Report No. 40027-MW Malawi Social Protection Status Report October 4, 2007 Sustainable Development Network, AFTCS Country Department 2, Malawi Africa Region Document of the World Bank Malawi Social Protection Status Report TABLE OF CONTENTS ACKNOWLEDGEMENTS ............................................................................................... ... 111 EXECUTIVE SUMMARY.................................................................................................. v 1 INTRODUCTION: SOCIAL PROTECTION ONTHE AGENDA INMALAWI....1 . Rationale......................................................................................................................... Aim ofthis rep0rt............................................................................................................ 1 3 4 Report approachand outline ............................................................................................ Limitations o f the stocktake ............................................................................................ 6 2. 7 Poverty and vulnerability inMalawi................................................................................ PROFILE OF POVERTY. RISKAND VULNERABILITY ...................................... 7 10 Poverty and vulnerability at household level ................................................................. Where are the poor and vulnerable?............................................................................... Poverty and Vulnerability Profile.................................................................................. 13 18 3 INVENTORY OF SOCIAL PROTECTION INTERVENTIONS ............................ . 21 Background and approach to the inventory.................................................................... 21 Project types and beneficiary coverage .......................................................................... 23 Objectives and target groups ......................................................................................... 26 28 Implementing agencies and period of implementation................................................... Geographic coverage o f safety nets interventions .......................................................... 30 Intervention costs.......................................................................................................... 32 4. Geography o f poverty. risk and vulnerability................................................................. LINKINGTHE INVENTORY TO THE PROFILES............................................... 37 37 Inventory and needs of poor and vulnerable groups ....................................................... 38 5 INSTITUTIONAL ARRANGEMENTS .................................................................... . 41 41 Institutional arrangements for safety nets and social protection ..................................... Key issues and questions ............................................................................................... 42 Progress: Towards a strong lead and institutional home................................................. 45 Remaining challenges for institutionalizing socialprotection......................................... 46 Social protectionmonitoring strategy ............................................................................ 51 6 BROADPROGRAM OPTIONSAND AFFORDABILITY OF . SOCIAL PROTECTION............................................................................................ 53 7 CHALLENGESFOR SP PROGRAM FORMULATION: CONCLUSIONSAND . RECOMMENDATIONS ............................................................................................ 62 References .................................................................................................................... 69 Appendix 1. Types of poor households.......................................................................... 72 1 LIST OFFIGURES.BOXESAND TABLES Box 1.1The emergence of social protection inMalawi ........................................................ 2 9 Box 3.1 Key M P V A findings on project types.................................................................... Box 2.6 Matching orphan status against an income profile ................................................... 25 Box 5.1 Examples o f donor coordination andharmonization inMalawi ............................. 47 Box 5.2 Social Protectionat DistrictLevel: Opportunities and Challenges ......................... 49 55 Box 6.2 Honduras Community based IntegratedChild Care Program................................. Box 6.1 Recommendations from the MPVA ...................................................................... Box 7.2 Social ProtectionFramework: GuidingPrinciples ................................................. 57 66 Table 2.2 Poverty and populationby region ....................................................................... Table 2.1 Humandevelopment indicators............................................................................. 8 10 Table 2.3 District poverty rates .......................................................................................... 11 Table 2.4 Intra-annual vulnerability by district ................................................................... Table 2.5 Vulnerability within districts .............................................................................. 11 12 21 Table 3.2 Project types 2003-2006 ..................................................................................... Table 3.1 A summary of `social protection' inMalawi....................................................... 24 25 Table 3.4 Interventions and target groups........................................................................... Table 3.3 Proportion o f households that reported benefiting............................................... Table 3.5 Number o f projects and related expenditure by district ....................................... 26 29 30 Table 3.7 Estimatedcosts of social protection costs in2003-2006 ...................................... Table 3.6 Implementers o f safety net interventions............................................................. 33 Table 3.8 Costs per beneficiary, district and project ........................................................... 34 Table 4.2 Poverty, vulnerability, population share and project concentration..................... Table 4.1 Poverty, vulnerability and regional coverage o f CCT and SSP............................ 37 -37 Table 4.3 Needs and current and potential projects............................................................. 40 Figure2 Relative share of social protection components .................................................... Figure 1Poverty. riskand vulnerability profile for Malawi ................................................ 20 24 Figure3 Cost per vulnerable person/ district .................................................................... 29 30 Figure 5 Period o f implementation ..................................................................................... Figure 4 Cost per poor person/ district............................................................................... 32 Figure 6 Financiers intotal SP cost .................................................................................... 35 43 Figure 8 Institutional arrangements for Social Protection 2006........................................... Figure 7 Institutional arrangements for Safety Nets 2003-2006 .......................................... 44 Figure 9 The scope o f social protection .............................................................................. 58 .. 11 ACKNOWLEDGEMENTS This status report provides a stocktake o f social protection in Malawi over the period 2003- 2006, and aims to help Malawi move towards a long-term social protection policy and program. The report examines the range, goals, and coverage o f existing social protection interventions, and how this matches the profile o f poverty, risk and vulnerability. Moreover, the report provides an assessment of the current institutional and financial arrangements for social protection, and suggests options for a more coherent national social protectionprogram. This report has been carried out jointly by the Government o f Malawi and the World Bank, and with the active involvement o f a group o f development partners providing support for social protection in Malawi, including DFID, UNICEF, WFP, European Commission, Norway, USAID, and representatives o f civil society. The report was prepared under the general guidance o f Bright Msaka, Chief Secretary, Office o f the President and Cabinet and Chairperson o f the National Social Protection Steering Committee (NSPSC), Frank K. Byamugisha (Acting Sector Manager, AFTS1, World Bank) and IanBannon (Sector Manager, AFTCS, World Bank). Tim Gilbo (Country Manager, World Bank), Michael Baxter (Country Director, World Bank) and Tijan Sallah (Lead Specialist, World Bank) also provided direction and guidance for the project. The report was prepared through a joint government-development partner process under guidance by a team o f government and World Bank officials led by Ms. Lillian D. Ng'oma, Secretary & Commissioner Department of Poverty and Disaster Management Affairs (DoPDMA), Office of the President and Cabinet, and Trond Vedeld (AFTCS, Task Team Leader, World Bank). The key authors o f the initial report were Rachel Slater, Overseas Development Institute (ODI) and Maxton Tsoka, Centre for Social Research (CSR) (consultants). The peer reviewers were A. Waafas Ofosu-Amaah, Will Wiseman, and Renos Vakis (all World Bank). The report benefited greatly from comments by members o f the National Social Protection Technical Committee (NSPTC); special thank to Harry Mwamlima inthe SocialProtectionUnitfor extensive andvaluable comments. BenBotolo, Director M&E, provided additional guidance as the Social Protection Unit was relocated to the Ministry o f Economic Planning and Development. Other core staff o f the Social Protection Unitproviding valuable input were A. Mzoma, Coordinator, and K.Kaphaizi, Coordinator. The report has also benefited from comments and valuable input from core donor representatives, including Sharon Kinsley (DFID), Mulle Chikoko (DFID), Mayke Huijbrechts (UNICEF), Bernd Schubert (consultant), Don Scalpelli (WFP), Dominique Blariaux (EU), and Oystein Botillen (Norway) - The World Bank team involved in developing the report and translating findings into the broader social protection policy process included Hardwick Tchale, Azedine Ouerghi, Antonio Nucifora, Nginya Mungai Lenneiye, Chrissie Kamwendo, and Wim Alberts. The report benefited greatly from additional analysis on labor and labor scarcity by Kathleen Beegle (DEC, World Bank). Comments were also provided by Harold Alderman, Khwima Nthara, David Rohrbach, and Paul Siege1(all World Bank). The work has also benefited from input by StephenDevereux and other professionals linkedto RHVP. Support from World Bank Institute (WBI) hasbeenprovided by Kaleb Tamiru and Danielle Carbonneau. 111 The report has received comments through two dissemination workshops with various stakeholders, and contributions from a wide range of stakeholders, includingmembers o f civil society and beneficiaries of social protection programs at household level. Meseret Kebede and Sharon Dotto Abu have undertaken the desk-top editing o f the final document, and provided excellent administrative support throughout the process. Such assistance was also provided by Annie Jere and Grace Soko in various aspects o f the administration o f the project. The work was funded by the World Bank with financial support from the Norwegian-Finnish Trust Fund for Environment and Socially Sustainable Development. This volume is a product of the staff of the InternationalBank for Reconstruction and Development/ The World Bank. The findings, interpretations, and conclusionsexpressedinthis paper do not necessarilyreflectthe views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guaranteethe accuracyofthe data includedinthis work. The materialin this publicationis copyrighted.Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The InternationalBank for Reconstruction and Development/ The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete informationto the Copyright Clearance Center, Inc., 222 RosewoodDrive, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750-4470, httu:/lwww.couvright.com/.All other queries on rights and licenses, including subsidiary rights, should be addressedto the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA, fax 202-522-2422, e-mailpubrights@worldbank.org. iv EXECUTIVESUMMARY Malawi is in the process of moving away from safety nets programming towards more long- term predictable social protection programming that helps poor households deal with risk and shocks through a more institutionalized and coordinated approach. There are significant changes underway to transform the way that social protection i s designed, funded and implemented inMalawi. Social protection is on the agenda o f Government, donors and NGOs in Malawi, and a National Social Protection Steering Committee and National Social Protection Technical Committees has been established. I n the period 2003-2006, including emergency aid and disaster response, the combined safety netshocial protection system amounted to an average of more than USD 134 million per year. There are, therefore, compelling reasons to ensure that the system for delivering social protection i s efficient and effective. This report provides a stocktake of social protection in Malawi for the period 2003-2006, and, in partnership with the development of a Malawi Social Protection Framework, aims to help Malawi move towards a long-term social protection policy and program. The report answers two specific questions: i) Do the range, goals and coverage of existing social protection interventions (inventory) match up with the existingprofile of poverty, risk and vulnerability? ii)Dothe current institutional and financing arrangementsmatch up with the need for institutionalized social protection in Malawi? The work feeds into a final section on options for a more significant and coherent national social protection program. For the purposes o f this report social protection i s defined as comprising: The policies and actions that protect and promote the livelihoods and welfare of poor and vulnerablepeople It is recognized that social protection overlaps with social policy, economic policies, and disaster reduction, and should not be considered as a stand alone policy. Profile of poverty, risk and vulnerability in Malawi Poverty in Malawi is widespread andpervasive, and over half(52 percent) of thepopulation - 6.4 million people in 2005 - live below the national poverty line, and one in five (22 percent) are defined as ultra poor, with income below expenditure of minimumfood requirements. The overall number o f poor has remained fairly stable over the last decade. Vulnerability seems to be increasing inrural areas, largely due to repeated shocks and depletion o f assets. On a district by district basis, poverty tends to increase southwards in Malawi but there are significant pockets of high poverty levels even in the north and pockets o f deep poverty in districts with low overall poverty levels. Rural poverty i s more pervasive than urban poverty but urban poverty i s growing faster. Proxies to identify poor households are difficult to establish but, drawing on the Malawi Poverty and Vulnerability Assessment, the stocktake suggests a focus on the following types o f households for formulating policies and approaches to social protection (a more detailed overview o f these groups i s provided inAppendix 1): V i) Ultra-poor households (including destitute) are usually larger than average households and with higher dependency ratio, and characterized by a lack of assets, especially limited landholdings (land per capita). Within the ultra-poor group i s a sub-group o f households that can be categorized as destitute. These are ultra-poor households that have the least assets and land (per capita), have no labor and / or highdependency ratios, for example due to chronic illness. Poor female headed households and elderly and child headed households would be more frequent inthis category; ii)Poorhouseholds ormoderatelypoor householdsarehouseholdsthatfallabovetheultra- poverty line, but are below the poverty line. They have some assets (land and labor) but are exceedingly vulnerable to further impoverishment. These.households are also heavily dependent on agriculture (or fishing) and are net consumers o f food; iii)Transient or at risk households are mostly found above the poverty line but are still vulnerable to adverse effects o f shocks and stresses, and may easily fall into poverty/ultra poverty. Inventory of Safety Nets in Malawi The current mix and outreach of thepresent safety nets system do not match up with theprofile of thepoor and vulnerable and their critical needs. The inventory shows that socialprotection is dominated by short-term term relief or emergency responses, the most extensive program being subsidizedyreefood distribution, and subsidized agricultural inputs. There have also been important investments inpublic works and school feeding. There are very few long-term developmental programs that are fully funded. The level of funding for different projects is not necessarily inadequate, but many projects do suffer from limited beneficiary coverage, mis-targeting and signiJicant leakages (especially the major transfer programs). Some o f the projects have less direct relevance for social protection as defined inthis report. Most programs are small and resources are spread thinly among many target groups. Some instruments are wrongly targeted, or there are various inclusion or exclusion errors (for example agricultural inputs provided for landless households). Different stakeholders recognize a very broad range o f poverty and vulnerable groups, creating a confusing overlapping o f target groups. Institutionalarrangement New institutional arrangements in Malawi give social protection a strong lead actor and institutional home. Under these new arrangements social protection will be largely government-owned and government-driven as long as donors can be better coordinated and harmonies their policy positions on social protection firmly behind government. The Social Protection Unit remains under-resourced, however, although very recently some progress has been made to improve staffing levels and technical support. Even so, the stocktake finds that the progress made so far by the SPU and committees to strengthen leadership and government ownership o f social protection has been critical. The vision o f the SP in Malawi i s to have separate, ring-fenced, basket funding for SP interventions and suggestions for institutional options are presented. The roles of district assemblies and NGOs in a decentralizedsocial protection program should be clearly thought through. Moreover, the extent to which NGOs or local governments can successfully ensure that the voices o f poor and vulnerable groups are heard inthe new National vi Social Protection Steering and Technical Committees i s not yet clear. It is also unclear whether working through all-party parliamentary committees can ensure a long-term commitment to social protection that i s not threatened by any future changes in government or electoral processes. Mainconclusionsof the mappingexercise Clearer dierentiation is needed between emergency responses and longer-term developmental social protection that seeks to enable / smooth consumption, and both protect and promote assets; and more efforts are required to understand risks and enable risk management and disaster reduction at different levels; `Clearly defined boundaries are required to identijj for what constitutes `social protection' beyond the fuzzy definitions that are currently used in different programs, and that frequently make socialprotection indistinguishable from `poverty reduction' or `development' programs; There is a needfor the government, through thepolicy design process, to determine how broad the scope of social protection should be, and howfar up thepoverty profile they want different kinds of instruments to reach. A decision should be taken on whether households that are labor scarce generally are poorer or not, and whether cash transfer can work across groups (labor scarce or not); Social protection programming should be evidence-based and draw on solid data such as the 2005 IHS2; Poverty and vulnerability have clear gender and generational dimensions, hence social protection policies need to be built on gender and age-disaggregated evidence, and be sensitive to gender inequalities; All key actors (governments, donors and NGOs) need to align themselves around a common language and understanding of social protection so that we can better understand who i s playing what role and to what end inthe existing system. Analysis of institutionalarrangements The changing institutional structure of the social protection unit and various committees has placed social protectionfirmly at the centre of the government's development agenda: Social protection policy is cross-cutting and multi-sectoral and requires inputsfrom different line ministries but the secretariat for social protection should sit in a non-line ministry I department. Social protection should not be a stand alone policy; Capacity issuesfor social protection need to be addressed both within the Social Protection Unit, and at district level to enable effective service delivery; Mechanisms for delivering social protection at district level need to be more clearly articulated; Systems of targeting of various instruments need to be continuously refined (now mainly through two or three steps community targeting); vii Effective monitoring and evaluation systems (in partnership with MoEPD) are critical if cost- effective social protection i s to be delivered by government; Thepolitics of socialprotection need to be managed, and not ignored; A key challenge is to determine the more precise mix of instruments (adjusted to the poverty and vulnerability profile), what funding levels would be affordable and appropriate, what trade offs are involved between investment in social protection versus investments in other related sectors, and what implications the new program would have for different development partners; Thesefindings resonate strongly with (and have, in part, guided) the principles and goals of socialprotection presented in the Malawi Social Protection Framework. Social protection program options and affordability Funding for social protection is considerable; the annual expenditure of about USD 134 million amounting to about 6,5 percent of GDP, 15 percent of annual budget, and about 34 percent of total development assistance. This i s significantly higher than previous estimates, and provides compelling reasons for improving the system. Options for a new national social protection program for Malawi should involve a new mix o f instruments, and introduce as the main new element a social transfer program for a majority o f the ultra poor through a combination o f e.g. cash transfershocial transfers and livelihood enhancing instruments. An overall goal would be to reduce ultra poverty and prevent moderately poor households from falling into ultra poverty. Such a program would also provide substantial additional support for poor, transient and vulnerable poor with priority to the 30-40 percent poorest, to promote livelihoods through productivity enhancing transfers that improve risk management, and limit people falling into ultra poverty. A social transfer program to reach the 10 percent poorest would cost in the order o f USD 40 millionannually, or about 2 percent of GDP, 4,5 percent o f the annualbudget, and 10percent oftotal development assistance. There appear to be two optionsfor a National Social Protection Pilot /Program that emerge from the stocktake related to a single instrument or a multi-instrument approach: One i s a social cash transfer pilot - based on experience in Mchinji with a single instrument approach. A second complementary option would be a multi-instrument approach; which is a more difficult and complex one, but potentially more valuable for learning and building a long-term and broader program adapted to the needs o f different groups. ... Vlll 1. INTRODUCTION: SOCIAL PROTECTION ON THE AGENDA INMALAWI Rationale Social Protection (SP) i s on the agenda o f Government, donors and NGOs in Malawi for a numberofreasons: 9 There is increasing recognition that Malawi is trapped in a vicious cycle of emergency appeals, reflecting a deep seated and complex crisis. The long-term problem of chronic and predictable hunger i s treated as if it i s an unpredictable emergency. To break out o f this cycle requires predictable resources, recognition o f risks and potential shocks, and more conscious efforts in risk management; ii) Recent years have seen signijkant innovation in managing risk and vulnerability, both within Malawi, in Africa and internationally, and there i s much that can be learnt from new instruments - for example weather-based risk insurance, long-term productive safety nets and cash transfer programs in other