f the benefit from government expenditure on secondary education. Enrolment intertiary education is associated almost exclusively with households from the richest decile. Considering that tertiary education i s many times more costly than secondary and primary education, and that tertiary students in reality pay no fees, it i s clear that the education budget and fee system i s currently highly inequitable. 227. The decentralization of the ORT budget is perpetuating past regional inequities in the distribution of education expenditures. Since 2005/06 the ORT budget has beendevolved directly to district level (inthe past districts received their ORT funds from the divisions).82 However, the funds have been allocated based on historic distribution figures, rather than objective criteria. Ideally the allocation o f finances to the various Districts should depend on variables such as the number o f schools, enrollment levels, number o f teachers, and geographical distances for the school-age population to attend school. Figure3.3 Benefitof education expenditureby incomequintile(IHS2 2004) Lorenz Curves for Education Enrolment 100% 90% -Line of Equality .- 80% - .:._ 70% -Cumulative per Capita 5 - I ConsumDtion 60% +Enrolment in Primary 3t2 50% 40% School 30% +Enrolment in 20% Secondary School 10% +Enrolment in Tertiary 0% lnsitutions Population distribution from poor to rich I Source: GoM andWorldBank, 2006 3.5 Impact of HIV-AIDS on the educationsystem 228. The HIV/AIDS pandemic makes it harder to achieve UPC. Experience from other countries as well as studies done in Malawi show that the education system i s seriously affected by HIV/AIDS, through increased teacher absenteeism and mortality, ** Expenditures onpersonalemolumentsremainthe responsibilityo f the central government. 66 and a growing number o f orphans who are at greater risk o f repeating or dropping out o f school (Kadzamira, K., et al., 2001). 229. The costs of teacher education are already USDl millionper year higher than they would be without HIV/AIDS, and could rise to USD3.2 million annually. Some 787 primary teachers, or 1.8 percent o f the teaching force, were reported to have died during the past year (EMIS) and secondary teacher attrition due to death was also 1.8 percent. Close to half o f adult deaths in Malawi are attributable to AIDS, so if we assume that death due to HIV/AIDS is currently 0.9 percent o f primary and secondary teachers per year, the present additional cost o f teacher education i s already USDl millionper yearg3.More than half o f this cost i s attributable to the education o f secondary teachers, because their education i s longer and much more costly. Moreover, experience from Uganda and Zambia indicates that teacher attrition due to death from HIV/AIDS could potentially increase to as much as 4 percent per year. Government would need to train approximately 1500 additional primary teachers and 225 additional secondary teachers per year to counteract this attrition, at a cost o f USD3.2 million. If teachers infected with HIV/AIDSwere to start receivingtreatment and antiretroviral drug therapy, attrition may never reach those highlevels, however. 230. Costs of absenteeism of teachers due to HIV/AIDS are hard to measure but likely to be considerable. The costs o f absenteeism due to HIV/AIDS are more difficult to establish, because we would need to know how many days, weeks, or months, teachers are absent for work for reasons to do with HIV/AIDS. The Pilot PETS analysis found that teachers were absent from work a total o f 17 percent o f the school year, corresponding to about 30 school days per year. Assuming a third o f the absenteeism, or 10 school days per year, i s related to HIV/AIDS, then the loss to the education system is about USD2.8 million. 231. Funeral expenses are a significant burden on the ORT budget, especially in primary. When a teacher or a teacher's dependent dies, the current (though unofficial) practice i s for the corresponding Education Division to pay the funeral expenses. These expenditures are difficult to track, as there i s no budget line for funerals. However, using the number o f teacher deaths and making certain assumptions about the average cost o f funerals, some 3-5 percent o f the divisions' ORT budget may currently be spent on funerals, with funeral spending in the whole MOE probably exceeding USD200,000/year.84Inthe primary sub-sector, where the available ORT budget i s small and the number o f teachers large, the cost o f h e r a l s as a share o f the budget i s likely to bemuchhigher. 232. Currently, 12 percent of primary pupils are orphans, and their numbers are growing (EMIS).A recent study projects that there will be 442,000 orphans enrolled inprimary education by 2010 (Hunter & Williamson 2002). The same study found that in Malawi, approximately 70 percent o f orphans are HIV/AIDS orphans. The share of 83The cost o f primary teacher education includes the cost o f training and a stipend for living expenses. For secondary teacher education, it i s assumed half take a 4-year university education, and half attend the more cost-effective D O M A S I teacher upgrading program. 84The assumptions are: half o f the funerals paid for by divisions are for staff, half for dependents, and estimated cost per funeral i s K10,OOO. Death benefits (pension and death gratuity) to dependents are out o f Treasury budget and therefore are not included. 67 orphans increases sharply with the age o f the children: 18 percent o f children in standard 8 are orphans compared to 9 percent instandard I. 233. The HIV/AIDS pandemic currently consumes roughly 4.4 percent of government spending on education, and the cost could increase three-fold. The current total costs o f HIV/AIDS to the education system are USD4.0 million, or about 4.4 percent o f total government education spending. This i s a conservative estimate because it excludes the impacts on higher education, and onthe MOEadministrative staff. In5-10 years, if teacher attrition due to HIV/AIDS rises to 4 percent the cost could rise to about USD8 million, in today's prices. Further, if the government were to introduce stipends for orphans, the cost may reach around USDl1.5 millionper year.85 3.6 Recurrentresource needs to achieve the MDGs 234. Progress has been made towards MDG 2 (to achieve universal primary education), but more effort is required. As mentionedabove, Malawihas improved its GER ratios inprimary education, but the Net Enrollment Ratio (NER) remains at around 80 percent, and only 60 percent o f children who start primary school actually finish (Le., 48 percent o f all Malawi children currently pass the PSLC exam). This suggests that Malawi will fall short o f achieving universal primary education by the year 2015. Both adequate policies and additional resources will be required, and part o f these resources needto be finded by development partners. 235. Malawi is well placed to reach Goal 3 (to promote gender equality and empower women). Notably, good progress has beenmade in equality o f enrollment in primary and secondary education (though more girls drop out and fewer pass the final exams) and inreducing gender disparity inyouth literacy. Much more progress i s needed in reducing the gender gap in tertiary education, however, and in increasing female participation inthe workforce and inpositions o f authority. 236. The recent Education Sector Plan is a comprehensive development plan covering primary, secondary and teacher training over the next 10 years. Once finalized, this plan i s expected to provide a basis for a future Sector-Wide Approach (SWAP)inthe sector, and may also allow for additional donor support through the FTI. Financial simulation 237. In order to identify financing needs for the plan, a financial simulations model has been used.As discussed above, one o f the key targets for primary education i s the achievement o f a pupil-per-qualified-teacherratio o f 60: 1. The simulation model first determines the resource requirements needed to meet this goal by 2015, given the current level o f remuneration o f teachers and spending on ORT. This i s referred to as the base scenario. The model then investigates the impact o f changing key policy variables, in line with education policy indicators set out inthe indicative framework o f the FTI; 85 It shouldbenotedthat the 2006 PovertyandVulnerability Assessment (PVA) highlightsthat orphans are nottypicallymore disadvantaged than non-orphanswho do not livewith their parents.Infact, fosteringis associatedwith a greater impact ofnon-enrollmentthanorphanstatus, when controllingfor other variables (GOM andWorldBank, 2006). Ingeneral, the evidencefromthe PVA would argue in favor o f a program to support childrenwho live inthe pooresthouseholds, independently o f their orphanstatus. 68 namely, to increase ORT spending in primary, improve the PTR to 40:l by 2015, and reduce the primary teacher salary relative to GDP. 238. Achieving a Pupil-per-Qualified-Teacher Ratio of 60:l by 2015 will be difficult due to the limited capacity of the teacher training colleges. As shown in Table 4.15, enrollment in public primary i s expected to grow to 3.4 million students in 2015. This forecast i s based on certain assumptions about population growth and developments in the repetition and dropout rates, also shown in Table 3.15. In order to reach a pupil: qualified teacher ratio o f 60:1, the number o f qualified teachers will needto increase from the current 33,000 to 57,000 by 2015; at the same time, the number o f unqualifiedteachers will fall from the current 11,000 to 0 by 2015. 239. The share or recurrent costs allocated to primary education would have to increase from 50 percent to 57 percent by 2015 to realize the Base Scenario. Assumingthe total recurrent education budget (including tertiary) remains at 5 percent o f GDP, the recurrent costs o f primary education will consume an increasing share o f the total budget.86This i s due to the growing salary envelope from the higher number o f teachers, and to a larger share o f the teachers being qualified (therefore earning higher wages). The recurrent costs o f primary will grow from the current USD46 million to USD64 millionby 2015. Table 3.15 Base scenario: Pupil-per-qualified-teacher ratio of 60:1by 2015 2003104 (Actual) 2015116 Enrollment: % enrollment inprivate primary 0.8% 2.0% Enrollment public primary (million) 3.1 3.4 Assumptions common to all scenarios: 6-13 year-olds, annual growth 2% 2Yo Education budget as % o f GDP 5% 5% Real GDP growth 2% 2% Average repetition rate 19% 11% Average dropout rate 15% 9% Annual teacher attrition 6% 6% Key policy variables: Number o f teachers (`000) 44 (60:l) 57 Qual. Teachers (`000) 33 (60:l) 57 Unqual. teachers (`000) 11 0 ORT budget, in% of salary envelope 23% 23% Salary qualified teacher, inmultiples of GDP per capita 5.2 5.2 Salary unqualified teacher, in multiples of GDP per capita 3.8 3.8 Recurrentcosts of primary: Recurrent costs, million USD* 46 64 % ofprojectedrecurrent education budget 50% 57% Source: Ministry o f Health and World Bank staff estimates Note: *In 2004 dollars 86 Itshould be emphasized that there needs to be a systematic approach in the distribution o f resources across all the education sub-sectors, given the linkages between them, and noting the bottlenecks that will be created by developing one sub-sector at the expense o f the others. 69 240. If average teacher salaries were held constant in nominal terms such that they gradually approached a level of 3.5 times GDP per capita, the cost of primary education would be reduced significantly even as more teachers are added to payroll. InTable 3.16, scenarios 2-4 each add one additional policy target (from the FTI Indicative Framework) to the base scenario.87 Scenario 2 reduces the average teacher salary from the current 4.8 (average o f the trained and untrained teachers) to 3.5 multiples o f GDP per capita by 2015. This represents a significant saving, whereby the recurrent cost o f primary education i s reduced from the current 50 percent to just 39 percent of the total recurrent education budget by 2015, even though more teachers are added to the payroll. Table 3.16 Recurrent cost implications of Scenarios 2-4 Base Scenario Scenario 2: Scenario 3: Scenario 4 Key policy variables: Pupil: teacher ratio 60:l by 2015 60:l by 2015 60:l by 2015 40:l by 2015 Pupil: qualified teacher ratio 60:l by2015 60:l by2015 60:l by2015 60:l by2015 Average teacher salary, inmultiples o f As now, 4.8 3.5 by 2015 3.5 by 2015 3.5 by 2015 GDP per capita ORT budget, % of salary envelope As now, 23% As now, 23% 50% by 2015 50% by 2015 Recurrent costs of primary: Recurrent costs by 2015, U S D million* 64 44 54 80 Percent o f projected recurrent 57% 39% 47% 71% education budget, by 2015 Source: Ministry o f Health and World Bank staff estimates Note: *In2004 dollars 241. The reduction in teacher salary under scenario 2 could fully finance a significant increase in the ORT budget. Scenario 3 increases the share o f ORT in the budget from the current 23 percent to 50 percent, holding everything else the same as in scenario 2. The recurrent cost o f scenario 3 by 2015 i s 47 percent o f the total recurrent education budget-still less than inthe Base Scenario. 242. Achieving a pupil-per-teacher-teacher ratio of 40:l would require 71 percent of the total recurrent education budget, even using some unqualified teachers. Scenario 4 builds on the assumptions in Scenario 3 to improve the pupil: teacher ratio to 40:1 while maintaining the pupil-per-qualified-teacher ratio at 60: 1 (it would be impossible to train a sufficient number o f teachers by 2015). This scenario, which incorporates all the policy targets o f the Indicative Framework, would require that 71 percent o f the total recurrent education budget be allocated to primary education. 87The FTIhas defined an indicative framework with reference values for a small number o f key structural parameters. For Malawi in2004, there are considerable differences with the reference values. Namely: i) the share o f the government budget allocated to education i s lower (17%) than inthe indicative framework (20%); ii)the PTR is muchhigher (72) than inthe framework (40); and iii)the level o f remuneration o f primary teachers is higher (5.2) than recommended (3.5). Malawi also has a relatively low level o f non- salary expenditure inprimary (23% o f salary envelope) compared to the reference (50 percent). 70 3.7 Implementingkey recommendations from the 2001 PER on education 243. Limitedprogress has been made in implementingreforms recommended in the 2001 PER. Below we briefly review progress made in each o f the five main recommendations from the 2001 PER: #I:Increasebudgetary allocations toprimary schoolsfor teaching andlearning materials, inspection and school maintenance. The government allocation to Teaching and Learning materials has increased, but remains far from the target o f USD5 per pupil. Primary and secondary education thus received a smaller share o f the budget in 2003/04 than in 1999/00, rather than a larger one. The education budget allocated to universities has increased, not diminished as recommended, squeezing the share to primary and secondary education. #2: Increase the deployment of teachers to lower grades, outlaysfor teacher training, and introduce an allowance to attract teachers to rural areas. There has been a significant improvement in the share o f qualified primary teachers, from 51 percent in 2000 to 75 percent in 2004. However, because teachers lost to attrition have not been replaced there i s an ever-diminishing pool o f primary teachers. 0 The ruralhrban teacher allocation in primary has worsened significantly. Rural Pupil: Teacher Ratios rose from 65 in 2000 to 77 in 2004, while the urban PTR improved from 46 in2000 to 44 in2004. #3: Improve efJiciency by restructuring exams and reorienting secondary school allocationsfrom boarding schools to day secondary schools. 0 The gap inthe pupil: teacher ratio among the different types o f secondary schools has closed, but there i s still a huge gap between CDSS and conventional government schools. #4: Increase cost sharing by raising secondary school and universityfees while increasing the size of the bursaryfund toprotect poor students. No progress has beenmade inimplementingthe set o f recommendations intended to increase cost sharing in secondary, and in particular, in tertiary education. Given that universities are consuming an increasing share o f the budget, education expendituresinMalawi are becoming more inequitable. #5: Address the impact of HIV/AIDS by introducing life skills as a separate subject in the school curriculum, through teacher training programs and by preventing funeral costs from cutting into teaching expenditures. 0 There has been progress in terms o f incorporating life skills and HIV/AIDS awareness into the curriculum at various levels o f education. 71 0 There has been no scaling down on funeral benefits as recommended inthe 2001 PER. Funeral costs are an increasing burdeno f the divisions' and districts' ORT budgets. 244. The recommendations from the 2001 PER not yet implemented are still priorities today, and there is no technical obstacle preventing their adoption. This PER has confirmed the validity o f the recommendations still outstanding from the 2001 analysis. No technical constraints have been identified that prevent implementation. Therefore, there i s a need to raise the level o f debate to increase national awareness on the needfor such reforms, inorder to reduce political obstacles: 0 increase the share o f expenditure to primary and secondary education, and reducing fundingfor universities; 0 improve the teacher allocation in rural primary schools in absolute terms, and relative to urban schools; 0 increase cost sharing insecondary, and inparticular, intertiary education; 0 eliminate the practice to funds funeral costs from the divisions' and districts' ORT budgets. 3.8 Recommended Policy priorities for Malawi's education system 245. Several `recommendations arise from the analysis carried out in this study. The full recommendations are summarized below, but a shorter and prioritized list o f recommendations i s presentedinthe executive summary to this report. #1: The government education budget needs to be rebalanced, both between sub- sectors and between regions, and from upper to lower primary: 0 A highershare o fthe budgetshould be allocated to primary education, inlight o f the significant financing needs o f achieving the MDGs. Substantial resources could be saved intertiary education and reallocated towards (lower) primary education. 0 Lower primary education needs to receive its fair share o f resources through internal reallocation of, particularly, classrooms and teachers. 0 The significant geographical imbalance inthe provision o fpublic education services needs to be redressed, inorder to give all Malawians equal access to education. The better performance o f the education systeminthe Northernregion needs to be replicated inthe other regions o f Malawi, which must benefitmore than average from future growth inthe education budget. 0 When budgets are decentralized, the geographical distributionmust follow an equitable financing formula, and not perpetuatepast geographical inequities. #2: Teacher development and management should be a cornerstone of primary education policy, as the vast majority of primary education expenditures finance teachers: 0 There i s a needto increase the output o f trained teachers to approximately new 6000 teachers every year; this requires: (i)review o f the pre-service teacher a 72 education program, such that the length o f the training i s balanced against the needfor a significant output; and (ii)additional investmentsto increase the numbero fplaces inTeacher Training Colleges. 0 The current trend o f teacher migration away from ruralposts mustbe reversed through teacher incentives and a financing mechanism (the money should follow the school, not the teacher). The excess number o funder-qualified teacher insecondary should be transferred back to primary schools. Teacher absenteeism can be eliminated by introducinglegislation to give communities the option o f managing their local school and the right to hirehire teachers (wherebythe government continues to pay for a certain number o f teachers, e.g. depending on school size, but cannot influence the choice o f teacher). 0 Teacher absenteeism can also be partially reduced by increasedmonitoring, and by raising the quality o f school management (including headteacher qualifications). #3: In secondary education, the emphasis of reforms should be to reduce the gap in education quality offered at CDSS versus conventional schools, and to reduce the cost burdenon government of boardingschools. All resources, including qualified secondary teachers, should be allocated equitably betweenthe different types o f public secondary schools to level out quality. The physical structures o f CDSS schools should beupgradedto an acceptable standard. #4: Improvethe equity of access and benefitsfrom education Government boarding schools should be phased out altogether, or boarding should be limitedto girls, where essential, and particularly vulnerable groups (special needs, orphans, etc.). The bursary scheme for secondary needs to be adequately targeted to reachpupils from the lower income quintiles, to increase their access to secondary. The higher repetition and dropout rates for girls must be addressed, by implementingthe recently approved `pregnancy policy' (including an information campaign and monitoring o f policy implementation), and increasing the number o f female teachers. The intake o f girls to government secondary schools should be increasedto 50 percent, by increasing the number of places inboarding facilities reserved for girls, as well as implementingmeasures to reduce girls drop out andrepletion(see previous bullet). #5: Introducereformsto reducethe large inefficiencyofthe Universitysystem: 0 The financing ofhigher educationneeds urgent reform, to reduce government subsidies(total and per student) and increase cost sharing with students considerably (students mostly belong to the richest income quintile). Increased 73 cost sharing requires a review o f the system o f allocation o f loans to target recipients o f government loans basedon means testing. 0 Universitiesneedto be made more accountable for the use o f public funds, and for progress in implementing neededreform, through settingup a reporting mechanism (such as semi-annual or annual reports). #6: The Ministry's HIV/AIDS policy should be implementedto help curb the spread of HIV/AIDS andto mitigateits impact on the educationsector: The GOM should prioritize providingthe fundingneededto implement the policy. There i s needto ring fence education money to limit the amount o f resources spent on funerals. As a short term, the funeral assistance policy should be simplifiedby replacingthe current practice o f covering the full cost with a fixed monetary subsidy. A more permanent solution would entail transfer o f the responsibility for funeral expenses to DHRMD (which i s responsible for payments to all government personnel). #7: More emphasis should be placed on improving management of the education system at all levels. The quality o f school census dataneeds to be further improved, particularly with respect to properly defining and assessing who are repeaters, dropouts, new entrants, etc. The number o f standard 8 students/PSLC graduates producedby primary education should be usedas a measure o f the performance o f this cycle. Studentrepetition needs to be curbed, for example through automatic promotion betweencertain grades (inas far as improvements inthe quality o f education allows). The higherrepetition and dropout rates for girls must also be addressed (see above).88 88InGuinea, the repetition rate was cut inhalfthrough this measure (instead o fallowing repetition after each school year, it was only allowed after each cycle, where a cycles lasts two years). 74 CHAPTER4: HEALTH 4.1 Introduction 246. This expenditure review concerns the activities o f the Ministry o f Health (MOH). For a detailed description o f the health sector in Malawi, see Annex 4.1. Because HIV/AIDS, though cross-sectoral in its impact, has enormous implications for public expenditure in the health sector, we analyze the long term sustainability o f Malawi's response to the pandemic insome detail. 247. This chapter is structured as follows. First, it discusses the key challenges faced by Malawi's health sector, and critical issues constraining the delivery o f better health services. The chapter then proceeds to examine the expenditure trends. The following section focuses on the expenditure implications o f the HIV/AIDS pandemic. The chapter concludes by proposing a set o f actions that the government and its development partners should take to improve sector performance. 4.2 HealthSector Issues Basic Health Indicators 248. The health status of Malawians continues to be poor and progress on basic healthindicatorsis mixed. Life expectancy declined from 48 to 39 years between 1990 and 2000, mainly as a result o f HIV/AIDS. The under-fivemortality rate (per 1000 live births) has improved from 258 in the 1980s to 133 in2004, and the infant mortality rate has declined from 138 per 1,000 live birthsinthe late 1980s to 76 in2004. Malawi's maternal mortality ratio (per 100,000 live births) has come down from 1120 in 2000 to 984 in 2004 but remains alarmingly high (DHS 2000 and DHS 2004). HIV/AIDS adult prevalence rate was 14.1 percent in 2005 (NAC, 2005), and appears to have stabilized. Malaria incidence has declined from the extremely high rate o f 812 cases per 1000 in 1992 to around 282 per 1000 in 2005 (HMIS), but continues to be a major problem especially among women and children. 249. Malawi has achieved good performancein vertical health interventions, but services that are highly reliant on health facilities have faltered. Malawi has achieved good rates of immunization, and as a result vaccine-preventable diseases have been dramatically reduced (e.g., polio, neonatal tetanus, measles). 89 The country's community-based tuberculosis control program, now nationwide in reach, i s cited internationally as a success story. Substantial improvements have been made in service delivery such as condom supply, HIV testing, increasing coverage o f insecticide-treated bed nets, and treatment with antiretroviral (ARV) drugs. Ranged against these achievements, however, performance in maternal and child health programs (which are more closely integrated into the overall health system) have lagged behind. There has been almost no improvement inthe proportiono f births with medical assistance, from 55 89The results o fthe latestDHS,however, indicatethat rate of fully immunizedchildrenhas declinedto 64 percentin2004, from 82 percent in2000, whichis a worrying development. 75 percent in 1992 to 56 percent in 2000 and 57 percent in 2004. Malnutrition remains alarmingly high for children: Malawi's 48 percent stunting rate in 2004 i s one o f the worst in Africa and has not improved in close to a decade. The rate o f maternal anemia (68 percent in 2004) i s appalling. Because o f the severity o f malnutrition and its deleterious long term effects, Chapter Five looks at nutrition inmore detail. Access to Healthfacilities 250. Malawi's health facilities are relatively close to households, but most of them are not capable of delivering essentialhealth services. Outpatient clinics are, on average, within 3.5 km to households as the crow flies-a short distance by African standards. However, just 9 percent (54 out o f 585) government and mission health facilities are capable o f providing the essential package o f health services (EHP)90 on-site (JICA/MOH inventory)." Indeed, Figure 4.1 shows that in each district, only 1 or 2 facilities have adequate EHP capacity. 