Report No. 40048-MZ Mozambique Beating the Odds: Sustaining Inclusion in a Growing Economy A Mozambique Poverty, Gender, and Social Assessment (In Two Volumes) Volume I: Main Report February 2008 Africa Region Poverty Reduction and Economic Management Document of the World Bank TABLE OF CONTENTS VOLUME 1.MAINREPORT PREFACE ..................................................................................................................................................... X ABBREVIATIONS .................................................................................................................................. XI1 OVERVIEW ............................................................................................................................................ XI11 THERECORDOF ACHIEVEMENTS ............................................................................................................. XIV Higher Incomes and Better Living Conditions through 2003......... Access to Services Increased.. So Did Holdings of Assets and Durable Goods Why Was GrowthSo Effective TheLooming Threat of AIDS........................................................................................................... PURSUING PROPOORECONOMICGROWTH-IN AGRICULTUREAND THE PRIVATE SECTOR ....................xxiii XXIV Supporting a Diverse andDiversified Agricultural Sector,.............................................................. xxvi Supporting Market Development and New InstitutionalArrangementsfor Stronger Market Participation......................... ....... .............................................................. xxvii Supporting Income Diversific Lab BUILDINGHUMAN Increasing Public Spending and Getting More ValueJi. CAPITAL-BY GETTTNG PUBLIC SERVIC Monitor and Evaluate Government Spending......................... IMPROVINGGOVERNANCE AND ACCOUNTABILITY-BY GETTINGTHE STATE CLOSERTO ITSCITIZENS POOR ............................................................................................................................. XXXIII Making Decentralization Workfor the Poor ........................................ xxxiv Increasing Access to Justicefor the Poor Making Poor People's Land Rights Real.......................... 1. POVERTY'SMANY DIMENSIONSINMOZAMBIQUE ......................................................... 1 MONETARY DIMENSIONS POVERTY OF ......................................... ........................................................ HOUSEHOLDCHARACTERISTICS BY INCOME GROUP .................................................................................. 4 1 HUMAN DEVELOPMENT-LOOKING INCOME BEYOND ................................................................................ ............................................................................... 6 VULNERABILITY, SHOCKS, AND COPING STRATEGIES 8 Risks, Shocks,and Coping Strategies ................................................................................................... 8 Especially Vulnerable Groups............................................................................................................ 10 DRIVERS HOUSEHOLD OF CONSUMPTION 14 PERCEPTIONSOF POVERTY IN2006 .......................................................................................................... ................................................................................................. 16 Defining Poverty in 2006 Perceptions of Poverty a ................... 2. THE MECHANISMSOF SHARED GROWTH:STRUCTURALCHANGEAND HOUSEHOLDLIVELIHOODS ............................................................................................................... 23 THE ENVIRONMENT FORPOVERTY REDUCTION, 1997-2003 24 STRUCTURAL CHANGES INTHE MOZAMBICAN ECONOMY ....................................................................... .................................................................... 28 BEHINDTHE GROWTH STORY: CHANGES INHOUSEHOLD LIVELIHOODS .................................................. 33 Trends in the Labor Market................................................................................................................ 35 Changes in HouseholdLivelihoodPatterns........................................................................................ 40 Earnings: Returns to EconomicActivity of Household Members.. Labor Market, Livelihood Changes, Earnings, and the Poor GENDERDIMENSIONS PROPOORGROWTH MOZAMBIQUE OF IN ............................................................... The Contributionof Female Education to Growth. Gender-Disaggregated Trends in the Labor M a r k Livelihood Trends...... ....................................... Gender Equity and Poverty................................................................................................................. 