parts o f the world - but there i s a need to think carefully about how to apply innovative instrumentsinthe Malawian context; iii)It is also clear thatperceptions of socialprotection are changing. Box 1.1 shows how safety nets and social protection have emerged in Malawi. More recently, changing institutional structures for social protection and the incorporation o f a pillar on social protection within the Malawi Growth and Development Strategy (MGDS) demonstrate that the Government o f Malawi i s serious about the potential for social protection to transform the lives o f poor people in Malawi. To this end it has established the National Social Protection Steering and Technical Committees and the Social Protection Unit (the former Safety Net Unit); iv) Donors are committing considerablefunding to social protection, but until there is a J;rm government commitment to social protection, they are reluctant to invest funds in social protection for periods of more than one or two years. NGOsare starting to review their own roles insocial protection. Whilst there is significant funding for NGOs engaging as implementingpartners in emergency humanitarianrelief, and NGOs have become skilled at delivering this ina timely and cost-efficient way, funding for long-term social protection i s scarce and NGOs have less demonstrated capacity in this area. There are also emerging questions about the roles that NGOs mightplay in long-term social protection - either as implementing partners where government capacity i s low, or as representatives o f poor people via civil society movements that lobby and play advocacy roles to influence government policies and programs. There is further evidence that key stakeholders have social protection on the agenda. Examples of these efforts are as follows: i) Government and stakeholders' analysis of safety nets whereby safety nets interventions were matched against the needs of actual and potential target groups; 1 ii)Governmentandstakeholders'mappingofinstitutionalstructuresforthecoordination and implementation o f an SP program; iii)Governmentandstakeholders'developmentofadraftSPpolicy; iv) Development partners funding o f independent analysis o f safety nets operating inthe country inorder to learn lessons for the development o f an SP program; v) Piloting o f some innovative safety nets or social protection approaches, for example, combining cash and food transfers to the food insecure, delivery o f benefits using vouchers, use o f an integrated approach in addressing the various needs of target groups and cash transfers; and vi) Increased training and capacity buildingof government staff. Box 1.1:The emergence of social protection inMalawi In Malawi, social protection had its formal origins in the Poverty Alleviation Program (PAP). PAP included the provision o f safety nets, in order to address the social dimensions o f structural adjustment programs, as one o f the strategies. The second stage came with the National Safety Nets Strategy (2000) and, subsequently the National Safety Nets Programme (2002) linked to the Malawi Poverty Reduction Strategy (2002) in which social protection was one of the four pillars. The third stage emerged in 2006, with the provision of social protection as one of the five themes in the Malawi Growth and Development Strategy (MGDS). These three `stages' also saw progressive changes inthe institutional structures that show increased seriousness on the part o f Government. From a peripheral PAP Coordinating Unit housed in National Economic Council (NEC) and operating under a presidential council on poverty alleviation, safety nets coordination moved to a Safety Nets Unit housed in the mainstream Ministry of Economic Planning and Development (originally NEC) and later Department o f Poverty and Disaster management Affairs (DPDMA) in the Office o f the President and Cabinet (OPC) and operating under a more formal steering committee on safety nets. Currently social protection i s coordinated by an SP Unit operating under a formal but an elevated National Social Protection Steering Committee whose chair i s Principal Secretary o f the OPC and the head o f the civil service. The space for the involvement o f the civil society has gradually improved, and their role expanded inthisperiod.' The stocktake i s one amongst several key activities underway to formalize a new Social Protectionpolicy and a national program inMalawi. ' Most recently, civil societyhas startedto review its role as both implementersand advocates. Inparticular, it i s strategizingon how to use the space createdby Government in the SP institutional set up (or lobby for more space). 2 Aim of this report This stocktake, alongside the development o f a Malawi Social Protection Framework, i s one o f the first steps in a process through which an SP policy statement and later national program will be developed.* The stocktake and framework development are informedby the activities around the social protection agenda, including lessons from stakeholder efforts in the recent past. The stocktake aims to derive lessons for future social protection deliveryby i)Developing a clear understandingofthe current system andwhat itisable, and unable, to deliver; ii)Mappingthecurrentrangeofsocialprotectionapproachesandactivitiesbydifferent stakeholders; and iii)Understandingthe current institutional and financial arrangements for delivering social protectionprojects andprograms. The report provides general policy guidance and broad institutional options, not detailed assessments o f individual instruments and programs, or detailed proposals for new institutional and financial arrangements. The main audience for the report i s government officials and development partners involved in social protection policy design or program development. The research undertaken for the production o f this report had the following tasks: i)Provideasynthesisdiagnosisoftheprofileofpoverty,vulnerabilityandrisk; ii)Develop an inventory of existing (recent) safety net interventions andassess the extent to which the profile of poverty and vulnerability is integrated with safety net coverage; iii)Assess the strengths andweaknessesofthecurrent institutionalarrangements that govern social protection (policy, planning, management and coordination); iv) Assess the cost and financial implications of the current system; and v) Identify areas for improvements and new options with reference to relevant international experience on social protection. The report i s closely linked to the development o f a framework for social protection that provides core principles for a subsequent social protectionpolicy and program (see below). Methodology The stocktake draws mainly on secondary data complemented by information drawn from interviews with key stakeholder^.^ These are supplemented by three Government databases: safety nets projects database under the Safety Nets Unit in the DPDMA, donor-funded The stocktake i s funded by the World Bank as agreed support to the Government of Malawi. The main sources include the most recent syntheses of safety nets most notably Kambewa (2005), Devereux and Macauslan(2006) and Devereux (2006), and Malawi Poverty and Vulnerability Assessment (GoMNorld Bank, 2006), `Poverty in Malawi - The Second Integrated Household Survey 2005: An Extract of Findings' (MEPD/NSO/WB, 2005) and Statistical Abstract of IHS2. Other main sources include projections from the 1998 Population and Housing Census (www.nso.malawi.net) and Malawi Vulnerability Assessments (www.fews.net)). Some of the stakeholders provided project-related reports and fliers and these also proved very useful in filling gaps. 3 projects database under the Department o f Debt and Aid in the Ministry o f Finance (MoF) and food security-related projects database under the National Joint Food Security Task Force inthe MinistryofAgriculture andFoodSecurity (MoAFS). The list of projects inthe stocktake reflects the broad and inclusive SP definitionusedby the SPU, i.e.: Social protection comprises thepolicies and actions that protect and promote the livelihoods and werfare of poor and vulnerablepeople. The adoption o f the broad definition enables the matching o f different target groups (i.e. those inpoverty and at risk of poverty) with the poverty and vulnerability profile. The stocktake i s interested in a number o f characteristics o f each project. These include its name, objectives, type, target group, type o f transfer and size, number o f beneficiaries, start and end dates, geographical coverage, costs (total and delivery), implementer and funding agen~ies.~ The stocktake team first occupied itself with collecting data for these characteristics. Some o f the characteristics for existing projects were in the two syntheses. However, some gaps that still existed were filled using the two other databases as well as project documents and interviews with implementers and other stakeholders. The MoF database was mainly used to fill cost gaps while the MoAFS database is used to fill cost as well as other gaps. Some interventions that were not in any o f the databases were added to the existing list usingprojects files and interviews with implementers and other stakeholders. An inventory of `existing safety net interventions' was defined as interventions that were operational at some time intheperiod 2003-2006. The list o f interventions that was collected, including those with gaps, i s provided in a separate overview (available with the World Bank). In some cases the project titles used in the stocktake may not be exact because they were not taken from official project documents. The number o f projects used inthe analysis o f each characteristic i s dependent on the number o f projects with the given characteristics.' The costs of interventions were mainly recorded in three currencies United States Dollars, - Euros and Malawi Kwacha. All costs were converted into United States Dollar using an internet-based foreign currency converter (www.oanda.orq) based on the start and end date o f the project. Projects without start month were assumed to have started in July o f the stated start year and ended inJune inthe stated end year, inline with the Malawi Government fiscal year. Likewise, projects indicated to have started and ending in the same year but without start and end months were considered to have runfrom July to December o f the year. These assumptions may make the calculations less accurate but without them it would not be possible to estimate costs/expenditures. Limitationsof the stocktake The stocktake benefited from a lot o f goodwill with many stakeholders providing significant quantities of data and information. Despite this there are a number o f limitations that need to be highlightedto putthe findings intheir right perspective. These are given below: The stocktakebuilds upon andupdatesthe inventoriescompiledby Kambewaand Devereux and Macauslan (2006). Some small-scaleNGO projectswere droppedduring analysis, inpart becausethere was not enough information available aboutthem. Similarly, the stocktake teamis aware that there are additional interventions implementedby NGOs over the period 2003-2007. 4 i)Thequalityofthematchbetweentheprofileandsafetynetstargetgroupsisaffected bynon-standard definitions o f target groups. Further, the safety nets under discussions operationally used the 1998 poverty profile as a basis, and not the 2005 poverty and vulnerability profile. Mismatches are likely and may be due to structural changes since 1998. However, given that each intervention assessed its target group before implementation (and not on the basis o f the 1998 IHS), we expect the 2005 profile to be more relevant for the 2003-2006 period; ii)Itisnotpossibletoproduceanexhaustivelistofinterventionsbeingundertaken.The main database by SPU has not been updated since early 2005. Further, the stocktake team was unable to visit all stakeholders and efforts to get informationby e-mail were also unsuccessful because not all implementers responded. Some small-scale and some new interventions have been omitted, but the effect o f these omissions i s likely to be minimal; iii)Robustanalysisishamperedbythereportingformatsofsomeprojects. Forexample, some interventions are multi-annual and with more than one safety net activity yet without disaggregated information by year and activity. This limits the analysis by year and activity. Other interventions are multi-donor and implemented by a consortium yet the reporting i s not disaggregated by donor or implementer. Again, this limits analysis by donor and implementer. Moreover, the social protection function o f ADMARC has not been included inthe analysis - since it was not part o f the original safety net strategy. This relates to its important role in providing subsidized foodgrain, as well as its social functions in relation to inpub'output marketing; iv) Where gaps existed, they were filled usingalternative sources. However, the different sources are not always consistent with one another; v) There is some information about geographical coverage at regional and district level but this is also patchy. This is due to the absence of project documents for most of the interventions. Analysis on intra-district coverage was not possible because there i s no accessible data onproject coverage by traditional authority (TA) areas; vi) It has been difficult to track exact implementation periods for programs (with subsequent implications for the analysis of costs). Inparticular, it has beendifficult to get start and end months for a good number of interventions. In such cases, we were forced to make assumptions as already indicated inthe methodology. This means that for some projects the months o f implementation are overestimated and for others underestimated. We assume they balance out; vii)In some cases, data on actual project costs are not available. In such cases we have used budgeted costs. Inmany cases, the costs exclude the cost o f delivery which is a major problem when comparing costs among different kinds o f instruments and projects6; ~~~~ Inothers, there is more than one cost figure for one project. Insuch cases, we take the highest.Insome cases, the cost of an intervention is in local as well as foreign currency. Insuch cases, the local currency figure i s used on the assumption that budgeting i s inU S Dollars but spendingi s mostly inthe local currency. 5 viii) The report provides general comments on the efficiency and effectiveness o f individual social protection programs, however, reference i s made to the MPVA for a more detailed assessmento f instruments(ref. chapter on social protection and disaster management). Report approach and outline The remainder o f this report seeks to address the issues outlined and each of the tasks outlined above. There are two critical questions for the study: i) Do the range, goals and coverage of existing social protection interventions (inventory) match up with the mistingproJile ofpoverty, risk and vulnerability? ii) Do the current institutional andJinancing arrangements match up with the need for institutionalized socialprotection in Malawi? Inorder to answer these questions, we beginbydescribingthe poverty, vulnerability andrisk profile in Malawi and by developing a profile in Section 2 against which the coverage o f existing interventions discussed in Section 3 can be matched. Section 4 matches the profile of poverty and vulnerability with the array o f interventions implemented as safety nets interventions. In Section 5, the current institutional arrangements for delivering social protection in Malawi are assessed. Section 6 presents broad program options in terms o f funding and directions for social protection, including lessons for the design and implementation o f social protection programs and pilots. Section 7 raises a set o f issues and challenges andprovides conclusions and recommendations. 6 2. PROFILEOF POVERTY,RISK AND VULNERABILITY Poverty and vulnerability inMalawi This section lays the foundation for understandinghow far existing safety nets in Malawi match up with poverty and vulnerability in Malawi. Understanding how appropriate the current system is, particularly how far it matches the needs o f poor and vulnerable people, will help inform the development o f a social protectionframework, policy and program. The section synthesizes the Malawi Poverty and Vulnerability Assessment (MPVA see GoM / World Bank 2006) and a number o f other sources. It deals with three main questions: who are the poor and the vulnerable, where are they and what are their basic pressing needs? Poverty in Malawi is widespread and pervasive. Over half (52 percent) of the population - about 6.4 million people - lives below the national poverty line, and one inJive (22 percent) are defined as ultra poor, with income below expenditure of minimum acceptable food requirements.' The overall number o f poor has remained fairly stable over the last decade. Poverty has both gender and generational dimensions. According to the MPVA, the prevalence o f ultra-poverty. i s higher for female-headed households, households headed by very young or old persons, located inthe rural areas o f the South and Central regions, larger households, and especially households with more young children and dependents, households with low levels o f education, limited economic opportunities, limited involvement in cash crops, and small landholdings. The gender dimensions o f poverty are reflected in the fact that: i)Povertyratesarehigherinfemale-headedhouseholds(58percentasopposedto51 percent for male-headed households) but in terms o f numbers there are more poor people inmale-headed than female-headed households; ii)Poorfemaleheadedhouseholdshavehigherdependencyratiothanpoormale-headed households; iii)Widowsarepoorerthanwidowers. Vulnerability seems to be increasing in rural Malawi, understood as susceptibility to shocks, largely due to repeated shocks and depletion of assets. The frequent and widespread existence o f shocks manifests itself into large movements into and out o f poverty inMalawi. Ninety five percent of the households in the national household survey (IHS2) reported exposure to at least one shock in the past 5 years. The most prevalent shocks relate to low crop yields (due to drought or floods) and increase in food prices (while fall insale prices for crops is also commonly mentioned; directing the focus to issues o fprice volatility). The poverty line o f MK16,165 / year i s equivalent to MK44.3 or USDO.50 per person per * deep. The average poor household subsists on an income o f around MK36.4 - 17.8 percent day, significantly below the `standard' USDl per day per person. Poverty inMalawi i s also below the MK44.3 daily poverty line while the ultra poor subsist on MK26.40, on average. '*This Malawi's populationwas estimatedat 12.3 million in 2005. includes imputedownproducedgoods andservicesandnon-foodpurchases. 7 Consumption inequality i s more pronounced inurban areas - the urban Gini o f 0.48 i s much higherthan the Malawiaverage o f 0.39 and the ruralaverage o f 0.34. According to the MPVA, poverty in Malawi is manifested through high mortality rates, low life expectancy, malnutrition and low education levels. Evenso, Malawi's child mortality and maternal mortality rates are better than average for SSA. Malnutrition is widespread among both poor and non-poor households and average caloric intake and dietary diversity are low for bothpoor and non-poor households. Chronic illnesses and HIV/AIDShave had a dramatic effect inreducinglife expectancy o f Malawians. See Table 2.1. On a district by district basis,poverty tends to increase southwards in Malawi but there are signijkant pockets of high poverty levels even in the north and pockets of deep poverty in districts with low overall poverty levels. Rural poverty i s more pervasive than urban poverty but urbanpoverty is growing faster. Source: GoMIWB 2006 The causes of this poverty are myriadbut two stand out: Limited livelihood sources and access to assets and services: most poor households earn their livelihood mainly from a single activity: household farming or fishing activity. Large seasonal fluctuations in agricultural production lead to severe fluctuations in consumption. Similarly, seasonality in time use leads to substantial underemployment for the most o f the year due to a dearth o f opportunities forced in part by limited access to transport infra~tructure.~Substantial portions o f the population remain away from services and markets; isolated from the rest o f the country both physically and in terms o f opportunities for diversified or more productive economic activity. Pervasive risks and high vulnerability to shocks." The frequent and widespread existence o f shocks manifests itself into larger movements into and out o f poverty. The very poorest households are risk averse and poor and richer households experience more shocks. But the poorest households may be very susceptible to particularly shocks especially crop failure. Most households have limited ex ante strategies to mitigate risks due to lack o f access to financial services and poorly functioning food markets which place a premiumon staples production. Households The seasonality of household labor requirements has ramifications for social protection options - especially those that require labor. loInthis regard, risks can be temporary orpermanent. They can also be idiosyncratic, affecting specific households (such as illness or death o f able bodied labor), or covariate, impacting communities and countries (like drought, price volatility, economic shocks) (ref. GoWWB 2006). 8 are therefore forced to resort to ex post coping mechanisms which often entail permanent damage to the household's ability to engage in productive activities (Source: G o M / World Bank 2006). Related to these two major causes o f poverty are dependency on rain-fed agriculture, limited access to land, input and producer markets and slow economic growth. It i s important to recognize that poverty and vulnerability are not the same. ' , Vulnerability implies the susceptibility o f individuals, households or communities to the negative impact o f shocks and stresses. This implies that households above the poverty line can be vulnerable to impoverishment ifthey face shocks and stresses that they cannot protect themselves against. It also implies that households below the poverty line can be vulnerable to destitution and death if they face shocks and stresses that they are not able to protect themselves against. The MPVA in this regard suggests that the social protection policy implicationof this two-sided poverty picture i s that social protectionpolicies are necessary to alleviate extreme poverty and assist the chronically poor to buildtheir assets. Given the role o f risk in preventing households Erom engaging in productive activities, social protection policies shouldbe redesigned to enhance the productivity and strengthen riskmanagement for the transient poor, protecting them against risks and economic shocks. The MPVA then highlightsthe keyimplicationsfor social protectionprogramming: i)Asocialprotectionprogram alonecannotbeleft todealwiththeneedsof thepoor (6.4 million in 2005) or even the ultrapoor (2.7 million). Social protection is but one o f the five themes requiredfor growth andpoverty reduction inMalawi; ii)Thedesignof theSPprogramsshouldhaveclearoperationallinkswiththefour other pillars of the Malawi Growth and Development Strategv (MGDS). An integrated approach is required to deal with pervasive poverty; iii)Equalizationofconsumption(or income) cannot be amajorobjectiveofasocial protection program but raising the consumption level o f the poor especially the ultra poor would be a progressive objective; iv) Social protection programming should recognize that Malawi households face the twin problems of limited sources of livelihoods andfrequent livelihood shocks. This implies that apart from the large pool o f the chronic poor there i s a large pool o f transient poor. These groups may require different types o f assistance; v) Nutrition programs on their own are likely to have limited impact on malnutrition since malnutrition is not only poverty-related. Pervasivemalnutrition across all social groups suggest a strong focus on addressing malnutrition across these groups; vi) A shift in focus from reducing ex-post poverty to ex-ante vulnerability is potentially cost-saving to poor people and the government. This view i s in line with the scope and spirit o f the MGDS. Moreover, the Malawi Social Protection Framework captures this within four key policy areas related to provision, prevention, promotion and transformative interventions. These four policy areas or outcomes are not clearly distinguishable categories, however. Each kind o f safety net or 9 social transfer type will have multiple potential outcomes, and will address these dimensions to different degree. For example, when poverty and vulnerability for some groups are rooted in social exclusion and marginalization and disempowerment, more so than to shocks such as drought or illness, the aim o f social protection would be to transform the living conditions o f those affected by addressing underlying causes (e.g. through social policies, upholding o f constitutional rights), while also provide or promote social transfer or productivity enhancing assistance. Where are the poor andvulnerable? Although poverty affects people everywhere in the country, it is predominantly rural (94.5 percent) and concentrated in rural Southern Region (49.7 percent) see Table 2.2. Southern Regionhas a greater share o f the poor -poverty and ultra-poverty incidences are higher there than anywhere else in the country. While the overall poverty incidence has not changed much between 1998 and 2005, there have been clear regional changes. Poverty incidence declined in the Southern Region from 68 percent to 64 percent but increased in urban areas from 18percent to 25 percent. Source: GoMlWB 2006 Poverty incidences at district level are diverse (and statistically uncertain) but still reflect the regional picture.l1The three districts with the highest rate o f poor are all in Southern Region (Nsanje - 76 percent, Machinga - 74 percent, and Zomba - 70 percent). The rural districts with the least relative number o f poor people inMalawi are all inCentral Region (Dowa -37 percent, Lilongwe - 38 percent and Kasungu - 45 percent). The urban area with the highest rate o f poor people i s Mzuzu (where 34 percent o f its population i s estimated to be poor) and Blantyre has the lowest rate (where 24 percent are considered poor). Table 2.3 presents the districts from the poorest to the richest. Districts in Southern Region are concentrated in the poorest category and Central Region districts inthe richest category. However, this pattern i s different when the focus i s specifically on vulnerability rather than poverty (for more indepth explanation of how vulnerability i s defined and measured, ref. MPVA). Some districts with low levels of poverty are highly vulnerable and vice versa. Much o f this i s related to the level o f diversification o f livelihoods systems: Non-poor but specialized districts (fewer sources of livelihoods at household level) can be more vulnerable than poor but diversified districts because o f the seasonality o f production and income. Table 2.4 shows the intra- annual dimension o f vulnerability. Districts with low vulnerability inthe first quarter, such as Ntcheu, may turn very vulnerable by the fourth quarter. Dependency on a single source o f "ItshouldbenotedthatIHS2povertyestimatesatthedistrictlevelhavefairlylowstatisticalaccuracy(seenext footnote). Hence, given the low accuracy, these district level estimates shouldnot be usedas abasis for policy design. 