251. The inability to provide the essential package of health services arises from supply stock-outs, a lack of basic utilities, and especially a lack of trained health workers. According to the J I C N M O H inventory, most health facilities lack adequate water, electricity, phone or radio communication. However, a lack o f trained health workers i s the biggest problem o f all, with a density o f skilled health staff per population that compares poorly with that in war-ravaged and post-conflict countries: Malawi's 599 physicians in 2003 translates into a per-total-population density o f 0.05, lower than Afghanistan's 0.19, Angola's 0.08, Sudan's 0.16, and D.R. Congo's 0.07. Similarly, Malawi's 3094 nurses and midwives in 2003 i s equivalent to a density o f 0.26, again much lower than Angola's 1.19, Sudan's 0.85 and D.R.Congo's 0.44 (JLI, 2004). Scarcity of TrainedHealth Professionals 252. The scarcity of physicians, nurses and midwives in Malawi prevents the delivery of essential health services. These low densities are manifested in high vacancy rates (percent o f unfilledposts) rangingbetween 30 to 80 percent o f skilled posts in government and CHAM health facilities (Table 4.1).92 The MOH has calculated minimumstaffing levels necessary to provide the essential health package (EHP). These alarmingly high vacancy rates are impeding delivery o f even basic health services. For instance, only 152 out o f 585 government and mission health facilities, including hospitals, currently meet even the "relaxed" staffing standard o f 2 nurses/midwivesper 90 The Ministry o f Health has prioritizedan integratedset o f cost-effective interventions to address the groups o f diseases which constitute the major causes o f illness and mortality. This Essential Health Package establishes a minimum standard o f health care to be provided free-of-charge for all. 9 1The JICA study applied the following criteria indetermining whether a facility could deliver EHP services: (a) it must be able to deliver outpatient care, family planning services, maternity services, and immunization; and (b) it must have the following staff complement - medical assistant or clinical officer (one per facility), and nurseimidwife (two per facility). 92 The MOH vacancy rates by cadre inend-2003 were 82 percent for specialist doctors, 32 percent for medical officers, 25 percent for clinical officers, 39 percent for medical assistants, and 67 percent for combined nursesimidwives. Inthe same year, the CHAM vacancy rates were 42 percent for medical officers, 36 percent for clinical officers, 45 percent for medical assistants, and 53 percent for combined nurse-midwives (Health Planning Services Department, 2003). 76 facility required to provide the EHP (Hozumi, 2003). There are 10 districts with no MOH doctor but with a non-government doctor, and four districts with none at all (Hozumi, 2003). 253. While the vacancy situation remains dramatic, action has recently been taken to address the problem. In addition to death due to HIV/AIDS, precipitating factors causing the skilled staff shortage include poor working conditions, and better- paying opportunities to work abroad. In this respect, starting in 2005 the GOM, with financial support from DFID and other donors, has boosted substantially the salary o f skilled health workers inMalawi (see below). The staffing crisis was also worsened by the historically inadequate production o f new workers (around 300 nurses and 20 doctors annually). An Emergency Training Program, launched a couple o f years ago, i s expected to produce, among others, 3950 new nurses up to 2009/10, to nearly double the 2004/05 stock of 4717. However, data on staff trained under the emergency training program show that only 2 percent o f the 309 initial set o f trainees joined the government health services (the rest went to C H A M and NGOs), so policies need to be put inplace to retain more o f the graduates (Staple, 2004). Figure4.1 Number of HealthFacilitiesand Number Capable of DeliveringEHP,by District,2002 I51 Nkhatabay Rumphi Dedza Dowa Kasungu Lilongwe I56 Mchinji Nkhotakh Ntcheu Nchisi Salima Balaka Blantyre Chikwawa Chiradzulu Machinga Mangochi Mulanje I Mwanza Nsanje Phalombe Thyolo I Zomba 1 0 10 20 30 40 50 60 Facilities capable of providingEHP R Facilities w/ at least 2 staff 0 Total facilities Source: Hozumi, 2003 77 254. While health facilities suffer from severe shortages of skilled staff, there are significant surpluses of unskilled staff. The proportion o f skilled and semi-skilled staff (Le., with two years or more o f training) to total staff in each o f the workforce areas i s low (Table 4.2).93 For instance, there i s more than double the establishment o f patient attendants. The result i s that, for instance, in nursing care in MOH, only 38 percent o f the staff are skilled or semi-skilled, while most o f the staff (62 percent) are hospital attendants or orderlies. This may signal inefficiency inthe distribution o f the workforce, or it may be an attempt to compensate for the lack o f more qualified staff. Table 4.1 Current and Required Human Resources for Government and C H A M Health Facilities, 2005 Cadre Target Basedon Malawi M O H Current Current % Actual Staff in % Vacant Tanzanian Levels Target Cadre Staff Vacancies Post to Target Posts Physicians 45 1 433 139 294 32 68 Nurses 9372 8,440 4,7 17 3,723 56 44 Clinical Officers n.a. 1,405 942 463 67 33 Medical Assistants n.a. 1,500 718 782 48 52 Lab Technicians n.a. 507 25 1 256 50 50 Pharmacists n.a. 285 93 192 50 50 Environmental n.a. 1,662 304 1,358 18 82 Health Officers Source: MOH; Staple 2004. Table 4.2 Distribution of Skilled and Semi-Skilled Staff, 2003 Cadre MOHP C H A M % o fTotal Within the Cadre, % % o fTotal Within the Cadre, % Workers Skilled Workers Skilled Clinical 7.6 100.0 10.3 100.0 Nursing 43.9 37.9 80.7 43.4 Preventive 45.5 5.6 0.4 75.0 Technical 3.0 n.a. 9.0 n.a. Total 100.0 100.0 Source: HealthPlanning Department (2003), Tables 8 and 12. Partnership with Private ServiceProviders 255. MOH is in the process of establishing service agreements with mission facilities, although current implementation is fragmented. MOHhas so far concluded 3 service agreements with C H A M hospitals to improve the efficiency and coverage o f service delivery. Some major problems exist in implementing service agreements however, including the fact that C H A M i s unable to negotiate as an "umbrella" organization. Since each mission hospital i s autonomous, MOH has to negotiate with every individual facilities, and this process has been management-intensive and time- consuming. 256. The private sector is emerging as a significant provider of services, but public private partnerships remain under-exploited. Around 18 percent o f all 93 The M O H staff establishment, which was revised as part o f the Functional Review in 1998 though not implementeduntil2001 (Martineau, Sargent, et al., 2001), i s not based on a detailed analysis o f workload, which i s imperative given the burden o f HIV/AIDS. 78 consultations are being done outside MOH and mission facilities, according to the 2002 CWIQ Survey. Moreover, more than half o f patients who go to government facilities first and do not receive adequate drugs or treatment end up going to private providers, according to the Malawi Economic Justice Network Survey. MOH has recently appointed a desk officer charged with the responsibility o f developing a public-private partnerships policy for the health sector. Service agreements could also be extended to non-CHAM NGOs, and the recent passage o f the NGO Law and the Procurement Law should assist inproviding the legal basis for such contracts. Quality of Care 257. The lack of skilled health workers and severe overloading of the available workers results in poorer quality and shorter consultations. Due to the shortage o f skilled medical workers, the workload has skyrocketed. For example, inantenatal care, a typical health worker faces as many as 100 patients per 8-hour day, implying an average encounter o f 4.8 minutesper patient, compared to the standard 30 minutes. Similarly, for family planning, a typical health worker faces as many as 50 patients per day, implying an average encounter o f 9.6 minutes per patient, compared to the standard 20 minutes (Rashidi, 2003). 258. I t is not uncommon for less-skilled workers to do technical and clinical work for which they are not trained. Inresponse to the heavy workloads, trained staff are "down-loading" clinical tasks, with weak and infrequent supervision, which also has serious effects on quality o f care. For instance, a study inone health facility found that 20 out o f 34 deliveries (or 59 percent) were attended by ward attendants (Ratsma and Ostergaard, 2003). 259. Health service utilization and client satisfactionis low. The low utilization o f health services and the low satisfaction expressed by households who usedthem i s due primarily to lack o f health staff, drug stock-outs, and poor treatment (Table 4.3; CWIQ, 2002). The poor quality o f health facilities i s starkly shown in the high fatality rate o f mothers reaching health centers and hospitals to deliver: in small-scale community audit o f all maternal deaths in Nankumba (population o f 63,000) 44 percent o f all maternal deaths occurred in hospitals and 12 percent in health centers (McCoy, et al., 2004), proving that although patients had physical access to facilities, these are ill-equippedor ill-staffed to deal with emergency obstetric cases. Table 4.3 Household Need, Use, and Satisfaction of Health Services in Malawi, 2002 (as percentage of survey respondents) Household Households W h o Households W h o Used Households Satisfied Residence Needed Health Services Health Services with Health Services Rural 18.4 16.9 11.9 Urban 14.8 16.6 12.7 Source: Malawi Core Welfare Indicators Questionnaire (CWIQ), 2002 260. Tertiary hospitals are hobbled by lack of autonomy. The granting o f autonomy to central hospitals has been in discussion since the mid-1990s. In addition to insufficient financial and human resources, the limited control that tertiary hospitals have 79 over their operations i s hampering their performance, notably in the key functions o f strategic management, procurement, financial management, human resources management, and administration. Delivery of Essential Drugs 261. Access to drugs is highly constrained. As many as 34 percent o f those who consulted a health provider duringthe past four weeks o f the CWIQ Survey in2002 cited unavailability of drugs as a reason for dissatisfaction. Unavailability o f drugs i s both a rural and urban problem (34 percent and 36 percent, respectively), with the northern region suffering the most (49 percent), compared to the central and southern regions (32 percent and 32 percent, re~pectively).'~ 262. The problems in drug availability are mainly due to Malawi's inability to reform the institutional structure of Central Medical Stores. Malawi remains almost solely reliant on the Central Medical Stores (CMS), which has been plagued for many years by institutional, management, and financing problems. Government facilities often use CMS on an "IOU" or "bursary" basis, because o f the often unpredictable and insufficient release o f funds for drugs by the Treasury and the absence o f a sustainable alternative financing mechanism. As a result, CMS i s frequently in arrears. CMS also suffers from a lack o f qualified staff, poor planning, poor logistics, and weak support systems. Poor stock management and storage often results in health centers receiving medicines not ordered, or outright non-delivery o f ordered medicines. Organizationally, CMS i s under the MOH and lacks the autonomy to operate as commercial enterprise. An action plan has been developed outlining the short-term steps to change CMS management, including international recruitment o f key positions. In 2006, the government awarded a contract to hire an external management team for CMS. 263. The country does not benefit from the efficiency gains of a more unified procurement system and bulk purchasing. Because o f CMS' inefficiency, several parallel procurements occur, each utilizing different logistics principles and software (e.g., SIGMED for CMS, and Supply Chain Manager for USAID-funded district pharmacy programs). This makes it difficult to ascertain at the facility level when to replenishsupplies. For other essential drugs within the purview o f CMS, the lack o f drug needs quantification, a poor information system, intermittent drug demand from facilities, and limited forecasting capacity all contribute to the making rational purchasing impossible. 264. Inadequate guidelines, misuse of drugs by patients and health facilities, high failure rates for certain drugs, and high vaccine wastage, characterize the use of drugs. Evenif access to drugs were to improve, much progress i s needed to enhance the effective use o f drugs. Practice guidelines are confusing or altogether lacking for 94Difficulties in accessing drugs i s confirmed repeatedly by surveys, such as the DHSwhich reports l o w use rates o f malaria bednets, Vitamin A to pregnant mothers, iron folate to pregnant mothers, oral rehydration salts to children with diarrhea, antibiotics to children with pneumonia, and STI drugs to those with infections. Drugstockouts are also corroborated by the Save the Children's health facilities survey in Balaka District where 4 out of 9 clinics reported stockouts (Patterson, Zakariya, and Kambalane, 2003) and the MSH(2003) baseline study on QECH. 80 some disease interventions covered under the Essential Health Package. The Malawi Essential Drug List (EDL) was updated in 2005, and the MOH already uses generics drugs in its health facilities. However, guidelines are lacking for some conditions, especially new programs such as ARV treatment, emergency obstetric care, and neonatal care (MoH POW,2004), and for others some minor changes could be made to simplify treatment and reduce treatments costs. Moreover, misuse o f drugs continues to be common. Patient knowledge o f drug dosing i s extremely low inMalawi (a little above 20 percent), compared to around 80 percent in other African countries (WHO, 1999). But misuse of drugs is also observed in health facilities (IMCI Baseline Survey, 2000). Compounding this, even when drugs are used correctly, the failure rate o f key drugs i s high.95Further, the vaccine wastage rates are very high for all antigens, except DPT.96 High wastage stems from poor maintenance of cold chain equipment, problems with erratic supply o f fuels (arising from erratic budgetary financing), non-compliance with the multi-dose vial policy, poor techniques and overstocking o f vaccines leading to their early expiry. Preventive Behavior and Knowledge aboutDisease 265. There are large gaps between possession of information and behavior change and practice among households. Although knowledge o f malaria prevention is high, ownership and use of insecticide treated nets is low; particularly for pregnant mothers. Similarly, while knowledge o f HIV/AIDS transmission and prevention i s very high,the use of condoms is low. Behavioraland cultural factors couldbe impeding good health practices and care-seeking, particularly in the diagnosis and treatment o f sexually transmitted infections (STI) even though these services are free. For both child and maternal health, household care-seeking i s generally delayed due partly to financial reasons and women's lack o f autonomy to make decisions. 4.3 Impact of HIV/AIDS 266. Malawi is the eighth worst affected country in the world from HIV/AIDS.97The 2003 HIV Sentinel Surveillance Report shows that the HIV prevalence for all antenatal care attendees was 19.8 percent; this figure has remained stable from 2001 (19.5 percent), but lower than 1999 (24.1 percent) (NAC 2005).98HIV prevalence for antenatal attendees in urban and semi-urban areas (21.7 and 20.8 percent, 95For instance, the antimalarial SP parasitological failure rate i s 25 percent (the failure rate should be zero (Lettington, 2004). 96Vaccine wastage rates are 70 percent for BGC, 50 percent for measles, 38 percent for OPV, 34 percent for tetanus toxoid, and 26 percent for D P T (the only antigen with a wastage rate within acceptable limit), according to the Immunization Program Financial Sustainabillity Report (MOH, 2003). 97Malawi's national incidence rate for H I V i A I D S is estimated at 14.1percent (NAC 2005), and the level o f HIV infection inthe adult populationhas remained fairly constant for the last nine years. For comparison with other countries, comparisons for HIV/AIDS rates are among prime age adults (aged 15-49 years). At 11.8 percent, Malawi is ranked after Swaziland (39 percent), Botswana (37 percent), Lesotho (29 percent), Zimbabwe (25 percent), South Africa (22 percent), Namibia (21 percent) and Zambia (17 percent). Data are from the Population Reference Bureau (PRB), 2004 and 2005. 98Due to the difficulty inobtaining accurate data inpart due to stigma, HIV prevalence is often estimated by extrapolating national estimates from antenatal clinic (ANC) surveillance data. 81 respectively) i s significantly higher than in rural areas (14.5 percent), but this gap i s narrowing because prevalence in the rural areas i s increasing (12.1 percent in 1999 to 14.5 percent in 2003). The south's prevalence rate continues to remain higher (23.7 percent) than the central and northern regions (19.2 and 19.1 percent, respectively). HIV prevalence i s especially high among adolescents (15-19 years old). Transmission remains largely through heterosexual contact (90 percent), and the rest through mother- to-child transmission (9 percent) and unsafe medical procedures (1 percent) (NSO and ORC Macro 2005, based on DHS 2004). 267. HIV/AIDS has generated a substantial increase in the overall burden of disease. HIV-related conditions now account for around 40 percent o f all inpatient admissions and about 70 percent o f admissions to medical wards (MGFCC, 2002). HIV co-morbidity (tuberculosis, other opportunistic infections) i s also increasing. About 77 percent o f TB patients are HIV positive (Kwanjana, et al., 2001) and reported cases o f TB have risen from 95 per 100,000 population in 1987 to 275 in 2001. The TB cure rate has declined to only 69 percent inthe late 1990s (compared to 90 percent inthe 1980s). The quantum increase in TB cases i s likely to be mirrored by increases in opportunistic infections. 268. HIV/AIDS reduces the availability of health workers. An estimated 2,000 health workers are living with AIDS (Ndyanabangi, et al., 2004). Although HIV/AIDS i s by no means the major cause o fthe drastically declining number o fMOHhealth workers, it is a significant contributor through staff deaths, incapacity, and absenteeism. For instance, 8 MOH staff died in 1990 compared to 270 in 1999 and 200 in 2000g9,with particularly high mortality rates among technical cadres and front line staff. About 498 days are lost per 100 nurses per year (or approximately 5 days each annually) through absenteeism (MOH, 2004). Absences due to illhealth, family problems, and attendance at funerals meant that total time worked only averaged 23.8 hours per lab technician per week, compared to the 44-hour norm (EMLS, 2002). Thus the full time equivalent (FTE) numbero fhealthworkers is evenless than the actualheadcount would suggest. 269. HIV/AIDS undermines the quality of services. Fewer clinicians are available as a result o f HIV/AIDS. Staff shifting from clinical to non-clinical services because o f their own HIV/AIDS status has increased the care burden for clinical service workers. In one study o f six districts (Ndyanabangi et al., 2004), 75 health workers left clinical services for non-clinicalwork, and o f these 95 percent were nurses, almost half o f whom were in their prime ages o f 30-39 years. In addition, staff shifts from general duties to ARV create an even greater deficit o f staff to deal with other medicalproblems. 270. Front line health workers are exposed to an unacceptably high risk of infection, further compounding staff attrition. HIV/AIDSprevention and care policies, guidelines, and protocols inthe workplace continue to be weak, and 96 percent o f service providers perceive risk to themselves from HIV/AIDS exposure, while 93.4 percent perceive a risk to their clients (JHPIEGO, 2003). These fears are well founded given poor health waste management and injection safety. Syringe reuse rates, particularly in immunization programs, were as high as 10.7 percent, and 49 percent o f health workers 99Though cause o f death i s not specified, 59 percent o f MOH deaths were those between the ages o f 30 to 44 (UNDP, 2002), the same age range where HIV/AIDS infectioni s concentrated. 82 giving vaccinations and 57 percent giving curative injections reported suffering at least one needle-stick injury during the last twelve months (Aitkenand Kemp,2003). 4.4 Expenditure Trends 271. Budget expenditures in health have been rising in real terms and as a share of total budget expenditures. There i s substantial volatility inhealth expenditures from year to year, but the trend is upwards. Government expenditures on health have risen in real terms from an average o f 1.5 percent o f GDP (USD28 million) in 2000/01 and 2001/02 to an average o f 3.0 percent o f GDP in 2003/04 and 2004/05 (USD56 million). Budget expenditures have also increased as a share o f total expenditures, from an average o f 4.8 in 2000/01 and 2001/02 to 6.9 in 2003/04 and 2004/05, reflecting an increased prioritization o f health in the national budget (Table 4.4). When including off- budget donor expenditures, health sector spending reached an average 7.1 percent o f GDP in2003/04 and 2004/05. 272. However, funding for the sector remains insufficient for the delivery of the Essential Health Package. Per capita budget expenditures on health were USD5 in 2004/05, which i s below the USD17 per capita necessary to deliver the Essential health Package (conservatively costed), and the USD34 per capita recommended by the Commission on Macroeconomics and Health in 2001. When including off-budget donor expenditures, health expenditures per capita were just below USD13 in2004/05. 273. A large share of the total expenditures in the health sector is provided by development partners. The share o f externally fundedexpenditureson health i s around 70 percent, and has remained fairly stable over the period. Almost all external finding from development partners was off-budget up to 2002/03, but the share included in the budget has increased substantially inrecent years. 274. Household spending on health is very large, amounting to one-quarter of total expenditure. The 2001 National HealthAccounts indicate individual out-of-pocket expenditure on health was USD3.30 per year, amounting to around one-quarter o f total health expenditures per capita. Although no information exists since 2001, it i s likely this figure has risen substantially due to the increasing privatization of care, and drug and supply shortages inpublic facilities. Most household spending is on outpatient care. Recurrent Expenditure 275. The growth inhealth expenditures has been driven mainly by increasesin recurrent expenditures. The health share of total government recurrent expenditures has increased from 5.1 percent in2001/02 to a substantial 7.4 percent in2004/05 (Table 4.4). Recurrent expenditures reached2.4 percent o f GDP in2004/05. 83 Table 4.4 Total Health Expenditures,by Recurrent and Development Classification, and by source of funding, 2000101to 2004105 2000/01 2001/02 2002103 2003/04 2004/05 NominalValues (MK Million) Actual Actual Actual Actual Actual Total Voted Expenditures 37266 42490 61260 80536 96625 Total Voted Recurrent Expenditures 25736 32675 49473 59537 71656 Total Voted Development Expenditures 11530 9816 11787 20999 24969 Total Voted Expendituresin Health 2055 1751 3645 6176 6009 Voted RecurrentExpendituresin Health 1912 1678 3480 4069 5319 Ministry ofHealthVote 1863 1618 3415 4002 5248 Othersupportorganizations 49 60 65 67 72 LocalAssembliesVotes (Total) 0 0 0 0 0 Voted DevelopmentExpendituresinHealth 143 73 165 2107 690 GovernmentofMalawi 114 32 56 98 71 Donors(Development Patners) 29 41 108 2008 618 Donors - 4282 6592 9934 9761 On-budget 29 41 108 2008 618 Off-budget 4241 6484 7926 9 143 Total Expenditure onHealth(on-budget andoff-budget) -- 5992 10129 14102 15152 Total domestically-fundedexpenditure on health 1710 3536 4167 5390 Total externally-fundedexpenditure onhealth - 4282 6592 9934 9761 GDP (nominal) 113871 136141 160137 189563 226272 Exchangerate (Malawi Kwacha/US$) 65.9 69.9 85.2 106.6 111.1 2000/01 2001/02 2002103 2003/04 2004/05 (USD million) Actual Actual Actual Actual Actual Total Voted Expenditures 565.8 607.9 719.0 755.5 869.7 Total VotedRecurrent Expenditures 390.7 467.4 580.7 558.5 645.0 Total Voted Development Expenditures 175.0 140.4 138.3 197.0 224.7 Total Voted Expendituresin Health 31.2 25.0 42.8 57.9 54.1 Voted RecurrentExpendituresin Health 29.0 24.0 40.8 38,2 47.9 Ministry o fHealthVote 28.3 23.1 40.1 37.5 47.2 Othcrsupportorganizations 0.7 0.9 0.8 0.6 0.6 LocalAssembliesVotes(Total) 0.0 0.0 0.0 0.0 0.0 Voted DevelopmentExpendituresin Health 2.2 1.0 1.9 19.8 6.2 Governmentof Malawi 1.7 0.5 0.7 0.9 0.6 Donors(DevelopmentPatners) 0.4 0.6 1.3 18.8 5.6 Donors 61.3 77.4 93.2 87.9 On-budget 0.4 0.6 I.3 18.8 5.6 Off-budget 60.7 76.1 74.4 82.3 Total Expenditure on Health(on-budget andoff-budget) 85.7 118.9 132.3 136.4 Total domestically-fundedexpenditureonHealth 24.5 41.5 39.1 48.5 Total externally-fundedexpenditure on Health 61.3 77.4 93.2 87.9 PercentShares Recurrentexpendituresinhealthas % of totalvoted recurrentexpenditures 7.4 5.1 7.0 6.8 7.4 TotalVoted expenditureson healthas % oftotal votedexpenditures 5.5 4.1 5.9 7.7 6.2 Total expenditureon healthas % of GDP 4.4 6.3 7.4 6.7 Total voted healthexpendituresas % of GDP 1.8 1.3 2.3 3.3 2.7 Totalvoted health recurrent expendituresas % ofGDP 1.7 1.2 2.2 2.1 2.4 Totalvoted healthdevelopmentexpendituresas % of GDP 0.1 0.1 0.I 1.1 0.3 Domesticallyfundedspending on Healthas % ofGDP 1.3 2.2 2.2 2.4 % Government to Total Voted DevelopmentExpenditures 79.7 43.8 33.9 4.7 10.3 % Donorsto Total VotedDevelopment Expenditures 20.3 56.2 65.5 95.3 89.6 % Donorsto Total HealthExpenditures 71.5 65.1 70.4 64.4 Governmentexpenditureon healthper capita 2.4 4.0 3.7 4.5 Total Votedexpenditure on healthper capita 3.1 2.5 4. I 5.5 5.0 Totalrecurrent spendingper capita 2.9 2.4 3.9 3.6 4.4 Per capita developmentexpenditures 0.2 0.1 0.2 1.9 0.6 Per capita total expenditures 8.4 11.5 12.5 12.7 GDP (nominal) 1728.7 1947.7 1879.5 1778.3 2036.7 Population(millions) 10.0 10.2 10.4 10.6 10.8 Source: Ministry of Finance,AnnualAppropriations Account. 