53 CAN MOZAMBIQUE BEATTHEODDS? THEPROSPECTSFORPROPOORGROWTH ...................................... 55 iii 3 . AGRICULTURALGROWTH.DIVERSIFICATION. AND MOBILITY-A CLOSER LOOKAT RURALPOVERTY ................................................................................................................ 61 AGRICULTURAL GROWTH AND POVERTY REDUCTION. 1996-2002 .......................................................... 62 Agricultural Markets and Household Welfare.................................................................................... 64 Changes in the Structure of Rural Household Income........................................................................ 65 RURALPOVERTYDYNAMICS THEROLEOFDIVERSIFICATION. AND 2002-05 .......................................... 66 Chronic Poverty and Vulnerability in Rural Mozambique Rural Poverty Transitions Consistent with Perceptions of TheRole of Agricultural Markets and Off-farm Diversification THE WAY FORWARD FOR SUSTAININGRURALPOVERTY REDUCTION ..................................................... 75 Supporting a Diverse and DiversiJed Agricultural Sector................................................................. 76 Supporting Market Development andNew InstitutionalArrangementsfor Stronger Market Participation....................................................................................................................................... 77 Supporting Income Diversificationthrough Wage-Employment Opportunities ................................. 78 4. GOVERNMENTEXPENDITUREPOLICIESAND THE POOR .......................................... 79 THE STRUCTUREOF PUBLIC EXPENDITURE 80 EDUCATION .............................................................................................................................................. .............................................................................................. 83 Government Policy Access to Education Servic .................................. 87 Eficiency in the Education Implicationsfor the Poor Perceptions on Cha HEALTH .................................................................................................................................................... 94 Government Policy and Spending Patterns ........................................................................................ 95 Access to Health Services ................................................................................................................... 97 Health Outcomes................................................................................................................................. 99 Implicationsfor the Poor............................ Perceptions of Changes in Health Services AGRICULTURALEXTENSION-FORTHEPOOR. NOTTHEA Government Policy and Spending in Extension........... Access to ExtensionServices....................... Implicationsfor the Poor .................................................................................................................. 112 WATER SUPPLY-A KEY OBJECTIVE OF GOVERNMENT POLICY ............................................................ 114 Despite Progress. Access to Safe Quality Water in Poor and Rural Areas Is Still Limited.............. 115 WHAT IS NEEDED AN EFFECTIVE FOR GOVERNMENT PROPOORSPENDING?........................................... 116 IncreasedSpendingand Valuefor Money in the Public Sector .............117 5. GETTING THE STATE CLOSER TO ITS POOR CITIZENS ............................................. 120 MAKINGDECENTRALIZATION FORTHEPOOR WORK ............................................................................. 121 Key Features of Decentralization So Far ......................................................................................... 122 Responding Better to Consumers and Citizens .................123 StrengtheningState and Local Capacity 126 Improving Representation and Participation................................................................. 127 Strengtheningthe Incentivesfor Participation and Implementation................................................. 128 Learningfrom Pilots......................................................................................................................... 129 INCREASING ACCESS JUSTICEFORTHEPOOR TO .................................................................................... 