10 livelihood - especially agriculture where incomes are highly seasonal - i s the main cause o f this. Chikwawa South 65.8 Karonga North 54.9 Mzuzuurban North 34.0 Thy010 South 64.9 Dedza Centre 54.6 Zombaurban South 28.7 Lilongwe Chiradzulu South 63.5 Ntcheu Centre 51.6 urban South 24.6 NkhataBay North 63.0 Mzimba North 50.6 Blantyreurban South 23.6 Differences in vulnerability also operate at a sub-district scale. Table 2.5 shows that some districts have diverse zones, and therefore different food security risks, while others are more homogenous.l3 12District levelpovertyestimates haveahighdegree of uncertainty surroundingthem (`high standard errors'); 95 percent confidenceintervalsaroundthese estimates are large, suchthat individual district povertyrates could beupto 20 percentabove or belowthe estimates inthe table. l3Note that diversity of zones within a district does not automaticallyimply districtresilienceto shocks and risks.For example, Machingahas sevenzones and all ofthemhavehigh levels of vulnerabilitywhile Rumphi, with only one zone, has relativelylow vulnerabilitylevels. 11 Blantyre 36.2 52.0 52.0 77.1 Zomba 17.7 33.5 45.8 61.9 Phalombe 12.2 29.7 50.0 50.0 Mulanje 32.0 42.0 76.2 76.2 Thy010 0.0 38.5 71.2 71.2 Chiradzulu 31.8 31.8 31.8 79.8 Chikwawa 79.6 79.6 79.6 79.6 Nsanje 58.8 81.1 81.1 81.1 Malawi 10.7 20.9 35.5 39.8 I Missing Table 2.5: Vulnerability withindistricts Food Entitlements as a percent of energy needs (average) Source: Malawi VAC Food Security Monitoring Report, Nov. 2005 Note: The averages arefrom ranges provided in the report. *Areas are direrent livelihood zones Lessonsfrom the geographyof poverty and vulnerability i) Whilst average poverty rates are lowest in the Centre and highest in the South of Malawi, the geography of poverty is complex. Poverty `hotspots' are spread all over the country. Poverty is widespread inrural areas but concentrated inurban areas. This implies, for example, that geographical targeting is likely to leave out the poor in untargetednon-poor areas and include the non-poor intargetedpoor areas; 12 ii)Whilstvulnerability andpoverty rates donotalwayscorrelatebecauseof dierent riskfactors, they converge in the Southern Region. For example, all the districts inthe Northern Region were among the ten least vulnerable districts while nine out o f the ten most vulnerable districts were in Southern Region. On this basis, Southern Region ought to have the greatest concentration o f safety net interventions; iii)Inter-districtpovertyratesarevariable-from 76percentto37percent.This,andthe fact that vulnerability and poverty do not overlap, suggests social protection interventions should not be generic in terms of types, coverage, and targeting. One size will not fit all. It i s important to distinguish between chronic and transient poverty, and how risk and vulnerability affect different groups; iv) Districts with the highest levels o f poverty require social protection that focuses mainly on supporting household consumption; v) Districts with lesspoverty appear to require afocus mainly on ex ante or preventative social protection to avoid the loss of assets after a shock and to promote assets in the long-term. Such interventions have the potential to help move households out o f poverty and out of the need for social protection;14 vi) However, because of the variation in poverty and vulnerability within districts, it would be useful to base shorter-term benefits, such as free food distribution or food- for-work programs on vulnerability assessment and longer-term benefits on poverty rates; vii) The'relative shift of poverty from rural to urban areas could reflect rural-urban migration and / or increased resource allocation in rural areas and Southern Region in response to the 1998IHS. Under the 1998 survey, urban areas were the least poor and probably received the least attention. But it could also reflect other factors. Social protection programming complemented by the creation o f livelihood opportunities in both rural and urban areas would be part o f an integrated approach. More attention i s required to the increase inurbanp ~ v e r t y . ' ~ Poverty and vulnerability at householdlevel Thepoor in Malawi are not a homogenous group and identifying the dimensions of poverty for different households is important for targeted safety nets implementation. Without a clear picture of these dimensions, we may end up with social protection instrumentsthat do not match the capabilities or needs o f the households that they are meant to support - for example, agricultural inputs for people with no land, public works for people with no spare labor. Such categorical distinctions between types o f households make more sense for social protection programming purposes than a simple cut-off line along a continuous income l4 This assumes low depth and severity of poverty in low incidence districts which could be wrong. No data analysis of IHS2 is done to test this assumption. This report remains weak on poverty and vulnerability in urban areas and has maintained a focus, largely in line with the W A C , on rural areas. More work is required to better understandthe implications of urban risk and vulnerability for social protection programming. 13 distribution. Hence, a distinction should not simply be made between poor and ultra-poor households.l6 Most attempts to identijj different groups of poor people draw on the distinction between poor and ultra-poor as measured by income (or consumption expenditure) but then include other characteristics of poverty and vulnerability. In the remainder o f this section we will consider previous and current attempts to identify different groups o f poor people. We will focus on: i)safety nets programming drawing on the 1998 IHS survey, ii)work by the NSPTC on income and labor, and iii)the findings o f the MPVA. Basedon these experiences, we then identify three groups o f households and their needs. Safety Nets programming following the 1998 I H S The 1998 poverty profile provided some basis for identifyingpoor households (GoM 2001). However, Milner and Tsoka (2005) report that these core proxy indicators were rarely used in practice because: i)Characteristicsfrom theprofile didnothelpoperations;practitioners hadtorefine these in order to have an operational definition of thepoorest. For example, the Joint Emergency Food Aid Program (JEFAP), a collaboration between Government of Malawi, donor organizations, the World Food Program and the NGO Consortium came up with an alternative list o f proxies; ii)Whencommunitygroupswereaskedtoidentifypovertyproxyindicators,indicators such as female-headedness or low landholding sizes were rarely mentioned although these were frequently stressedas correlates of poverty by `outsiders'. The overall lesson from previous experience i s that the current system for identifying poor people that focuses on categorical targeting rarely used the proxy indicators from the 1998 poverty profile without refiningand ground-truthing them with targeted communities because they were too broad. The difficulties in identifying the poorest o f the poor brought in the idea o f a three-prong approach; use o f a short-list o f proxy indicators, use o f community-developed poverty manifests and verification o f identifiedbeneficiaries usingindependent assessors or the entire community. Thus, there are currently no standard target groups even for similar programs because of the need to refine the proxy indicators andmake them community relevant. Work by the NSPTC in2006 Inan effort to characterize the poorhltra-poor, the NSPTC disaggregated the households into groups and used income and availability o f labor and land in the household to create a two- dimensional profile o f poverty.l7A categorization into four groups was identified: poor with l6Povertylines for poor and ultra-poorhouseholdsare defined as annual per capitaconsumption expenditure levelbelow MK16,165 and MKl0,026 respectively.The ultra-poorare definedas those with per capita expenditurelevelsthat are belowthe foodpovertyline whilst estimates for the poor includeanon-food component.For more detail onpovertylines, see above. l7These proxy indicators are not strongly backedby the IHS2 or its derivativethe MPVA (see below) and the extent to which labor availability is a useful proxy was debated, and modified through later analysis (see discussionson labor scarcity anddependency ratiobelow). 14 labor (group A), poor without labor (Group B), ultra-poor with labor (Group C) and ultra- poor without labor (Group D). The potential needs o f these groups were then identified as follows: Grour, I Needs A Inputs,employment, markets& extensionservices B Land andthen inputs,markets & extension services C Employment, cash for consumption and assets & productive assets D Social assistance, free food & education for their children Both analytical work on the IHS2 and the operationalization o f the IHS2 findings show the problem o f utilizing a set o f proxies indicators for targeting that are not grounded. The work o f the NSPTC shows the difficulty o f focusing only on a small number o f proxy indicators (income and labor) whose influence over poverty i s not necessarily greater than a whole range o f other factors, as further analysis o f the IHS2 data for this report suggested. If labor and income were to be accepted as necessary and sufficient proxies for defining target households for the SP policy, the NSPTC suggested the following match o f needs o f poverty groups and different types of social protection interventions (See Annex 4 o f the minuteso fthe NSPTC heldon 24thand 25thOctober, 2006): Social cash transfers meet all the needs o f all poverty groups, except the provision o f extension services; Credit schemes match all the needs of all the poverty groups, except the need for community support; Social support projects respond to all the needs o f all poverty groups, except market access; Targeting inputsprojects are important for all groups for various needs; Targeted food distributions meet the needs o f all groups, except the richest poor group; School feeding projects meet the needs o f all groups, except the richest poor group; Targeted nutritionprograms meet the needs o f all groups; Inputssubsidy projects are suitable for the richest andpoorestpoverty group. Note that the short and qualitative assessment by NSPTC did not dwell much on the quality and extent o f the matches between needs and interventions although it provided an overview o f the main problems associated with each type o f intervention, in terms o f effectiveness in meetingthe needs o f the poor (based on the selected proxies). The debate about the relationship between poverty, labor and household dependency ratio as basis for defining key target groups has lead to some uncertainties for planning and policy- making by government and donors, and i s further dealt with insubsequent sections. 15 Evidencefromthe MPVA (2006) According to the MPVA, which draws on the IHS2, poor households differ from non-poor households along many different indicators (hence making the selection o f a few proxies for definingtarget groups difficult): Interms ofthe gender dimensionofpoverty: i) Poverty rates are higher infemale-headed households (58 percent as opposed to 51 percentfor male-headed households), due inpart to higher dependency ratios, but in terms o f total numbers there are more poor people in male-headed than female- headedhouseholds; ii) Female-headed households are more likely to be poor compared to male headed households becausewomen; - are less likely to get access extension services; - are less likelyto engage incash crops like tobacco andhybridmaize; - spend less time inincome generating activities; - are paid less than men for the same task / period when employed; - are less likely to be employed than men; - are less likely to get poverty-reducing credit - they mostly get smaller amounts. Interms ofdemography i) Poor households are generally larger than non-poor households. In particular, households in the poorest decile are more than twice as large as households in the richest decile; ii)Poorhouseholdshavealargerdependencyratiocomparedtonon-poorhouseholds with the poorest decile having a dependency ratio four times that of the richest decile households. Inparticular, poor households have more children than non-poor households; iii)Percapitaconsumptionislowerinhouseholdsheadedbytheagedthaninhouseholds headed by younger members of households. Interms ofeducation i) Poor households tend to be headed by persons with little or no education; ii) Poor households host less literate members than non-poor households. The literacy rate for the poorest decile i s 50 percent while that o f the richest i s 87 percent; 16 iii) Childrenfrompoorhouseholdsarelesslikelytoenroll,attendandcompleteschool than those innon-poor households; almost 30 percent o f poor children do not even start primary school and fewer complete. Interms ofhealthandnutrition i)Poor households are not necessarily disadvantaged in terms of morbidity and malnutrition; it i s the non-poor households that report higher morbidity and incidence o f chronic illness than poor households and there i s no significant difference between poor and non-poor households interms o f malnutrition. Interms ofaccess to markets i)HouseholdslocatedfurtherawayfromthrivingtradingcentersortheBoma(district headquarters) are more likely to be poor. Interms oflivelihoodsources i)Poor households rely heavily rely on household farming and fishing for their livelihoods but the majority o f them rely on ganyu for their cash income (68 percent) as compared to salary/wage income (21 percent) or self-employment in household enterprises; they own the minority o f household enterprises (27 percent); ii)Poorestruralhouseholds, despite their heavyinvolvementinganyu, deriveonly9 percent o f their total household consumption from ganyu income because most of their consumptionis derived from ownproduction. Interms ofownership ofassets i)Poorhouseholdsownlimiteddurableassetsandlivestock,includingchickens; ii)There is not much variation in average landholdings between rich and poor households. However, since poor households tend to be larger (as they have more children), poor households generally own very little land in per capita terms. For example, non-poor households own almost twice as much land per capita than poor households (0.42 hectare compared to 0.23 hectare) and households in the poorest decile own an average o f 0.17 hectare per capita compared to 0.59 hectare per capita for the richest decile households. However, there i s no significant difference between female and male headed households. Interms ofvulnerability i) Poorerhouseholdsaresusceptibletofewershocksthanricherhouseholds, becauseof their avoidance of high-risk, high-return activities, but are very susceptible to particularly shocks especially crop failure; ii)Poorhouseholdsarelesslikelytobeexposedtonon-agriculturalbusinessshocks. Interestingly, orphan status is not consistently correlated with a higher probability of being ultra-poor, possibly due to the purposive placement of orphaned children in better-off 17 households in the extendedfamily network. Not surprisingly, households incommunities with better infrastructure also have lower probabilities o f being ultra-poor (although living in a community with ADMARC i s associated with higher rates o f ultra poverty), and households in remote areas, or in communities not accessed by a tarmac road, are more likely to be `ultra-poor' (GoM/WB 2006: xxiv). The MPVA also shows that `the frequent and widespread existence of shocks in Malawi manifests itselfin large movements into and out of poverty: even though the income poverty level has stayed constant since 1998, about two-thirds o f households have moved into or out o f poverty duringthe past decade. Such large movements also reflect the fact that a quarter o f Malawians have income levels within 20 percent points o f the poverty line, who could be forced into poverty by even slight misfortune' (GoM/WB 2006: xxiv). This suggests that there i s another group o f households, just above the poverty line, for whom social protection i s also critical to prevent them from sliding into poverty. These may be termed the transient poor - which for various reasons are very vulnerable to shocks and stresses, and may slide towards (or move out of) poverty. Povertyand vulnerability profile Before presentingour poverty and vulnerability profile, two important caveats / explanations are required. First, matching the group characteristics and needs against a poverty incidence line i s difficult because o f the complex interplay o f different factors. Box 2.6 gives an example o f why. Second, a distinction between households with more or less labor scarcity may be important to make when identifying the best instrumentsto support households, but the relationship between poverty, labor scarcity and dependency ratio in the households i s complex because: i)Laborisnottheonly, northemostcritical, factorindefiningultra-poverty(MPVA 2006, Devereux et a1 2006); ii)DatafromtheIHS2005analyzedbytheWorldBanksuggeststhatlaboravailability inultra-poor households is notworse thanthat innon-ultra-poor households; '* iii)Focusing socialprotection efforts onultra-poor households without labor would, according to the IHS 2005 data mean reaching about 5.5 percent o f Malawi's households but there is no certainlythat these would be the very poorest h~useholds.'~ A category o f `destitute'households may still be retained defined as a sub-group o fthe ultra- poor with few assets, including little or no labor and land. '*On average, 5.5 percent of householdsare labor scarce and ultra-poor whilst 7 percent of householdsare labor scarce andpoor and 12percentare labor scarceandnon-poor. Labor scarcity is defined here as ahousehold with EITHER no householdmember inthe age group 19-64who is fit for productive work OR (inthe case of the householdsthat has one or more fit adults aged 19-64) has a dependency ratio of more than 3. These figures are for householdsandnot for total population. The share of population that is labor scarce and ultra-poor may be larger than that of householdsbecauseultra-poor householdsare larger than poor andnon-poor households. These figures draw on additional analysis ofthe IHS2 data for this report. l9 We recognizethat in some districts the figure may be muchhigher but as yet there has beenno rigorous qualitative analysis of this. 18 Box 2.6: Matchingorphanstatus against an incomeprofile According to the MPVA, the IHS 2005 shows that orphan status i s not categorically associated with ultra poverty. The MPVA reportsthat an orphan i s equally likely to be inapoor household(57 percentofall orphans) as anon-orphanedchild (58 percentof all non-orphans), possibly because orphans are deliberately placed inwealthier extended family households. Yet, Devereux et a1(2006) also draw on the IHS and findthat: poorer households care for more orphans than richer households (butpoor households also have more children than non-poor households); female-headed households have, on average, more orphans across the income range; female-headed households inurbanareas, on average, host more orphans than anywhere inthe country (but they also host more non-orphanedchildren too). Poorer female-headed households care for many orphans including double orphans. Whilst these findings may appear inconsistent, the `differences' mainly arise from the fact that the MPVA compares orphaned children with non-orphanedchildren, while Devereux et a1 (2006) comparedhouseholds with and without orphans. Ingeneral all children (orphaned or not) are more likely to be inpoor households and infemale headedhouseholds. Thus, making statements about the characteristics o f households along the poverty line is exceedingly complex. Theprofile presented matches up objective, material indicators of poverty with wider social indicators of poverty, and risk and vulnerability. On the basis o f what has been discussed so far, we provide an operational grouping o f the poor, their vulnerability and needs. Three different poverty groups are proposed: i) Ultra-poor(anddestitute); ii) Poor or moderately poor;and iii) Transientpoor(atrisk). The characteristics o f each, the needs and the appropriate social protection instruments to support them are shown inFigure 1. See also Appendix I for more detailed characteristics of each o f these groups. Within the ultra-poor group are a sub-group o f households that are categorized as destitute. These are ultra-poor households that have the least assets and / or highdependency ratios, for example due to chronic illness, and among these female / elderly /child headed households are more prevalent. For these households, the likelihood of graduation from social protection support out o f poverty i s least likely. The numbers o f destitute households i s not clear but they are a category that, given the commitment to expanding the Mchinji social cash transfer pilot, are highon the agenda for social protection. Given the potential interventions for the various needs o f the poor and vulnerable, the next section analyses the interventions that have been implemented inthe recent past to determine whether they meet the needs or not. 19 3. INVENTORY OF SOCIAL PROTECTION INTERVENTIONS Backgroundand approachto the inventory To place the current range o f safety nets inMalawi incontext, it i s necessary to first consider earlier activities that have had a strong influence o n the current system. Whilst social protection invarious forms and under different banners predates Malawi's independence, we can identify four important periods inMalawi's recent history (Table 3.1). These are critical indefining Malawi's current situation. Table 3.1: A summarv of `social protection'inMalawi Period Types of interventions Comments 1964-1981 Inputandoutput pricecontrols There were few formal safety interventions Universalinputs subsidy (beyond price controlshubsidies). Market based policies and practices dominated. This can be Farmer clubs and inputs credit best be described as `those good old days', in facilities terms o f protection from exposure to effects of poverty and covariate shocks. 1981-1990 Inputandoutput pricede-controls SAPS under stabilization forced Government to Phasing out universal subsidies dismantle the `SP' systemwith no replacements. The period had no effective protection o f the TargetedNutritionPrograms poor and vulnerable, except in serious Food transfers (relief) emergencies. 