84 276. Primary and secondary health services are being given more prominence in recurrent expenditures, compared to tertiary services. Over the past few years, secondary health care services have accounted for the largest share o f recurrent expenditures, between 32 percent and 37 percent. Primary health care has captured between 22 percent and 27 percent o f expenditure. The share of tertiary health care services has correspondingly declined to less than a fifth o f total recurrent spending (Figure 4.2). A slightly different picture emerges when a similar analysis i s done for the district level alone, where over 50 percent o f the expenditures go to secondary health care services, while primary health care services take up between 30 and 40 percent o f recurrent expenditures. Figure 4.2 Percent Distribution of Recurrent Health Expenditure, by Level of Care and by Economic classification, 2001/02 - 2004/05 I I I I 100% 80% 60% 0% Secondary care 40% rn% Rimary care 20% Q % Admnistration 0% and support +Q +Q +Q +Q ,@ +Q? ,Vk ,." Source: GOM Consolidated Appropriation Accounts, various years; MOH returns to MOF Note: Figures for primary and secondary care are combined in FY 2004105 due to a change in accounting svstems Nevertheless, the level o f Figure 4.3 Distribution of Recurrent spending on administration and Expenditure by Level, 2001/02-2004/05 support services as o f 2004/05 i s highand suggests the existence o f 100% 90% inefficiencies. 80% 70% 278. The share of 60% 50% rn Central Hospital headquarters recurrent 40% expenditures has increased 30% 20% significantly. The district 10% 0% expenditures fluctuated around 61 2001/02 2002/03 2003/04 2004/05 percent On average (Figure 4'3)' The allocation to tertiary hospitals iource: G o M Consolidated Appropriation Accounts, various ' years. MoHreturns to MoF looExpenditures on administration and support for district and tertiary levels were largely presented as secondary and tertiary health services inthe initial years. 85 gradually declined from 25 percent in 2001/02 to 21 percent in 2004/05. On the other hand, the share o f the recurrent budget allocated to headquarters has increased substantially over the past few years, from 12 percent in 2001/02 to 20 percent in 2004/05. The reasonbehindthis trend i s unclear. 279. The method of financing referrals and medical evacuation abroad through direct budget support is inefficient, unsustainable, and lacks transparency. MOH currently has a line o f direct budget expenditure on referral o f patients abroad for specialist treatment, usually in the form o f medical evacuations to hospitals in South Africa and Botswana. The cost o f such referrals i s enormous and historically represented a significant budget outlay. For instance, in FY03/04, it accounted for around MK60 million (around USD550,OOO) o f MOH expenditure, though it has been drastically cut to M K 3 million (USD28,000, for 15 referrals) in the current financial year. (But even at this low number o f referred patients, the average cost is still considerable at USD1,866). This scheme is evidently non-transparent and prone to political interference. More importantly, the scheme i s inefficient in that it operates on a fee-for-service reimbursementbasis, thus givingproviders a `blank cheque' for charges. A more efficient way o f dealing with this problem i s for the government to actively negotiate for discounted fees or, better still, convert this program into a contributory health benefit for civil servants. 280. The share of personal emoluments (PE) to total health recurrent expenditures has declined, after increasing for most of the 1990s. PE expenditures took a sharp increase in the past decade, from 23 percent in 1990/91 to 39 percent in 1997/98, but it has fallen again in Figure 4.4 Teacher and Health Sector Salary recent years to only 28 percent in Index, Relative to Other Civil Servants 2004/05 (Figure 4.2). This may be explained by acute staff shortages and low salaries, but has also been exacerbated by a sharp increase in other recurrent transactions spending in recent years. 281. Staff salaries were very low, but have increased substantially following the wage reform in October 2004 and D S F G Y I ? L hf 2: 0 P Q R Cndr recent donor-funded initiatives. OTerbm .Rn:&aciterr Monthly remuneration inclusive Note: The index measures the gross salaries o f teachers and o f allowances o f the most senior- health workers divided by the salaries o f other civil servants (as o f February 2006). There are no teachers or health level specialist doctor in2003 was workers inthe top grades from A to C. Health sector staff is only around MK25,603 widely dispersed through the grades, though most nurses are (USD234), while a senior nurse ingrades K,L,M,and doctors are inhigher grades. earned MK12,930 (USDll7) per Source IMF 2007, based on M o F data month. Relative to other countries in the region, civil servants' 86 salaries in Malawi were extremely low (Valentine, 2003).'O' The civil service wage reform started in 2004 has provided top grades with a 50-70 percent nominal salary increase.102 282. Salary top-ups of 52 percent, financed by donors from the Health SWAP pool, are critical to reduce the human resources crisis inthe health sector. As part o f the SWAp, a large grant from DfID is also contributing to redress the low level o f remuneration. The six-year grant aims to improve incentives for recruitment and retention o f Malawian staff through a salary top-up o f 52 percent (based on October 2004 salary levels) for eleven selected professional and technical cadres. As a result health workers salaries are now far larger-typically 50-100 percent above other civil servants insimilar grades (Figure4.4). lo3 There has also been a slow down inthe out-migration o f nurses from Malawi. The wage differentiation has occasionally been the source o f dissatisfaction in the rest o f the civil service, but has not been a discernible source o f overall wage pressures in the civil service. While there i s a consensus that the top-ups are necessary given the critical situation o f human resources in the sector, the sustainability o f this arrangement is doubtful and the scheme represents a potential fiscal liability for the government inthe future. 283. Expanding the training capacity is also critical to resolve the human resources shortage in Malawi. Within the context o f the SWAp, government and development partners have agreed on an Emergency Human Resources Program. In addition to providing the 52 percent salary top-up, the program will expand training capacity by over 50 percent on average, and more in key cadres, through infrastructure expansion and increasing teaching staff at training schools, over six years. The plan aims to triple the number o f doctors and double the number o f nurses in training. While more Malawians are beingtrained, the program i s paying for volunteer doctors and nurse tutors to fill vacant posts critical for training and delivering health services. In addition, the M o H has started preparation o f a Strategic Framework for human resources including a draft deployment policy and incentive package for hard to reach places. Development Expenditure 284. Development spending in health has fluctuated widely, due to poor capture of donor expenditures in the budget and under funding of budgeted government allocations. Coverage o f donor funding in the budget has been improving from around 20 percent in 2003/04 to 62 percent in 2005/06 (JPCR 2007 briefing note), but a substantial share of external funding is still off-budget. In addition, changes in Io' The median salary earner in Malawi MOH used to get only USD45 equivalent per month, compared to USD145 inZambia and USD251 in Tanzania. The only exceptions to the l o w salaries in Malawi are health staff working under the Performance Contract Scheme, whose contracts are quite attractive. A 50 percent increase inthe minimumwage in2004 meant a substantial salary raise for the lower grades o f the civil service, and the February 2006 wage adjustment provided lower grades an increase o f 13-15 percent. The health salary premiums are substantial and-bearing inmindthe low GDPper capita-should constitute an attractive remuneration for Malawians. However, this i s probably warranted by the high demand for health workers from neighboring countries as well as in developed countries (e.g. the United Kingdom). 87 classification o f funding (from development projects to recurrent program support), lo4 means that analysis o f the development account i s difficult. It should also be noted that the definition of capital expenditureis not clear ingovernment budgeting.Inmost cases it refers to external sources o f funds (including recurrent expenditures) and not necessarily expenditure on capital items. As such, there i s a lot o f recurrent expenditure classified as capital, simply because it i s from an external source. As mentioned above the total external funding to Malawi accounts for about 70 percent o f total health sector expenditures, which leads to concerns over the financial sustainability o f the sector. 285. "Pooled financing" among a few donors has started but substantial resources are managed as parallel funds, contrary to the Government'svision of a more unified financingthrough SWAp. MOH has collaborated with its development partners to develop a Program o f Work (POW) intended to finance the delivery o f an essential health package (EHP) under a sector-wide approach (SWAp). Beginning in 2005/06 the UK DFID, NorwayBida, UNFPA, and the World Bank, are pooling their funds with the Ministry o f Health as part o f the health SWAp. The rising importance o f the Global Fund and other disease initiatives has created difficulties in planning and managing the health sector. Funds from these sources are large but are highly unpredictable due to protracted preparation and negotiations, as was the case with the USD39.5 million malaria grant. Moreover, activities are highly 'projectized', many o f them occurring outside M O H and through 'demand-driven', fragmentary, and poorly coordinatedNGO and community-based activities. 286. The inflowof externalassistance is increasingdramatically,but a significant gap exists inthe financingof the essential healthpackage (EHP). The POWi s costed at USD735 million over six years, and involves increasing the per capita (government and donor) health spendingfrom USD7.7 in2004/05 to USD12.6 per year in2009/10. For FY04/05, it has beenestimatedthat as muchas USD13 millionin pooled funds already supported the POW,although pooled commitments were as highas USD23.1 million. Available data on donor commitments show, however, that expected contributions from SWAPparticipating donors plus the expected government health budgetare not enough to cover the resource requirements o f the outer years o f the POW, with the expected gap reaching USD38.5 millioninFY06. 287. The stalled fee reformsin centralhospitalsleave themwith little source of alternativefunding. The decline inbudget expenditures going to tertiary hospitals has not been accompanied by more intensive internal resource mobilization. The Treasury approved the retention o f hospital fees in 2003 on the understanding that fees recovered from private patients will cover (or at least significantly support) the cost o f the services to poorer patients. The existingpatient tariffs were gazetted in Parliament in 2000. Five years have passed and they have not been updated. These tariffs ("current fees in force" inTable 4.5) are very low and do not correlate with the actual cost of services. They do not include capital costs or operational overheads and have been overtaken by the kwacha depreciation and domestic inflation. For instance, although the actual cost o f an outpatient department visit in 2004 inpublic hospitals has reached MK325, the fee level has remained at MK40. The stagnant fee levels mean that private patients and those with 104A shift in above andbelow the line spending in2003104 howevermeans this is not a comparableyear. 88 insurance coverage such as MASM are being charged fees much lower than the actual economic cost o f the hospital services they receive. I t should be noted that both Blantyre and Lilongwe do not have district hospitals, and therefore a significant portion o f their patient base are self-referred primary- or secondary-care patients, many o f them poor. To address the variety o f patients that tertiary hospitals have, a sliding scale fee should be considered. The stalled fee reforms have been due, inpart, to the little substantive action inthe granting ofautonomy to central hospitals so that they can set their own fees. Table 4.5 Illustrative Comparisonof Actual Cost of Outpatient Visit versus Current Fees in Force and ProposedFees (Not Implemented), 2001 and 2004 Cost vs. Fees, in MalawiKwacha 2001 2004 Actual Cost o f Service Public hospitals 231 325 Mission hospitals 245 344 Current Fees inForce inPublic Tertiary 40 40 Hospitals (since the 1990s) Proposed Fees in Public Tertiary Hospitals 170 if seen by 214 if seen by clinical practitioner; (not implemented) general practitioner 408 if seen by general practitioner Sources o f basic data: Levin, et al. (1999); .TIP Subcommittee (June 2004). Note: The actual cost o f service was derived from the study, "Costs o f Maternal Health Care Services in Blantyre District, Malawi". The 1999 data were adjusted to 2001 and 2004 figures by multiplyingthe implicit GDP deflators for those years (1999=1.00). Cost implications of HIVIAIDS 288. With a national prevalence rate of 14.1 percent in 2005, HIV/AIDS has substantial adverse effects on the government's fiscal situation. The epidemic is already a drain on scarce domestic resources, compounding the constraints to development that the country faces. It i s estimated that up to 60 percent o f health expenditures is already devoted to HIV/AIDS-related programs (Republic o f Malawi, 2005). Further, treatment activities absorb the largest share o f expenditures (Table 4.6). This implies that it will be hard to reduce the dependence on a large share o f external resources, though the domestic private sector could potentially increase resources devoted to tackling the epidemic. Table 4.6 Shares of Funds Spent on SpecificHIV/AIDSActivities 2004/2005 Share o f HIViAIDS funds Treatment, Care & Support 40% Advocacy and Prevention 24% Leadership and Coordination 11% M&E 3% Capacity Building 16% Impact Mitigation 6% Source: Republic o f Malawi (2005) 289. The sizable financial commitment must be viewed in the context of a Malawi's heavy reliance on external aid and the volatility of external assistance. As seen in Chapter One, Overseas Development Assistance (ODA) as a share o f GDP has fluctuated wildly since 1980, ranging from 40 percent at its peak to just 10 percent in trough years. While volatility in ODA seems to be declining it i s still substantial. Given the heavy reliance o f the health sector inparticular on external assistance, this volatility 89 could pose particular difficulties in the context o f HIV/AIDS since continuity o f treatment, prevention and mitigation are all essential to success in combating the epidemic. A large proportion o f funds spent on HIVIAIDS are external, and the Global Fund contributed some USD38.5 million or 69 percent o f international funds for HIV/AIDS programs inthe period 2004/05. 290. The GOM set a target for 80,000 ever started on ART by 2006 and 240,000 ever started on ART by 2010, which requires significant financial resources. In2005, it is estimated that 169,000 people inMalawi neededART, whilst 31,000 were actually receiving it, implying a coverage rate o f around 20 percent (UNAIDS 2006b). Achieving the target for ART delivery of near universal coverage implies significant financial commitments inthe near future. 291. There is an inadequate funding stream to meet the projected cost of Malawi's current goals and commitments towards HIV/AIDS programs up to 2013. The financial requirementsnecessary to meet the government's goals and commitments have been estimated according to two distinct scenarios and based on specific assumptions. lo5 The `baseline scenario' considers the resource requirements needed to meet the UNAIDS' targets (that from 2010 onwards 75 percent o f those inneed o f ART will have access to it). The `low scenario' assumes current levels o f coverage o f HIV/AIDS related services will remain unchanged for the whole time span (at 20 percent). The GOM would need to spend USD61 million in 2007 on prevention, treatment and mitigation, increasing to USD210 million by 2013, in order to meet its targets (baseline scenario). Inthe low scenario, total resource requirements would amount to USD57 million in2007, increasing to USD95 million by 2013 (Table 4.7). Table 4.7 HIV/AIDS ResourceNeeds inUSD million, Baselineand Low Scenarios BASELINE SCENARIO LOW SCENARIO 2007 2010 2013 2007 2010 2013 Prevention Youth 1.6 3.8 1.9 1.6 1.8 2.0 Workplace Programs 1.1 12.1 13.0 CondomProvision 3.8 15.1 16.4 3.0 3.3 3.6 STI Management 1.9 5.9 5.9 1.9 1.9 1.9 VCT 1.2 2.2 2.4 1.2 1.3 1.4 PMTCT 2.0 2.2 2.3 2.0 2.1 2.3 Mass Media 0.4 2.0 2.0 0.4 0.4 0.4 BloodSafety 2.0 2.2 2.4 2.0 2.2 2.3 Sub-Total 13.9 45.4 46.3 12.1 12.9 13.8 Treatment and Care ART 27.4 87.8 153.7 22.5 39.0 56.1 Non-ART 16.7 6.3 6.9 19.3 20.8 22.7 Sub-Total 44.1 94.1 160.6 41.8 59.8 78.7 ovc `05A detailed descriptiono f assumptions and data sources i s providedinAnnex 4.2 (also see: David2007, "Fiscal Space and Fiscal Sustainability of HIV/AIDSPrograms in Sub-Saharan Africa" unpublished manuscript,ACT Africa, WorldBank, WashingtonDC). Cost estimates were obtainedusingthe EPP, SPECTRUM, AIM and RNMmodelsanddependon assumptionsmade about futurepaths o f the epidemic, costs and coverage rates of interventions. 90 School Expenses 3.0 3.0 2.9 3.O 3.O 3.O TOTAL 61.0 142.5 209.9 56.9 75.7 95.5 Source:World Bank staff calculations 292. The largest single expenditure component under both scenarios is ARTs. In the baseline scenario, ARTs representaround45 percentof the resource needs in 2007 and 73 percent in 2013. In the low scenario, ARTs are still the largest simple component accounting for 40 percent o f the resource needs in 2007 and 57 percent by 2013. 293. Administration, management, advocacy and monitoring and evaluation could raise the cost even further. Bollinger and Stover (2006) point out that items not included inthe projections could add about 7 percent to the costs o f the programs every year. 294. The estimates presented above, likely underestimate the real resource needs for universal access. The purpose o f the exercise is to give a rough idea o f resource requirements in the future and whether adequate fuding will be available to cover the needs. The precise estimates are indicative, as they depend on numerous assumptions and factors, and can change significantly if different assumptions are made such as regarding future changes in drug prices, development o f drug resistance and externalities associated with increased access to treatment, implementation bottlenecks, health system deficiencies, accuracy o f the data used, among others. For comparison, Malawi has recently developed a second generation National HIV and AIDS Action Framework (NAF). NAF targets for moving towards Universal Access (UA) to ART presume a total budget o f USD1.17 billion from 2006 to 2011.lo6 Table 4.8 Current InternationalCommitmentsof for HIV/AIDS, USDmillions Source Start Date Total Budget - Already Amount Disbursed Undisbursed GFTAM Oct-2003 178.6 41.4 137.2 Oct-2006 19.0 2.0 17.1 MAP (WB) 2003 35.0 26.0 9.0 PEPFAR 2004 14.5 N.A. N.A. PEPFAR 2005 15.2 N.A. N.A. PEPFAR 2006 16.4 N.A. N.A. Total 281.4 67.1 168.2 Source:GlobalFundWebsite, PEPFARWebsite, World Bank. 295. The medium-termsustainability of HIV/AIDS programs are not assured, even using rough estimates that arguably understate the resource requirements. Table 4.8 lists funding from the three major international donors for HIV/AIDS, the Global Fund, PEPFAR and the World Bank towards HIV/AIDS inMalawi. Inaddition, the government has committed to spending at least USD2 million annually towards the fight against AIDS (Republic o f Malawi, 2005). The figures indicate that international funding and domestic availability for HIV/AIDS related expenditureswill be adequate to meet the needs as estimated inboth scenarios above until 2008. Nonetheless, by 2009 in lo6 The largestsingle expenditureinthe UAbudgeti s for ART and relatedcosts (provisionof ART alone amounts to 35 percent of the budget, andHIV testingis 17 percent). The secondhighest cost i s for preventionandbehaviorchange(34 percent of the budget). 91 the baseline scenario and 2010 inthe low scenario, Malawi would needto raise additional finds, either from international sources or by reducing expenditures in other areas, or from alternative sources such as other donors, domestic revenue, the private sector or a domestic health insurance mechanism. None o f these alternative sources appears likely to yield significant amounts, however. 4.5 Povertyfocus and Incidenceof Expenditurein the HealthSector 296. Spending and access to health services is distributed with considerable equity across socioeconomic groups. Access to health services and monetary benefits from government curative health services are slightly biased towards the rich, although not in a very pronounced way. The poorest 20 percent received only 15 percent o f the government subsidy that went to curative health services. On the other hand, the richest 20 percent received 21 percent o f the subsidy (Figure 4.5).'08 Whilst the distribution o f benefits from the provision o f government health centers was fairly equitable, the poor receive a considerably lower share o f the benefits from the provision o f government hospitals. Figure4.5 Access to Health Services and BenefitIncidenceAnalysis of Public Spending inthe HealthSector Access :::I Lorenz Curves for Access to Health Services Lorenzs Curves for Benefit Incidence of Treatment Fi -Lineof Eoualltv . iI :::e i-Lineof Equality I ~~. 80% 1 80% - 70% 70% - -Cumuiatlvs per Capita; 60% Consumption 50% - Consumption 50% I 50% - -4-Benefit Incidence of 40% +Treatment at Government Facility 40% - Treatment 30% 30% - 20% +Hospitalisstion at 20% 0% Government Facility 0% .111111111~ .- Populationdistnbutionfromp o r t o nch bpulatin distribution frompoor to rich I 1 Source: GoMandWorldBank, 2006 Note: Unfortunately,expenditure datadoes not allow a differentiationon the basis oftype o f treatment; hencethe benefit incidenceanalysisincludesboththe benefitsof a visit or day treatment, with the benefits o fthose who are hospitalized.Access can be disaggregatedby type of healthfacility, however. 297. The equitable distribution of benefits from public health expenditures may reflect the prioritization and provision of essential health services without charge. The overall fairly equitable distribution contrasts with trends identified inother developing countries. Although the reason for an equitable distribution o f benefits cannot lo7These estimates are inlinewith an IMF study which finds that by 2010, Malawicouldbe devotingup to 6.5 percentof its GDP to HIV/AIDS relatedhealthservices, up from the estimated 4.3 percentin2000. (Haacker,Marcus (2003), ProvidingHealthCare to HIV PatientsinSouthernAfrica. IMF Policy DiscussionPaper). lo*There are no significant differences inthe patterno f subsidy distributionbetweenmales andfemales. However, the distributiono fthe health subsidyinurbanareas was morebiasedtowards the richthanthe poor, with only 6 percento fthe subsidygoingto the poorest quintile but42 percentfromthe richest quintile. 92 be determined from the benefit incidence analysis, it i s likely the result o f the recent prioritization o f essential health services and provision o f public health services without charge. 298. The utilization of specific health services is unequal across wealth groups, but the gap is narrowing. The 1992 and 2000 DHS surveys allow an analysis of the utilization o f selected health service by wealth quintiles (Table 4.9). For use o f contraception, all wealth groups registered increased rates and the gap between the poorest and the richest quintiles has narrowed between the two periods. The same i s true for women's knowledge o f HIV/AIDS. However, for child immunization and antenatal care for pregnant women, coverage worsened across all wealth quintiles. Similarly, large differences between the richest and poorest quintiles continue to be observed for such key indicators as under-fivemortality. Table 4.9 Selected Health Service Utilization Indicators by Wealth Quintiles in 1992 and 2000 Year Poorest Middle Richest PoorestMchest Indicators Quintile Quintile Quintile Ratio Percent o f children hlly immunized 1992 73 80 89 0.82 2000 65 69 81 0.80 Percent o f married women using 1992 4 6 17 0.23 contraception 2000 20 23 40 0.50 Percent o f pregnant women with 1992 82 89 93 0.88 antenatal care visits 2000 77 78 86 0.90 Percent o f births attended by medically 1992 45 50 78 0.58 trained personnel 2000 43 51 83 0.51 Percent o f women knowledgeable about 1992 61 56 68 0.90 sexual transmission o f HIV/AIDS 2000 88 92 98 0.90 Source: Population Reference Bureau (2004) and Carr (2004), both based on Gwatkin, Rutstein, Johnson, Suliman, and Wagstaff (2003) from analyses o f the 1992 and 2000 DHS data. 299. Increasing the share of benefits to the poor will require improving their health awareness and facilitating greater utilization of health services, as well as improving quality of services. The distribution o f benefits from health expendituresare largely explained by differences in the utilization o f health services and the lower reported incidence o f illness amongst the poor.'09 This implies that if the Malawi Government wants to increase the share o f the benefits reaching the poor, it would be important to understand what factors affect the utilization o f government health services, as well as individual decisions about health care. Any attempt to further increase the share o f the benefits from public health spending reaching the poor should focus on improving health awareness and facilitating greater utilization o f health services, particularly among the rural poor. lo9Analysis o f the IHS2 data suggests that the poor may have higher thresholds before classifying themselves as ill(GOM and World Bank, 2006). 93 4.6 Implementingkey recommendationsfrom the 2001 PER on health 300. Good progress has been achieved in implementing some reforms, but in other areas there has been little progress or even backtracking. Below we briefly review progress made ineach o f the five main recommendations from the 2001 PER: #I:Reallocate expendituresJFom Ministry Headquarters, andfacility construction towardspreventative health careprograms (e.g. malaria and HIV/AIDS control). Re- prioritize National Health Plan. The NationalHealthPlanhas beenreprioritized, incollaboration with development partners, including a focus on preventative health care programs and interventions to combat malaria and HIV/AIDS. 0 There is too higha share o f expenditures allocated to Ministry Headquarters. #2: ExpandpublicJinancing-private provision arrangements by increasing government subventions to CHAM. Limitedprogress has beenmade inexpanding public-private partnerships by increasing government subventions to CHAM. #3: Introduce cost recovery at tertiary hospitals, andfacilitate the creation of autonomous hospital Boards. 0 No progress has beenmade inintroduce cost recovery at tertiary hospitals, and facilitating the creation o f autonomous hospital Boards. #4: Reform Central Medical Stores (CMS) by taking the necessary stepsfor granting institutional autonomy to CMS and introducing measures such as reviewing the Essential Drug List and cost-sharingfor drugs. 