130 DISSEMINATING INFORMATION ON RIGHTS AND RESPONSIBILITIES ....................................................... 131 Making theJustice System Fa EstablishingProcedures MAKINGPOORPEOPLE'S LA ....................................................................................... 140 Delimiting Community Land............................................................................................................. 144 Improving the Consultation Process................................................................................................. 145 iv Making WomenBetter off in ClaimingRights .............................................................. 147 RecommendedActions ......... .............................. 149 6. THE IMPACT OF HIV/AIDSINMOZAMBIQUE ................................................................ 150 HIV/AIDSPREVALENCEINMOZAMBIQUE............................................................................................ 150 STRUCTURAL DETERMINANTS OF HIVlAIDS IN MOZAMBIQUE ............................................................. 153 THEFEMINIZATION OF HIV/AIDS INMOZAMBIQUE .................................................................................................. .............................................................................. 154 DEMOGRAPHIC EFFECTS HIV/AIDS OF 157 Orphans ......... .............................................................................. ................ 159 HIV/AIDS and on the Livelihoods of Poor Households.... MACROECONOMIC OFHIV/AIDS ................................................ ANTIRETROVIRAL TREATMENT.............................................................................................................. IMPACT 165 SUMMARY AND CONCLUSIONS ............................................................................................................... 166 7. THE POVERTYREDUCTIONSTRATEGY (PARPA):EVOLUTION AND PROSPECTS 168 PARPA I................................................................................................................................................ 169 PARPAII ............................................................................................................................................... 171 CIVIC ENGAGEMENTTHEPARPAPROCESS IN ....................................................................................... 173 Structure andFunction of the Poverty Observatories......................... ThePlethora of Monitoring Mechanisms ........................................... Participant Frustration ............................................................................... 177 RecommendedActions ...... .............................................................................. 178 REFERENCES ......................................................................................................................................... 181 VOLUME2. APPENDIXES APPENDIX A. CHAPTERTABLES APPENDIXB. METHODOLOGY REFERENCES V LIST OF TABLES TABLE1 ALLTHREE STANDARDMEASURESOFPOVERTYDECLINED XVI TABLE2 ANNUAL GROWTHRATESOF GDP. CONSUMPTION. INVESTMENT.EXPORTS. .. SIGNIFICANTLY ............................... AND IMPORTS. 2000-08 TABLE3 HOUSEHOLD . ..................................................................................................................................... XXIV CONSTRAINTSONACCESS JUSTICE TO TABLE 1.1.POVERTY MEASURES, BY AREA, 1997 AND 2003 .......................................................................... 2 ............................................................... XXXVII TABLE1.2.COPINGSTRATEGIESINCOMMUNITIES VISITED, 2006................................................................ 10 TABLE 1.3. SCHOOL ATTENDANCE ORPHAN STATUS, AREA,AND WEALTH TERCILE,2006 BY .................... .................. 12 TABLE1.5. NUMBERMEALS CONSUMEDDURINGPREVIOUSDAY,2006 ................................................. TABLE 1.4. SELF-ASSESSMENTWELFARE RELATIVE TO OTHERS INTHE COMMUNITY, 2006 OF 19 OF 20 TABLE 1.6. PERCEPTIONS OF CHANGEINHOUSEHOLD POVERTYINRURALAREASOVER LAST FIVE YEARS, 2006 ........................................................................................................................................ 21 TABLE1.7.PERCEPTIONSOF CHANGEINHOUSEHOLD POVERTY INURBAN AREASOVER LAST FIVEYEARS, 2006 ......................................................................................................................................... 