1990-1994 Interventions under the Social Inspired by `adjustment with a human face' Dimension o f Adjustment including calls. Politics in the early 1990s weakened the credit schemes (Malawi Mudzi farmers clubs which led to the collapse o f the Fund) credit system. The effect was that the little that Targeted NutritionPrograms and remained for the smallholder collapsed. Food transfers (relief) 1994-2006 MSMEs Credit schemes Dominated by Government and donor Public works programs (PWPs) initiatives. Gradual increase in NGOs offering CFW, FFW, IFW SP interventions. Inputstransfers took advantage Inputs transfers (SIP, Starter pack, o f the collapse o f the credit scheme, escalating TIP) inputs prices and frequent droughts. Food Food transfers (relief and otherwise) transfers and school feeding also took advantage School feeding o f frequent food insecurity to become almost Cash transfers (pilots) annual. Most o f the interventions were in the Targeted inputs subsidies spirit of safety nets. There was only partial and Targeted nutritionprograms selective protection. Integrated livelihoods support Socialprotection in independent Malawi took theform ofprice controls and subsidies. By the late 1980s, these measures had yielded little and had become financially unsustainable. From 1981, the Structural Adjustment Program rightly called inearnest for their removal. The late 1980s / early 1990s saw the removal of price support and subsidies and nothing established in their place under structural adjustment. During the period o f structural adjustment, the only visible programs were targeted nutrition programs (for example therapeutic and 21 supplementary feeding for children, and pregnant or lactating women). Food aid was prevalent duringyears o f significant food shortages. I n early 1990s,projectized safety net programs came to thefore under the social dimension of adjustment (SDA). New programs were introduced, many o f them market-oriented. The Social Action Fund (MASAF) emerged from SDA thinking as a mechanism to support community-based activities that would alleviate poverty. This shift corresponded to a change in government which brought a refined focus on poverty. The policy framework for the poverty alleviation program specifically provided for the work o f an institutionalized social action fund and space for civil society organizations like NGOs and FBOs to undertake safety net interventions in aid o f the poverty alleviation program. However, the policy framework did not provide for an effective institutional framework for coordinating the poverty alleviating safety net activities. In the late 1990s,government and development partners, mainly WorldBank, DFID and EU, recognized that the safety net interventions were having little impact on poverty. An inventory and analysis o f social protection activities were undertaken in 1999 as a background to a drive by Government to develop a national safety nets strategy (ref. World Bank 1999: Malawi A Safety Net Strategy for the Poorest, Africa Region). The exercise showed that the safety nets interventions, including the Government-inspired public works and inputs transfers, were too short-term, ad hoc, patchy and uncoordinated. The National Safety Nets Strategy, program and institutional structure emerged from the exercise in 2000 (NEC 2000). This strategy and the institutional structure were adopted inthe Malawi Poverty Reduction Strategy as pillar three with four key components: i)Public works program; ii) Targeted inputs program, iii)Targeted nutrition program, and, iv) Direct transfers program, including relief food and support to secondary school going OVCs. A Safety Nets Unit (SNU) was established to coordinate and prioritize safety nets through four technical sub- committees. The system provided support through welfare support interventions, productivity enhancement, and disaster management. These earlier systems for dealing with poverty, risk and vulnerability have essentiallybrought about the present safety nets system inMalawithat focuses on: i)Sectoral approaches to addressing food security and maintaining agricultural production, i.e. especially Starter Pack and TargetedInputsPrograms; ii)Community-basedapproachesaimedatmitigatingthenegativeimpactsofstructural adjustment on the poor people, i.e. MASAF's social action fund; iii)Expandedsupplementaryandnutritionalfeedingtoaddressthecommonmalnutrition o f vulnerable groups like children below the age o f 5 years, pregnant women and lactating mothers; and iv) Frequent free food distribution for short-term as well as long-term food insecurity, especially between 2001and 2006. In this section we analyze the inventory of safety nets in terms of key areas and specific questions that are relevant to the subsequent social protection framework and policy development. 22 Project types and beneficiary coverage According to Kambewa (2005), there are three broad categories inMalawiunder which most o f the interventions fall: i)Productivity-enhancingprograms(inputstransfersandpublicworks); ii)Welfaresupportprograms(targetednutrition-includingtherapeutic,*' supplementary and school feeding - direct welfare transfers and capacity building for social welfare services); and iii)Disasterhkmanagementactivities. In the latest analysis of safety nets interventions.lin Malawi, the NSPTC identified twelve different types o f safety nets interventions: Public Works MASAF type School-feeding SocialCashTransfer Public Works EUtype IITargetedfood distribution MASAF Social support Targeted Inputs DonationsInkind SocialFunddMicro projects Fertilizer subsidy IITargeted Nutrition II1IMicro-credit schemes This list excludes other public works like food-for-work and inputs-for-work and includes micro-credit schemes which are not considered as safety nets interventions although they mostly target the poor and vulnerable. Inthe period2003-2006 the productivity enhancingprojects were the most common, interms o f number of projects, beneficiaries and total cost. This was followed by welfare support projects. There was only one disaster management project and this has been grouped under direct welfare transfers. In terms o f number o f beneficiaries and expenditure, direct welfare transfers or social transfers at the time o f data collection were very small. Most o f the projects are onpilot basis. See Figure 2. In terms of outreach, the most extensive programs have been; i)subsidizedfree food distribution, most importantly food aid through WFP's Regional program ,ii)subsidizedfree agricultural inputs (especially fertilizer), and iii)public works (through e.g. EU funding and MASAF).The fertilizer subsidy has been a key element of the Government's present policy- with a clear but not well targeted social protection dimension. School feeding is andther important program, which has a clear productivity enhancing element (boosts health and supports school enrolment), but a less clear social protection function. Cash transfers are becoming increasingly important, but have over the period analyzed been provided only on a very limitedbases to specific vulnerable groups e.g. orphans, disabled, and elderly. *'Stakeholders made compelling arguments during the stocktake for removing therapeutic feeding from the stocktake on the basis that it is a clinical treatment. Despite broad agreement, because o f the difficulties of disaggregating data on therapeutic feeding from other social protection activities (for example in the WFP PRRO), it remains inthe stocktake. 23 I Figure 2: Relative share of SP Components 100 90 80 70 Y 60 50 a B 40 30 20 10 0 Source: Authors estimates Interms of actual type ofprojects, cash-for-work were the commonest projects during 2003- 2006 in terms o f number o f projects and amount o f money spent. However, in terms o f number of beneficiaries, the most prominent projects were targeted inputs transfers and subsidies. Numbers of cashprojects were small (see Table 3.2). Source: MVACIAuthors data and estimates According to IHS2, very few households reportedbenefits from most o f the interventions. On the basis of the proportion o f households that participated in the safety nets, some o f these projects are relatively insignificant interms o f impact. While public works programs appear to have been available here and there, cash transfers were indeed very rare (See Table 3.3). The critical question is whether the current project types adequately cover the needs o f the poor and vulnerable. Our conclusion i s that the project types are not necessarily inadequate (although some o f them are not very relevant for social protection) but suffer from limited beneficiary coverage, mistargeting and significant leakages (especially the major transfer programs). Most o f them are too small in scale to have any real impact, as such they just 24 `scratch the surface'. According to the MPVA (Box 3.1), fewer households reported participating invarious projects due to leakages. At least once Project 2001 2002 2003 2001-2003 Free food distribution percent percent percent percent Free food distribution 10.1 15.1 12.5 26.5 Targeted nutrition program 2.5 3.3 3.6 7.0 Supplementary feeding 0.4 0.7 0.8 1.4 Source: GoMIWB 2006 Table 9.1 Box 3.1: Key MPVA findings on projecttypes The two largest programs (inputs transfers and food transfers) are prone to excessive leakages; number o f households recorded to have received the transfers is far more than the actual households that report receiving the transfers. Most probably some end up in non- deserving households as `bargain' purchases andnot transfers. Almost all safety nets interventions have targeting problems. The worst are the two largest programs; they have high errors o f inclusion and exclusions due to a. poor geographic targeting - W A C is objective but ground realities force district level b. biased beneficiary targeting - Community targeting has proved useful at times but works officials to spread the benefits across the district. under pressure to include non-deserving and therefore forced to exclude deserving households. School feeding is a good education program. It falls short o f being an effective safety net intervention. Its inclusion o f non-poor pupils and exclusion o f poor out-of-school children is not an operational error but a fundamental flaw. A pure social protection intervention would have worried about ths. Public works programs have the potential to alleviate poverty but have very low coverage. Moreover, they are riddled with design challenges because a. highwages attract non-poor but low wages keep participants poor, b. implementing theminnon-lean periodmisses the time when they are really needed; and c. implementing them inthe leanperiod is technically difficult (too wet for most works) and regressive (the poor would be attracted away from working on their own land) Source: GoM/WB 2006 25 Objectivesand targetgroups According to the National Safety Nets Strategy the overall objective of safety nets is to protect the livelihoods of the most vulnerable. An analysis o f the objectives o f the safety nets projects shows that most o f them match the general objectives of safety nets. Almost all projects have food security and poverty alleviation as elements o f their objectives. Cash for work programs have the most diversifiedobjectives, possibly reflecting the versatility o f cash and the expected use o f the created community assets. The long list o fproject objectives also reflects the multiplicity o f implementers and lack o f coordination amongst them as well as lack o f common understandingo f what each type o f intervention canrealistically achieve. Recently objectives of someprojects include elements to deal with chronic illness (HIV/AIDS and TB). Chronically ill people are increasingly included in the objectives o f targeted nutrition programs thereby expanding the range o f target groups. Thus beyond under five children, lactating and pregnant mothers as vulnerable groups, nutrition programs are now extended to include TB patients and HIV-infected persons. Related to this, households hosting the chronically ill or orphans and headed by the aged or a child are becoming common target groups. This comes from the perception that the households are too poor to cope with the burden o f hosting the illand orphans. However, we understand that this perception i s not supported by IHS2: child-headed households are almost non-existent and households hosting the chronically ill or orphans are, at the worst, as bad as other poor households but certainly not necessarily poorer than those hosting no orphans and the chronically ill. Again, several of the interventions have very broad target groups. Example o f these include `vulnerable households', `(ultra) poor with labor', `work-constrained', `disadvantaged people', `poorest households' and `ultra poor households'. Can these be easily identified on the ground? See Table 3.4 for a list o ftarget groups as given bythe implementers. 26 cash/fod/inputs for work; SF=Supplementary feeding; TF=Therapeuticfeeding Of all the types of projects, transfers have the longest list of target groups; cash hasfifteen, food fourteen and inputs ten. However, beneficiary groups for cash transfers are generally more specific than for other interventions. Supplementary and therapeutic feeding programs have generally narrowly defined target groups as compared to public works which have the broadest of target groups. Transfers with attached incentives (conditional transfers) are rare and are mainly associated with transfers targeting school going age children as a way o f encouraging them to stay in school. The Mchinji pilot social cash transfer has an additional payout for every child in school and the take home food under school feeding programs i s contingent on the child consistently attending classes. This can be seen as a `soft' condition. The MPVA suggests the use o f conditional transfers for direct welfare transfers in certain contexts. *' Many of theproject objectives are very ambitious, and often do not`appear to be achievable in the current timeframes for theprograms. Evaluation and reviews o f various interventions have consistently shown that the objectives are rarely achieved due to a number o f reasons, '*The debate about conditional and unconditional cash transfer is high on the agenda in Malawi, especially in relation to whether the National Cash Transfer Pilot (the Mchinji scheme) i s a Conditional Cash Transfer. The two sides of the argument are: i)The extra stipendthe pilot provides for school-going children is claimed not to be conditional becauseit i s not monitored; ii)Giving moreto householdsthat havechildreninschoolconstitutes a condition, whether it is formally monitored or not. An alternative and non-conditional approach, which could also be tested, would be to give additional cash to households that have children of school-going age, irrespectiveof whether they are enrolled or not, andto arrange for teachers/ education officials to give speeches when beneficiariesreceivetheir cashto encourageparents/ guardiansto enroll their childrenin school. 27 not least the small sizes o f the transfers and the period a beneficiary gets the transfer. The lesson for social protection programming i s that there i s need to consider social protection as a long term commitment - programs with lifespan o f more than five years are required. This i s where a shared SP framework, policy and program would combine the scattered goals and targets into a few common objectives, help guide designers o f SP projects and enable longer- term commitments. Regarding target groups, there is an urgent need to use evidence when designing SPprojects and selecting the target groups. Use o f proxy indicators must be preceded by credible analysis because some o f the proxy indicators have significant errors o f inclusions and exclusion as the discussion in Section 2 shows. Further, analyses o f the projects show that some o f them target inappropriate groups in terms o f the needs o f the target group and the type o f transfers/payment. Since the characterizationo f the four groups i s more symbolic than real, as already discussed in section 2 above, there is a need to design specific SP projects based on expressed needs in given contexts. This means that if an SP program has to be designed, it should be bottom up and should have clear objectives that point to addressing specific needs and risks o f specific target group (and utilize a combination o f proxies and community targeting and verification). Geographic coverage of safety nets interventions Analysis of geographic coverage of safety nets in relation to the poverty and vulnerability profile is dfJicult, and the results should be interpreted with caution. For example, an analysis of project concentration using number o f projects in a district i s very rudimentary. First, projects come by different scales, interms ofbeneficiaries and size ofthe transfer or the period a beneficiary enjoys the benefits. Second, even if the scale o f the project i s given, the scale of the project needs to be related to the levels o f poverty or vulnerability inthe district. Third, the exact implementation area inthe district is critical due to intra-district variations in poverty and vulnerability. Most of theprojects do not give number of benejciaries per district. Inorder to get a feel o f geographic concentration o f the safety nets interventions, we relate the district cost (a sum of average cost o f each project implemented in the district) and number o f people at risk (as given by MVAC) and number o f poor people (as calculated from IHS2) for each district (See Table 3.5 and also Figures 3 and 4). Ideally there i s supposed to be no major difference between districts in terms o f the cost per vulnerable or poor because all vulnerable or poor ought to be protected equally. Rumphi, Ntchisi, Phalombe, Dowa and Karonga are the topJive districts, in terms of cost per vulnerable and Salima, Phalombe, Nkhotakota, Dowa and Nsanje in terms of cost per poor. Comparing these with the levels o f poverty and vulnerability presented in Section 2 shows that there i s no matching. None o f the top five districts interms o f cost per the poor i s among the poorest 10 districts and only one o f the top five districts interms of levels o fvulnerability were the most vulnerable. This demonstrates, to some great extent, that there i s geographic mistargeting which i s a recipe for beneficiary mistargeting.22 ~ 22Notethat althoughimperfect, usingnumber of projectsper district, wouldhave ledto similar conclusions. 28 Table3.5: Numberof projects and relatedexpenditure byDistrict District I Projects I Cost/ I Cost/ District I IProjects1 Cost/I Cost/ Source: Authors estimates Figure3: Cost per vulnerablepersonper district(USD) Figure 2: Cost pervulnerable Source: MVAC/IHSZ/Authors data and estimates 29 I Figure4: Cost per poor personper district(USD) Figure 3: Cost per poor 350 300 250 200 150 100 50 0 Source: MVAC/IHSZ/Authors data and estimates Implementingagenciesand periodof implementation Most of the major interventions and Government-inspired interventions are implemented through the district assemblies. Table 3.6 presents implementing agencies mainly for the interventions that were operational in the period 2003 and 2006. For example, district assemblies implemented seven different public works programs that had cash as a transfer. In total they had implemented thirteen different interventions. CARE International inMalawi i s the leading NGO interms of number of interventions it has implemented inthe periodunder focus. It i s closely followed by consortia partners namely Africare (6), Save the Children Federation (6), Emmanuel International (5), RedCross Society (5), World Vision Malawi (5), and The Salvation Army (5).23 Major donor funded, but NGO implemented interventions mostly involved a consortium of NGOs. The WFP's food aid program to the chronically illinvolved the highest number o f agencies. A total o f fourteen NGOs were involved. Other interventions requiring a good number o f NGOs include Improving Livelihoods through Increasing Food Security (I-LIFE) and Sustaining Productive Livelihoods through Inputsfor Assets (SUPLIFA). Most of the NGO activities involve district assembly staffas well. The most commonly used personnel are community development officials for community mobilization. In some cases, NGOs use other technical staff depending on the type o f interventions. For example, various public works require some technical expertise at district level. These personnel are only available if demand for their time for district assembly work (and other higherpayingNGOs) i s not high. As long as district assemblies are under-funded, NGOs have full access to the use o f personnel at district and sub-district levels. The implication i s that the timely and efficient implementation o f NGOs' work is dependent on the under-funding o f district assemblies. This may obviously create a problem if Government starts to fully utilize the capacities at district level. 23 Notethat this was only on the basis ofthe projectsincludedinthe inventory.As already indicatedthere could beprojectsthat were implementedinthe periodbut were not reportedto SPU or the stocktaketeam duringits datagatheringphase. 30 Table 3.6: Implementersof safety netsinterventions Implementer *]Projects1 Interventionsrunfrom2003 to 2006 ' Note: CT=Cash transfer; FT=Food transfer: IT=Inputs transfer: CFW/FFW/IFW = cash/fod/inputs for work: SF=Supplernentary feeding; TF=Therapeuticfeeding: TNP=Targeted Nutrition Program (therapeutic and supplementaryfeeding;) Most interventions are implemented within a year (57 percent of the projects); very few run for three or moreyears. The average runningperiod was 21 months for a selection o f typical projects.24There i s some variation if the period i s analyzed by type o f project (see Figure 5). I-Life's five years, i s the longest period o f funded implementation. This i s followed by TNP which runs close to three years. Often, longer programs do not have confirmed lifetime funding (creating uncertainty for beneficiaries over what benefits they will be entitled to). Of course, food transfers are not expected to run for periods beyond a few months if they target the transient poor during emergencies. It i s when food transfers shifted to the chronic poor that the period o f implementation started to go beyond twelve months. Cash for work projects should operate for a longer period to allow participants to benefit long enough to build assets or save for investment. Unfortunately, they are operated as safety nets for the transient poor who are expected to recover quicker than the chronic poor. 24Note that a few of these projects are on-going. We also included projects that predate 2003 for this analysis. For those with specified end dates, the future date is used in the calculation on the assumptionthat they will continue as planned. For those without end date but are on-going, December2006 is taken as the enddate. 31 Months Figure5: Periodof implementation 70 60 50 40 30 20 10 0 Integrated TNP CFW FT FFW IFW FACT IS IT CT Source: Authors estimates The period of implementation for most of the interventions is too short to achieve poverty reduction.25 Worse still, it i s rare to have the project target the same beneficiary for the entire period o f implementation. This i s mostly true for public works programs. The high demand forces implementers to allow as many participants as possible in order to spread the benefits to as many as possible, thereby dilutingthe potential impact o f the projects. Interventioncosts The analysis in this sub-section i s based on twenty-five projects. They include all major safety nets interventions implemented in the 2003-2006 period.26For some types o f analysis some o f the projects are broken into their sub-projects. In such cases the number o f projects increases to 28 but the cost drops because only components with required information are included. This is the case for I-LIFE,WFP's Country and regionalprograms. Estimatedcosts The total costfor the 2.5projects in thefour year period is USDS38 million. This translates to USDI34millionper See Table 3.7. Clearly, Malawi can achieve a lot even with current levels of spending if only theprojects are rationalized. This i s more so considering that the projects analyzed here are not exhaustive. The rationalization can be in terms o f coverage and project types and mix. It i s apparent that improved outcomes can be achieved if the country prioritizes projects that directly deal with 25There are good examples of program s with longer durations- for example, the WFP PRRO. However, these are dependent on external funding. At this moment, the WFP PRRO is not fully-funded. 26 Estimating the amounts of money spent on social protection was the most problematic task in the development of the inventory. The main issues are (i) difficulties in obtaining data, due in part to reluctanceto share financial data for activities by some actors; (ii)difficulty in sharing financial data for activities among all actors; (iii)different budgeting and reporting timelines for different donors and NGOs; (iv) differentiating between budgeted spend and actual spend; (v) disaggregating spend by different partners in consortia; (vi) disaggregatingbetweengovernment and donor spendbecause of preferreddonor aid modalities; (vii) identifying annual spendfor multi-annual program s; and (viii) some projectswere active before our start date of 2003. 