0 Limitedprogresshas beenmade inreformingthe Central Medical Stores (CMS). Progress has been made inreviewingthe EssentialDrugList and reviewing cost- sharing for drugs. #5: Create apay and incentives package to recruit, train and deploy healthpersonnel to rural areas. Substantial progress has beenmade to improve the pay incentives to recruit, train and deploy health personnel, including inrural areas, largely thanks to funding from development partners. 301. The recommendations from the 2001 PER not yet implemented are still priorities today. N o technical constraints have been identified inthe analysis presented inthis chapter preventingthe granting of autonomy to hospitals and reforming the CMS. Therefore, there i s a need to raise the level o f debate to increase national awareness on the need for such reforms, andproceed with implementation. 94 4.7 RecommendedPolicy Prioritiesfor Malawi's HealthSector 302. Several recommendations arise from the analysis carried out in this study. The full recommendations are summarized below, but a shorter and prioritized list o f recommendations i s presentedinthe executive summary to this report. #1:The ratiooftrainedto untrainedhealthworkers mustbe dramaticallyimproved by investing heavilyin skilledworkers. Although the Six-Year Emergency Training Plan has already received significant government and donor support, funding for the salary top-ups beyond the six-year period remains to be secured. GOM should negotiate with the Global Fund to include a major HRH component in its Round IV proposal, and advocate that other donors to support the emergency training and recruitment plans. The human resource management systems to recruit and deploy workers, track their presence at post, supervise their work, and evaluate their performance need to be dramatically improvedto achieve commensurate workers' productivity. The current HRHplanning and policy framework needs to be amended to include a system for tracking trainees and recruiting them at the completion o f their training. Conditions on those being subsidized by training programs, such as formal bonding arrangements, should also be designed. Design an incentive program to promote staffing in hard-to-fill posts in rural settings through staff incentives (notably, improve accommodation in rural posts) and a financing mechanism (Le., the money should follow the health post not the staff). #2: The potentialfor public-privatepartnershipsshould be better exploited. The government's experience with establishing service agreements with C H A M hospitals should be evaluated, specifically on issues dealing with the predictability o f government funding, use/misuse o f funds, actual cost o f services, and conflict resolution. A more unifiedand comprehensive contract betweenthe government as financier and the C H A M facility (or other NGO) as service provider, covering staff salary subventions from MOF, provision o f drugs, and service agreements for other recurrent transactions (ORT), would help to ensure greater cohesion in the financing o f government and CHAM facilities. #3: Tertiary hospitals should be reformed to improve quality of care and cost recovery. The hospital autonomy bill, which envisions making tertiary hospitals legal corporate entities under the direction and control o f a hospital board, should be enacted and its provisions implemented. Cost recovery at tertiary care facilities should be improved through design o f a sliding scale fee structure that keeps up with current prices. The fee structure should account for the variety o f patients that tertiary hospitals have, and ensure 95 continued access to the poorest. A Poverty and Social Impact Analysis should be carried out to informthe new fee structure. #4: The procurement and deliverysystems for essential drugs must beimproved. CMS should undergo comprehensive institutional reform. Donors have supported GOM to convert CMS into a non-profit trust with an independent board o f trustees, but government i s yet to make a decision on the status o f CMS. Moving this agenda forwardrequires the highestlevelo fpolitical support. Procurement and drug tracking systems should be rationalized to improve drug stream and realize the cost saving benefits o f bulk procurement. Switching to a demand-driven (requisition) system embodying principles o f a "cash-and-carry" financing which define the Tanzanian and Zimbabwean reformed drug systems would achieve this. Drug registration and quality assurance and control mechanisms should be clarified and revised. Guidelines for essential drugs should be improved to decrease use o f inappropriate therapies #5: Healthexpenditures and outcomesshould be rationalizedand better monitored. 0 The level o f spending on administration and support services is one-third o f all recurrent expenditures; similarly the share o f the recurrent budget allocated to headquarters i s one-fifth. The reason for such high levels o f expenditures on administration and headquarters i s unclear, and suggests the existence o f significant inefficiencies The increasing flow o f resources to the health sector, much of it off-budget, necessitates better monitoring o f expenditures and outcomes. Development expenditures need to be comprehensively and consistently captured in the government budget. Consideration should also be given to a formal quantification o f the efficiency o f public expenditures using a nationwide sample o f health facilities. #6: Additionalfinancing needs to be secured urgentlyfor HIV/AIDS. The sustainability o f HIV/AIDS ART programs is in doubt. There is a need to secure medium-term financing to maintain current levels o f provision o f HIV/AIDS programs. The funding gap i s even larger when considering the GOM's target to scale up substantially provision inHIV/AIDSprograms. Inthis context, the government should apply for the new Rolling Continuity Channel o f the Global FundinNovember 2007. This could potentiallyprovide up to six years (3+3) o f further funding for the HIV and AIDS national response (including anti- retrovirals). The Global Fund staff have signaled that they expect Malawi to present a budget for scaling up the current response. #7: Community-basedhealth services should be expanded and strengthened Successful expansion o f community-based programs for essential health services (e.g., community-based `directly observed treatment, short course' DOTS for tuberculosis, family planning services relying on community-based distributors, 96 distribution o f bednets) will depend on the availability o f health commodities, so their success hinges critically on CMS reforms. 97 CHAPTER NUTRITION 5: 5.1Introduction 303. This public expenditure review looks at the composition and amount o f expenditures on interventions that are intended to reduce the level of malnutrition in Malawi. Malnutrition has been identified as an underlying cause in over 50 percent o f child deaths under age 5 in the developing world. Improving the nutritional status o f children under-5 years o f age i s therefore key to reaching the MDG #4 o f reducing by two thirds the under-5 mortality rate. Malnutrition in under-5s i s also a key indicator o f MDG #1 to eradicate extreme poverty and hunger. Further, malnutrition has a profound impact on the ability o f individuals to learn and work throughout their life. Thus, the nutritional well-being o f young children directly and indirectly contributes to the country's future development. 304. This chapter begins by looking at key issues pertaining to malnutrition. It then analyses expenditure on nutrition before proposing some key policy recommendations to effectively address the scourge o f malnutrition inMalawi. 5.2 Key NutritionIssues Undernutrition 305. One in every two children under age 5 are chronically malnourished in Malawi. Data from the 2004 Demographic Health Surveys indicates that 48 percent o f children under-5 i s stunted. Further, an exam o f trends from the three Demographic Health Surveys carried out in 1992, 2000 and 2004 shows that there has been little to no progress incombating malnutrition since 1992.'loStunting, the best indicator for chronic malnutrition, has hovered between 48-49 percent for the last 14 years for children under five years o f age (Table 5.1). Underweightand wasting have also improved little in the same time period. These data are broadly consistent with those from the National Micronutrient Survey for 2001, according to which 53 percent o f children under age 3 suffer from stunting,31percent are underweight, and 4.7 percent suffer from wasting. Table 5.1Childrenunder 5 with MalnutritioninMalawi, 1992-2004 Stunting Underweight Wasting (low height for age) (low weight for age) (low weight for height) 1992 48.7% 27.2% 5.4% 2000 49.0% 25.4% 5.5% 2004 48.0% 22.0% 5.0% I10Stunted children(low height for age) suffer from long-termor chronic malnutrition. This indicator identifiespast undernutrition, or chronic malnutrition, but misses short-termchanges inmalnutrition. Wasting (low weight for height) is an indicator o f acute or recentnutritional deficits and is usedfor identifying severelymalnourishedchildren. Wasting inindividual children andpopulationgroups can change rapidly and shows markedseasonalpatterns associatedwith changes infood availability or disease prevalence.Underweight(low weight-for-age) individuals weigh significantly less thanawell-nourished peer of a specific age. This indexreflects bothpast (chronic) and/or present(acute) undernutrition. 98 Source: Malawi Demographic Health Surveys 1992,2000 and 2005. 306. Stunting is pervasive in all regions and across all income groups, but is most severe in the Central Region. Both urban and rural households in all regions are afflicted alike, though rural stunting i s slightly higher, and the Central Region has the highestprevalence o fboth stuntingand severe stunting (GoM and World Bank, 2006). 307. 35 percent of all households do not obtain sufficient energy (measured by calories). This share i s marginally higher inrural areas (36 percent), and i s most acute in the Southern region (40 percent). Moreover, the magnitude o f the difference in caloric intake i s disconcerting: individuals in the lowest welfare quintiles consume about 40 percent o f the calories o f individuals in the top quintile (NSO, IHS2). Calories are not distributedproportionately within the household to individuals with the highest nutrition requirments. 308. Child malnutrition is not highly correlated with poverty (and with energy intake). Children in all income quintiles show similarly high incidence o f malnutrition, even though non-poor households receive substantially more calories. This i s most clearly seen in Central region, which has both the most calories per capita and the most underweight children. The Poverty and Vulnerability Assessment o f 2006 found that a mother's age and level o f education have positive and significant effects on reducing malnutrition (GoM and World Bank, 2006). Living in a female-headed household; mothers being incharge o f feeding practices, and access to improved sanitation all have a positive impact on the nutritional status o f older children. Finally, participation in a targeted nutrition program (for children under 5) seems effective in reducing child malnutrition. These findings underline the cross-sectoral nature o f the problem and the importance o f good coordination intackling it successhlly. Micronutrient malnutrition 309. As many as 59 percent of children between 6 and 36 months of age (preschool age) suffer from vitamin A deficiency and 80 percent have anemia (low hemoglobin). The 2001 National Micronutrient Survey (NMS) found that micronutrient malnutrition i s common among both younger and older children, and women (Table 5.2). A more recent school health and nutrition survey has shown that prevalence o f anemia among school age children i s 54 percent. Anemia i s also a considerable problem among pregnant women (with 47 percent anemic according to the 2004 Demographic and Health Survey). The 2001 NMS survey found that while vitamin A supplementation coverage was excellent for children 6-11 months (98 percent had received vitamin A in the previous six months) because they visit the health system for immunizations, coverage i s much lower for children 12-23 months (66 percent received vitamin A inthe previous six months) and for children 24-36 months (only 44 percent received vitamin A in the previous six months). Non-pregnant women and school age children both have highrates o f vitamin A deficiency, at 57 percent and 38 percent, respectively. There are no national programs to reduce vitamin A deficiency in either o f these groups until the recent launch o f the National School Health and Nutrition Program implemented by the Ministry of Education. Women are supposed to receive vitamin A within 8 weeks after delivery but, according to this survey, only 35 percent o f women reported receiving vitamin A during 99 their postpartum period. Iodine deficiency disorders are also a problem. I t is estimated that 64 percent o f children have low I Q inareas with highiodine disorders. 310. One of the factors contributing to pervasive micronutrient malnutrition and food insecurityis the over-relianceon maize in the Malawian diet. Malawi is a food insecure country for maize annually (duringthe lean season) and often duringlonger periods o f food shortages due to climatic crises. However, even in settings with recurring food shortages, effective nutritionprograms can reduce stunting inchildrenunder 5 years o f age. The per capita consumption o f maize in Malawi i s one o f the highest in the world, with 93 percent o f cereal consumption deriving from maize (GoM and World Bank, 2006). combat malnutrition. 1 Conse19uently, dietary diversification must be part o f the response to Table 5.2 Prevalenceof MicronutrientMalnutrition,2001 Vitamin A Deficiency Anemia Children 6-36 months 59% 80% School children 38% 22% Nonpregnant women 57% 27% M e n 38% 17% Source: National Micronutrient Survey (NMS) 2001 Lack of coordination across government ministries and donors 311. The cross-sectoralnature of the problemmakes nutritioninterventionsa challenge to coordinate and measure. Many ministries, donors and NGOs are involved in nutrition-related activities, which may be part o f health, agriculture or other sectoral programs. This has made coordination difficult, and has also made the collection and analysis o f information for this Chapter more difficult. The full table o f expenditure data compiled for this analysis i s presentedinAnnex 5.1. 5.3 Expenditureon nutrition'12 312. Though malnutritionin Malawiis very severe, nutritionhas not received increased funding or attention in recent years, and government spends less than 0.05 percent of GDP (USD0.9 million) on alleviating malnutrition in Malawi. Malnutrition diminishes future productivity, perpetuating people's vulnerability to poverty traps in the future and reduces the chance o f Malawi attaining the MDGs. Nevertheless, the government only spends annually approximately USD300,OOO on nutrition programs. An additional USD600,OOO are accounted for under the Health SWAP (which 001s together Ministry of Health budget and some donor hnding to the health sector)." For comparison, the government spends about USD1.5 million on the 'I'Inaddition, diet relatednon-communicable diseases such as obesity, hypertension, gout, arthritis, coronary heart diseases are on the increase. '12The analysis o f spending on nutrition i s mostly based o n expenditures in2004 or 2005 (or for government, the fiscal year 2004/05). However, M O H provided planned expenditures for 2005106 under the SWAp. Government (and the Health SWAp) also spends an estimated USD3 million per year on feeding patients admitted to public hospitals. Although this i s a necessary public expenditure, it i s not considered a direct nutrition intervention and i s excluded from the analysis in this chapter. 100 day-to-day feeding o f students in boarding facilities at higher education institutions (at the two public universities and in elite secondary boarding schools) than on alleviating malnutrition in Malawi. This i s in spite o f the fact that secondary and, in )?articular, tertiary students are known to belong to the highest socio-economic quintiles in Malawi. Failure to address this knding inequity will continue the cycle o f malnutrition, prohibiting the majority o f malnourished children from ever reaching secondary school. 313. Total annual expenditures on direct nutrition interventions were approximately USD25.6 million, or just above 1percent of GDP in 2004/05, and almost entirely funded by donors. Out o f the total nutrition expenditures, almost 96.5 percent i s accounted for by NGOs and donors, the largest contributors being the European Union and WFP. The balance o f 3.5 percent o f total nutrition expenditures i s the amount spentby government (of which, about 2.5 percent accounted for by the health SWAPand government accounted for another 1percent, as discussed above). Targetingof expenditures 314. There has been little attention given to programs that will prevent malnutritionat the household level. Most o f the attention and resources has gone into programs that are known to have little impact on nutrition. Also, there i s very little information available that can give an overview o f what type o f nutrition interventions are currently being implemented, and how well they are targeted to the most vulnerable populationgroups. 315. Micronutrient fortification and supplementation and community based nutritionprograms can be extremely cost effective, while feeding programs are least cost-effective. A 1996 World Bank study on the cost-effectiveness o f strategies to reduce under-5 child mortality found that highly cost-effective interventions include: nutrition education, iron fortification or iron supplementation o f pregnant women, iodine fortification, and vitamin A fortification or supplementation o f children under-5 (McGuire, 1996). Provision o f food supplements (feeding) or food subsidies are the least cost-effective interventions (Table 5.3). 101 Table 5.3 Returns on Nutrition Investments* Costper life saved Returns toprogram Costper discounted (1996 USD) dollar (in wages) healthy life year gained (1996 USD) Undernutrition interventions Food supplements 18,337 1.4 234 Nutrition education 797 32.3 10.2 Integrated PHC-N** 9,966 2.6 127 Food subsidies 42,552 0.9 375 School feeding 2.8 534 Iron deficiency Supplementation o f pregnant women 800 24.7 12.8 Fortification 2,000 84.1 4.4 Iodine deficiency Supplem. repro-aged women 1,250 13.8 18.9 Supplem. all people<60 4,650 6.0 37.0 Fortification 1,000 28 7.5 Vitamin A deficiency Supplementation child<5 130 50 4.0 Fortification 400 Source: McGuire, 1996. Notes: * While not listed here, growth monitoring andpromotionis also acost-effectiveinterventionbecauseit integrates and promotes meaningfulnutrition counselingand healthinterventionswhen growth faltering is detected.Early childhood development programs can also be good entry points to introducenutritionandother activities for young children, see Box 5.1. **Nutrition integratedinto Primary Health Care 316. The current portfolio of nutrition interventions in Malawi does not prioritize cost-effectiveinterventions. Table 5.3 provides a breakdowno f the nutrition expendituresby type o f intervention. Inspite o f the fact that supplementary food i s not a cost-effective strategy to combat malnutrition, more than one third o f all expenditures are on therapeutic and school feeding program~.''~This compares to expenditure o f just 1 percent on the highly cost-effective programs o f micronutrient supplementation (of vitamin A to children under 5 years o f age and iron to pregnant women) and fortification. Moreover, it i s not known what proportion o f the population i s actually being reachedby these interventions. The fortification o f salt with iodine i s mandatory in Malawi at the point o f entry (borders) and i s supported through the Health SWAP and various donors (monitoring o f its enforcement, and provision o f training and equipment). The cost to government and donors o f this program i s negligible, as the fortification costs are borne by consumers (included inthe price o fiodized salt). 317. Becausea large proportion of the population buys very little of their food, fortification programs may be very challenging in Malawi. Identification o f a proper vehicle to fortify may a daunting task. Currently, there i s a law that all salt that enters Malawi should be adequately iodized. The Inter-Ministerial Committee on Nutrition (the National Fortification Alliance i s a sub-committee o f this) i s currently looking into the possibility for micronutrient fortification o f sugar and/or cooking oil. One NGO, World Vision, ran a program for community-level fortification o f maize meal (which i s mostly Other benefits are derived from school feeding, namely, reduced rates of dropout, and improved learning. Nevertheless the cost o f such programs is very high. 102 milled locally in Malawi), and the other option, depending on the cost-effectiveness o f this program, could be to scale upthis maize mealfortification program. Table 5.4 Breakdownof Nutrition Expendituresby Type of Intervention Expenditures % distribution (USD millions) Undernutritioninterventions: Growth promotion 0.4 1% HIV/AIDS and nutrition 0.6 2% Lactation 0.1 0% Nutrition education 0.2 1% School feeding 3.7 14% Therapeutic feeding 5.1 22% Malnutrition interventions: Fortification 0% Interventions inhealth linked to nutrition 1.4 5% Supplementation o f micronutrients 0.2 1% Iron Iodine Vitamin A Other: Dietary diversificatiodFood security 12.8 50% Otheriadmin. 0.6 2% Total 25.6 100% Source: Author's calculations. 318. Interventions are not targetedat the most vulnerable under age 2. Young children, particularly children less than 2 years o f age, and pregnant and breast feeding women have the highest nutritional requirements o f any age group and are generally more vulnerable to malnutrition than the rest o f the population. In Malawi, no interventions specifically target the under-2s. However, it i s possible to analyze the allocation o f the nutrition interventions (typically the health sector interventions) by the following life-cycle target groups' 15:Highpriority groups include children under 5, and pregnant and postpartum mothers; children aged 5 to 18 are medium priority; and other adults are lowpriority. 319. About a third of nutrition expenditures target young children, pregnant women and mothers of young children, but half of all expenditures are not targeted to any group. Table 5.5 shows that 32 percent o f nutrition expenditures target children under age 5, and a further 2 percent target the other high priority groups of pregnant women and mothers o f young children. However, 52 percent o f expenditures are not targeted to any specific life-cycle target group. It i s possible that the funds that are not targeted to any specific life-cycle target benefit children more than adults, and young children more than older children, but there i s no information to ascertainthat1I6. '15N o t all expenditures can be assigned to a specific life-cycle target group, either because the intervention is o f a more administrative or general nature (e.g., research and policy development), too little information is available, or because it benefits everyone ina household or community. 'I6Inyears of severe food crisis money spent onthe procurement and distribution o f emergency food aid dwarfs direct nutrition interventions. This study does not include those emergency expenditures, or the recurrent expenditures o f maintaining the strategic grain reserve. Nutrition interventions such as gardening 103 Table 5.5 Breakdown of Nutrition Expenditures by Life-cycle Target Group Expenditures (million USD) % distribution <5 years exclusively 8.1 32% Pg and pp mothers, <5 years 0.4 2% 5-18 years 3.7 15% All ages 13.4 52% Source: Author's calculations. 320. A number of donors reported they were preparing to launch new programs, but no donors have reported major new programs targeting the under 5s. Donors reported new programs planned for HIV/AIDSand nutrition, GTZ i s starting up a large school feeding program, and the World Bank will work on School Health and Nutrition starting this year. Other donors, such as JICA, have expressed interest in supporting school feeding. While these are all important initiatives, care should be taken not to move the attention and resources away from the highest priority groups: infants, young children and pregnant/post partum women. All development programs should include a component that helps address the problem o f malnutrition. For example, providing messages about adequate infant and young child feeding practices should be channeled through school children and parent teacher associations. 321. Some degree of socio-economic targeting is incorporated in practically all donor and NGO programs, and in some government sponsored programs. Most donors report using the annual Vulnerability Assessment for Malawi, to identify the communities that are worst hit by food shortages. Other interventions, such as the therapeutic feeding o f severely malnourished children (sponsored by government and multiple donors and NGOs), select their beneficiaries basedon objective criteria (such as low weight for height) that can be expected to be closely associated with level o f poverty and or food insecurity o f the household. l7 322. When funding to `feeding programs' and other curative interventions are subtracted, the resources spent to prevent malnutrition from occurring are even more limited. Almost half o f nutrition expenditures(44 percent) inMalawi are curative innature, and can thus not expect to have any lasting effects on feeding ractices. Table 5.6 provides a breakdown o f total nutrition expenditures by preventiveE8and curative interventions. and livestock projects, supplementary and therapeutic feeding o f malnourished infants, are included, however. *I7There i s not much information available on the socio-economic targeting o f other government sponsored programs. However, a very fine socio-economic targeting is not always possible and may be impractical in a country like Malawi, where poverty affects the majority o f the population. `I8Preventative programs include the distribution o f micronutrient supplements though they are both curative (for people who have clinical signs o f deficiency) and preventive. Typically Vitamin A is promoted to all children under 5 to reduce under-five mortality (preventive), while iron i s distributed only to pregnant women who have anemia (curative). 