21 TABLE2.1. GDPBY EXPENDITURE CATEGORY, SHARE, AND GROWTH RATE, 1997-2003 ........................... 25 TABLE2.2. GDP, LABOR TABLE2.3. DECOMPOSITIONCHANGEINPOVERTYBYLOCATION SECTOR, 1997-2003 ..................27 FORCE, PRODUCTIVITY, AND POVERTYBY SECTOR, 1997 AND 2003 .................... OF AND 28 TABLE2.4. TYPEOFEMPLOYMENT,AREA,1997-2003 BY ........................................................................... 36 AREA,2003................................................................................................................................................... TABLE2.5. DISTRIBUTION THELABOR OF FORCE SECTOR, TYPEOFEMPLOYMENT, BY AND 36 TABLE2.6. HIGHESTLEVEL EDUCATION OF COMPLETED, BY TYPE OF EMPLOYMENT, 1997 AND 2003 ..................................................................................................................................................... 43 TABLE3.1. ESTIMATED ACTUALAND POTENTIAL CROPYIELDS, 1998 ........................................................ 64 ..................... TABLE3.3. RURALINCOMEPOVERTYINCIDENCEAND TRANSITION MATRIX, 2002-05 ............................... TABLE 3.2. SOURCES OF GROWTH RURALHOUSEHOLD BY QUINTILE, 1996-2002 IN INCOME 65 68 TABLE3.4. RURALINCOMEPOVERTYINCIDENCEAND POVERTYTRANSITIONMATRIX, 2002 AND 2005 ...................................................................................................................................................... TABLE3.5. RURALPOVERTYDYNAMICSAND OFF-FARM INCOMEDIVERSIFICATION, 2002-05 ...................69 TABLE 4.1. SECTORAL EXPENDITURES A PERCENTAGEOF TOTALEXPENDITURES, AS ............... 74 82 TABLE4.2. GROSS AND NET ENROLLMENT RATES, BY EDUCATION LEVEL, 1996-2005 .............................. 1999-2006 88 TABLE4.3. PRIMARY SCHOOL(EP1) COMPLETIONAND DROPOUT RATES, BY PROVINCE, 1997 AND 2003 ...................................................................................................................................................... TABLE 4.4. BENEFIT INCIDENCE:EDUCATIONEXPENDITURE ALLOCATION BY QUINTILE, 2003 ................... 90 TABLE4.5. BENEFIT INCIDENCE:EXPENDITURE ALLOCATION BY QUINTILE AND GENDER,2003 .................92 93 TABLE4.6. NUMBERLIVEBIRTHSATTENDED HEALTHFACILITYAND VACCINATION OF AT COVERAGE, BY PROVINCE, 1997-2003 ......................................................................................................... 99 TABLE 4.7. REGIONAL COMPARISON OF WELFARE INDICATORS 100 TABLE4.8. SELECTEDSOCIAL OUTCOMEINDICATORS ............................................................................... ................................................................. 101 TABLE4.9. PERCEPTIONS OF CHANGEINHEALTH SERVICES, BY AREAOFRESIDENCEAND GENDEROFTHEHOUSEHOLD 2006.................................................................................................. HEAD, 106 TABLE4.10. PROAGMBUDGET EXPENDITURE BY C~MP~NENT, 1999-2005 110 TABLE 4.11. EXTENSIONWORKERDENSITY IN MOZAMBIQUE, 2004 ........................................................... ........................................... 111 TABLE5.1. HOUSEHOLD CONSTRAINTS INACCESS JUSTICE TO .................................................................. 134 TABLE 5.2. RESOLUTIONOF PROBLEMSBY ENTITYINURBANAREAS,BY GENDEROF HOUSEHOLDHEAD AND TERCILE, 2006 ...................................................................................................... 136 TABLE 5.3. RESOLUTIONOF PROBLEMSBY ENTITYINRURALAREAS,BY GENDEROF HOUSEHOLDAND TERCILE, 2006 HEAD ...................................................................................................... 137 TABLE 5.4. AWARENESS HOW TO OBTAIN LAND OF TITLE, BY AREA,GENDEROF HOUSEHOLD HEAD, AND TERCILE, 2006 .......................................................................................................................... 143 TABLE5.5. REASONSFORLACK TITLE, BY AREAAND GENDEROFHOUSEHOLD 2006 OF HEAD, ...............144 TABLE6.1. AWARENESSHIV/AIDSAND PREVENTIONMETHODS, 2003 OF ............................................... 156 TABLE6.2. RATIO OF CURRENT SCHOOL ATTENDANCE ORPHANSVERSUS NONORPHANS, OF BY TYPE OF ORPHAN, 2003 .............................................................................................................................. 160 TABLE6.3. DIFFERENCE-IN-DIFFERENCES OFRURALHOUSEHOLD ANALYSIS COMPOSITION BY GENDEROFDECEASED PRIME-AGEADULTSIN MOZAMBIQUE, 2002-05 .................................................... 162 vi TABLE6.4. PERCENTAGERECEIVING ANTIRETROVIRALCOMBINATION THERAPY(HAART) IN MOZAMBIQUE. BY GENDERAND REGION. OCTOBER2005 ................................................................ 