27The estimated cost for the period 2003 to 2006 is calculated by using the proportion of the period a project operatedduring this period. The proportion is appliedto the given total cost. 32 the needs of target groups and re-works the geographic targeting. An analysis of cost per district and cost per beneficiary confirms the needfor rationalization. See Table 3.8. 3 200516 Fertilizer subsidy 60.000 4 EmergencyDrought ReliefProgram - PWP (thru MASAF) 50.300 5 200415 TIP 35.085 6 GoM/EU Public Works Program 34.930 7 WFP Country program [CP 101061 34.400 8 ImprovingLivelihoods throughincreasingfood security (I-LIFE) 31.500 9 Inmtsfor assetsDroeram under (IRLADP) v 15.000 10 200314TIP 13.420 11 Local Authority ManagedProjects - MASAFI11PWP 12.500 12 ReliefCashfor Work - Conditional cash transfer PWP (thruMASAF) 12.100 13 SustainingProductiveLivelihoods throughInputs for Assets (SUPLIFA) 8.860 14 Consortiumfor Southern Africa Food Emergency(C-SAFE) 6.680 Source: Authors estimates The highest cost in terms of cost per beneficiary, when direct comparison is done, isfor relief cash and in-kind transfers, food transfers and cash for public works.28 Public works generally transfer less to beneficiaries than the cost of the works. The high cost o f food transfers i s probably due to the size o f the transfer (which i s effectively higher than the cash from public works projects) and cost o f transport and handling. The costs increase considerably through possible leakages (See also MPVA 2006). I n terms of allocations per district, the highest costs arefor cash-for-work, food transfers and inputs-for-work. These projects are generally implementedon the basis o f vulnerability levels and concentrate inthe geographic areas targeted. Inmany cases, a program or combination o f programs have limited coverage across districts, and reach only a few districts over a given period. The mainconcern is that several vulnerable districts mightbe left out. 28Obviously, these projectshavevery differentobjectives and outcomes, hence, such direct comparisononly providea crude comparativeassessment of cost per beneficiary. 33 The three largest project types (highest cost per project) are food transfers, inputs subsidies and public works. 40.9329 School (WFP feeding 3 610,000 20 25.0 only 9.40) 1.248 8.3 Cashtransfers 2 7,065 2 0.5 66.38 0.235 0.2 Inputssubsidy 1 2,000,000 28 60.0 30.00 2.143 60.0 Inputs transfers 2 3,701,350 28 49.0 13.23 1.748 24.5 Funding agencies Majorfunding agencies of the interventions include WFP (and its many donors), EU30World Bank, DFID and USAID (see Figure 6). The Government contribution does not include its core-financing o f TIP, M W C D bursaries, M o H contribution to Targeted Nutrition Program and the substantial funding o f the 2005/6 fertilizer subsidy. GTZ i s a new player in SP in Malawi and i s supporting school feeding. UNICEF, beyond its contributions to nutrition rehabilitation units i s providing significant financial and technical support for the Mchinji Cash Transfer Pilot and has included support to future social protection inW A F . The Irish development aid agency (Development Cooperation Ireland, now called Irish Aid funded a food and cash transfer pilot through Concern Worldwide, while OXFAM GB funded a pilot cash transfer inThyolo. 29This figure represents morethan one projectandfor abroader set of activities than only feedingthe school children.The WFP school feeding projectcosts are significantlylower - calculatedat less than $10per beneficiaryper year by WFP. The MPVA calculatedschool feedingprogramsto be USD 15 per pupil. 30It has not beenpossiblethrough the data collectionprocessto get fully updatedand correct figures from EU. 34 Figure 6: Financiers intotal SP cost Others, 0.2 percent GOM, 3 percent, - DflD, 18 percent II Source: Authors calculations Key issuesfrom the funding arrangements i)Donors dominate thefunding of SP interventions. Donors fund most of the operational costs and activities in Government project implementation unitdagencies like MASAF, GOWEUPIU and NAC. Donors fund most o f the activities o fNGOs; ii)Thereisverylittlecorefunding -NGOheadoffices onlycoverkeypersonnel.Most o f personnelcosts are absorbed by project budgets; iii)Governmentfunding hashistorically beenverylimited. However, Governmenthas steppedup its fundingcommitment to SP interventions inthe recentpast as evidenced by its increased fundingo f PWPs and inputtransfers and subsidies; iv) Government is yet tofund direct welfare transfers. Ithas fundedvery few bursariesin a year. The highest was 615 bursaries in 2003/4. In comparison, N A C funding covered 38,855 bursaries in2005/6; v) Donors are increasingly funding NGOs in consortia to ensure consistency and uniformity of approaches. It started with C-SAFE and extended to SUPLIFA, CO- GUARDand I-LIFE; vi) The mostpopular project types are among the most costly (high costper bene$ciary). As a corollary, the least costly project types are not the most common. Rationalization will require careful analysis o f donor orientations as well as evidenced-based promotion o f the most cost-effective types which also seem to deal with the needs o f the poor more directly. For example, it i s unlikely that the biggest donor - WFP - would flexibly switch to cash transfers in the immediate term, though cash / food combination pilot design i s underway. Likewise, it may take some convincing for EU to switch to inputs and food transfers because its orientation has been social protection tied to food production and infrastructural development; i.e. productivity 35 enhancing interventions that deal with the immediate as well as future needs. What thismeans is that careful analysis ofdonor orientations andneeds ofthe various types o f the poor and vulnerable should lead to a mix o f interventions that can be `bought' by various stakeholders, including Government that has hitherto shunned direct welfare transfers, except indisaster cases. 36 4. LINKINGTHE INVENTORYTO THE PROFILES Geography of poverty, riskand vulnerability As already seen (Table 2.2), the region with the highest incidence of poverty is the Southern Region and the region with the lowest poverty rate is the Central Region. Poverty does not increase from North to South. What generally increases from North to South i s vulnerability (Table 2.5), a main reason being that food production in the North i s more diversified than elsewhere in the country. In general projects are concentrated in areas where poverty and vulnerability are the highest but in the case of specific projects there are problems with the geographic targeting. This i s true o f district coverage (Table 3.5). Region Population CCT PWPProjects SSP SP Poverty Share (percent) Share In=MK1.4b I Share Average Incidence ~ 1 2 .m 3 n=1,831 n415,956 II Beneficiaries 1 Spending ,' n=367 1-1-28 n=12.3m North 12.4 7.9 8.7 5.9 17.4 11 56.4 Centre 40.9 52.4 38.5 33.5 37.9 15 46.7 South 46.6 39.7 52.9 60.6 44.7 13 64.4 Malawi 100 100 100 100 100 13 52.4 Case studies of two MASAF implemented projects namely Conditional Cash Transfers (CCT) public works and Social Support Program (SSP) demonstrate the point further. CCT projects were over-represented in the Central Region but under-represented in the other two regions. ShareintotalSP Percentof Poverty Populationshare projects(n=40) pop. at risk incidence (percent) . Top 5 Districts Lilongwe 60.0 17 37.5 9.4 Dedza 45.0 30 54.6 4.9 Dowa 45.0 16 36.6 3.9 Salima 42.5 63 57.3 2.6 Phalombe* 40.0 50 61.9 2.3 Chitipa 27.5 27 67.2 1.3 Rumphi 25.0 14 61.6 1.2 Karonga 25.0 7 54.9 1.9 Mwanza 22.5 76 55.6 1.4 Neno 22.5 86 55.6 Malawi 100 100 However, in terms o f number o f direct beneficiaries, the Southern Region had 53 percent o f the beneficiary population and received 61 percent o f the total amount allocated. This is in 37 line with poverty and vulnerability levels. SSP projects were over-represented inthe Northern Region(see Table 4.1). Inthe absence of detailed district by district data to provide an assessment of whether the existingsafety nets match current risk and vulnerability indifferent locations, we can only go as far as comparing the number o f projects and poverty and vulnerability profiles. For simplicity and demonstrative purposes, we show the top and bottom districts in terms o f number of projects implemented in their areas. Table 4.2 presents the related poverty and vulnerability levels. Again, it is clear that some districts do appear to have a signijkant larger share ofprograms than others that are equally, if not more needy. For example, Lilongwe district, which has low poverty and vulnerability, hosted the highest number o f projects, and more than for example Mwanza or Neno, which has higher poverty and vulnerability levels. Inventory and needs of poor and vulnerable groups Section 2 discussed a number o f the proxy indicators for poverty and vulnerability, and the likely needs each group o f poor households have. Possible interventions to meet these needs were also highlighted. This section discusses how the range and coverage o f the existing safety nets system matches the profile o f poor and vulnerable groups and their likely needs. There are limitations to how precisely this can be done. First, most interventions are targeted on vulnerable group categories rather than on a broader or more speciJic assessment of poverty, risk and vulnerability. For example, programs that target female-headed households assume ,that they are poorer than other households - the PVA suggests this i s partially true but there are also many households that are poor that are not female-headed. Other examples of where poverty does not match up well with vulnerable groups include the nutrition status of children, and number of orphans in households. However, many projects appear to use these indicators as proxies. Poverty and vulnerability assessments (for example the PVA and Devereux et a1 2006) suggest that these are unreliable proxies and may result insignificant mistargeting. Second, there have been very few in-depth evaluations ofprograms that can tell us which kinds of households received what support in practice. This means that we would not know how successfully the projects reached the target groups even if reliable proxies like household size or number of dependents were used. However, drawing on assessments of vulnerability andreviews and reports about program coverage, the likely actual coverage o f individualprograms inthe profile i s as follows: i)Cashforworkseemstotargettherighttypesofhouseholds; ii)Instrumentsthataimtosmoothorenableconsumption, especiallyfooddistribution, cover a very big range o f households - ultra-poor, poor and transient - and have targetingproblems which are at the expense o fother riskmanagement instruments; iii)Sometransfers(especially inputs)seemtogotohouseholdsthatmaynotmakethe best use o f the inputs being either labor-constrained or land-constrained or even too poor. A case in point i s the targeting o f coupons to the poorest households that may not have cash to pay even the reducedprice; 38 iv) Some food for work projects target labor-constrained households like elderly, child and female headed households. They even target households caring for chronically ill patients; v) Some inputs for work interventions target land-constrained households and towards orphan-headed households. The inventory demonstrates that there is little strategic plan and coordination of the overall range and coverage of social protection interventions in relation to needs across the poverty/vulnerability profile, although these findings cannot be further substantiated without extensive field research. The targeting o f so many differentkinds o f vulnerable groups across small projects and many locations has created a messy tangle o f social protection coverage. There is also often a lack o f consistency in approaches (for example regarding payment o f different wage rates for public works). A different mix of interventions is required to meet the needs of the diferent groups in poverty. There are a number o f possibilities regarding how the needs o f the various groups in poverty can be met. As a first step towards rationalizing the existing project types, the safety net interventions should be targeted to different poor groups as follows: i)Ultrapoor households benefitbest fromdirect transfers (e.g. foodandcash) that enable them to ensure consumption and invest inthe productive assets and productive capacity. PWPs have created assets like irrigation structures that are eventually utilized for the promotion of their livelihoods strategies on top o f cash or food transfers to enable them meet basic needs and take promotional risks for transition out o f ultra poverty, but an improved quality o f assets created by public works i s required. Building productive capacity in the longer-term means paying high enough direct transfers to keep children inschool; ' ii)Productivity-enhancingprogramslikepublicworksthatencouragedirectinvestment inlivelihoods and smart subsidies that reduce the risk of investmentinproductionare appropriate for poor households. Providing insurance against risk, for instance by supporting the adoption o f weather-based insurance, i s also likely to increase households' productivity-enhancing investments. Direct transfers also encourage investment inlivelihoods and would be effective for this group; iii)Social insurance - i.e. protecting households against risk- like weather based insurance i s appropriate for preventing the transient poor slidinginto poverty, but may also be relevant for other groups o fpoor/ultra-poor; iv) Integrated SP projects with activities addressing more than one need (multi- instrument) are often more appropriate for poor groups than single-instrument approaches because o f the multiplicity o f household needs. But single-instrument approaches may be less costly to administer. 39 Examples of activities and instruments that can be integrated are asfollows: i)Householdstargetedwithcashtransferscanbenefitfromsmallbusinessmanagement training, participation incommunity savings and loans; ii)Poorhouseholdsparticipatinginlong-termcashforworkprogramswouldalsobenefit from linkages to micro-credit institutions, small business management training and even membership inlocal savings and loans clubs; iii)Participants in targeted nutrition programs can benefit from food security and nutrition activities like community-based therapeutic feeding, home-gardening and public works linkedto winter cropping. able 4.3: Needs and current and potential projects Key Major Projectsdealingwithneeds 1 Potentialprojects Characteristics needs Projecttype ICoverage I Destitute or near destitute; Very small L3 0 asset base; a Small 0 5E landholdings per capita; Single livelihood - activity Limited asset base Single activity a"0 L3 Vulnerableto shocks Vulnerableto shocks Weak asset base Table 4.3 presents how the current safety net instruments and projects match up with the three key poverty groups in relation to their likely needs. The table also suggests in the last column what potential projects might be considered to address weaknesses with existing instrumentsandbetter address localneeds. 40 5. INSTITUTIONALARRANGEMENTS The institutional environment can determine the efectiveness and eflciency of social protection systems. The design o f a new nationalprogram would needto take into account the roles of, and relationships between different institutional actors and the interests and incentives they have in the design and delivery o f the program (World Bank 2003~).One o f the key findings o f the MPVA i s that poverty continues to be widespread inMalawi, and that there has been little or no progress in reducing poverty and inequality over the last decade, despite the fairly extensive safety nets systems. As indicated, there are both design and institutional problems related to the mix o f instruments and the way planning; coordination and management i s handled. It i s therefore important to ask to what degree the current institutional arrangement i s appropriate or adequate for effective and efficient delivery o f social protection. Key issuesand questions This review o f the institutional arrangements for the delivery o f social protection serves three purposes. First, it enables an analysis o f appropriateness of the current arrangements. Secondly, it allows an identification o f which institutional linkages and organizational factors are critical for social protection implementation. Finally, it enables an assessment o f current capacity to deliver social protection mechanisms and ensure that the future social protection framework, policy and program include measures to build required capacity in various institutions if necessary. The section i s mainly an assessment o f existing institutional arrangements, and provides only general guidance on selected institutional issues related to coordination o f future programs at national and district levels. Mathauer (2002) identifies the following issues associated with assessing institutional arrangements for social protection: i) The multi-dimensionality and multi-sectorality of safety net interventions means that coordination is required at both thepolicy and theprogram level -both a lead actor and an institutionalhome are requisite for social protection delivery; ii)Safetynetactorsandsocialservicesministriestendtobeweakandareoftenunder- resourced, have little voice, and this creates a problem for implementation; iii)Beneficiariesof socialprotection arediversebuthaveonlyweakrepresentation.At the same time, information asymmetries affect access to social protection by different households so that strong voice and monitoring systems need to be built into institutional arrangements; iv) Socialprotection benefits arepublic goods but have the consumption characteristics of private goods. Hence, it i s difficult to create competitive pressures and signals within social protection delivery; v) Delivery o f social protection needs to be located within existing decentralization reforms. 41 Utilizing these issues as a framework o f analysis, the following questions arise: i)DoessocialprotectioninMalawihavealeadactorandinstitutionalhomethat enables efficient and effective social protection delivery (strong coordination and management)? ii)Aresocialprotectionandsafetynetactors,ateachlevel,adequatelyresourcedand do they have enough voice? iii)How far are beneficiaries of social protection represented in the current institutional framework? iv) How well do current information systems for social protection work? v) I s there a strategy for M&E to ensure effective service delivery? I s it implemented? vi) How far does the delivery of social protection follow the process of decentralization and devolution o f service delivery to district assemblies? Institutional arrangementsfor safety nets and social protection The institutional arrangements for safety nets from 2002-2006 are shown in Figure 7. In 2002, the Safety Nets Unit was located in the Ministry o f Economic Planning and Development and its work was guidedby two committees: the National Safety Nets Steering Committee; and the Project Management Unit. In 2004, the institutional home o f the Safety Nets Unit was moved to the Department o f Poverty and Disaster Management Affairs within the Office of the President and Cabinet, ostensibly to give more profile to the unit. However, at this time, the roles and membership o f the two committees driving policy for social protection went unchanged. The focus o f the Unit, its name, and government policy more widely, remained on safety nets rather than social protection and the implementation systems for safety nets (via four committees each responsible for one component o f the system) were not changed. With the safety nets unit sitting in the Department o f Poverty and Disaster Management Affairs, the need to move from a safety nets approach to a social protection approach was re~ognized.~~ Further changes were made beginning of 2006 and the new institutional arrangements for delivering social protection are shown in Figure 8. The shift in language (from safety nets to social protection) reflects a commitment on the part o f government to move towards longer-term and developmental activities to both alleviate poverty and reduce it. 3'As the report is being edited, the Social Protection Unit has beenrelocatedto the Ministry of Economic Planningand Development. 42 u 0 0 ?I vi 0 0 PJ m .. b 4cd .3 0 m0 CElL Progress: Towards astrong leadandinstitutionalhome The inventory shows as many as 24 actors engaging in social protection. The problems faced in developing the inventory - for example overlapping activities and consortia of implementing partners across government, donors and NGOs - are an indication o f how difficult it i s to effectively coordinate all these activities. If finding out who i s doing what, when and where on social protection inMalawi i s difficult, coordinating all these activities i s a challenge of much greater magnitude. Similarly, the constitution of previous and current committees for safety nets and social protection demonstrate the wide range of stakeholders, both within government and beyond, that are involved in social protection. Within government there i s a very wide range o f sectors - with interests inboth social policy and economic policy. Trying to coordinate the activities o f a large range o f government departments with different (and potentially competing) sectoral priorities and budgets requires a strong (government) lead that can manage and prioritize between differentsectors. Thefirst implication for Malawi is that the Social Protection Unit is better located in a central ministry, and not a line ministry. Both the Ministry o f Economic Planning and Development and the Office o f the President and Cabinet are central ministries but the lesson i s that, even ifproblems are identified with the current location o f the unit, it should not move to a line ministry. The second implication is that the social protection framework, policy and program need signijkant political weight and resources behind it. Even when the unit was housed in the central Ministry o f Economic Planning and Development, coordination remained problematic. Why? International experience shows that social service ministries tend to be weak and under-resourced. Drawing on Kambewa (2005), a comparison o f prioritization, resources and expenditure under the four components of the safety net system (Figure 7) show that expenditure on components 1 (targeted inputs delivered by MoAFS) and 2 (public works delivered by MASAF and District Assemblies) were significant, whilst the direct welfare program delivered by the MoWCD (in spite o f the fact that it ought to form the core o f support to the pooresthouseholds) was much, much smaller. The new institutional arrangements are a step towards achieving greater authority and clout for social protection. The most important changes have been those at committee level. The new National Social Protection Steering Committee has new membership (see Annex 4.1) comprised mainly o f permanent secretaries, and the committee i s chaired by the Chief Secretary (the head o f the Civil Service and senior civil servant inthe Office o f the President and Cabinet). The National Social Protection Technical Committee also has revised membership and has new TORSthat represent a shift from project management to playing a technical policy advisory role. The composition of both committees also reflects the need for a government-driven and government-ownedprocess. For example, out o f 18 members o f the new Steering Committee 13 are from Government. Similarly, out o f 23 members o f the new Technical Committee, 14 are from government, whilst the others are donor agencies andNGOs. How far the committee chairs will be able to manage: i)the different priorities o f various government stakeholders and; ii)the (sometimes peremptory) influence o f donors, remains to be seen. 45 Remaining challengesfor institutionalizingsocial protection Whilst signi9cant progress is being made to strengthen the institutional arrangements for social protection, there are still challenges ahead. Inthe case o f some o f these challenges, progress cannot be made until the social protection framework i s finalized and becomes operational. However, it i s still important to consider various challenges in order to make recommendationsthat may influence the policy process. Implementation systems and stakeholder roles are not fully clarified Despite the establishment of the NSPSC and NSPTC, there is no clariJied and agreed new institutional framework for implementation of social protection. Thus in Figure 8, the implementation box i s empty. Without an implementation strategy and framework, there remain a range o f uncoordinated, overlapping and sometimes contradictory activities. This has meant that actors engaging in social protection are continuing to work on a project-by- project basis. All new social protection activities are not yet placed before the NSPSC/NSPTC for decision making. It is expected that discussions on appropriate institutional arrangements for coordination and implementation will become more pronounced as the policy and program formulation process proceeds. Donor roles remain unclear in this regard. Whilst the social protection policy process i s largely government-owned and government-driven, donors are influential and not always so well coordinated. Donors have different approaches to and activities within social protection. Whilst there are some positive examples o f donor coordination and harmonization, for example inM V A C and JEFAP (see Box 5. l), there are also examples o f lack o f coordination between donors that may be holding back steps towards a long-term social protection program inMalawi. i)Some donorsfocus on particular instruments (that they often call `poverty reduction programs' rather than `social protection') and therefore work through different institutional arrangements to those associated with the SPU / NSPSC / NSPTC; ii)SomedonorsseeWFP'sschoolfeedingprogramaspartofsocialprotectionwhilst others regard it as an educationprogram; iii)UNICEFfocusesontheverypooresthouseholdsandensuringtheirconsumption, whilst DFID and the World Bank focus more on enhancing productive activity and ex-ante riskmanagement; iv) Donor policy positions are sometimes out of sync with one another and those of government (the EU Public Works program pays more than MASAF creating confusion at local level; the general WB policy on public works i s to pay below the prevailing agricultural wage rate); v) Different donorsfocus on different elements or dimensions of thesocialprotection agenda, and each tries to pushforward its own priorities rather than attempting to speak ina unifiedvoice with Government. 