104 Table 5.6 BreakdownofExpenditureby Preventiveand CurativeInterventions Expenditures % distribution (million USD) Preventive interventions: dietary diversificationlfood security, 13.6 53% growth promotion, lactation, nutrition education, micronutrient supplements Curative interventions: interventions in health linked to nutrition, 11.3 44% H I V i A I D S and nutrition, school feeding, therapeutic feeding Adminiother spending 0.6 2% Box 5.1 HondurasCommunity-Based IntegratedChild CareProgram The Community-Based Integrated Child Care Program (Atenci6n Integral a la Niiiez-Comunitaria, AIN- C) in Honduras is widely regarded as a model program in preventive health and nutrition care. AIN-C relies on volunteers to pro-actively engage both the families and communities to monitor and maintain the adequate growth o f children less than 2 years o f age. AIN-C also treats and refers sick children under 5 to health services. For the under two's, the Program employs inadequate monthly growth as a triggering device for applying a diagnostic decision-tree analysis to identify the causes o f inadequate weight gain, and combines it with formative research-based protocols that address the causes o f the problem, rather than simply treating its short term symptoms. The volunteers use a simple, uniform, highly structured but personally relevant counseling approach with families, while helping their communities and municipalities analyze and act against the causes o f poor child growth that are beyond a family's ability to improve. The critical local level implementing unit o f the Program consists of a team o f about three community volunteers responsible for about 25 children. The Program has been implemented in roughly 1,800 communities. A 2000 evaluation found that the program was reaching 92 percent o f children under two and was effective in improving mothers' child-rearing knowledge, attitudes and practices, including feeding practices and appropriate care-giving and care-seeking practices for children with diarrhea and acute respiratory illness. The program i s already being replicated in several countries around the world. The long-term, annual, recurrent cost o f a child under two that participates in the program is USD6.82. This cost also includes some curative care services provided to children between two and four in the same communities. However, not all these costs are additional, as the Ministryo f Health was carrying out some o f these services before, and because the better nutrition and preventive care may reduce the need for more costly health services such as hospitalization. The Ministry o f Health estimates that the annual incremental cost o f the program i s about USD4 per child under two. The preventive care component makes up 78% o f the cost o f the program, and curative care the remaining 22%, so while the program integrates preventive and curative care, it emphasizes preventive care. For additional informationsee: Fiedler, John L.,2003, A CostAnu1,vsis of the Honduras Community-BasedIntegrated Health Cure Program. Health,Nutrition andPopulation(HNP) DiscussionPaper.The World Bank. 5.4 RecommendedPolicy Prioritiesfor Nutrition 323. A few recommendations follow from the analysis carried out in this chapter. The full recommendations are summarized below, but a shorter and prioritized list o f recommendations i s presented inthe executive summary. #1:Reduce considerably the share of nutrition expenditure invested in programs that are costly and not expected to have a lasting impact in place of more cost effective strategies Feeding programs that are untargeted or targeted to non-priority groups, such as adult food distribution or school feeding, have been shown not to be cost effective 105 interms of nutrition and should be scaled down (infavor ofprograms targeted to under-ss,andparticularly under-2s -see below). Nutrition education programs are needed in order to improve the distribution o f food among household members and give caregivers and other family members information about good practices for feeding young children (generally, people are not aware how many times per day an infant should be fed, etc.). Micronutrient supplementation i s highly cost-effective and should be boosted. It may not be sufficient to deliver the micronutrient supplementation to young children through the primary health care system, as not all children under 5 come into regular contact with clinics and other health care providers. Moreover, any national fortification program must take into account the fact that many Malawians buy only a limitednumber o f food stuffs. 0 Diet diversification must be part o f the response to combat malnutrition. This needs to be addressed through a multisectoral approach that addresses both the production and marketing o f crops that are rich in micronutrients and small livestock, and the dissemination o f knowledge about the benefits o f consuming new foods andhow to prepare them. #2: Refocus programs to combat malnutrition towards improving the nutritional status of the under-5s. Develop and roll out a national community nutrition programto target the priority age groups in a low cost and effective manner. Such a program, modeled along the lines o f the Honduras Community-Based Integrated Child Care Program (Box 5.1) could channel more resources into the prevention o f malnutrition among under-5s (and in particular under-2s), and would be possible to sustain due to its relatively low cost. There are approximately 900,000 children under the age o f two in Malawi, so such a program could be implementedin Malawi for about USD6-7 million per year. This would not all be additional costs, however, as it would replace some existing services. #3: Improved coordination among government and donors, and across sectors i s neededto tackle the dauntingchallenge of malnutritioninMalawi. The recent appointment o f a Principal Secretary for Nutrition, HIV and AIDS, directly under the Office o f the President and Cabinet, has yet to produce new programs and more government funding to the sector, but could undertake this role. Inaddition, in order to implement the nutrition policy and national nutrition programs effectively, there i s a need to build capacity at all levels. #4:. A research agenda on the determinants of malnutrition in Malawi needs to be pursuedin order to design evidence-based policiesandprograms. Evidence from the 2005 IHS2 highlights that child malnutrition i s not highly correlated with poverty levels (and with caloric intake), suggesting that other factors need to be tackled to eradicate it. However, there i s a need to deepen the understanding o f non-income factors, and what policies can be introduced to mitigate them. 106 CHAPTER ROADS 6: 6.1 Introduction 324. The Ministry o f Transport and Public Works (MOTPW) determines overall transport policy in Malawi. The overall policy, strategy and,investment plans for the sector have recently been updated and are presented in the government's `Road Sector Program' of April 2007. Responsibility for the management o f the network was handed over to the National Roads Authority (NRA) in 1998. In 2006 new legislation was approved to improve the institutional framework for the sector, and the National Roads Authority has been separated into two distinct bodies: the Roads Authority (RA) i s responsible for the construction, maintenance and rehabilitationo f public roads, while the Roads Fund Administration (RFA) is responsible for raising, administering, and accounting for funds for construction, maintenance and rehabilitation o f public roads. The new set-up has the merit to separate road financing and road management functions o f the sector. The Ministry retains responsibility in all areas o f road safety and axle load control. Enforcement i s done in conjunction with the Malawi Road Traffic Police. The overall res onsibility for the management o f the road network lies with the Roads Authority.'" However, there are other actors in the sector including the Ministry o f Local Government and Rural Development (MOLGRD) and the National Construction IndustryCouncil. 325. The primary source o f revenue for roadmaintenance activities comes from the Road Fund; financing for development projects comes from the donors and the government. The Road Fund was set up primarily for maintenance activities on the public road network. It i s hnded mainly by a fuel levy on diesel and petrol, which accounts for more than 90 percent, and road transit fees and overloading fines. The fuel levy charges per liter o f fuel are currently at MK6.70 for diesel and MK8.75 for petrol, corresponding to around 4.9 centdliter and 6.3 centdliter respectively in 2006. Until 2006 the Road Fundused to be managed directly by NRA; it i s now the responsibility o f the RoadFundAdministration. 326. The public road network, comprising the main, secondary, district and urban roads under the management and responsibility o f RA, comprises 17,910 kilometers o f roads (Table 6.1). It is estimated that the entire network of roads, including other community roads under responsibility o f District Assemblies and private roads i s about 25,000 kilometers.12' Road transport carries over 70 percent o f the internal freight traffic ` I 9The RA has a 10 member Board o f Directors and i s responsible for the main, secondary, district and urban roads, while community roads are the responsibility o f the District and City Assemblies which are placed under MOLGRD. I 2 OThese numbers are from the road reclassification study which has recently been completed in 2006, and i s awaiting formal approval. An additional 9,000 km o f roads previously `undesignated' have been proposed for inclusion as part o f the `designated' network, i.e. the network o f roads that are officially gazetted and for which responsibility is assigned to a particular agency for maintenance. About 63 percent o fthe `designated' network is considered maintainable, i.e. ingood or fair condition. Rural roads (defined as district and community roads) comprise about 60 percent o f the network. 107 and 99 percent o f passenger traffic in Malawi. Road transport also dominates the other modes o f transport interms o f employment creation. Table 6.1: Public Road Network in 2006 Road Class Km Y O M a i n 4,006 16.1 Secondary 4,249 17.0 District 8,095 32.5 Community 7,019 28.2 Urban 1,560 6.3 Total 24,929 100.0 Source: Road Sector Program 2007 6.2 Key Road Sector Issues 327. The road network is heavily imbalanced: a small network of main roads mostly paved and in good condition, and the rest of the network unpaved and mostly inpoor condition. The paved network i s almost entirely ingood or fair condition (98 percent). However, the greater part of the network i s unpaved (74 percent), and about 40 percent o f the unpaved roads remain in poor condition (Table 6.2). In fact, the majority o f the roads in poor condition are the rural roads. Overall, about 40 percent o f the total network is still inpoor condition, and would require large amounts o f money to bring it to maintainable condition.'" More attention and finances will need to be concentrated on the unpaved roads network inorder to bring it to an acceptable standard. 328. The condition of the network has not improved. The condition o f the road network has remained stable over the last 5 years (Table 6.2). The condition o f the network, whilst not deteriorating, has not improved, however. A properly functioning network will need to have at least 70 percent o f its roads in good to fair condition. This can only be achieved with a greater injection o f funds into appropriate rehabilitation and maintenance. Table 6.2: Road Network Condition 200o/o1 2002103 2003/04 2004/05 2005/06 G F P G F P G F P G F P G F P Paved 71 24 5 I O 24 6 12 22 6 12 22 6 19 19 2 Unpaved 12 41 47 13 40 47 13 39 48 13 39 48 14 46 40 Note: G: Good condition; F: Fair condition; P: Poor condition. Source: NRA,various years, and Road Sector Plan 2007 329. Government contributions to the Road Sector continue to fall short of recommended levels. Recommendations from the 2001 PER suggested that domestic contribution to road sector expenditures should be increased, especially towards maintenance. With the exception o f the Road Fund, however, allocations o f Government funding for road construction and maintenance have been uncertain and irregular. Road Fund revenues have increased, but planned targets were only partially met due to inconsistencies inthe adjustment o fthe fuel levy (Table 6.3). 12'Only roads ingood or fair condition are considered maintainable. 108 330. Funding for the maintenance of the network continues to be insufficient, although there has been some improvement recently. The official policy in the road sector i s to first maintain the existing road network followed by rehabilitation, upgrading and construction o f new roads. Since its inception, however, the NRA has managed to maintain at best up to 30 percent o f the total road network annually with revenues from fuel levy. Part o f the problem, i s that the Road Fundrevenues, which fund maintenance, are limited in scope. Inaddition, as discussed below, there continues to be a bias towards financing rehabilitation, upgrading and construction o f new roads at the expense o f systematic under funding o f maintenance. Table 6.3: Revenuesfrom the RoadFund, 1998/99to 2003/04 Exchange Projected Revenue Rate Revenue Revenue Surplus/(Deficit) Year MK million MK: US$l.OO US$ million US$ million US$ million 1998/99 302.00 45.00 6.71 6.71 1999/00 375.00 75.00 5.00 9.50 (4.50) 2000/01 601.OO 67.00 8.97 12.00 (3.03) 2001/02 1,I13.00 75.00 14.84 12.90 1.94 2002103 1,217.00 80.00 15.21 13.80 1.41 2003/04 1,285.00 90.00 14.28 15.30 (1.02) 2004105 1,461.OO 110.00 13.28 15.25 (1.97) Source: NRA 331. There is inadequate planningand prioritizationin roadmaintenance and construction activities. Decisions on road maintenance and construction have been made on an ad hoc basis, through selecting from a list o f activities and projects compiled by the staff of NRA, and no consistent criteria or method o f analysis, and controls for implementation, have been used. Project selection and planning are not systematic, with frequent deviations and additions, often from higher levels of government and/or political interference. Also, the choices are limited by tied funding from government and donor sources. In 2006, the NRA created a Road Data Management Unit with capacity for strategic analysis and investment planning using dedicated software (the `Road Data Manager'), to base decisions on road maintenance and construction on thorough project analysis and appraisal. The unit i s now operational and able to calculate and compare Internal Rates o f Return across roads in various geographical areas, and various types o f expenditures (such as maintenance, rehabilitation, upgrading, or new construction). The intention i s that the analysis and prioritization from the Road Data Manager should feed into the business and strategic plan for sector. Hence, it i s critical that the Road Data Manager results become a core part o f the `procedures agreement' which i s envisaged by the new legislation as the instrumentwhich will enable the government to assess whether funds accruing to the RA are beingused efficiently. 332. Monitoringand evaluationis virtually inexistent. Monitoring and analysis o f sector expendituresare done irregularly and inconsistently. There i s a needto improve the monitoring and analytical capacity o f the MOTPW, MOLGRD and RA in areas o f planning, programming, and effective monitoring. Sector performance has been compromised with little or no allocation o f funding for the monitoring and evaluation o f all maintenance, rehabilitation and construction projects at either Ministryor NRA level. 109 Efforts to improve monitoring and evaluation o f sector performance has been made in recent years and a set o f possible indicators have been recently identified, which need to be operationalized. A complete road inventory and condition survey has been completed in2006. 333. Important institutional reforms have been adopted. As discussed above, in order to improve management and governance of NRA, the government drafted legislation to separate the hnctions o f financing (Road Fund) and management (Operations) within the National Roads Authority, and the new Act has been approved by Parliament in 2006. It i s expected that the separation o f the functions will allow greater efficiency in the sector, and enhance objectivity in resource allocation. By introducing performance agreements and instituting technical audit procedures, it can shield the implementers from political influence with respect to road allocation decisions. It is essential that this reform goes alongside a more effective decentralization o f local road planning management responsibilities at the district level. 334. Local authorities have very limited capacity inthe management of district roads. Very little policy, planning and management capacity exists at the district level. These authorities are, however, best placed to assess local needs, and need to be strengthened inthe areas o f planning and management o f roads under their jurisdiction. The Government Malawi Rural Transport and Travel Program, in conjunction with National Construction Industry Council, and donors are training technical personnel from District Assemblies to handle basic civil works contracts and better understand management o f roads. 335. Malawi is still one of the worst countries inthe world with respect to road safety. Data from the SADC National Road Safety Council shows that Malawi records an estimated 106 fatalities per 10,000 vehicles, compared to 60 per 10,000 in Tanzania and 33 per 10,000 in Kenya. The cost to the economy has been estimated at 3.2 billion Kwacha, equivalent to about 2 percent o f GDP in 2004.122Road safety audits are not conducted. Such audits need to be instituted post design for newly constructed roads or roads under construction, and at the design stage for new projects. The establishment o f the road safety database at the National Road Safety Commission will aid inidentifying accident black spots for appropriate remedial action. The Government o f Malawi also intends to commission the development o f a national road safety master plan to address all areas relating to road safety education, engineeringand enforcement. The government has also announced the intention to harmonize the operation o f the National Road Safety Council, the Road Traffic Directorate, and the Traffic Police through the creation o f a Roadtraffic Authority. 336. Malawi lacks a well developed local construction industry. Whilst the national construction industry has a considerable number o f local contractors and consultants registered, the level o f participation and performance needs improvement, through capacity building, training and access to financial resources for equipment purchase or hire. The NRA 5 year strategic plan estimates that about USDSOO million will needto be spent over the next five years to bringthe network up to a standard where 122Based on preliminary estimates o f a combination o f medical costs, lost output and property damage costs. 110 70 percent o f the network i s in good condition. This requires about 11,000 kilometers o f roads to be rehabilitated. The local contracting capacity will need to be strengthened and opportunities for their participation inthe works needto be created. 6.3 Expenditures Trends 337. Total road sector expenditure declined over the last five years as a percentage of GDP, and maintenance expenditures have declined even further. Total expenditure in the roads sector has decreased steadily from 3.3 percent o f GDP in 1999/00 to an estimated 2.2 percent o f GDP in 2004/05, but have been increased substantially in 2005/06 to 3.6 percent o f GDP (Table 7.4 and Table 7.5). The share devoted to maintenance expenditures has decreased steadily over the last five years from over 55 percent o f total road expenditures in 2000/01 and 2001/02, down to around 33 percent in 2003/04 and 2004/05, and only around 17 percent in 2005/06. The balance o f expenditures on roads has been taken up by upgrading & construction, and rehabilitation o f new roads which have increased substantially, both as a share o f roads expenditure and as a percentage o f GDP. 338. Funds spent on maintenance, have been inadequate to meet the maintenanceneeds of the network,and to clear the sizeable backlog of maintenance. It is a well accepted principle that the priority in the road sector should be adequately maintaining the existing network, rather than new construction. Recommendations from the 2001 PERrequired that maintenance o fthe `core network'123 be fully fundedwithin 4 years, and that donor funds be used to clear the backlog maintenance. However, neither o f these recommendations has been implemented. More recently, the 2007 `Road Sector Program' raises the issue o f the funding gap for maintenance and presents options for closing it. 339. Starting in fiscal year 2006/07, a new system to carry out routine maintenance has been introduced. The 2005 NRA Five Year Strategic and Business Plan estimates that an amount o f USD6.6 million would be needed annually to cover routine maintenance on the entire network. In principle, therefore, the cost o f routine maintenance can be fully covered by the revenue inflows into the Road Fund (since inflows into the Road Fund regularly exceed USD14 million, see Table 7.3). Unfortunately, however, actual expenditure on routine maintenance over the past few years has still fallen short o f budgeted amount, as a significant share o f the Road Fund finances have been diverted to rehabilitation and construction o f new roads. In2004/05, about 4500 kilometers o f routine maintenance was undertaken, representing coverage o f only about 50 percent o f the core public road network and about 25 percent o f the total public roads network. It i s critical that the cost o f routine maintenance be fully met in future. Routine maintenance contracts have been awarded to two companies in each district. Hence, it could be argued that as o f 2007, the whole network i s being maintained. The maintenance interventions carried out by the contractors have to be approved on an ongoing basis, but the total value i s limited to the amount o f the contracts. While these arrangements appear to constitute an improvement from previous 123The `core network' i s definedas that part of the publicroadnetwork carrying more than 30 vpd, estimated at around 8000 kilometers. 111 years, there will be a need to increase the amount o f the contracts, however, as in2006107 the available resources were usedup by the middle o f the fiscal year. Table 6.4 Expenditure allocations to maintenance, rehabilitation, and upgradinglconstruction,1999100-2005106 1999100 2000101 2001102 2002103 2003104 2004105 2005/06* MK millions Routine Maintenance 394 695 1368 1653 905 1604 1300 Periodic Maintenance 336 947 963 494 725 112 308 Rehabilitation 353 155 742 232 118 120 1966 Upgrading /Construction 1967 799 1277 1940 2157 3255 6105 T O T A L 3050 2596 4350 4319 3905 5091 9679 USD millions Routine Maintenance 7.6 10.6 19.6 19.4 8.5 14.4 10.1 Periodic Maintenance 6.5 14.4 13.8 5.8 6.8 1.0 2.4 Rehabilitation 6.8 2.4 10.6 2.7 1.1 1.1 15.3 Upgrading /Construction 38.0 12.1 18.3 22.8 20.2 29.3 47.5 T O T A L 58.9 39.4 62.2 50.7 36.6 45.8 75.3 As % of GDP Routine Maintenance 0.4 0.6 1.0 1.o 0.5 0.7 0.5 Periodic Maintenance 0.4 0.8 0.7 0.3 0.4 0.0 0.1 Rehabilitation 0.4 0.1 0.5 0.1 0.1 0.1 0.7 Upgrading and Construction 2.1 0.7 0.9 1.2 1.1 1.4 2.3 T O T A L 3.3 2.3 3.2 2.7 2.1 2.2 3.6 As % of TotalRoads Expenditures Routine Maintenance 13 27 31 38 23 32 13 Periodic Maintenance 11 36 22 11 19 2 3 Rehabilitation 12 6 17 5 3 2 20 Upgradingand Construction 64 31 29 45 55 64 63 T O T A L 100 100 100 100 100 100 100 As % of Total GoM Recurrent Expenditures Routine Maintenance 1.4 1.9 3.2 2.7 1.1 1.7 1.1 Periodic Maintenance 1.2 2.5 2.3 0.8 0.9 0.1 0.3 Rehabilitation 1.3 0.4 1.7 0.4 0.1 0.1 1.6 Upgrading and Construction 7.2 2.1 3.0 3.2 2.7 3.4 5.1 T O T A L 11.2 7.0 10.2 7.1 4.8 5.3 8.1 Total Governmentrecurrent expenditures 27220.8 37266.4 42490.3 61260.3 80536.3 96625.0 119736.0 GDP (MK millions) 91,809 113,871 136,141 160,137 189,563 226,272 270,989 Exchangerate (Malawi Kwacha/US$) 51.82 65.87 69.9 85.2 106.6 111.1 128.6 Source: NRA Note: (*) 2005106 are estimates 340. The full estimatedmaintenance and rehabilitationcosts for the networkis around USD5O million/year,which is much higher than can be providedfor by the RoadFund. The cost o fmaintaining andrehabilitating the unpavednetwork to its target condition are estimated at USD20 million per year, while the costs related to the paved network are an additional USD23 million. Inaddition, the operational costs for the Road Authority and Roads Fund Administration are estimated at USD5 million per year (including operating expenses, capital expenditures, planning activities, design, studies, and surveys). Hence the total annual costs o f maintain the existing network i s in the region o f USDSOmillion. 112 341. There is a need to revise upwards the level of the fuel levy, or to identify alternative sources of funding for road maintenance. Since the fuel levy was established in 1998 there have been 3 increases inits level inMalawi Kwacha. However, since construction costs are have been roughly constant in dollar terms, the value o f the fuel levy inreal terms i s better representedby its value indollar terms. Infact, since 2001 the value o f the levy in dollar terms has decreased by about 40 percent. A quick calculation shows that inorder to cover all maintenance costs the fuel levy would need to increase from the current MK6.70 / liter o f diesel and MK8.75 / liter o f petrol to a level o f MK20.0 / liter o f diesel and MK26.0 / liter o f petrol (equivalent to USD 0.14 per liter of diesel and USD0.19 per liter o f petrol). Ifincreases o f this order are unacceptable the fuel levy should be supplemented with direct funding through the government budget. Table 6.5: Expenditure allocationsto maintenance, rehabilitation, and upgrading/construction, by funding source, 1999/00-2004/05 (MKmillion) 1999100 2000/01 2001102 2002103 2003104 2004105 RoadFund MK mil YO Routine maintenance 394 695 1,368 1,653 905 1,243 24.4 Periodic maintenance 45 1 361 7.1 Sub Total - 394 695 1,368 1,653 1,356 1,604 31.5 Developmentpartners Periodic maintenance 309 854 906 324 274 110 2.2 Rehabilitation 353 155 742 232 118 70 1.4 Upgrading & Construction 1,711 602 1,116 1,502 1,430 2,509 49.3 Sub Total - 2,373 1,611 2,764 2,058 1,822 2,689 52.8 GoM Periodic maintenance 27 93 57 170 2 0.0 Rehabilitation 50 1.0 Upgrading & Construction 256 197 161 438 727 746 14.7 Sub - Total 283 290 218 608 727 798 15.7 Total 3,050 2,596 4,350 4,319 3,905 5,091 100 Source: NRA 342. The backlog of periodic maintenance has not reduced and is estimated at USD106 million as of 2005. A considerable amount o f donor funds have gone into funding backlog maintenance over the five year period from 2000/01, but due to the insufficiency o f the funding, there i s still a large uncleared backlog. The NRA Five Year Strategic and Business Plan (for the period 2005/06 to 2009/10) estimates it would require around USD106 million to clear the backlog o f maintenance over the 5 year period. The imbalance between expenditure on maintenance and rehabilitation/ construction i s increasing. Duringthe last three years, on average over 60 percent o f total road sector expenditures have been allocated to new road construction and rehabilitation (Tables 6.4 and 65). Routine maintenance expenditures have accounted on average for 23 percent o f road sector expenditures during the past three years, while periodic maintenance for 8 percent, and only around 3 percent in the last two years. Hence, rehabilitation and construction account for three-to-four times the amount o f money spent on maintenance operations. Further, this imbalance has been getting worse during the past years, as expenditure on new construction works and rehabilitation continued to increase with funding from donors and government. On the other hand, the amounts 113 allocated to routine maintenance have been decreasing over time. In fact, following an increase to about 1 percent o f GDP in 2001/02 and 2002/03, expenditure on routine maintenance has decreased to around 0.5 percent o f GDP inthe last three years. 343. Donors contribute large amounts to all activities except routine maintenance. Funding for the road network comes from three main sources: the government budget, the donors and the Road Fund. Around 32 percent o f the total expenditures came from the Road Fund, 16 percent from the government budget. The Government budget allocation i s used mainly for paying salaries for the staff o f MOTPW, and for undertaking some development projects. The Road Fund i s used to cover routine maintenance. Development partners account for the balance o f 52 percent o f sector expenditures (Table 6.5). Donors have concentrated on development programs inthe Road Sector, i.e. rehabilitation, upgrading and construction of new roads. Donors account for about 77 percent o f expenditures on rehabilitation and upgrading/construction o f roads, and government accounts for the balance 23 percent. However, due to the outstanding backlog maintenance the donors have obliged by financing periodic maintenance activities. Donors have provided a portion o f the fimding for rehabilitation (59 percent) and periodic maintenance (23 percent). All funds for routine maintenance are internally generated and funded from the Road Fund. The Road Fund accouits for almost all routine maintenance and some expenditure on periodic maintenance. 344. Physical measurements of the amount of works carried out indicate a disconnect between actual outputs and plannedtargets. On average, between 1999/00 and 2004/05 only about 70 percent o f the maintenance budgeted was actually carried out and only about 50 percent o f the rehabilitation and upgrading/construction (Table 7.7). The data show a market improvement in 2004/05, both an increase in the number o f kilometers and improvedperformance with respect to targets. Table 6.6 PhysicalOutputs 1999/00 to 2004/05 in Kilometers Year Routine Periodic Rehabilitation Upgrading Maintenance Maintenance /Construction Planned Achieved Planned Achieved Planned Achieved Planned Achieved 1999/00 4500 3560 340 220 280 150 130 79 2000101 5200 2580 450 230 250 90 140 70 2001/02 7200 5832 596 250 200 105 136 23 2002103 2300 1231 180 165 62 47 90 60 2003104 4245 4200 115 96 285 112 82 14 2004105 4251 4599 70 66 15 10 188 142 Source: NRA 6.4 Poverty Focus and Incidence of Expendituresin the Road Sector 345. Access to roads heavily skewed in favor of the richest households. The poor tend to live inremote areas with limitedroads and/or means o f transportation, which constrains their access to markets and trading centers, and limits their economic opportunities. Rural communities on average are located 20 k m s from a tarmac road, and this distance is higher at about 40 kms on average inthe North region (GoM and World Bank, 2006). Further, roads inrural areas are often impassable, on average for up to four months in the year. Sizeable portions o f the population remain isolated from the rest o f 114 the country for substantial amounts o f time. In the North region, on average roads are passable by minibusfor only 5 months inthe whole year. Inthe South region, on average they are passable for less than 8 months. This emphasizes the need to be addressed to increase the conditioninthe overall network and ensure rural accessibility and mobility. 346. There has not been a poverty focus in the road sector expenditure. As discussed, more funds are being spent on rehabilitation, upgrading and construction o f new roads compared to periodic maintenance. There i s thus an imbalance between funding for the paved and unpaved roads. More expenditure is focused on paved road in order to preserve the hi h value o f the asset much to the apparent detriment o f the unpaved road network.12' Limited funds are allocated for feeder road construction and maintenance (under the `Pro-Poor Expenditure'allocations). However the amount varied from year to year with no clear rationale for the amount allocated. On the other hand, 60 to 80 percent o f funds from the MASAF public works program are channeled into roads projects, indicative o f the demand at district level. 6.5 Implementingkey recommendations from the 2001 PER on roads 347. Good progress has been achieved in implementing some of the reforms; but some of the recommendations remain to be implemented. Below we briefly review progress made in each o f the four main recommendations from the 2001 PER, in light ofthe analysis presentedinthis chapter: #I:Increase outlaysfor road sector through a combination of a higherfuel levy, other user charges and through HIPC resources. Little progress has been made inincreasing resources for roads use from fuel levy and other user charges. #2: Amend RoadAct topermit concessioning toprivate sector. The Public Roads act has not been revised, and little progress has been made in increasingprivate sector involvement. The National Roads Authority act has recently been replaced to increased independence o f the RoadAuthority. #3: Fullyfund routine and backlog maintenance Limited progress had been made in fully funding roads maintenance. The new system for funding routine maintenance has improvedthe situation as o f 2006, but the amounts allocated remain inadequate. Further, there is still substantial under fundingo fperiodic maintenance, infavor o f construction o f new roads. #4: IntegrateDistrict Assemblies, M S A F andNRA roadprograms in order to rationally plan the rural roadsprogram Progress has been made in developing an organic plan for the sector, with the adoption o f the Roads Sector Plan in2007. Iz4A Cabinet directive from 1998 set the level of expenditures at 65 percent for maintenance of main, secondary and tertiary roads, and 35 percent for maintenance of district and urban roads. 115 348. In sum, key recommendations from the 2001 PER are still valid today, and should be implemented. Specifically, as discussed above, there continues to be a needto increase the fuel levy to a level adequate to fully fund maintenance activities, and to prioritize funding for all road maintenance. 6.6 RecommendedPolicyPrioritiesfor the Road Sector 349. Six recommendations arise from the analysis carried out in this chapter. The full recommendations are summarized below, but a shorter and prioritized list of recommendations i s presented inthe executive summary. #1:Allow for regular review for adjustments in fuel levy to increase funding in RoadFund. The fuel levy i s the most consistent source for maintenance funding. InMalawi the equivalent o f 16.5 US$per liter (and average o f 14US$ for diesel and 19 US$ for petrol) i s needed to cover (routine and periodic) maintenance o f the full network. Such a large increase in the fuel levy should be introduced gradually over a few years. As an intermediate step, a review o f the fuel levy up to the equivalent o f at least 10 US$per liter will ensure adequate funding for routine and periodic maintenance for at least 70 percent o f the entire network. 0 Inaddition, the government should carry out a poverty and social impact analysis o f such a large increase inthe fuel levy, and to assess whether alternative options to raise the required funding for the maintenance o f the road network might be preferable. I t i s also important that funds levied from the road users through the fuel levy are utilized for the sole purpose o f maintainingthe roads. Further,there is a needto operationalize the annual formal review mechanisms of the fuel levy, inaccordance with the 2006 RoadFundAct. #2: Funding for routine maintenance must be adequate and consistent, and if domestic resources are not adequate donor funding should also be channeled towards maintenance. 0 Fully fund routine maintenance o f at least the core network and carry out systematic periodic and backlog maintenance on the entire network, to bring the whole network to maintainable condition. If there is not enough funding available, decisions will need to be taken as to what size o f network can realistically be managed based on available resources, and on the amount of rehabilitation to be undertaken. The sustainability o f the sector i s in question when what is beingrehabilitated cannot or is not maintained. 0 If that internally generated funds for the sector are insufficient, it may be necessary to prioritize any external funding towards covering backlog and periodic maintenance before reconstruction. Currently most donor hnding i s negotiated bilaterally, with emphasis on particular roads rather than the improvement o f the network as a whole. 116 0 A sector wide approach will create opportunities for donors to have a global picture o f the sector and focus funding on priority programs; this i s now being planned through the Road Sector Program approved inearly 2007. #3: Adopt a system to ensure planning and prioritization in road maintenance and construction activities. The introduction o f the Road Data Manager system (by the Road Data Management Unit at RA) should assist in prioritization and allocation o f funds more efficiently, by basing decisions on road maintenance and construction on thorough project analysis and appraisal. This i s especially important as government i s planning to expand expenditures on roads. It is important that the evaluation based on the Road Data Manager (produced by the Road Data Management Unit) become a core part of the `procedures agreement' envisaged inthe 2006 RoadAuthority Act. #4: M o r e attention and finances needto be spent on the rural roads. A more equitable balance o f finance needs to be established between rural and main roads. MOTPW and the RA need to work with MLGRD to identify an appropriate mechanism o f sourcing and allocating funds for road maintenance at district level. The formula used to channeling maintenance resources i s biased in favor o f the paved network, and needs to be revised. #5: Better road sector data i s are essentialfor setting policy and for monitoring and evaluation. Analysis o f the road sector expenditure was difficult because neither the Ministry nor NRA keeps adequate and consistent data on expenditure on roads. The data available from the Treasury is not reflected in the financial statements available from NRA. The two sources of data were not consistent and road sector expenditure incurred by MLGRD and MASAF were not reflected at either source. MOTPW and NRA should establish a unified system for reporting sector expenditures. Data on outputs (e.g. kilometers o f roads covered by the various activities) i s not systematically recorded, making it impossible to accurately calculate unit costs for maintenance, rehabilitationand construction. Build on recent efforts to improve monitoring and evaluation o f sector performance by operationalizing the indicators which have recently been identified. Complete the recently carried out road reclassification study, by preparing the relevant legislation and submittingit to Parliament. #6: Increase the capacity of domestic construction industry to meet demands Consider expansion o f duration and coverage o f the routine maintenance contracts (started in 2006), in order to provide firms a more stable income stream which would allow increase ininvestments and capacity. Maintain preferential treatment inpublic works tender for foreign companies that go into ajoint venture with a local company, to facilitate know-how transfer. 117 ANNEXES 118 ANNEX1.2 NATIONAL DROUGHT INSURANCEFOR MALAWI'*' 350. Inrecent years the Government of Malawi has been pursuing innovative approaches towards a comprehensive national food security strategy. Inthis context, the World Bank i s proposing the use o f ex-ante market-based instruments, which would assist the government to manage the financial risks associated with volatility in maize production. 351. The proposal is to pilot the use o f an index-based weather derivative contract to transfer the financial risk o f severe and catastrophic national drought that adversely impacts the Government's budget to the international risk markets. The aim o f such a contract would be to secure timely and reliable funds for the Government if a contractually specified severe and catastrophic shortfall inprecipitationoccurs during the agricultural season, as measured by weather stations throughout the country. Access to such contingency funds in a time o f crisis would generate a supplemental source o f emergency financing in May to complement existing budget resources, giving the Government more flexibility in its drought and enhancing the Government's ability to launch an efficient and cost-effective drought response. 352. Such a new approach to financing a Government response to drought promises several additional advantages. These include greater autonomy for the Government o f Malawi, minimizing the dependence on an appeals-based model that relies on unpredictable and often untimely donor emergency relief funds, with more emphasis on an ex-ante risk management and financing model. Establishing event- specific, contractually guaranteed contingency funding also creates fiscal stability for the Government and has the added advantage o f providing risk price information to assist Government, and donors, in its investment decisions with respect to managing and mitigating drought risk. 353. Donors, particularly the UK DFID and the EC, have expressed an interest in supporting an initial pilot. Financial donor support in the range o f US$1 million to US$2 million i s requiredto pay for the insurance premium on behalf o f the Government, in addition to a US$300,000 communication and infrastructure support package for the Malawi Meteorological Office to ensure any initial initiative has long-term potential and scope for development and refinement. The objective o f a pilot would be to: (a) establish ifrisktransfer from Malawito the international weather market is feasible; (b) establisha market price for Malawian weather risk; (c) determine the operational and legal protocol for the Government-level risk transfer; (d) demonstrate the potential role of a weather risk management instruments, in conjunction with other risk management strategies - particularly market-based price risk management instruments - in ex-ante Government food security risk management. Several piloting seasons beyond an initial phase will be necessary to fully grasp the role and flexibility o f weather risk management instruments within the Government's strategy, contingency planning and operational drought response as outlined by objective. However securing the strength o f the Malawi 125This annex has beenpreparedby Joanna Syroka, incollaborationwith ErinBryla, AntonioNucifora, and DavidRohrback.The appendixi s basedon the reportby Hess and Syroka(ZOOS), and subsequent studies by Ibarra,Hess, Syroka, andNucifora (2005), and Syroka andWilcox (2006). 120 Meteorological Office and its network will enable such development and, if the programme moves beyond the pilot phase, it i s suggested that the Government allocates premium funds within the budget to protect its own fiscal exposure to drought in hture years. Malawi Maize Production Index 354. The success o f a weather risk management program for the Government o f Malawi depends on the design o f the underlying rainfall index on which the contract i s based. Index-based insurance i s not insurance in the traditional sense where the insured party i s compensated for every dollar loss that can be proved to have occurred. Rather the coverage is based on the performance o f a specified index duringthe insurance period where the index i s designed to correlate as closely as possible with the underlying risk. In the context of maize production, index-based insurance does not cover the actual production loss o f maize; instead it covers the shortfall in rain during a drought year. Because rainfall and maize yields are highly correlated, changes in rainfall - its cumulative amount and distribution - can act as an accurate proxy for maize losses. Often these contracts are legally structured as derivative, rather than insurance, contracts although they still perform an insurance hnction. Risk mitigating payouts are made from the insurer if the index crosses a specified trigger level at the end o f the contract period, indicating situations where a loss i s most likely to have occurred, i.e. payouts are made based on the index and not on the actual loss itself. 355. The benefits o f such an approach i s that by usinga proxy measure o f risk, such as rainfall that i s available on a real-time basis, payouts can be made as soon as the rainfall data i s collected and the index calculated, rather than waiting for the time- consuming and often subjective loss assessments to determine the magnitude o f losses (and therefore payout). Moreover, if based on an independent, objective, verifiable and replicable dataset, an index also creates an opportunity to transfer the risk to the international markets, thus removing this risk from Southern Africa. The Malawi Meteorological Office's rainfall data i s o f excellent quality, satisfying a key prerequisite for risk transfer. The disadvantage o f such an approach i s that proxy indicators are not perfect predictors o f actual events on the ground and there will always be some element o f mismatch between the index and actual experience. Inaddition it i s important to note that not all production deficits are caused by drought: excessive rainfall and flooding, civil strife, poor farm-management, pest infestations and inadequate seed and fertilizer supplies may be as important as deficit rainfall in triggering food emergency situations. These risks are not captured by an index-based rainfall cover and cannot be objectively indexedfor risk transfer. Despite these shortcomings such an index-based approach can, for the purposes o f risk management and budgetary support, provide the Government o f an estimate o f losses due to deficit rainfall; it can thereby be used in an underlying weather risk management contract to reduce the Government's overall fiscal exposure to drought in the critical extreme years when national maize production i s severely impacted. 121 FigureAl.1: The MMPIversus actual and Malawi Meteorological Office-predicted total Malawi national maize production (MT) 1 3000000 T-"-- 500000-- -National Actual Maize Production -National ___-__ Met Office Model Estimated Production National MMPi Regressed Against Historimi Yieids ~ 0 4 ' 356. The index proposed for Malawi - the Malawi Maize Production Index (MMPI) - i s constructed using rainfall data from 21 weather stations throughout the country and is based on the Malawi Meteorological Office's national maize production forecasting model, a modified version o f the FAO's Water Requirement Satisfaction Index adapted for Malawian conditions which uses daily rainfall as an input to predict maize yields and therefore production throughout the country. 126The M M P I has also beenusedinthe recent DFIDand USAID-fundedstudy to evaluate the 2006/7 agriculture input supply program in M a 1 a ~ i . lFEWSNET use this model for their staple crop ~ ~ production early warning system for Africa.128 The model captures not only the total amount o f rainfall received at each station, but also its distribution during the agricultural season and how that impacts maize yields. Using such an underlying rainfall index, a contract can be structured to reflect conditions which would impact national maize production and therefore food security, resulting in significant shortfalls o f maize available for national consumption. The inter-annual variations in the current M M P I have a correlation coefficient o f 74% with inter-annual variations in historical national maize yields for 1984-2006 and 85% with inter-annual variations the Malawi Meteorological Office's national maize production forecasting model for 2000-2006 (see Figure Al.1). The M M P I estimate for the 2006/7 season so far indicates the current season i s one o f the best rainfall years for agriculture on record, in agreement with the Government's first-round maize production assessment and the Malawi Meteorological This is an improvedversion of the index initially outlinedinHess and Syroka (2005). Study conducted by ImperialCollege, Wadonda Consult, MichiganStateUniversity, Overseas DevelopmentInstitute, undertakenfor the Ministry of Agricultureand FoodSecurity.Fundedby: DFID, USAID and FutureAgricultures Consortium(2007). 12'http:liearlywarning.usgs.govladds 122 Office's national maize production forecasting model. Inparticular the index picks up the well documented historical drought events in 2005, 1994, and 1992 and a weather derivative contract based on such an index would have triggered cash payouts to the Government inthose years. 357. Consider the following simple example where the Government enters into the weather derivative contract structure, based on the MMPI, illustrated inFigureAl.2. The contract has a trigger level o f 1.7 millionMT, i.e. a payout is only made if at the end o f the agricultural season inMay the M M P I- the rainfall-based index that i s calibrated to national maize production and therefore i s in MT units - i s calculated to be below 1.7 million MT. Ifthe M M P I i s below 1.7 million MT, the Government will receive US$ 30 for every MT the M M P I i s below 1.7 million MT, up to a maximum payout limit o f US$ 20 million. Ifthe M M P I i s above 1.7 million MT no payment i s made. Such a contract would have paid out US$ 20 million in 1992, US$11.5 million in 1994, US$12.1 million in 1995 and most recently US$4.5 millionin2005. The average payout, or expectedloss o f the contract, over 45 years i s US$1.2 million. For entering into this contract and transferring this expected loss to the market and securing the right to access contingency funds of up to $20 million if there is a drought-related crisis as measured by the MMPI, the Government mustpay an annual premiumo f approximately US$2 million.'29 From a donor perspective this i s can be viewed as an opportunity to leverage US$2 million premium contribution into a $20 million drought-response contribution in a worst case scenario where a severe and catastrophic shortfall inprecipitationhas occurred. However this ex-ante commitment enables the Government and its donor partners to plan for and programme these funds within a greater drought contingency plan well in advance o f a drought event, which ensure a more efficient and cost-effective use o f the contribution. Furthermore by tracking the M M P I throughout the season, as real-time rainfall data i s received, the Government and partners can monitoring the likelihood o f a payout at the endo f May and assess their operational and financingpreparedness. 358. Further work is recommended to improve the M M P I ifthe Government is to use the index with confidence as a basis for budgetary capital allocation. Inparticular the Malawi Meteorological Office should continue its research, assisted by FA0 and other experts, to improve its national maize forecasting model, and consider potentially incorporating the negative impact o f excess rainfall on maize yields into the methodology. The Malawi Meteorological Office's national maize production forecasting model i s based on 75 weather stations and rain gauges throughout the country and arrives at a national maize productionnumber by aggregating from the RDP level, as there are on average almost two weather stations per RDP. Because only 21 stations within the Malawi network satisfy the strict underwritingcriteria ofthe weather insurance market the M M P I only uses these 21 primary stations and arrives at a total maize production estimate by aggregating from the regional level, as the distribution o f these stations i s not sufficient to produce estimates as a higher spatial resolution. The remaining 54 stations utilized by the Malawi Meteorological Office are rain gauges which would have to be secured and made into real-time reporting stations ifthey were to beusedinan index for risktransfer into the international risk market. Such an upgrading 12' This is a simplepremiumrate estimate and only through a competitivemarkettender will the true price o fMalawianweather risk be discovered. 123 o f the network, allowing additional stations to be added to the index, would greatly improve the MMPI's ability to capture spatial variations in rainfall and its agreement with the MeteorologicalOffice's model. Figure A1.2: An example of a hypothetical MMPI-based weather derivative payout structure $25,000,000 $20,000,000 c v) 2. - $15,000,000 n ..$ s + e $10,000,000 V $5,000,000 1.7 million MT $- 500,000 700,000 900,000 1.100,OOO 1,300,000 1,500,000 1,700,000 1,900,000 MMPl (MT) The Weather Market 359. Weather market players, from both the reinsurance and financial communities - a growing market, valued at US$45 billion o f outstanding risk inthe most recent industry survey - are extremely interested in new transactions such as the proposed pilot for Malawi. Weather i s an uncorrelated risk that can enhance the portfolio positions o f traditional risk management firms and professional investors such as alternative hedge funds that are looking to differentiate themselves from other funds that deal in traditional financial markets. The new risks, from locations outside OECD countries, allow for more diversification and hence enhance the riskheturncharacteristics o f commercial risk portfolios. Ultimately this should lead to more aggressive pricing o f weather insurance products in the global market, which inturn should lead to more firms entering the sector attracted by greater market liquidity. In due course this should result in greater business growth and expansion through broadening product offerings and increasing global networks that will benefit the end user customers seeking risk management products, such as the Government o f Malawi. 360. Recently the United Nations World Food Program, with technical assistance from the World Bank, entered into the first-ever humanitarian aid weather 124 insurance contract with a leading European insurer.130 The contract provided contingency funding in case o f an extreme drought during Ethiopia's 2006 agricultural season. The policy was based upon a calibrated index o f rainfall data gathered from twenty-six weather stations across Ethiopia. Payment would have been triggered when data gathered over a period from March to October indicated that rainfall was significantly below historic averages, pointing to the likelihood o f widespread crop failure. While the experimental pilot transaction only provided a small amount o f contingency funding - a maximum payout o f $7.1 million for a premium o f $930,000 paid by USAID - the funds would have been available to WFP at harvest-time which would allow for an intervention four to six months earlier than the traditional appeals- based system. If a catastrophic drought had occurred in 2006, WFP would have used these funds to assist 65,000 households in November 2006. WFP are currently working with DFIDand the World Bank on second-phase program inEthiopia for 2008-2010. Conclusions and recommendations 361. To initiate the piloting process - the testing o f a financial instrument, over several seasons, that could ultimately enhance the Government's national food security strategy - it is recommended that the Government o f Malawi enter into an initial pilot risk management transaction that would provide contingency funding in the event o f extreme drought during the next crop season. The cost o f the 2004/2005 drought response inMalawi was over US$ 110 million for the Government and this pilot will not aim to cover the entire financial magnitude o f such a potential drought response in the country. Rather it will provide coverage for a definedpercentage o f this risk as part o f a pilot program and will test the feasibility o f transferring Malawi's drought risk to the international risk markets. The objective o f a pilot therefore i s not only to secure limited drought financing in a timely and objective manner but moreover, by establishing a market price, to allow the Government and donors to determine the best use for this type o f instrumentintheir food security strategy in subsequent years in conjunction with other available risk management and mitigation tools. Inaddition piloting this approach will also help define all o f the operational and legal implications for entering into this type o f transaction and securing contingency funding for Malawi and other LDC Governments. 