46 Box 5.1: Examplesof donor coordinationand harmonizationinMalawi In recent years, the allocation of food aid has been rationalized, under the overall guidance o f the Joint Emergency Food Aid Program (JEFAP). The JEFAP includes representatives o f various government ministries, the donor community and a consortium o f NGOs. The objectives o f JEFAP were to source and coordinate the distribution o f relief food items to the most vulnerable and food insecure households in Malawi in a spirit of transparency and accountability. The creation of this government task force at national level promoted agreement and coordination among stakeholders o f the humanitarian response to emergencies. The JEFAP structure has also provided a crucial forum to develop and refine operational targeting criteria. The Malawi Vulnerability Assessment Committee (MVAC) i s the main institutional process generating information for national-level targeting decisions. The M V A C carries out an annual livelihoods analysis to assess which areas are most in need o f assistance. The main methodology adopted by MVAC i s a Livelihoods-Based Vulnerability Approach, known as the Household Economy Approach (HEA). This livelihoods-based vulnerability approach generates information and analysis that provides a foundation for better understanding the dynamics o f change and vulnerability within households. Source: GoM / WB(2006): 216-217 Social protectionpolicyprocessesand implementationare under-resourced Previous analysis of the institutional arrangements for delivering social protection in January 2005 showed that the work of the safety nets unit was severely compromised by resource limitations -particularly staffing (Jones et a1 2005). The analysis suggested that changes in senior management, training and other commitments resulted in little progress being made in the inception phases o f the development o f a more strategic and policy influencing approach to social protection. In January 2005, the government commitment for senior level support for safety nets appeared to be `in limbo' due to change in senior management. Resolution o f senior management appeared to be unlikely for a period o f months. The safety nets unit was operating with three staff members. It was suggested that the range of work coveredby the safety nets unitwas very large and there were concerns that the unit's activities were spread too thinly, preventing them from making significant progress on any issue. At the time of writing, whilst signiJicantprogress has been made regarding the commitment to social protection at a senior level, staff resourcing issues have not yet been resolved. In December 2006, the Social ProtectionUnitwas operating with only one team member (out o f an agreed IO). Staff turnover remains a problem with staff either being moved to other departments, or returning to studies. Untilthis problem is addressed, there i s little likelihood o f the social protection framework moving forward without the support o f external consultants. The use o f consultants (including those producing this report) detracts significantly from attempts to move towards a government-owned and government-driven process, but has been important to enable progress to be made in policy and program formulation. Regarding new staff recruitment, the Civil Service Commission has advertised for three posts and one postinghas beenapproved. 47 Institutionalcoordinationand roles of decentralizedinstitutions Beyond the Social Protection Unit itseK there is an important set of questions about the structuresfor operational coordination between the large implementing agencies and sector agencies for the new national program. Obviously, the existing Technical Committee and Steering Committee for Social Protection will play an important role in the planning and coordination o f programs. Linked to this i s the role o f the district level and the district level assemblies, and to what degree design and delivery o f social protection should and can be decentralized. While design o f program activities might be part of program activities o f district development plans, full decentralization might not be feasible (because o f lack o f capacity for designing) or desirable (due to likely quality control issues, loss o f economics o f scale, and coherence o f programming across districts leading to fragmentation). It would be important however to find a way o f involving the districts not only in implementation o f programs but also intheir strategic design. As part o f Malawi's Poverty Reduction Strategy, decentralization, via devolution, has formed a key component o f Malawi's strategy to combat poverty. This included the decentralization o f service delivery to district assemblies as a cornerstone o f Government development policy in recent years. Two issues have emerged regarding the role o f district assemblies insocial protection delivery.32 First, whilst there is a legalframework for a comprehensive decentralization of government functions, reviews of theprocess suggest a lack of genuine internal commitment at the centre of government in terms of genuine transfer of powers and resources (Cross and Kutengule, 2001). Ellis et a1 (2003) argue that PRSPs are heavily infused with the rhetoric o f local ownership, participation and the empowerment o f the disadvantaged, but in reality, decentralization inMalawi has beenpromoted by donors, pursued with limitedenthusiasm by central government and i s not a response to spontaneous popular demand for local autonomy indecision-makingprocesses. They also suggest that the decentralization of fiscal processes inMalawi are viewed largely as a mechanism for increasingthe potential tax yield onwhich district assemblies can rely. Sectoral analyses o f decentralization (for example, Kaarhus and Nyirenda (2006) on agriculture, Davies et a1 (2003) on education) mirror these general concerns about the influence and commitment / reluctance o f various parties involved. They also highlight ambiguity over the exact roles o f different parties, and the need for clear leadership, strategy and reliable support (especially funding) for successful and efficient decentralization. The second issue is the question of district level capacity to design and deliver social protection. A synopsis o f findings and critical questions emerging from interviews, especially at district level in Nkhotakota and Mchinji i s given in Box 5.2. It i s clear that capacity at district level i s highly variable with implications for the implementation of social protection programming. However, in various interviews at district levels it has been argued that capacity for implementation i s often present, and that capacity issues are being used as a smokescreen by: i)Centralpowerswhoarereluctanttodevolvepowerandresources; 32There are additional questions about the links between,different actors and operations on the ground, about the coordination of policy and operations at district and local level that are not addressed here (due to limited field assessment). It is particularly unclear how far relationships between the state, district assemblies and the district commissioner are changing, and how this will affect social protection delivery. 48 ii)Variousotherimplementersofsocialprotection,particularlysomeNGOswhoare involved indelivery o f services themselves; and iii)Donorswhodonottrustdistrictassembliestoaccountforfunds. In theface of concerns about limited capacity, some actors are by-passing local government, rather than investing in the capacity of local government. Whilst there may be arguments for doing so during emergencies, i.e. when there i s not time to buildcapacity, it i s less clear what the rationale for doing so is for long-term, developmental social protection programming. The major lesson is that actions under social protection programs, including those by donors and NGOs, should seek to support district level implementation, rather than bypass it. Box 5.2: Social Protection at District Level: Opportunities and Challenges National and district level vulnerability assessment There i s more disagreement than agreement on the national assessment o f district level vulnerability. In many cases, district assemblies disagree with the national vulnerability assessment o f their districts. There are two issues here. The first i s that where social protection interventions are directed by vulnerability assessment it i s difficult to map resource flow if the national assessment is ignored. The second issue i s on the methodology o f the assessment. Why do district assemblies often disagree with the assessments?Is it because they don't know the methodology or because they are not involved in the assessment? Whatever the case, it i s vital for the framework and policy to ensure uniformity o f the `evidence' used to guide SP interventions. This leads to the next critical issue Theplace of the district assemblies in the SP institutional set up All major SP interventions including those by NGOs are implemented by district assemblies or use district assembly staff. These district level staff have implementation lessons that needto be fed into the policy making and program design processes. While NGOs (represented inthe Technical Committee) can present their lessons and views, district assemblies cannot. Whilst the Ministry o f Local Government i s part o f the Technical and Steering Committees, it i s not clear how the arrangement allows the DAs to provide their inputs in the central Government processes. Some representation from the Malawi Local Government Association at national level would help. For example, the Malawi Local Government Association could also create a forum where lessons learnt from implementing various SP interventions can be fed into the process. Design of SP interventions at district level Again related to the first point, who designs SP interventions and on what basis? Given that DAs are in charge o f development issues at the district level, they should be more firmly involvedindesigning also SP interventions suitedto the needs of their vulnerable and needy. If such interventions came out o f local government structures like VDC and ADCs it might help incorporation o f beneficiary views in the design. However, some level central control of planning and design would be required. Funding arrangements of SP interventions at district level 49 Ideally, SP interventions should be designed and implemented at district level. However, regardless of where the design o f the interventions i s coming from, there i s the issue o f funding o f SP interventions and the flow o f funds to the district level. The vision o f the SP inMalawi i s to have separate, ring-fenced and basket funding for SP interventions. The question i s how can these resources reach the DAs?There are two possibilities. The first i s that resources can be housed in the DoPDMA but controlled by the NSPSC. Transfer of resources would be direct to the DAs using a special SP window in the Local Government Development Fund (LGDF - possibly under the current `committee' window - inparallel with MASAF). The other alternative i s that Treasury transfers the resources directly to DAs and the SPU provides policy advice and support to the development of SP programming in District Development Plans. Practitioners preferred the latter to enable the SPU concentrate more on policy issues than financial management. A third option is that the resources are controlled by the Local Government Finance Committee and distributedto district on the basis o f a formula that i s remotely related to vulnerability and dynamic poverty. Coordination of SP interventions at district level Interviewsat district level suggest some frustration inthe relationshipbetweenNGOs and DAs. These are largely around a lack o f accountability of NGOs to the district assemblies where they work and a similar lack o f reporting. These implementation coordination problems should be addressedmore firmly as the process moves towards definingthe new nationalprogram. Beneficiary voices are not heard in social protection policy and programming Poor and vulnerable groups in Malawi tend to have weak representation in decision making arenas and policy processes. They are also poorly organized. Mathauer (2002) argues that beneficiary groups tend not to be vocal - they don't voice their concerns, they don't demand safety net services, they rarely complain about inadequate services. Instead they tend to be passive beneficiaries o f social protection. Hence, there i s a need to facilitate mechanisms to ensure that beneficiaries are better represented in decision making about social protection - both in policy formulation at national level and in programming o f District Development Plans at the local level. The Social Protection Steering and Technical Committees have attempted to achieve this by inviting the NGO umbrella, CONGOMA, to be a member o f the Steering Committee and a number o f NGOs that seek to represent the poor, tojoin the Technical Committee. Beyond representation on these social protection committees, some NGOs in Malawi, for example, ActionAid and PLAN, are actively seeking to support poor and vulnerable groups' voice in social protection dialogue in a number o f other ways - including articulating a harmonized NGO position on social protection and lobbying for greater representation o f vulnerable groups within the main structures for social protection policy-making. NGOshave also recently agreed on the need to create more political space for dialogue about social protection. Similarly, Mathauer raises questions about the politics o f social protection and 50 argues that there is a danger that beneficiaries may be taken over by more poweqfiul, but less needy, groups advocatingfor their own interests (2002, p.7). But this raises questions about who can claim to represent potential beneficiary groups in dialogues around social protection: i)IsitNGOsorotherformsofcivilsociety? ii)Isitrepresentativesfromdistrictassembliesandtraditionalleaders? iii)Orpoliticians? At present there is limited space for NGOs on the various social protection committees and even less space for political representatives. It i s clear that social protection i s a highly politicized issue - the political debate and posturing over the implementation o f the government's input subsidy voucher program are testimony to this. The way forward is not clear, however, and needs more discussion. Suggestions at stakeholder workshops were to explore options for working with all-party parliamentary committees to try and garner cross- party political support for social protection. The involvement o f civil society has also been strengthened and three regional workshops and one civil society workshop will take place in 2007 as part of the planningprocess. Social protection monitoring strategy Theframework for analyzing institutional arrangements emphasizes that strong monitoring and evaluation is critical to ensure the cost-effective implementation of social protection programs. The review shows that monitoring and evaluation i s rarely taken as seriously as it needs to be in many o f the projects, and the coverage o f M&E i s very patchy.33 Many programs equate monitoring with reporting on how much food / cash / inputs etc were delivered where, when and how, rather than seriously evaluating the extent to which programs enabled risk reduction, mitigation and coping among the poor and vulnerable households. This is, inpart, symptomatic o f the prevailing emergency response which'tends to report on the outputs o f programs rather than the outcomes. There are examples of good practice in monitoring and evaluation, for example, Oxfam's cash transfers program and Concern Worldwide's food and cash transfer program. InMASAF, some progress was made in 2005 with the establishment of an M&E system (Jones et a1 2005). However, whilst specific sub-programs, notably public works, and fertilizer subsidies, have been the subject o f regular evaluation, others have received much less attention. For example, the impact o f social support programs on the poorest households is not well understood. There i s no formal mechanism to show how the proceeds from social support programs are transferred to beneficiaries. Regular, strategic and in-depth monitoring and evaluation of social protection interventions should underpin the social protection framework, policies andfuture programming. Efforts around M&E should be focused not on establishing a whole new M&Esystem, rather efforts should aim to harmonies and strengthen the existing range o f different information and M&E systems (e.g. linking the Joint Task Force Database on agricultural policies and programs with the SPU database on safety nets). 33With the exceptionof MASAF and some NGOs(e.g. Oxfam, ConcernWorldwide) little systematicwork was found on M&Eamongagencies involvedto support findings. 51 There are other conclusions too: Donor reporting should be better harmonized,particularly where there is basketfunding and where a consortia delivers social protection interventions. Currently, the capacity to do good M&E i s compromisedby the range and timing o f reporting requirements o f different donors. M&E can also be used to try and address the problem o f beneficiary voice identified above - for example, good M&E can move beyond leading questions about whether beneficiaries found transfers useful, to clearly understanding how transfers are used, their ramifications for wider livelihoods and local economies, and the gaps (risks) that remain in spite o f transfers. Questions remain about the relationship between the M&E group within MEPD and the Social ProtectionUnit and how far a coordinatedprocess for M&E can be established. 52 6. BROADPROGRAMOPTIONSAND AFFORDABILITY OF SOCIAL PROTECTION This section attempts to lay the basis for a discussion o f costs and targeting, by providing some broad program options for possible social protection and social transfers. The section does not, however, attempt to answer what level of finding i s affordable or advisable for Malawi. An assessment o f affordability would need to consider total annual costs o f social protection inrelation to GDP, government budget, donor assistance, and trade offs involved infunding socialprotectionversus other forms ofinvestment. Thepresent safety nets system costs USD134 million annually, which is about 6,5 of GDP, I 5 percent of annual budget, and about 34 percent total development assistance. It i s important to keep in mind that this amount includes emergency food aid and fertilizer input supply (among others). To put this figure in perspective, MPVA estimates that close to USD60 million would be needed annually to move the ultra poor (22 percent o f the population in 2005) out o f poverty through a cash transfer program (assuming no transactiodadministrative costs), or about 3 percent o f GDP and close to 7 percent o f annual budget.34Hence, a main issue in social protection i s to rationalize the existing system and move towards a different mix o f instruments and programs, rather than increasing the total spending on social protection. This would also require a consideration o f present level o f fertilizer subsidies and food subsidies (through ADMARC), which is not included in these costs. To this end, one could, as an example, consider raising the incomes o f a significant share o f the ultra poor to a minimumsubsistencelevel (food poverty line), or alternatively for some estimated proportion o f essential food requirements (e.g. to cover shortfalls during the four-month lean season). Given the number o f poor involved, a cash transfer program would probably have to reach a minimumo f 10-20 percent ofthe populationto be meaningf~l.~' It is important to be aware of the opportunity costs involved infunding social transfers at the expense of investments in other social sectors or more directly productivity enhancing activities (such as education, irrigation, roads), or other expenditures that could improve welfare of the poor at current income levels (such as rural health or water supply). An assessment o f such trade-offs between investments in different sectors i s difficult. But one may, for example, compare the investment in such a program with the current budgets and spending of the Ministry of Health or Ministry o f Education. There are important under- investments in these sectors. For example, it is argued that the recurrent costs o f primary education would need to grow with USD22 million by the year 2015 in order to achieve universal primary education (from the current costs o f USD46 million to USD64 million) (ref. Public expenditure review, 2007). Such trade-offs are important to think through, while determining a sound level o f investment. There is obviously no "right amount" to choose to spend on social protection. 341t.wouldcost USD315 million to similarly move out of poverty all the poor, i.e. 52 percent of the population (GoM/WB 2006:234) or 17percentof GDP. This would clearly not be financially feasible underpresent fiscal constraints. 35For illustrative purposes, a cashtransfer programdesignedto reachthe poorest 15 percent of the population, andtransfer USD3/month to eachperson(MK 420/month) would cost a bit less than USD 70 millionper year (assumingno transactionand administrative costs). This is aboutthe same level as provided monthly per person under the National cashtransfer pilot project (Mchinji pilot scheme). Suchaprogramwould represent about 3,5 percent of GDP, 8,5 percent ofthe annualbudget, and abit less than 20 percentof annual development assistance. This is clearly affordable. 53 Given three key target groups (destitutehltra poor, poor and transient poor), the following mightbe seenas significant andmeaningful options and scope for socialpr~tection.~~ First, there is a need to undertake a major rationalization of existing safety nets system, with a new mix of instruments andprograms, possibly at about the same level of investment. Some options to this end are proposed in Box 6.1 based on MPVA recommendations. Second, the new program would make the shift from (short term) safety nets to long-term and predictable social protection programs. Third, an overall goal could be to reduce poverty and enable the poor to escape from poverty and vulnerability. i)A majorobjectivecouldbeto significantly reduce therateofultrapoorand provide welfare support to those who are unable to construct a viable livelihood through a combination o f social transfers (conditionalhnconditional), and productivity enhancing instruments andmeasures to prevent asset erosion. Priority could be given to the "destitute" or chronic poor mainly by providing cash transfers or food transfers - as well as ensure complementary long term investment in human capital, access to health and education, support of asset transfers, labor-saving technologies, or securing rights over productive assets e.g. land. A cash transfer program to 10 percent poorest would cost in the order o f USD 40 million annually, or about 2 percent of GDP, 4.5 percent of the annual budget, and 10percent o ftotal development assistance; ii)Suchaprogramwouldprovidesubstantialadditionalsupportforpoor, transient and vulnerable poor with priority to the 30-40 percent poorest, by promoting livelihoods through productivity enhancing transfers that improve risk management, productive capacity, and asset bases. This would limit people falling into ultra poverty, and help poor people to graduate out of poverty or vulnerable circumstances (through e.g. a combination o f unconditionallconditional cash transfers, access to services, micro-credit, agricultural input, inputs for work, and direct asset building for communities or individuals (livestock, tools, equipment, infrastructure). Moreover, improve risk management and disaster reduction by preventing the erosion o f assets when households or communities are hit by adverse shocks, and enabling risk management that help them mitigate against, or cope with shocks. Additional objectives might be to address e.g. assistance to solve collective problems, and access to affordable services and markets. This social protection program would likely be the main responsibility o f several sector ministries and attract most key donors/NGOs involved in the safety nets program today - dependingon choice o f funding mechanisms and type o f instruments involved. Since few safety nets instrumentshave so far been adapted to the needs o f the ultra poor and the destitute, it i s suggested that this group would be prioritized through a new social transfer program.37An important share o f the poor and ultra poor groups i s located in urban areas (where there are also less potential benefits from public works and input subsidies.). The difficulties in identifying the poorest o f the poor has brought in the idea o f a three-prong 36 Even if such categories of householdscan be distinguishedthrough the MPVA, these are not necessarily discretecategories on the groundand easyto distinguishthrough implementation, whichtypically involvesa communitytargeting process. 