362. Index-based weather risk management instruments should be viewed as one o f a growing number o f tools and options in the Government's arsenal for managing food security and creating fiscal stability in the face o f exogenous shocks. The appropriate use o f this instrumentwill need to be evaluated each year and several seasons o f piloting beyond this initial season are recommended. To support an initial pilot for the next agricultural season in Malawi and to instigate this greater engagement on national weather risk management, financial donor support inthe range o f US$1 million to US$2 million i s recommended to pay for the risk-transfer premium on behalf o f the Government. Such a premium could secure approximately US$lO million to $US20 million o f drought-specific contingency funding. Note as an alternative a donor may choose to "insure" the Government themselves by ear-marking and indexing contingent I 3 OFor further information see WFP (2005) and Syroka and Wilcox (2006) 125 fund within their own donor portfolio, without the need for risk-transfer and the associated premiumcost. 363. In order maximize the efficacy of the MMPI as a proxy for maize shortfalls in Malawi, for an initial pilot but more importantly for subsequent seasons, investments in the observing and measurement infrastructure o f the Malawi Meteorological Office are also necessary. A minimum financial support package o f US$300,000 is recommended for the Malawi Meteorological Office (MMO) to enable it to improve its network.13* This would also give the MMO the required capacity to communicate rainfall data reliably in real-time to the World Meteorological Organization's Global Telecommunications System (GTS). This will help ensure that the Government will be able to access the international markets and benefit from such risk management products. It is important to remember investing in the Malawi Meteorological Office network and its communication capabilities has significant benefits for Malawi and the greater international community, which are over and above those required for weather risk management and successful risk transfer. In particular securing the real-time reporting capability o f the Malawi Meteorological Office to the GTS will significantly improve the satellite rainfall estimates and therefore early warning systems, e.g. FEWSNET's, for the country and region; satellite rainfall estimates critically depend on station-level data feeds to adjust for known biases in their computational algorithms. 13'Cost breakdown includes: $3000 x 54 = $162,000 for Casella automated and GSM-enabled rain gauges, installation, spare parts and communication costs, plus an additional $138,000 to strengthen the IT support system for the Malawi Meteorological Office including laptops for field engineers, computers to receive data and software and training to quality control and clean incoming data. Finally, US$20,000 are needed for M M O to complete its communications hub project in Blantyre. Not all o f this investment i s needed in the first year for the initial pilot transaction. I26 AIVNEX1.3 CORE CHARACTERISTICS OF THE MAQUETTE FORMDGSIMULATIONS (MAMS)MODELTHE MALAWIAN OF ECONOMY 364. The MAMS i s an economy-wide simulation model created to analyze development strategies. The model integrates a relatively standard dynamic recursive computable general equilibrium (CGE) model with an additional module that links specific MDG or poverty-related interventions to poverty and other MDG achievements. The relatively detailed treatment o f government activities in M A M S makes this link possible. 365. The core CGE model includes a public sector that is disaggregated into 9 government activities: three types o f education (primary, secondary, and tertiary cycles), health, water-sanitation, government agriculture, irrigation, (other) infiastructure, and other government services. Like other productionactivities, these government sectors use production factors and intermediate inputs to produce an activity-specific output (in the case o f the government this means different types o f services). The private sector is dividedinto agriculture, industry,andprivate services. 366. The factors o f production inthe model include three types o f labor: those with less than completed secondary education (unskilled), with completed secondary education but incomplete tertiary (semi-skilled), and with complete tertiary (skilled). Each o f these labor types i s therefore linked directly to the education sectors/cycles, and thus the growth inthe labor force will inpart depend on the finctioning of the education system inthe model. The remaining factors o f production include public capital stocks by government activity, a private capital stock, and land. 367. The government finances its activities from domestic taxes, domestic borrowing, and foreign aid (borrowing and grants). The provision o f education, health, and water-sanitation services contribute directly to the MDGs. Growth in the stock o f public infrastructure capital (in the case o f Malawi this refers to roads) contributes to overall growth by adding to the productivity o f other production activities. In addition, growth in the stock o f the government agriculture capital stock and the irrigation capital stock contribute to growth inthe private agricultural sector by adding to the productivity o f this sector. 368. These different MDGs are covered in an additional set o f functions that link the level of each MDGindicator to a set of determinants. The determinants include the delivery o f relevant services (in education, health, and water-sanitation) and other indicators, also allowing for the presence o f synergies between MDGs, i.e. the fact that achievements in terms o f one MDG can have an impact on other MDGs. Outside education, service delivery for other MDGs i s expressed relative to the size o f the population. In education, the model tracks base-year stocks o f students and new entrants through the three cycles. In each year, students will successfully complete their grade, repeat it, or drop out o f their cycle. Student performance depends on educational quality (quantity o f services per student), household welfare (measured by per-capita household consumption), the level o f public infrastructure, wage incentives (expressed as the ratio between the wages for labor at the next higher and current levels o f education for the student in question; an indicator o f payoff from continued education), and health status (proxied by MDG 4). 127 Key assumptionsand results of the modelunderthe baseline simulation 369. The MAMS model has been used to analyze the trade-off between spending more on human development than on infrastructure. The base year for the simulation i s 2004. The base simulation, BASE-PRGF, has been used as the benchmark against which two additional simulations are compared. These two simulations, SIM- SOCIAL and SIM-INFRASTR,analyze the possible trade-offs between allocating more public resources to "economic services and infrastructure", on the one hand, or to "human development" on the other, given the expenditure envelope available in any one year. For both simulations, spending on "other government services" remains at the same growth rate as under the ba~e1ine.l~~Details o f the assumptions under each simulation are presented inTable Al.2. 370. The BASE-PRGF simulates a baseline for the Malawi economy, reflecting the policies that the government has been pursuing since mid-2004, inline with the IMF PRGF. The choice o f growth rates for exogenous variables i s broadly in line with the assumptions made under the IMF PRGF program. The PRGF program envisages that Malawi can grow by 6 percent per year on average in the next few years (as a result o f a short-term catching up effect from years o f poor economic management), and that growth will drop to 4.5 percent over the medium term (from year 2011) converging to the regional average. Inthe MAMS simulations, however, we assume that growth will remain at 6 percent throughout the simulation period to 2015. Therefore, in this baseline simulation, the exogenous part of TFP growth by sector is adjusted to generate an annual growth rate o f 6 percent for real GDP at factor cost.133 371. The choice o f growth rates for the exogenous variables i s in line with the assumptions made under the IMF PRGF program. The PRGF program entails substantial changes compared to the unsustainable policies pursuedbefore 2004. On the income side o f the fiscal accounts, foreign borrowing i s assumed to increase exogenously at an annual rate o f 3.3 percent. Financing inflows from the Rest o f the World to the government grow exogenously at 5 percent annually. Taxes are maintained as a fixed share o f GDP, however domestic government borrowing i s not endogenous. It i s instead fixed at zero for 2005-2015, implyingthat the government cannot borrow any more domestically. Lastly, complete foreign debt relief is assumed such that the value o f foreign debt i s brought to 132 Intotala set of six simulationshasbeenrunto analyze the trade-offbetweenspendingmoreor less on human developmentthan on infrastructure, but theseresultsare notpresentedhere. The simulations SIM- INFRASTR+l, SIM- NFRASTR+2, and SIM- INFRASTR+3 show progressively morespendingon the infrastructuresectors, whichimplies, giventhe budgetenvelope, progressivelyless spendingonhuman development.The simulations SIM-SOCIAL+l,SIM-SOCIAL+2, and SIM-SOCIAL+3 are preciselythe opposite:progressivelymorespending on humandevelopmentand less on infrastructure. Also simulations havebeenrunto test the robustnessof the model. The full set of simulationsis availableuponrequest; also see: Diaz-Bonilla,et al., 2007, "Trade-offs between humandevelopmentandinfrastructurespendingin Malawi" unpublishedmanuscript,World Bank, WashingtonDC. ' 3 3To assess the impact o f lower assumptionregardingGDP growthrates, however,we also ranthe simulation(s) assuminga 4 percentGDP growthrate throughoutthe period.This approach allows us to better assess the robustness of the modelto assumptionsregardingGDP growth.The resultsindicatethat the assumptionon the levelof GDP growth has a substantial effect on the outcomes o f the simulation in terms of levels,but does not changesthe direction o fthe tradeoffsbetweenthe differentstrategies. 128 zero in 2007. From 2008 onwards, the stock o f foreign debt again begins to increase by the amount o f foreign borrowingmentioned above. 372. The baseline for all government income parameters i s exogenous in BASE-PRGF, and thus government expenditure i s made endogenous. The resulting annual growth o f government expenditures i s set to be uniform for all sectors except for education, in which the target assumption i s to keep government consumption per enrolled student constant (the model definition for educational quality). The PRGF program imposes sound fiscal discipline which only allows government consumption by sector to grow at 4.8 percent annually (compared to 14.4 percent on average during2000- 2004). As mentioned, the only exception i s for education, in which the target i s to keep government constant expenditureper enrolled student. 373. Underthe PRGF, Malawi's efficiency is expected to increase as a result o f macroeconomic stability. This simulation therefore incorporates also a few additional productivity increases for the baseline. The elasticity o f TFP for all activities with respect to the public infrastructure capital stock equals 0.006. The capital stocks from the irrigation sector and from the government agriculture sector have a stronger direct impact on agriculture. The elasticity o f TFP for the agriculture sector with respect to both these capital stocks i s increased to 0.1. In addition, the simulation includes an elasticity o f factor productivity for labor in all activities with respect to health. Health in this case i s measured by the change in the value o f the under-5 and maternal mortality rates relative to the base year. The elasticity i s changed from zero to -0.1, which implies that labor force productivity in every activity increases when the under-5 mortality rate decreases relative to its value inthe base year 2004. 374. Expenditure on education grows by more for secondary and tertiary as compared to the primary cycle because o f the relatively much larger number o f students in primary school in Malawi in 2004. An improvement in graduation rates and continuation rates inthe primary cycle implies a relatively large number o f new students in secondary school given the very small number of students in this cycle in 2004. Therefore, since educational quality i s assumed constant, a relatively large increase o f students in secondary school implies that government spending must also increase by a relatively large number. This explains why the growth in government spending i s higher for the higher cycles than for primary school. 375. For the other macro aggregates, household consumption increases 4.9 percent annually, while public and private investments increase 5.1 percent and 8.2 percent per year, respectively. Public investment grows by 5.1 percent per year. The real exchange rate slightly depreciates, by a mere 0.02 percent per year. On the other hand, exports increase by a strong 7.6 percent, and imports by 5.1 percent per year. 376. The stock o f foreign debt is zero by construction in 2007 and grows slowly after 2007. As a result o f the low levels o f debt by 2015 under BASE-PRGF, foreign debt for the whole period actually decreases at what would be on average a -10.4 percent rate per year. With the decrease in the foreign debt stock, foreign interest payments as a share o f GDP decrease from 1.5 percent in 2004 to 0.2 percent in 2015. The stock o f government debt as a share o f GDP decreases from about 26.5 percent in 129 2004 to about 11 percent in 2015. Domestic interest payments thus also decrease, from 9.2 percent of GDP to 3.8 percent by 2015. 377. In the growing economy, wages and employment increase for all three labor types. Employment growth i s in the range o f 1.9-2.2 percent per year. Due to the decline in unemployment in the model, the labor force growth rate i s lower-between 1.6-1.9 percent per year. Wages increase between 4-7.6 percent per year. Both employment and wage growth i s highest for the better educated. This may be due to the following. Growth in any public sector implies the need for more skilled labor. Due to the almost insignificant size o f the skilled (Le., completed tertiary) labor force inthe total labor force, any increase in the number o f these workers will be large relative to their base year employment levels. In addition, the increase in public sector spending i s increasing the demand for skilled workers, but the education sector i s not providing fast enough increases to the supply in the given model period, therefore resulting in strong wage increases for this labor type. 378. The poverty rate decreases, directly due to the increase in household per capita consumption. Monitoring all other education, health, and water-sanitation measures also shows improvements by 2015. The net completion rate for primary school doubles, although remaining at a very low level, while the maternal mortality rate i s cut by almost half. Although access to safe drinking water and to safe sanitation improves, the results are quite small. Simulation of alternative expenditure strategies 379. When the focus i s on infrastructure, government expenditures on public infrastructure, agricultural, and irrigation services increases from a 4.8 percent per year growth rate in BASE-PRGF up to 9 percent per year for SIM-INFRASTR (Table Al.1). Growth in government spending in health and water and sanitation decreases from 4.8 percent per year to 3.3 percent per year for SIM-INFRASTR,while for primary education the growth in consumption decreases from 2.2 percent to around 0.7 percent per year. Spending inother government services maintains its growth rate over time as determined. 380. Total public recurrent expenditure on the three infrastructure sectors analyzed are relatively small, increasing from less than 2 percent o f GDP in 2004 up to 2.8 percent o f GDP in 2015 for SIM-INFRASTR. However, the amount o f investment requiredinthe infrastructure sectors for a given increase inrecurrent expenditure inthese sectors i s higher relative to the requirements in the HD sectors. Therefore, the results show that government investment spending also increases, from an annual growth rate o f 5.1 percent to 5.6 percent. TFP i s positively affected as the infrastructure capital stock increases. Absorption, household consumption, and private investment all grow at higher rates. With a growing economy, wages also grow by more for all labor types. The exchange rate appreciates at a rate o f -0.17 percent per year. Exports grow at an annual 8.3 percent at its highestpoint versus the 7.6 percent per year for BASE PRGF. Imports grow by 5.5 percent per year, 0.4 percentage points higher than for BASE-PRGF. Both foreign and domestic debt decrease as a share o f GDP in 2015, as do domestic and foreign interest payments. 381. The results are the complete opposite when the focus i s on increased spending in the HD sectors (SIM-SOCIAL) rather than infrastructure. Government 130 recurrent consumption o f health and water and sanitation grows at an annual 5.6% in SIM-SOCIAL, versus 4.8% in BASE and 3.3% in SIM-INFRASTR. Primary education expenditureincreases from a 2.2% annual growth rate inBASEto 3% in SIM-SOCIAL. As explained earlier, in SIM-SOCIAL the growth ingovernment infrastructure spending i s exogenously decreased, while the growth rates for the HD sectors are endogenously scaled up so as to stay within the fiscal space limits. The results show that GDP growth (5.6%) i s now lower than for BASE (6%), and much lower than in SIM-INFRASTR (6.3%). Both government and private investment grow by less, as do household consumption and absorption. The real exchange rate depreciates, and both exports and imports grow by less. Inan economy with slower GDP growth, wages for all labor types grow by less as well. Poverty decreases by less than under the BASE and less than under a scenario that focuses on infrastructure spending(SIM-INFRASTR). 131 TableAl.1: Results of the MAMS simulations: PovertyReductionGrowthFacility Baseline plus Trade-off Simulations sim-social base-prgf sim-infrastr Values Units Absorption" 2367 bn Kwacha 540 5 18 4 88 Household consumption 1787 bnKwacha 5 18 4 94 4 61 Governmentconsumption (total) 308 bn Kwacha 584 5 85 5 81 Governmenteducation services (total education) 7 5 bnKwacha 7 21 8 65 9 48 Governmentprimary educ services 5 1 bnKwacha 0 71 2 17 3 03 Governmenthealth services 3 6 bnKwacha 331 4 78 5 64 Governmentirrigation services 0 2 bnKwacha 904 4 78 -0 10 Governmentagricultural services 1 3 bnKwacha 904 4 78 -0 10 Governmentwatwsanitation services 0 2 bnKwacha 331 4 78 5 64 Governmentpublic infrastructure services 2 2 bnKwacha 904 4 78 -0 10 Governmentother services 159 bn Kwacha 4 78 4 78 4 78 Governmentinvestment 21 0 bn Kwacha 555 5 13 4 64 Private investment 6 3 bn Kwacha 835 8 18 7 93 Exports 499 bnKwacha 829 7 60 6 73 Imports 982 bn Kwacha 551 5 12 4 65 Real exchange rate (price-level deflated) 1000 indexed -0 17 0 02 0 19 -Kwacha per Foreign Curr. Unit to 100 GDP at factor cost 1673 bnKwacha 632 6 00 5 59 Wage of labor with less than secondary education" 14584 8 Kwacha 4.17 3.97 3.60 Wage of labor with secondary education 69027 4 Kwacha 6.41 6.29 6.03 Wage of labor with tertiary education 167075 8 Kwacha 7.72 7.55 7.27 Direct Taxes 8 4 % of GDP 8.41 8 41 8 41 Import Taxes 2 7 % of GDP 2.70 2 70 2 70 Other IndirectTaxes 8 5 %ofGDP 8.48 8 48 8 48 GovernmentIncomefrom ROW 11 5 %ofGDP 9.73 10 16 10 73 Total Gov Income 34 2 % of GDP 31.99 32 44 33 03 GovernmentSpending (by sector) Primary Education 2 7 % of GDP 2.15 2 52 2 81 Rest of Education 1 3 % of GDP 4.32 5 00 5 53 Health 1 9 % of GDP 1.63 1 94 2 18 Irrigation 0 1 % of GDP 0.17 0 11 0 07 Agriculture 0 7 % of GDP 1.08 0 71 0 43 Water and Sanitation 0 1 % of GDP 0.08 0 10 0 11 Public Infrastructure 1 1 % of GDP 1.64 109 0 67 Other GovernmentServices 8 4 % of GDP 8.48 8 62 8 88 Domestic Interest Payments 9 2 % of GDP 3.64 3 75 3 91 Foreign Interest Payments 1 5 % of GDP 0.22 0 23 0 24 Gov Savings 2 5 % of GDP 5.11 4 81 4 49 Total Gov Spending 34 2 % of GDP 31.99 32 44 33 03 Investment 14 5 % of GDP 13.92 13 90 13 96 Foreign Savings 9 3 % of GDP 6.37 6 65 7 03 Trade Deficit 28 4 %of GDP 17.81 1880 20 10 Direct Taxes 8 4 % ofGDP 8.41 8 41 8 41 Foreign Debt 162 3 % of GDP 24.22 25 29 26 73 Domestic Debt 26 5 % of GDP 10.51 10 83 11 29 MDG 1: headcount poverty rate 52 4 Yo 29 61 31 45 33 94 MDG 2: primary (1st cycle) net completion rate'" 8 0 % 10 82 1544 18 54 MDG 4: under4 mortality rate (share of live births) 133 0 per1000 11341 108 12 10644 MDG 5: maternal mortality rate (share of live births) 960 0 per 100000 587 02 499 00 472 57 MDG l a : acess to safe drinking water 66 1 % 67 39 67 89 68 06 MDG 7b:acess to safe sanitation 63 7 /D 6466 64 86 64 84 Foreign aid per capita 28 9 US$2004 38.12 37.59 37.09 Present value of total foreign aid 2004-2015 US$bn2004 3.74 3.72 3.70 Simulations base-prgf base Poverty ReductionGrowihFacilityscenario sin-economic Spendingon educ, health,and water-sanitationdecreasesby 1.5 timas the growth rata in base-prgfsimulation, while spendingon "Irrigation","AgricultureGcv",and "PublicInfrastructure"adjustsas required. sim-social Spendingon educ, health,and water-sanitationincreasesby 1.5timesthe growih rate in base-prgisimuiation while spendingon "Irrigation","AgricultureGov",and "Public Infrastructure"adjustsas required. For both sim-economicand sim-scialsimulations,spendingon "OtherGovernmentServices"remainsat the same growthrate, *Thedifferentmacroaggregates (absorption,consumption,investment, exports, imports,and GDP at factorcost) are all in realterms (at constant2005 LCU) "Wages are realin a settingwith constantCPi. "'MDG 2 here refersto the rate of primaryschoolcompletionfor studentsof the correctage cohort AND within the 8 year time span (avery stringentdefinitionof MDGP). 132 Table A1.2: MAMS modelassumptionsacross simulations BASE SIM-iNFRASTR SlMSOClAL GDP growth rate 6% enuoyeiious sndogerioiis growth rata of oxpondituro(n gror,ih rato of exparditwe :n tiilras sectors !ricreasej by irifrss seclors decreases by i . 5 linies the growth in BASE: 1.S tirnsa thb growth iii BASE: "other gov sewces" maintaw "other gov services" nimritiiins Government Consumption (non- same expend!ti:rP grovd!i saiiie sxpe!-diti! re 'jrnwth education) clearsfiscai account rate: wtiiie growth rate o! ra!e: while growth rate of expenditureiii t4D sectors expeno!li!re in t.10Socfors nridogenotisly decreases so endogeriousiy irr?rases so as :o keeu over~ilyov as to keep cveraii gov consimp:ion rmthinfiscai consumptionwithin fiscal space iiriit. srxce lllilit adjusts such that total adjusts such that totai adjusts such that total Government Consumption expenditureper enrolled expenditureper enrolied expenditureper enmlled (education) student remains constant student remainsconstant student remainsconstant between2004-2015 between2004-2015 between2004-2015 Domestic government borrowing set at zerofor 2005-2015 set at zero for 2005-2015 set at zerofor 2005-2015 Direct and indirect taxes fixed share of GDP fixed share of GDP fixed share of GDP Unrequited official transfers (grants) 5% 5% 5% Foreign Borrowing 3.3% 3.3% 3.3% Completein2007 (i.e.,foreign Completein2007 (!.e.,foreign Completein 2007 (Le., foreign Foreign Debt Relief debt equalszero in 2007);for debt equalszero in 2007);for debt equalszero in 2007), for otheryeas: same as in SIM- other years: same as in SIM- other years: same as in SIM- HISTORICAL HISTORICAL HISTORICAL Private Investment Demand Endogenous;savings-driven Endogenous:savings-driven Endogenous;savings-driven investment investment investment FDI 2% 2% 2% Elasticity of TFP for all activities w.r.t. the public infrastructure 0.006 0.006 0.006 capital stock Elasticity of TFP for Agriculture activity w.r.t. the government 0.100 0.100 0.100 agriculture capital stock Elasticity of TFP for Agriculture activity w.r.t. the irrigation capital 0.100 0 100 0.100 stock Elasticity of of factor productivity for labor in ail activities w.r.t. -0.100 -0.100 4.100 health 1.95%per year through2008; 1.95%per year through2008; 1.95%per year through 2008; Population growth 1.4 1.7%per year for the ~ 1.4-1.7%peryearforthe 1.4- 1.7% peryearforthe remainingperiod. remainingperiod. remainingperiod Note,Ali numbers reoresentexoaenousvalues or Darametervalues fi.e..fixed throuah time and simu!ation) 133 ANNEX2.1 BRIEF DESCRIPTIONOFSELECT FINANCING INSTRUMENTS Floating rate bonds: These instruments are generally long-term bonds that offer an interest rate that i s peggedto a benchmark, such as the short term 3-months Treasury bill rate, and adjusted periodically depending on the movements in the chosen benchmark rate. InMalawi, the Treasury bill rates have been declining steadily in recent years from an average o f 42 percent in 2002 to around 25 percent in 2005. Given the sustained decline in the Treasury bills rate, it would be useful to issue these instruments on an experimental basis. The government would benefit as long as the Treasury bill rates continue to decline. A further advantage would be to lengthen the maturity o f the debt portfolio, thereby reducing the refinancing risks. However, a serious disadvantage o f these instruments i s that the interest cost for government would increase if the interest rates reverse their downwards trend. Inflation-indexedbonds: These bonds linkboth the principal and interest to an inflation index. Such instruments present the advantage to fix the real interest rate on the bonds, by allowing the nominal rate to adjust in line with the inflation rate. By reducing the inflation-related risks such instruments allow the government to offer a lower interest rate and a longer maturity, thereby reducing interest costs as well as the refinancingrisks. On the other hand, a major risk to the government is that any increase in inflation, whether due to poor macroeconomic policy or the occurrence o f external shocks, leads to an immediate increase in the interest bill, firther compounding the need for financing at a time when there is a need to tighten monetary policy. Hence, such instruments are advantageous to the government only when the government i s pursuing a strong anti- inflationary policy. In Malawi, inflation measured through the consumer price index (CPI) has declined from 35 percent in 2000 to around 10 percent in 2003. However, inflation subsequently increased to around 15 percent in 2005, largely as a result o f the global increase in oil prices and the 2005 food crisis. Given Malawi's high vulnerability to exogenous shocks, therefore, this instrumenti s probably not appropriate. Bonds with embedded call/put options: Such instruments consist o f plain fixed rate bonds with "call" and "put" options that are exercisable either by the investor or the government after a specified period o f time. The investors would "call" if interest rates rise above the coupon rate on the date o f option and the government can "put" if interest rate falls below the coupon on the date o f option. Hence, these instruments offer greater flexibility than traditional bonds. Such contractual arrangements would allow the government to take advantage o f interest rate movements. These instruments also help to reduce the uncertainty premium attached to the interest rate, and so help in lowering interest rates. 134 ANNEX3.1 BRIEFDESCRIPTION EDUCATION OF SYSTEM INMALAWI Level o f education Student enrollment Number o f schools Primary 3,166,786 5,103 Public 3,140,440 4,98 1 Private 26,346 122 Secondary 180,157 967 Public CDSS 83,492 561 Government boarding 16,223 35 Government day 19,828 59 Open school 1,957 16 Grant aided 16,322 63 Private 42,335 233 Primary teacher training 4,800 6 Face to face - 1 522 Distancelearning 578 Tertiary" 5,902 2 Public University o fMalawi 5,520 1 University of Mzuzu 382 1 Private NIA NIA 383. The primary cycle i s 8 years and children are expected to enroll in the first grade, standard 1, at six years o f age. At the end o f standard 8, pupils sit for the Primary School Leaving Certificate (PSLC) exam. Since 1994, there are no fees for attending public primary education. Enrollment inprimary, whether public or private, was a total o f 3.2 million pupils in 2004. The overwhelming majority, more than 99%, attended public schools. The number o f schools was 5,103; 98% o f these were public (this category includes a large number of schools built and owned by religious agencies or communities, where the government provides teachers and other inputs). In the largely rural Malawi, as much as 96% o f schools are categorized as rural schools, where average 135 enrollment i s 585 pupils compared to an average enrollment o f 1,580 pupils at urban schools. 384. The secondary cycle is four years. After two years, students sit for the Junior Certificate o f Education (JCE) exam. After two additional years, students can obtain the Malawi School Certificate o f Education (MSCE). The results in this exam determine eligibility to public university. Enrollment in secondary was about 180,000 in total in 2004. A characteristic o f secondary education i s that a large share (23%) o f the student body i s enrolled inprivate secondary schools. There were 967 secondary schools in2004; 24% o f the schools were private, 7% grant-aided (government provides some funding), and the remainder 69% o f schools were government-run. There are four types o f government-run schools: government boarding, government day, community day secondary schools (CDSS), and open schools. The government boarding and government day schools are nationalhegional schools that enroll the students that obtained the best results in the PSLC exam nationwide or within a larger region. The CDSS serve a local catchment area, were often built by the communities themselves, and are less well endowed in terms o f the number and quality o f teachers and other resources. The CDSS account for almost 70% o f enrollment ingovernment-run secondary schools, as shown in Table A3.1. 385. Higher education is mainly provided at the two public universities, University o f Malawi and University o f Mzuzu. These are autonomous and the government subsidy to these two institutions comes under a separate vote on the government recurrent budget, independent o f the budget o f the MOE. University o f Malawi i s the oldest and largest o f the two. It is organized in five colleges: Bunda College o f Agriculture, Chancellor College, College o f Medicine, Kamuzu College o f Nursing and the Polytechnic, and had a total o f about 5,500 students enrolled in 2004. The number o f graduates, however, i s only around 350-400 every year. Most degree programs have a duration o f four years, leading to a bachelor degree (the university accepts a few master degree students, but they have to be fblly sponsored by an outside agency or donor, as they are not sponsored by government). 386. The University o f M z u m is a relatively new and small university. It first enrolled students around 1999, but continues to have a total enrollment o f only about 400 students. Interms o f private tertiary education, a Church of Central Africa Presbyterian sponsored university has been established recently and a Catholic fbnded university i s expected to commence soon inMontfort; thus, private universities i s a small but growing subsector. 136 ANNEX4.1 BRIEFDESCRIPTIONHEALTH OF SYSTEM INMALAWI 387. Health services in Malawi are organized into four levels. At the most rudimentary level, the household itself produces health services through "self-care" and use o f appropriate health information and goods available within the immediate community through a range o f suppliers including health surveillance assistants; social marketing programs; village drug revolving funds; and drug shops. The first level o f the households contact with the facility-based network i s the primary care health posts, dispensaries, clinics, and centers. The second level i s representedby district hospitals, which act as referral units for primary-level facilities. In general, there is a district secondary-level hospital for each district. In places where there are no government district hospitals, Christian Health Association o f Malawi (CHAM) hospitals act as default district hospitals. The tertiary level consists o f the four central hospitals (Queen Elizabeth Central Hospital or QECH in Blantyre, Zomba Hospital, Lilongwe Central Hospital or LCH, and MzuzuCentral Hospital). In Blantyre and Lilongwe, there are no district hospitals so the QECH and L C H also act as district hospitals. 388. Although most health facilities in the country are under the MOH, the non- government sector i s growing apace. Malawi has a total o f 617 health facilities, or 1 facility per 17,800 people (Table A4.1). Out o f the total number o f facilities, 64 percent are operated by the MOH; 5 percent by urban councils/local governments (LG); 26 percent by the CHAM, Banja L a Mtsogolo (BLM), and other NGOs. The table below does not include for-profit clinics. In recent years, the non-government sector has increased the number o f its facilities more than the government sector. Between 1997 (usingNational Health Plan data) and 2002 (usingJICNMOH data), government health facilities increased from 354 to 424 (20 percent increase), while non-government health facilities increased from 146 to 193 (32 percent increase). Table A4.1 Number of Health Facilities in Malawi by Type and Ownership, 2002 Type Government Nongovernment Total MOHP LG CHAM BLM NGO Central hospital 4 4 Mentalhospital 1 1 2 District hospital 22 22 Hospital 19 27 46 Health center 288 12 115 1 416 Maternity center 2 12 1 15 Rehabilitation center 1 1 Clinic 2 4 8 21 1 42 Voluntary counseling clinic - 3 3 Dispensary 54 4 8 66 Total 392 32 161 28 4 617 Source: MOHiJICA Health Facilities Survey 2002 389. MOH is currently establishing service agreements with mission facilities, although current implementation i s fragmented. MOH has so far concluded 3 service agreements with C H A M hospitals to improve the efficiency and coverage o f service delivery. Some major problems exist in implementing service agreements however, including the fact that C H A M i s unable to negotiate as an "umbrella" organization. Since 137 each mission hospital i s autonomous, MOH has to negotiate with every individual facilities, and this process has been management-intensive and time-consuming. 390. The private sector is, indeed, emerging as a significant provider o f services and is drawing an increasing number o f Malawians. Inresponse to service deficits inthe public sector, people are being drawn to these new providers. Around 18 percent o f all consultations are being done outside MOH and mission facilities, according to the 2002 Core Welfare Indicators Questionnaire (CWIQ) Survey. Moreover, more than half o f patients who go to government facilities first and $annot have drugs or treatment there end up going to private providers, according to the Malawi Economic Justice Network (MEJN) Survey. 138 ANNEX 6.1 GLOSSARYKEY TERMSINTHE ROAD SECTOR OF Main roads: Inter-territorial roads outside cities or towns unilaterally designated by Government providing high degree o f mobility connecting provincial capitals and/or servingas international corridors; Secondary roads: Roads outside cities or towns unilaterally designated by government providing a high degree o f mobility linking main centers o f population and production and connecting to the main roadnetwork; Tertiary roads: Roads outside cities or towns unilaterally designated by government linking collector roads to arterial roads accommodating the shorter trips and feeding the arterial road network; District roads: Roads outside cities or towns designated by Government after consultation with the District Authorities providing intermediate level o f service connecting local centers o f population and linking districts, local centers o f population and developed areas with the principal arterial system; Urban roads: Any other road in an urban area other than a designated road including arterial and collector roads crossing city boundaries. Main h c t i o n i s provision o f accessibility over relatively short trip lengths at low speeds and providing services to smaller communities. Paved Roads: All roads which are bituminized or with asphalt concrete surface Unpaved Roads: All earth and gravel roads DesignatedRoads: All gazettedroads which are managedby government and are allocated to appropriate authorities Un-designated Roads: All roads which are not gazetted. Routine Maintenance: These include grass cutting, grading o f earth roads, re-gravelling, pothole patching and edge repairs, replacement o f timber on timber decked bridges, culvert and mitre drain cleaning, replacement o f road signs etc. Periodic Maintenance: Covers shoulder and edge repairs, patching o f potholes, slurry sealing o f areas with crocodile cracks, followed by a complete slurry seal. These works are very necessary for the protection o f the roads investment from further deterioration. Periodic maintenance i s usually carried out in the sixth; tenth and fifteenth year after the road was upgraded from earth to bitumensurface Rehabilitation: Carried out where the road has deteriorated to such an extent that periodic maintenance i s no longer feasible. This is mainly because o f prolonged deferring o f periodic maintenance, which if carried out at an appropriate time would arrest further road deterioration, hence reduced cost o f maintenance. These works include the ripping off o f the road surface to base and the sub base layers followed by reconstruction o f base, sub base and the road surface ConstructionKJpgrading: Construction or upgrading o f new road from eartwgravel road to bitumen standard; as well as upgrading from earth to gravel standard; construction o f new bridgesand replacement o f timber decks with reinforced concrete CoreRoadNetwork: This includes selected main, secondary, tertiary urbanand district roads which form a backbone o f the public road network. 139 REFERENCES Aitken, Jean-Marion and Julia Kemp(2003). HIV/AIDS, Equityand Health Sector Personnel in Southern Africa. EquinetDiscussion Paper No. 12. Regional Network for Equity inHealth in Southern Africa. Fundedby DfID. Ayala-Ostrom, Beatriz, Marita Van Rooyen, and Tony Odendaal (May 2004). Output to PurposeReview: Strengthening Malawi Central Medical Stores Project. DffD Health Systems Resource Center. London. Baekey, Carole A. (August 2004). HospitalAutonomy Bill, version 1.3. Working Draft. Barron, Peter (December 2004). Relationship Betweenthe NGOs, Researchand Teaching Institutions Operating inthe Two Central Hospitals inMalawi. Draft. Bollinger, L and Stover, J. (2006) "The Potential Impact o f HIV/AIDS Interventions on the Epidemic inAfrica" FuturesInstitute, mimeo. Bollinger, Lori, John Stover, and Martin Palamuleni (July 2000). The Economic Impact o f AIDS inMalawi. The Futures Group International. Washington, D.C. Calcon (2003). Malawi Health Facilities Inventory Survey. Final Report. Fundedby JICA. Castro-Leal, F., J. Dayton, L.Demery and K.Mehra (2000). "Public Spending on health care in Africa: do the poor benefit?". Bulletin o f the World Health Organization, 78(1) 66-78. Collier and Gunning, (1999). "Explaining African Economic Performance", Journal of Economic Literature, Vol. XXXVII, March 1999. David 2007, "Fiscal Space and Fiscal Sustainability o f HIV/AIDS Programs inSub- SaharanAfrica" unpublishedmanuscript, ACT Africa, World Bank, Washington DC Demery, Lionel (2000). "Benefit Incidence: apractitioner's guide". World Bank DFID2004, The budget as theater, DFIDstudy, 2004 EdData (DHS)2002 Survey.NSO and ORC Macro. Reports available on the internet at: http://www.eddataplobal.org/documents/index.c fin?fiseaction=showdir&pubcoun try=MW&statusID=3&sho~pcs=O 140 EMC Jatula Associates (January 2003). Review o f cost-effectiveness and design standards, Malawi. (referenced in "PAD - Third Social Action Fund". Report 25565-MW. April 2003.) Fiedler, John L.(2003): HNP Discussion Paper: A Cost Analysis o f the Honduras Community-Based Integrated Health Care Program. HNP Discussion Paper. The World Bank Fozzard et al., 2002,ODI Working Paper N.166 Gelb, A., Eifert, B.,2006, Improving the dynamics o faid : towards more predictable budget support. Policy ResearchWorking Paper no. WPS 3732. World Bank, Washington DC. GoM and World Bank, 2006, Malawi Poverty and Vulnerability Assessment: Investing in Our Future, Volume I: Synthesis o fMainFindings And Recommendations. World Bank Report No. 36546-MW GoM and World Bank, 2006, Malawi Poverty and Vulnerability Assessment: Investingin Our Future, Volume 11:Detailed Background Papers.World Bank Report Government o f Malawi (2004); Consolidated Annual Accounts for the financial year ending June 2004. Government o f Malawi (GOM) (May 24,2004). Country Procurement Assessment Report. Volume 3: Executive Summary. With support from the World Bank. Gwatkin, Davidson (2003). "Free govement health services: are they the best way to reach the poor". World Bank GwatkinDR,RutsteinS, Johnson K, Suliman EA, Wagstaff A. (2003). Initial country- level differences about socio-economic differences inhealth, nutrition, and population. 2nd ed. Washington, DC: World Bank. Haacker, Marcus (2003), ProvidingHealth Care to HIV Patients inSouthern Africa. IMF Policy Discussion Paper). Haacker, Markus (2002). The Economic Consequences o f HIV/AIDS in Southern Africa. IMF Working Paper (WP/02/38). Health Planning Services Department, Malawi, 2003. Hess, U.and Syroka, J. (2005) Weather-based Insurance in Southern Africa: The Case of Malawi. Agriculture and Rural Development Discussion Paper 13.World Bank, Washington D.C. Paper available on the internet at: http://www- wds.worldbank.ora/external/default/WDSContentServer/WDSP/TB/2006/08/23/00 141 0310607 20060823 17123l/Rendered/PDF/3705 1OMWOWeath1dOinsurance01P UBLIC1.pdf HospitalAutonomy Program (December 2003). Executive Summary o f the Report on the Baseline Survey o fManagement Systems at Lilongwe and Queen Elizabeth Central Hospitals. Management Sciences for Health. Fundedby USAID. Hozumi, D.(2003), Status o f HealthFacilities inMalawi, study fhded by JICA and. submittedto the Malawi Ministry o fHealth andPopulation, JICA, Tokyo. Hunter,S.; Williamson, J. (2002): Childrenonthe brink.Updated estimates and recommendations for interventions. USAID.Washington, D.C. Ibarra, H.,U.Hess, J. Syroka, and A. Nucifora. 2005. Use o fWeather Insurance Markets for Managing Food Supply Risk: Malawi Case Study.Available online at: http://www.passlivelihoods.org.uk/site files%5Cfiles%5Cre~orts%5Cproiect id 240%5CWeather%2OIiisurance%2OMarkets%2OVIII Hess.doc IMF, 2007a. Strengthening the Policy Context and Technical Framework o f Budget Preparation. FiscalAffairs Department. IMF,2007b, Treasury Reforms - Implementation ofa CashFlow Planning System. EAST AFFUTAC. JHPIEGO (2003). InfectionPreventionHospital Study. Ministryo f Health and Population. JIP Subcommittee on Hospital Autonomy (September 2004). Draft Summary o fNational Policy on Hospital Autonomy. Ministry o f Health. JIP Subcommittee Task Team on Patient Tariffs (June 2004). Revision o f Patient Tariffs for Government Hospitals. Draft report. JLI, 2004: The Joint LearningInitiative (JLI) for improving human resources for health, a global initiative o f the Rockefeller Foundation. Kadzamira, K., et al. (2001). The impact o f HIV/AIDS on Primary and Secondary Schooling inMalawi: Developinga comprehensive strategic response Kadzamira, Kamlongera, et al. (2001). The impact o f HIV/AIDS on Primary and Secondary Schooling inMalawi: Developinga comprehensive strategic response. Kathyola, Janet (2003). The Status o f Human Capacity Development for an Effective Responseto the HIV/AIDSEpidemic inMalawi. Malawi Institute o f Management. March. 142 Kengueand Mingat (2002). Analyse comparative internationale de la feminisation du corps enseignant et de l'impact du genre de l'enseignant sur la performance des systemes Cducatifs primaires en Afrique. Kengue, V.M. andMingat, A42002). Analyse comparative internationale de la fkminisation du corps enseignant et de l'impact dugenre de l'enseignant sur la performance des systCmes Cducatifs primaires en Afrique. Communication Colloque AFEC, Caen, May 23-25,2002. Kwanjana, J.H., A.D. Harries, F. Gausi, D.S.Nyangulu, F.M.L.and Salaniponi (2001). TB-HIV SeroprevalenceinPatients with Tuberculosis inMalawi. Malawi Medical Journal, 13, pp.7-10. Laevens, Anne (September 2002). Case Studies o f Intermittent Preventive Treatment (IPT)/Sulfadoxine-Pyrimethamine(SP) Implementation (Malawi and Kenya). From Malaria inPregnancy: From Policy to Implementation. Levin,Ann, RonaldMangani, MarkMcEuen, Regson Chaweza, andNysonChizani (July 1999). Costs o fMaternal Health Services inBlantyre District, Malawi. Special Initiatives Report No. 17. Partnership for Health Reform. Abt Associates. Mahal, Ajay et a1(2000). "Who benefits from public health spending inIndia?". National Council of Applied Economic Research, New Delhi. Malawi Core Welfare Indicators Questionnaire (CWIQ), 2002: Report available online at: http://www.nso.malawi.net/data_on_line/agriculture/cwiq/cwiq,pdf Malawi Demographic Health Surveys 1992, 2000 and 2005. Reports available online: http://www.measuredhs.com/countries/count~~main. cfm?ctry_id=24 MGFCC (2002) An Integrated Responseto HIV/AIDS and Malaria: Application to the Global Fund.National AIDS Commission, Lilongwe. Malawi HealthEquityNetwork (2002). BudgetMonitoring Initiative, FY2001/2002. Firstand SecondQuarter Report. Malawi ResearchTeam, The (2003): The Impact o f HIV/AIDSon the education sector in Malawi. Study 1: Examining policy, leadership, and advocacy responsesinthe education sector. MOESTAIEP-UNESCO, Malawian Ministry o f Health (2004). A Joint Programme o f Work for a Health Sector Wide Approach (SWAP)[2004-20101. Malawian Ministry o f Health (2005). Ministry o f HealthDraft Expenditure Returnfor the financial year ending June 2005 143 Management Sciences for Health (2003). Report on the Baseline Survey o f Management Systems at QueenElizabeth Central Hospital, Malawi. MSHMalawi. Reducing Child Morbidity and StrengtheningHealth Care Systems inMalawi. Fundedby USAID. Martineau, T, Sargent, K, et al. (2001) Institutionaland human resourcesappraisal (Phase 11)of SexualandReproductive HealthProgramme, MinistryofHealthand Population, Government o f Malawi. Mission Report August 2001. Liverpool School o f Tropical Medicine, L A T H & JSI, UK. McCoy, David, Esther Ratsma, and Mike Rowson(2004). Going from Badto Worse: Malawi's Maternal Mortality. Commissioned by Task Force 4 o fthe U.N. MillenniumProject. McGuire, J. 1996. The Payoff from Improving Nutrition, World Bank, Medecins Sans Frontieres (MSF) (July 2004). Antiretroviral Therapy inPrimary Health Care :Experience o f the Chiradzulu Programme inMalawi. MSF/Malawi. MEFMI,the SouthAfrican Debt Office, andthe World Bank (2000): "Case Study: Sovereign Debt Management, Cash Management & Domestic Debt Development Malawi". MinistryofHealth (MOH) (2004). MinistryofHealth(March2004). A Joint Programme o f Work for a Health Sector Wide Approach (SWAp), 2004-2010. Department o f Planning. MinistryofHealth (MOH) (2004). Treatment ofAIDS: The 2-Year Planto Scale Up antiretroviral Therapy inMalawi, 2004-2005. Lilongwe. Ministry o f Health and Population (February 2002). The Malawi Essential Health Package.Annex 3: Summary o f February 2002 Dissemination Meetings. Lilongwe. Ministry o f Health and Population (November 2003). Malawi National Immunization Programme: Financial Sustainability Plan. Lilongwe. MOEST, 2000. Policy & InvestmentFramework 2000-12 (PIF). Government o f Malawi. MOEST. EMIS school census: Basic Education Statistics for the years 1990-2003. M o HPOW,2004 M O H has collaborated with its development partners to develop a Program o f Work (POW) intendedto finance the delivery o f an essential health package (EHP) under a sector-wide approach (SWAp). 144 MOH/DELIVER Project (October 2003). Health Commodities Logistics Management SystemProceduresManual. Fundedby USAID. National AIDS Commission (NAC) (Jan. 2004). HIV and AIDS inMalawi. Republic o f Malawi. NationalAIDS Commission (NAC) (Nov. 2003). HIV Sentinel Surveillance Report. Republic o fMalawi. NationalMicronutrient Survey (NMS) 2001. Report o f the NationalMicronutrient Survey 2001. Lilongwe, Malawi. MinistryofHealthand Population (MOHP) Malawi. Unpublished. National Statistics Office andMeasure/DHS (June 2005). Malawi Demographic and Health Survey 2004: Preliminary Report. Zomba, Malawi and Calverton, Maryland, U.S.A. Ndyanabangi, B.A.,M.Makuti, A. Gonani, A. Macheso, S. Kinoti, and S. Shongwe (, 2004). The Impact o f HIV/AIDS on the Health Work Force inMalawi. XV International AIDS Conference. Barcelona, Madrid. NEC/NSO/IFPRI (2001): The Determinants of Poverty inMalawi, 1998.National Economic Council//National Statistical Office inMalawi and the International FoodPolicy ResearchInstitute inWashington DC. NSO (2005): Malawi Public Expenditure Tracking Survey inEducation, Pilot Study 2004. Zero Draft. O'Donnell, Owen et a1 (2005). "Who benefits from public spending on health care in Asia?" Equitap Project: Working Paper #3. Patterson, David, Allison Zakariya, and Maggie Kambalane (2003). Final Health Facilities Assessment inBalaka District, Malawi. Partnership for Equity, Access, and Quality Project. Save the Children U.S. Pearson, Mark (2002). "Benefit incidence analysis: how can it contribute to our understanding o f health systems performance?". Department for International Development: Health Systems ResourceCentre. Phuka, J., S. Taulo, M.M.Faye, and A. Mazick (April 2003). Health Care Waste Management and InjectionSafety: Assessment Report. Department o f Preventive Health. Ministryo f Healthand Population. Lilongwe. Quick, JonathanD. (November 2000). DrugFinancing Strategies. Slide presentation. World HealthOrganization. Essential Drugsand Medicines. Geneva. 145 Rashidi, Tambudzai (2003). Deployment o f Reproductive Health Care Personnel: The Case for Malawi. Presentation during the Leadership Forum, JHPIEGO. Baltimore, Maryland. December 4. Republic o f Malawi (2005) "Malawi HIV/AIDS Monitoring and Evaluation Report" Office o fthe President and Cabinet. Stanecki, KarenA. (July 2002). The AIDS Pandemic inthe 21st Century. UnitedStates Bureau o f Census. Staple, Anne L.Martin (May 2004). Proposed 6-Year HumanResource Relief Programme for the Malawi Health Sector. Duke Center for International Development, DukeUniversity. Stover, J., and M. Fahnestock. (2006) "Coverage o f Selected Services for HIVIAIDS Prevention, Care, and Treatment in Low- and Middle-Income Countries in2005" Washington, DC: Constella Futures, POLICY Project. Syroka, J. and Wilcox, R. (2006) Rethinking International Disaster Aid Finance. Journal o f International Affairs, "The Globalizationo f Disaster", Volume 59, 2, Spring/Summer 2006. Tessa Tan-Torres Edejer, Moses Aikins, Robert Black, Lara Wolfson, Raymond Hutubessy, David B Evans (November 2005): Achieving the millennium development goal for health. Cost effectiveness analysis o f strategies for child health indeveloping countries. BMJ, doi:10.1136hmj.38652.550278.7C. UNAIDS(2006a) "Report onthe GlobalAIDS Epidemic" UnitedNations, Geneva. UNAIDS (2006b) "Progress on Global Access to HIV Antiretroviral Therapy" United Nations, Geneva. UNDP and the Government o fMalawi (February 2002). The Impact o f HIV/AIDS on HumanResourcesinthe Malawi Public Sector. Fundedby UNDP. van de Walle (1996). "Assessing the welfare impacts o f public spending". World Bank Policy ResearchWorking Paper 1670. Van der Veen, Fred(July 2002). Assessment o f Existing STI Care Services and Recommended Strategies to Improve STI Care for SelectedTarget Groups. Family Health International. RegionalHIV/AIDS Program. Fundedby USAID. Wagstaff, Adam (2001). "Poverty and health". Commission on Macroeconomics and Health:Working Paper WG1.5. I46 WFP (2005) Pilot Development Project: EthiopiaDrought Insurance. Available online at: htt~:llwww.wfp.or~/eb/docs/2005/~~077056-5.pdf Whitworth, Jamilah (2005): Update o f the education chapter o f the public expenditure review 2000. Final draft. DFID, Malawi. World Bank (2002): Poverty and Nutrition inBolivia. A World Bank Country Study. World Bank (2003): A Chance For Every Child. The other two groups were: 1. Relatively successfulcountries (GER>85%, PCR>70%), and 2. Low coverage countries (both GER and PCR<60%). World Bank (2003): Bruns,B., Mingat, A,, Rakotomalala, R.:Achieving Universal Primary Education by 2015.A ChanceFor Every Child. Data are from between 1997 and 2000. World Bank 2003, Implementation CompletionReport for the Third Fiscal Restructuring and Deregulation Program (FRDP 111).Report N.26151. World Bank, Washington DC. World Bank (2004). Malawi Education Country Status Report: Cost, Financingand School Effectiveness inMalawi -A Future o f LimitedChoices and Endless Opportunities. World Bank (2004a). Malawi Education Country Status Report: Cost, Financing and School Effectiveness inMalawi -A Futureo f LimitedChoices and Endless Opportunities. Africa RegionHumanDevelopment, Working Paper No. 78. World Bank (2004b): Project Appraisal Document. Malawi Education Support Project 1. World Bank 2004 Country Status Report ineducation, World Bank, 2005:Education inEthiopia-Strengthening the Foundation for Sustainable Progress. World Bank/IMF 2006, HIPC completion point document World HealthOrganization (2003). Improvingthe Affordability and Financing of Artemisinin-Based Combination Therapies. Malaria Control Department. Geneva. 147 148