37This concerns inparticulare.g. poor elderly, disabled, or infirm (iflimited support); labor and/or landand asset constrainedhouseholds,many amongwhichwould bepoor femaleandlor elderly-headed. 54 approach to targeting (a special version tested also in Mchinji) which uses a short-list o f proxy indicators, use o f community-developed poverty manifests, and verification o f identified beneficiaries using independent assessors or the entire community. The new social transfer program could be funded from a combination o f sources, such as rationalization o f the existing safety nets program, earmarked and additional government funding, and some additional donor funding. This would be inline with existing policies as laid down inthe MGDS, as well as the socialprotection framework. The program would include mechanisms for joint funding or basket funding- although many smaller schemes would likely continue to coexist with larger multi-donor program(s). The program would be decentralized in relation to planning and implementation with substantial participation o f people through appropriate social accountability measures. The vision o f the SP in Malawi is to have separate, ring-fenced and basket funding for SP interventions. How mightthese resources flow to the district assemblies? There are various possibilities of which two stand out: i)ResourcesarehousedintheSPUbutcontrolledbytheNSPSC. Transferof resources would be directed to the DAs using a special SP window in the Local Government Development Fund(LGDF - possibly under the current `committee' window -inparallel with MASAF); ii)TreasurytransferstheresourcesdirectlytoDAsandtheSPUprovidespolicy advice and support to the development o f SP programming in District Development Plans. Practitioners tended to prefer the latter of these approaches to enable the SPU concentrate more on policy issues than financial management. More specifically, the MPVA brought out the following recommendation for a national programs, the core of this being supported by this stock take (even if a set o f issues are pointed to - see next section). Box 6.1: Recommendationsfrom the MPVA 1, Put inplace a long-term, regular andpredictable system o f social protectionthat: i)Facilitategraduationoutofpovertybyactingasaspringboardthatallows households to take advantage o f opportunities for wealth creation while protecting against risk-induced hardship i.e. breaking the poverty trap for the chronically poor; ii)Preventerosionofhumanandphysicalcapitalthataccompaniesextreme iii)poverty; householdvulnerabilitiesbroughtaboutbyfrequentshockslike Address food crises and chronic illness; 2. Strengthen the role of diagnosis and evidence in the design of SP interventions as well as increase resources for monitoring and evaluation o f the interventions; 3. Adopt better targeting and prioritization inlight o f the pool o f poor people and 55 fiscal constraints; 4. Increase government ownership and leadership by establishing a single government-controlled SP program ; Introduce a mix o f interventions that would reach all vulnerable groups and helphouseholds better manage risks: i) Gradually introduce long-term, regular, reliable and predictable system o f direct transfers to the 10 percent o f the ultra poor in the form o f conditional cash (after harvest) and food (lean period) transfers using a two-stage targeting method - geographical community targeting (using a few categorical indicators); ii) Scaleupprogramsinnutritionandpreventivehealthtochildrenunder two years o f age usingthe Honduras model; iii) Scaleupbursariesprogramsfortheextremepoortobuildtheirassets; iv) Re-introduce private sector friendly TIP or a combination of targeted fertilizer subsidy-vouchers coupled with social cash transfers for purchase o f fertilizer; 4 Scale up inputs-for-work voucher programs in areas with inputs availability for the transient poor; vi) Scale up weather-based index insurance including the possibility o f re- insuringinthe internationalmarket as well as for riskmanagement o fthe transient poor; vii) Scale up public works program to increase its coverage because it is well targeted to cater for cover households faced with shocks; viii) Provide emergency aid (food during drought and disasters, blankets, shelter and water during disasters) and/or food price support when households are faced with covariate shocks; ix) Take up an index based insurance policy or purchase call options on the South African futures exchange market to cover maize production failure; x) Address the social protection function o f ADMARC -as it relates to marketing -as well as food subsidies; xi) Build on best practices in program design, adjusted to the country context, such as: i)pure transfer program should target very specific groups; ii)safety net expenditure should finance some productive investment to help graduation out of poverty; iii)programs should have multiplier effects and help e.g. risk reduction or diversification; iv) timing of intervention is important; v) simpler programs are often less administratively costly (Smith and Subbarao 2003 in MPVA, World Bank 2003, DFID2006b). Source: GoMlWB 2006 A key issue would be to determine the moreprecise mix of instruments, whatfunding levels of government and donors would be afordable and appropriate, and what implications this would havefor different development partners. This would need to start from the state o f need ingivenregions, risk factors, the nature o f the target groups, and the role o f allied social policies, and other politico-institutional factors. For example, the new mix o f instruments would need to address the fact that female headed households are much more prevalent among the ultra poor, and assess potential gender-biases inexisting instruments (e.g. types o f 56 public works) and how new instrumentscanbe made more gender neutral (e.g. well-designed cash transfers). An assessment o f trade offs between investments in social protection versus investments in other related sectors would also be necessary to determine this. To this end, theory and evidence suggest that the trade-offs between traditional safety net goals and growthlefficiency have probably been exaggerated. Safety nets often have the potential to enhance growth - with careful design, follow-through and monitoring (World Bank 2003, DFID2006b, IFPRI2004). Figure 9 assists in setting the scope of social protection inline with the recommendation of the National Social Protection Technical Committee that social protection should have clear boundaries and that the interventions envisioned should favor a shift away from relief (NSPTC Minutes 21-09-2006). One may identify a core o f social protection activities that include direct welfare transfers (such as cash transfers) and productivity enhancing transfers (for example inputs distribution and public works programs that create assets) (ref. e.g. Kambewa, 2005, Devereux 2006, as well as the G o W B 2006). Coherent and progressive social protection interventions have strong positive linkages with other economic and social policies, and with disaster management, and through synergies with these policies can contribute to economic growth. As reflected in the Social Protection Framework, one should take into account the potentially broad scope o f social protection objectives and interventions as outlined earlier in the report, and their overlaps with other areas o f national policy. Social protection and social policy: Policies for improved social service provision, for example, primary education and health, are not directly concerned with social protection but really constitute mainstream social policy. Box 6.2 HondurasCommunity-based IntegratedChildCare Program This program implementedby the Ministry o fHealth inHonduras is widely perceived as a `good practice' program inpreventive health and nutrition care. Itrelies on volunteers to pro- actively engage boththe families and communities to monitor and maintain growth of children less than 2 years o f age. Sick children under 5 are also treated or referred to health services. The volunteers use a simple and structured counseling approach with families - based on a diagnosis system to identify inadequate monthly growth and their causes. The treating relates to causes, not short term symptoms. The volunteers also interact at community and district levels to analyze and create awareness about the causes o f poor child growth that are beyond the family's ability to address. The long-term annual recurrent cost o f a childunder two that take part inthe program i s USD 6.8. Source: Fiedler, J. L. (2003) cited in GoM/WB (2006) However, social policies specifically aimed at tackling poverty, exclusion and vulnerability, such as interventions to combat chronic malnutrition, improve nutrition awareness and address an over-dependence on maize, may fall withinthe social protectionremit. Moreover, newer evidence from the international literature suggests that for example cash transfers and vouchers can improve human development, by addressing demand-side issues (e.g. through soft conditionality, DFID2006). A significant national and targeted nutritionprogram should be part of, or closely associated with the new social protection, possibly adapting the Honduras model to the Malawi context (see Box 6.2). 57 Figure 9: The scope of social protection nutritionprograms, school inputssubsidies; maize options, feeding, integrated SP programs weather-based risk insurance, integrated SP programs Health, education Macro-economic management, trade, Cashandin-kind donations and Social protection and economic policy: There are also aspects o f economic policy that overlap with social protection but overall are likely to fall outside the scope o f social protection policy for example agricultural extension, and general aspects o f food security, food market management in support o f social protection, such as futures markets, weather insurance, well-managed financial or food reserves, and support to private traders. Moreover, economic policies that foster markets and the private sector may be essential to make social transfers functioning efficiently. Hence, agricultural and trade policies need to be conducive. Social protection can be designed to support long-term growth as reflected in the Malawi Growth and Development Strategy (and in the literature, World Bank 2005, IFPRI 2004, DFID2006). 58 Social protection, risk and disaster management; These are closely related areas. It i s essential that emergency responses are underpinned by an effective long-term social protection program, since the same type o f interventions that protect and promote livelihoods in a routine way can help reduce their vulnerability to hazards that cause disasters and accelerate recovery after disasters. The overlaps between these different areas are shown inFigure 9 and include activities such as school feeding programs to increase enrolment and improve attendance, and inputs subsidies that aim to both improve access to inputs whilst simultaneously strengthening private sector fertilizer markets (ifthe private sector i s involved). Inpractice, it will be necessary to attribute responsibility for a particular social protection intervention through mutual agreement between government sector agencies involved, taking account of whether the primary objective is a social protection one or not. Optionsfor a nationalsocialprotectionpilot program Inorder to determine the best options for the National Social Protection Program, a national pilot program has been started. Certain concerns can been raised, however, about the degree to which the existing national pilot (the Mchinji scheme) can provide all the answers required. The approach i s based on a single-instrument approach (cash transfer), and initially focused only on one specific district. The recent expansion o f the pilot to other district may help bring broader experiences, while assessment o f only one instrument may still limit the lessons that can be drawn. The pilot i s now being evaluated. The findings o f the stocktake suggest a need to differentiate between a `pilot project' and a `pilot program'. A pilot program may in this regard involve actions at different levels (district and national) and draw on different instruments and experiments with different institutional arrangements (multi- instrumentapproach). National cash transferspilot project The Cabinet's approval o f the expansion the Mchinji social cash transfers pilot might be one way forward. The main advantages of this approach are that: i) The pilot can be expanded very quickly (steps to expand from Mchinji have already been agreed inCabinet); ii) It can help us understand how well cash can work under different levels of poverty, and with different levels o f institutional and implementation capacity, and involve less transactioncosts; iii) Additionalfactorscouldbeassessed-forexampletheeffectofthesizeandtimingof transfers (e.g. monthly, quarterly, or biannual). An expandedpilot wouldprovide an opportunity tofollow households through thepilot, and facilitate a discussion about choice of target categories (for example, help discussion about what proxy indicators are most suitedfor identifyingthe ultra poor). This will help to identify objectives and target groups for cash transfers in the future. Overall, this cash pilot would enable government to make informed policy decisions (based on economic efficiency and general effectiveness rather than ideology about whether people should do public works in 59 return for income support), about whether to target using labor availability in the household, and whether to target only the ultra-poor or most poor households. The disadvantage o f this single-instrument approach i s that it i s less likely (than the alternative below) to enable improved coordination between different instruments,programs and stakeholders. National social protection pilot program Integration o f instruments to address complex context could be done either through separate single-instrumentprograms that come together at local level, or through programs that adopt an integrated approach. Ad hoc uncoordinated and sometimes competing interventions are common in Malawi. Each one o f these generally addresses one o f the many needs of the target group leaving many (sometimes more pressing) needs. Evaluations have shown that they are often not effective indealing with poverty even inthe short-term. Kambewa quoted by the MPVA recommends L "There is a need to have a clear structure at district level that will coordinate safety nets programs to avoid having ad hoc type of programs which are likely to have little impact in isolation yet they havepotential of having signijkant impact if they combine efforts" (GOM/WB (2006):214). The case for integrated and synchronizedinterventionsand programs The key challenge here is institutional i.e. working in synchronized manners especially if programs are not implemented by a consortium or ifthey arefinanced by different donors or designed independently. District assemblies are not currently in a good position to negotiate for synchronization unless they are also involved in the design o f the SP Program at district level. So far there i s no talk o f district-level design o f safety nets or program design coordination. The need to use national level vulnerability assessment for equity sake and limited scope of involving district assemblies in the design o f national level a SP Program makes the synchronization o f interventions at community level rather remote. However, for a successful SP program, the involvement o f district assemblies i s crucial. The MPVA arguesfor simple interventions in order to deal with administrative constraints (ref. GOWWB 2006, World Bank 2003~).A number o f not-so-simple interventions have also been implemented in Malawi on pilot basis. Evaluations o f these indicate that effects o f these interventions may be more sustainable at household level. However, the programs are relatively more expensive (but not as expensive if each component i s implemented separately). In some cases, it i s simply the use o f extension services or civic educators or the linking o f the participants to the market or financial services that make a difference. Ifwell- designed, and after learning from projects like the I-Life, an integrated intervention is likely to be better suited for moving households out o f poverty and into a sustainable, resilient and independent livelihood. In conclusion, an alternative to an expanded cashpilot would be apilot program with two or more coordinated activities in a number of districts. At district level a social protection committee would be established with membership from the district assemblies, local government officials and civil society (NGOs). There would also be technical support from national government (SPU, line ministries and MVAC). Supported by capacity building, this committee could design a program o f activities under the social protection pillar in their district development plans (or outside o f the district development plans if the timing did not 60 allow inclusion). Gradually, as capacity increases, districts could decide their own mix of social protection activities but these should meet the needs o f each identified group. These programs could then be funded in various ways. At a national level, donors could contribute to a basket fundthat is inpart dispersed to the district level. Each district could be monitored throughout the process by a national monitoring system, and, after a period, evidence from each district about the mix of instruments, coverage, implementation arrangements, and funding mechanisms would be collated to assess achievements and degree o f uniformity across districts. This would help to identify best practice and to define a common but flexible national program, with large degree of involvement of the district level. Onthe negative side, this would be a long and complex pilot process and possibly carry higher administrative costs. It would require real commitment to decentralization on the part o f the national government and longer-term fundingfrom donors. The benefits though would be increased capacity at district level and the devolution of decision-making closer to communities themselves. At the same time, the program would yield a much greater understanding o f the most appropriate mix of instruments and mechanisms than would be possible through a single-instrument pilot project. Such a program would include systems of participation and social accountability to foster civil society and community actor involvement at local level. 61 7. CHALLENGES FOR SP PROGRAMFORMULATION: CONCLUSIONS AND RECOMMENDATIONS The stocktake has taken advantage o f work already done and attempts to provide information to help move the agenda forward. The key conclusions and recommendationsare as follows: Matchbetter the rangeof social protectioninterventions with the povertyand vulnerabilityprofile i) The main message is that the current mix of interventions does not match up with the critical needs of the poor - both chronic/ultra poor and transient poor. The following i s a summary o f the suggestions made on the mix o f instruments gleaned from various works, includingthe assessment inthe MPVA o f efficiency and effectiveness o f specific safety net instruments (see Box 6.1); ii) Public works programs should not necessarily target the chronically poor households but the poor and transient poor. The wage rate should relate to their needs and should assist them graduate out o f poverty faster to reduce the pool o f households in SP. Ifpublic works programs are to benefitthe poorest households, evidence suggests that they need to be longer-term programs to provide some income assurance for chronically poor households. Specific measures would be required to adapt public works to the needs and time constraints o f women and female headedhouseholds; iii) Cashasaform of directwelfaretransfer toboththeultrapoor andthepoor is likely to be most effective for social protection where there arefunctioning food markets. However, most cash transfers thus far have been short-term emergency responses. The Mchinji pilot is still in its early stages and i s presently being evaluated. We will soon know more about the impact on markets, prices, consumption and investment that result from cash transfers in Malawi. Wider external monitoring and evaluation o f the Mchinji pilot is critical and was being planned at the time o f the stocktake in November whilst internal M&E concentrated solely on operations, i.e. `on the monitoring o f activities, outputs, costs and cost-effectiveness' and not on impacts (Schubert 2006a); iv) Combiningfood and cash transfers using variable levels of transfers on the basis of food prices has proved successful in separate pilots in Malawi. But the approach requires a lot o f flexibility if food prices escalate over the lean period. Once a good mix o f interventions i s in place, including those that can stabilize grain prices even in the lean season, the variable cash transfer option may not be too expensive or problematic; v) A scaled up cash transfer system has the potential of attracting private sector participation, especially financial institutions/rural banks. It i s therefore imperative that the option o f involving the financial sector inthe design o f the SP interventions be considered seriously; vi) Inputs subsidies have good objectives but are expensive due to leakages, on top of mistargeting. The MPVA has an extensive discussion o f pros and cons, and includes them in its list o f a possible mix o f interventions but highlights the need 62 to ensure that the current subsidy-voucher program be redesigned to address the gap in coverage, as the current scheme omits cash-constrained small holders (see Box 6.1). The monitoring and evaluation o f the 2006-2007 season inputs voucher scheme will provide more up to date data about the effectiveness and relative cost o f subsidies as part o f social protection; vii) School feeding has positive impacts, and deserves to be part of an education policy. But its costs are comparable to primary education As a social protection instrument it is not well targeted, and there is little hard evidence about its impact on nutrition. Therefore, the link between school feeding and social protection might make more sense by viewing school feeding as a response to specific covariate shocks (those felt by many households at the same time) and individual household shocks (loss o f parents) that effect school enrolment and attendance; viii) A nutritional program aimed towards mothers and children under 2 years, based for example on the Honduras model, might be more effective as a social protection program than schoolfeeding: ix) It makes little sense to argue that there are too many or too few diflerent interventions. It is more important that each intervention has defined a clear target group. It i s very difficult though to identify different target groups as they are scattered through the poverty profile and geographically. Moreover, the transient poor group is fluid judging from the MPVA, MVAC data and Sharma (2005); x) Targeting remains a key problem. More work i s needed to minimize targeting errors during geographic and beneficiary targeting. It is critical that targeting considers the gender and age specific issues o f poverty, and choice o f approaches. There are likely to be both exclusion and inclusion errors in most programs. Systems o f targeting are being developed and adapted to local context (e.g. Mchinji, Concern, Oxfam for cash transfer), even if lessons can also be drawn from other countries; xi) Thepolitics of geographic targeting at district level could be dealt with by linking the SPU, MVAC and district assemblies together more closely. Stakeholders should be assured that all groups in poverty will be dealt with using appropriate SP instruments. Both within district and between district, the pressure to include non-deserving participants and exclude deservingparticipants can be minimized if few community-derived and easily understood proxy indicators `are used, and selected participants are certified by the entire community before the transfer o f the benefits; xii) Social protection has moved up on thepolicy agenda of government and donors. At the time o fthe stocktake, whilst experience with cash was small and limited, it 38The MPVA reportsthat school feeding cost USD15 per pupil, excluding the cost of the iake-homepack given to girls and double orphan boys comparedto USD17.3 per child for primary educationitself. WFP can provide a nutritious breakfast to each child for a whole school year for USD9.40 per year per child, rising to USDlS/child/year for those girls and orphan boys who get the take home ration as well. These costs includethe direct food transfer to the children, but also all the direct and indirect operational and support costs (transport, GoM support, admin, etc). 63 was hugely influential in current dialogues about the new social protection program. There was, however, also considerable attention accorded to other safety nets instruments, reflecting recent evaluations o f e.g. public works, targeted nutrition, school feeding programs, and fertilizer subsidies. Identify a core set of social protection activities that are supplemented by economic, social and disaster management policies A critical issue is to determine the scope of social protection goals and activities within the context of the national policy and program. Defining social protection i s generally difficult, and i s especially so in Malawi. As the inventory shows, social protection i s defined in a plethora o f ways by different actors through a wide range o f projects. Make the scope o f social protection too wide, it simply becomes an alternative term for `poverty reduction' or `development'. Set the scope too narrowly and social protection equates to a very narrow set o f instruments, and does not provide an adequate response to the various risks that different kindsofpoor households face. A long-term social protection program should prioritize core activities but should also consider interventions in the overlapping circles as complementaiy for the sake of meeting all the needs of thepoor (ref. Fig 9). However, interventions inthe overlapping circles should mostly be included as priorities in the mainstream economic and social policies and programs. In particular, it i s recommended that `core social protection' includes a modified targeted nutrition programs (perhaps along the lines of the Honduras model) - developed in relation to the Food Security Policy. Excluded from the core activities should be fertilizer subsidies, and possibly also school feeding (depending on its design and targeting). Market- based instruments like weather-based insurance and grain storage options may also be covered under sector policies, yet important risk management and disaster reduction measures need to find a proper institutional home. This also goes for the social function o f ADMARC. Hence, these kinds o f measures should probably be considered elsewhere in the greater MGDS other than core social protection. The more narrow scope proposed here i s on the basis o f political and fiscal sustainability, but also in consideration o f the capacity o f the institutional arrangement. 39 Move programmingbeyond safety nets to social protection Whilst the W A C and other analyses o f vulnerability have been critical in demonstrating how vulnerable households are to impoverishment and how susceptible they are to shocks and stresses, it i s clear that there are many households that are not vulnerable to further impoverishment because they are already very poor. The implication o f this i s that social protection should not be solely about preventing people from sliding (deeper) into poverty (a safety nets approach) and helping them recover from shocks, it must also be able to address the chronic poverty by supporting the consumption needs o f the chronically poor and destitute. At the same time it is necessary to ask whether social protection should only enable households to meet their consumption needs, or whether it should enable them to move out o f poverty. Devereux and Macauslan (2006) refer to this issue as the need to identify whether 39The analysis of the institutional arrangements for delivering social protection highlights some, though not all, of ways in which politics permeates social protection programming. We hope that differentiating between core social protection and wider economic and social policies will help to get a commitment to a long-term core program and thereby reduce the scope o f social protection to become a political tool. 64 the objective o f social protection is to alleviate poverty or to reduce it. Many of the activities that are described as social protection by stakeholders in Malawi have been implemented as short-term and fairly ad hoc poverty alleviation responses - they have provided safety nets to households during times o f stress or shock. These could be defined as emergency responses and considered as activities more associated with disaster management and response although they have formed a significant part o f safety nets policies inMalawi inrecent years. Box 7.2 Malawi socialprotection framework - guidingprinciples4' Principle 1. Leadership - Strong national ownership and clear government leadership o f an agreed National Social Protection Policy and Program should provide for the coordination and alignment o f partners' long-term support under a single strategic framework, with an effective voice for civil society. Principle 2. Evidence-based Social Protection - Social Protection must be based on evidence and analysis o f who needs what type of assistance, when (and for how long), where and why? It shouldbe driven by needs, not by instruments; and should include an analysis of the cost-effectiveness o f alternative interventions. Principle 3. Beneficiaw preferences prioritized - Civil society, in general, and beneficiaries, in particular, should be consulted and closely involved in the design, planning, and implementation of Social Protection interventions. Principle 4. Protection and promotion of secure livelihoods - Social Protection should encompass provision for the very poorest, as well as protection andpromotion of secure livelihoods that aim to help as manypeople as possible to graduate out o fpoverty andreduce their vulnerability to livelihood shocks. Principle 5. Tackling social exclusion and marginalization - Social Protectionshould address social as well as economic vulnerabilities, by protecting disempowered groups such as vulnerable women and children, the disabled, andHIV/AIDSvictims, against discrimination andexploitation. Principle 6. From reactive to predictable social support - The Government and its partners should replace reactive and weakly coordinated short-term social support with timely, adequate, and predictable social support to the poorest and most vulnerable Malawians. Principle 7. Observance of constitutional riphts - Social Protection should promote the realization o f related constitutional rights enshrined in Chapter IV o f Malawi's Constitution, including rights to development (Section 30), access to information (Section 37), and administrative justice (Section 43); it should be easy-to-understand, simple to administer, accessible in terms of procedures, methods of communication, and the attitudes o f deliverypersonnel to beneficiaries inboth rural and urban areas, and should incorporate effectiveprocedures for handlinggrievances. Principle 8. Accountabilitv and transvarencv - Social Protection should be transparent, incorporating principles o f accurate and timely dissemination o f information; publicity o f instances involving abuse o f the system; disclosure o f the contract terms and unit costs o f non-governmental organizations ("NGOs") or private agencies selected to administer social grants; and transparency in the procurement process for the provision o f Social Protection services. Any targeting must be fair, and seen to be fair. Principle 9. Joint financial management and Monitoring & Evaluation puidelines - Government agencies and development partners involved in Social Protection should commit (to the extent that it i s practicable under the development partner organization's charter and operational procedures), to a common set of financial management, audit, progress, and monitoring & evaluation reporting formats, consistent with national and internationalguidelines and indicators. Principle 10. Common funding - Mechanisms should be put inplace to ensure that Social Protection receives predictable and institutionalized funding, under Government control and sustainable over the longterm. 40 This framework of guidingprincipleswere preparedthroughajoint government-developmentpartner process with DFIDas leadagent and funder. This draft version includesafew minor changesthe World Bank has proposedto the frameworkthat was discussed inthe National SteeringCommitteeinApril 2007.. 65 Continue steps to institutionalize social protection through the technical and steering committees and the role of the SPU i) The changing institutional structure o f the Social ProtectionUnit and various committees has placed social protection firmly at the centre o f the government's development agenda; ii) Socialprotectionpolicyiscross-cuttingandmulti-sectoralandrequiresinputsfrom different line ministries but the secretariat for social protection should sit ina non- line ministry/ department; iii) Articulatingtheprinciplesofsocialprotectionimplementationshouldbeapriority once other aspects o f the social protectionframework are inplace; iv) Capacity issues for social protection needsto be addressedbothwithinthe social protection unit, and at district level to enable effective service delivery; V I Mechanisms for delivering social protection at district levelneedto be more clearly articulated, and effective monitoringand evaluation systems developed; vi) There i s a need to find ways o f addressing / managing the politics o f social protection, rather than ignoring them. Current efforts to harmonize and better coordinate social protection in Malawi are on the right track and largely agreed by stakeholders The findings o f the stocktake also provide robust support for the principles o f coordination laid out inthe Malawi Social ProtectionFramework (Box 7.2). Challenges and ways forward for social protection i) The move of the safety nets unit to become the Social Protection Unit -first in OPC and now in MEPD - is a critical way forward in ensuring clear government leadership of social protection. The predominance of government stakeholders in the steering and technical committees, with additional members from the donor community and civil society have the potential to provide better coordination and alignment o fpartners' long-term support; ii) Thefindings emergingfrom thisstocktakearestrongevidenceof theimportanceof developing a more coherent national policy and program for social protection. Current range of activities and projectized approaches are costly, yet uncoordinated and projectized and ineffi~ient.~~Disparities in regional and district coverage, coupled with overlapping targeting systems and different policies regarding benefit 41It is not suggestedthat the activities of individual stakeholdersare not well-planned(for example, it would be unfairto makethat claim about the former safetynets unit programmingfor safety nets) but when all activities (of Government, donors andNGOs) are put together they appear adhoc anduncoordinated- in coverage, targeting, andimplementation. 66 rates all make for a systemthat does not match the poverty profile and the needs of differentvulnerable people. Itresults incompeting andreplicatedefforts; iii) The dijficulties faced in deriving evidencefor the stocktake demonstrate the importance of ensuring that decisions about social protection are based on evidence. The stocktake has been able to show that the current range o f safety nets does not adequately match up with the needs o f different households. The new social protection policy and program i s an opportunity to change this. Similarly, the stocktake also demonstrates the need for a synchronized, coordinated and effective set of M&E systems and reporting requirements that enable transparency and accountability inthe social protectionprogram; iv) Concerns that are widespread in Malawi about thefocus on emergency relief for dealing with chronic problems are wellfounded. A large proportion o f expenditure on social protection i s addressing long-term poverty and vulnerability with emergency resources. This means that social protection i s focused more on ensuring consumption than on longer-term asset building. There i s a need to ensure that future social protection programming i s more predictable and developmental and that emergency or disaster responses are built on top of social protection programming, rather than the other way around; V I The ways in which the voices of beneficiaries are represented in the social protection programming process need to be clariJed and articulated more clearly. At this stage, the current mechanism (NGO membership of committees) does not yet appear to be working and this mechanism i s disputedby district assemblies; vi) A long-term commitment is essentialfor turning the ad hoc safety netsprojects into the needed social protection program. Stakeholders agree that long-term commitment i s conditional on a credible social protection program emanating from a shared SP policy and framework. Government's long-term fundingcommitment i s conditional to a social protection program that attracts other stakeholders, especially donors. Donors, on the other hand, would commit long-term funding once Government commits long-term funding o f a social protection program. NGOs also would welcome long-term funding for social protectionprojects; vii) There is a needfor the government, through thepolicy designprocess, to determine how broad the scope of social protection should be, and how far up the poverty profile they want different kinds of instruments to reach. Inthis regard, a decision should be taken on whether households that are labor scarce generally are poorer or not, and whether cash transfer can work across groups (labor scarce or not); viii) Key challenges are to determine the more precise mix o f instruments (adjusted to the poverty and vulnerability profile), what funding levels of government and donors would be affordable and appropriate, what trade offs are involved between investment insocial protection versus investments inother related sectors, and what implications the new program would have for different development partners; ix) Finally, poverty and vulnerability have clear gender and generational dimensions, for example, female headed households are more prevalent among the ultra poor and destitute; young women are more susceptible to HIV/AIDS and risks o f falling 67 into destitution. Hence social protection policies need to be build on gender and age-disaggregated evidence, and be sensitive to gender inequalities. 68 References Cross, S. and Kutengule, M. 2001: Decentralization and Rural Livelihoods in Malawi, LADDERWorking Paper No.4, University o fEast Anglia, Norwich. Davies, L.,Harber, C. and Dzimbadzi, C. 2003: Educational Decentralization in Malawi: A Study ofProcess, Compare 33 (2): 139-154. Devereux, S. 2000: Social protectionmechanisms inSouthern Africa. Regional Hunger and Vulnerability Program www.waheng;a.net Devereux, S., Baulch, B., Phiri, A. and Sabates-Wheeler, R. 2000:Vulnerability to Chronic Poverty and Malnutrition inMalawi: A Report for DFID Malawi' Devereux, S and Macauslan, I.2006: Section 2 Review o f Social Protection Instrumentsin Malawi. In Vulnerability and Social Protection in Malawi, IDS DISCUSSION PAPER 387 by Stephen Devereux, Bob Baulch, Ian Macauslan, Alexander Phiri and Rachel Sabates- Wheeler, November 2006 DFID 2006: Using Social Transfers to Improve Human Development, DFID Briefing, February 2006, DFID, London DFID2006b: Social ProtectioninPoor Countries, Social Protection Briefing Note Series,No. 1,January 2006, DFID,London Ellis, F., Kutengule, M.and Nyasulu, A. 2003: Livelihoods and Rural Poverty Reduction in Malawi, WorldDevelopment 31 (9): 1495-1510. Fiedler, J. L. 2003: HNP Discussion Paper: A Cost Analysis o f the Honduras Community- BasedIntegrated Health Care Program, HNP Discussion Paper, World Bank GoM (1995): Policy Framework for Poverty alleviation program , Ministry o f Economic Planning and Development, Malawi. GOM/WB 2006: Malawi Poverty and Vulnerability Assessment - Investing in Our Future, Vol Iand 11, Report no. 36546-MWYRepublic o f MalawiNorld Bank, June 2006 GoM 2005: Malawi Growth and Development Strategy. Ministry o f Economic Planning and ' Development. GoM 2001: A relative profile o f poverty in Malawi, 1998: A quintile-based poverty analysis o f the Malawi Integrated Household Survey, 1997-1998 Government o f Malawi Poverty Monitoring System. http://www.nso.malawi,net/data on line/economics/ihs/relative poverty profile.pdf IFPRI 2004: Linking Safety Nets, Social Protection, and Poverty Reductions -Directionsfor Africa (edited by M. Adato, A. Ahmed, and F. Lund), 2020 Africa Conference Brief 12, IFPRI, Washington D.C. 69 Jones, S., Lumsden, F. and McCord, A. 2005: SNU / DoPDMA Support Consultancy Visit: Januav 2005 VisitReport Kaarhus, R. and Nyirenda, R. 2006: Decentralization in the Agricultural Sector in Malawi: Policies, Processes and Community Linkages: A Status Report, NORAGRIC Report N o 32., NorwegianUniversityo f Life Sciences Kambewa, P. 2005: A review o f Safety Net Programs for 2003-04. Prepared for the Safety Net Unit, Department o f Poverty and Disaster Management Affairs and Ministry o f Economic Planning and Development. Mathauer, I.2002: Institutional Analysis Toolkit for Safety Net Interventions: Institutional Assessment, Institutional Design, Institutional Capacity StrengtheningHDNSP, World Bank MEPDINSOIWB 2005: Poverty in Malawi - The Second Integrated Household Survey 2005: An Extract ofFindings,MEPD, National Statistical Office, World Bank Milner and Tsoka 2005: Targeting the Poorest o f the Poor for Insecticide Treated Nets in UnderstandingHealth and Education- Social Scientists perspectives NEC (2000): Safety Net Strategy for Malawi, 1 September, 2000, Lilongwe NEC, NSO & IFPRI 2000: The Profile o f Poverty in Malawi, 1998. Poverty Monitoring System. MalawiGovernment. NEC, NSO & IFPRI 2000 : A Relative Poverty Profile o f Malawi, 1998. Poverty Monitoring System. MalawiGovernment. Republic o f Malawi I World Bank 2006: Malawi: Poverty and Vulnerability Assessment: Investing in ourfuture, Draft for Discussion. Schubert. B. 2006a: Guidelines for Internal Monitoring o f the Mchinji Pilot Social Cash Transfer Scheme, Report prepared for the UNICEFI Mchinji Pilot, Lilongwe. Schubert, B. 2006b: Designing a pilot social cash transfer scheme for Malawi: Third and Final Report 1 August to 15 September 2006 Report on a consultancy for the Malawi Social Protection Steering Committee financed by UNICEF.42 World Bank 1999: Malawi A Safety Net Strategy for the Poorest, Report No: 19974 -MAI World Bank 2003: Targeted Transfers in Poor Countries: Revisiting Trade-offs and Policy Options, Social Safety Nets Primer Notes No. 13, World Bank World Bank 2003b: What Role for Safety Net Transfers in Very Low Income Countries? Safety Nets Primer Notes, No. 11, World Bank World Bank 2003c: Incentives and the Role of Institutions in the Provision o f Social Safety Nets, Social Safety Nets Primer Notes, 2003, World Bank World Bank 2005: Social Safety Nets, November 2005, World Bank 42There are four reports intotal but only those referenced inthe text have beenlisted here. 70 Appendix 1.Types of poor households The following householdtypology draws upon the Malawi Poverty and Vulnerability Assessment, andrepresents potential target groups for different social protection approaches. The typology may be utilizedfor policy formulation purposes (GOMNorld Bank 2006).43 Ultra-poor households (including destitute) are usually larger than average households and with higher dependency ratio, and characterized by a lack of assets, especially limited landholdings (land per capita). Many households are affected by or vulnerable to chronic illness, often the result o f HIV/AIDS, and a few are de facto or de jure child-headed households.M Others are female-headed households, often elderly women, with many dependents. Ultra poverty also comes more frequently with low education o f head of household, low degree o f diversification or formal employment/wage salary, and less involvement in tobacco farming. In these cases the main need o f the household i s to ensure consumption - particularly to ensure household entitlements to food - and this, alongside maintaining investments in human capital, should be one o f the main objectives of social protection for this group. A second priority i s protecting asset status. Direct transfers and inputs programs are core instruments. Households with labor could benefit from public works but it i s critical to ensure that the level o f payment and the work requirements do not become a tax on the ultra-poor that prevents them investing in their own assets and livelihoods activities. The timing o f public works is also critical. This group is heavily dependent on agriculture as the main source o f livelihood, and subsequently there i s greatest variation inlabor requirementsat different times of the year. It i s important that public works activities do not compete with the peak o f agricultural activities, even if this corresponds directly with their hungry months. Ifpublic works are not productivity-enhancing - i.e. they are poorly timed and compete with household activities, they create assets that are o f limited use to the community or that cannot be maintained - then it i s certainly better to give transfers without labor requirements. Withinthe ultra-poor group is a sub-group of households that canbe categorizedas destitute. These are ultra-poor households that have the least assets and land (per capita), have no labor and / or high dependency ratios, for example due to chronic illness. Poor female headed households and elderly and child headed households would be more frequent inthis category. For these households, the likelihood o f graduation from social protection support out of poverty is least likely. The numbers o f destitute households i s not clear but they are a clearly a category that are highon the agenda for social protection. Poor households or moderately poor households are households that fall above the ultra- poverty line, but are below thepoverty line. They have some assets (land and labor) but are exceedingly vulnerable to further impoverishment. These households are also heavily dependent on agriculture (or fishing) and are net consumers o f food. They enter the market for food when the price i s highest, and sell when the price is lowest. Foodproduction shocks 43 These categories are not necessarilyclearlydistinguishable as discretehousehold"types" onthe ground, and may"mask" specific groups. For exampleregardingadistinctionbetweenchronic andtransient poor, Devereux arguesthat householdsthat aremost vulnerableto shocks are those who are already poor, who face total destitutionwhen the next drought or shock comes along. Householdsabove the povertyline mayface the same shocks but be better placedto resistthese-their incomesand assets buffersare moreresilient-which is why they "churn" aroundthe povertyline.Those likely to bemost at risk are those facing"composite food insecurity" i.e.chronicallypoorwho are also exposedto livelihood shocks (Devereux, pers.message, 2007). 44 According to IHS2, there are very few child-headed households in Malawi but the few are among the very pooresthouseholds. 71 resultingfrom drought or lack o f inputs, and subsequentvolatility inthe price of staple foods are the major shock faced by poor households. Children inthese households may register for school but their attendance rates are poor. Child malnutrition remains a significant problem. The vulnerability o f this group o f poor households is increasing due to HIV/AIDS and other chronic illness. This leaves households with little resilience to shocks so that even small shocks, such as very slight variations in rainfall, can have serious long-term impacts on the well-being of the household. This makes poor households risk averse. The objective o f social protection for this group should be enabling them to deal with shocks or mitigate the impact o f risks. Examples include making fertilizer more affordable and, therefore, a less risky investment. Transient or at risk households: Data from Malawi also show that we need to be concerned both about people who are already poor, but also about those that are above the poverty line but are likely to be adversely affected by shocks and stresses. A panel study tracking 291 rural households shows that many o f them move in and out of poverty over time (Sharma et a1 2002). Similarly, the MPVA suggests that transient poor households can be identified using characteristics highlighted inthe determinants o f poverty analysis, complemented by a households - the transient poor - the objective of social protection should be mitigating the measure o f household's exposure to shocks (GoM / World Bank 2006: xxiv). For these impact o f shocks - for example through social insurance instruments and micro-finance. 72