165 TABLE6.5. NUMBERPEOPLERECEIVINGANTIRETROVIRALCOMBINATIONTHERAPY(HAART) IN OF MOZAMBIQUE. 2006.......................................................................................................................... TABLE6.6. ANTIRETROVIRALCOVERAGEINMOZAMBIQUE ANDNEIGHBORING COUNTRIES. 2005 ...........165 166 TABLE 7.1. KEYFEATURES PARPA 1AND I1 OF .......................................................................................... 179 vii LISTOFFIGURES FIGURE1.URBANPOVERTYI S 52PERCENT. RURAL 55 PERCENT............................................................... XVII FIGURE2 .POVERTY RATES VARIED CONSIDERABLYBY PROVINCEIN2003 XVIII FIGURE3 ACCESS TOSERVICESIMPROVED FIGURE4 INURBANAREASEMPLOYMENTAGRICULTUREI SDOWN, SELF-EMPLOYMENT OUTSIDE .. ............................................. ................................................................................................. XIX .............................................................................................................................. IN FIGURE1.1.POVERTY RATES, BYAREA,1997AND 2003.................................................................... AGRICULTUREUP xxv FIGURE1.2.POPULATIONANDPOVERTY, BYPROVINCE, 2003 .. ............................................................... 4 FIGURE1.3. HUMAN DEVELOPMENT INMOZAMBIQUE, BY PROVINCE, 2000 .................................................. 8 FIGURE1.4.REGRESSIONRESULTS: DETERMINANTSCONSUMPTION,2003 OF ............................................. 15 FIGURE2.1. AGRICULTURE'S SHAREOFLABOR FORCE REMAINS ABOVE THE SHARE OF GDP, BUT NOT INDUSTRYOR PRIVATE SERVICES ........................................................................................................ 30 FIGURE2.2. AVERAGE ANNUALGROWTH RATES OF OUTPUT AND EMPLOYMENT,SECTOR, BY 1997AND 2003 ....................................................................................................................................... FIGURE2.3. INURBANAREASEMPLOYMENTAGRICULTUREI S DOWN, SELF-EMPLOYMENT IN OUTSIDEAGRICULTURE U P ...................................................................... ............................................... 37 FIGURE2.4. MORERURALHOUSEHOLDS HAVEINCOMEFROMHIGH-VA SOURCES ............................... 41 FIGURE2.5. FORBOTH WOMEN AND MENEDUCATIONI S THE MOST IMPORTANT DETERMINANT OF WAGES, 2003 .......................................................................................................................................... 45 FIGURE2.6. ALL SECTORSARE STILL GROWINGRAPIDLYINMOZAMBIQUE, ................................ FIGURE3.1. GROWTH AGRICULTURALGDP, 1992-2003 ......................................................................... 2000-08 56 OF 62 FIGURE3.2. AVERAGEANNUALGROWTH RURALHOUSEHOLD OF INCOMEPERADULTEQUIVALENT, INCOMEPERADULTEQUIVALENT QUINTILE, FIGURE3.3.SEASONALVARIATION INWELL-BEING AND VULNERABILITY INANGOCHE,2006 1996-02AND 2002-05 .................................................BY 66................... 70 FIGURE3.4. SEASONALVARIATION INWELL-BEING AND VULNERABILITY INCHOKWE, 2006 ..................... 70 FIGURE3.5. RURALPOVERTYTRANSITIONSTATUSAND WELFAREPERCEPTIONS IN2005 .......................... 72 FIGURE4.1, TOTALEXPENDITUREPARPAPRIORITYAREAS-AS A PERCENTOF GDPANDTOTAL IN EXPENDITURE, 1999-2006 .................................................................................................................. 83 FIGURE4.2. NUMBEREPl SCHOOLSINURBANAND RURALAREAS,1996-2005 .................................... FIGURE4.3. NUMBEREP2SCHOOLS INURBAN RURALAREAS,1996-2005 .................................... OF 85 OF AND 86 FIGURE4.4. TOTALNUMBERTEACHERS PROPORTIONOF TRAINED TEACHERSINEPl, OF AND 1996-2005 .................................................................................................................................................... 86 FIGURE4.5. PUPIL-TEACHERRATIOSINEP1RURALAND URBANAREAS,1996-2005 ................................. 87 FIGURE4.6. ENROLLMENT BOYS AND GIRLS, BY AGE,2003 OF .................................................................... 88 FIGURE4.7. TOTALHOUSEHOLD EDUCATIONEXPENDITURE SHAREOF EDUCATION AND EXPENDITURE TOTALCASHEXPENDITURE, AREAOFRESIDENCE,2003 IN BY ............................................. 91 FIGURE4.8. PERCEPTIONSOF CHANGE INTHE EDUCATION ............................ 94 FIGURE4.9. HEALTHEXPENDITURESHARES, BY SOURCE, 1997-2002 ..... SECTOR, BY AREA,2006 ............................................. 95 FIGURE4.10. NUMBERINHABITANTS PERPHYSICIANANDNUMBERINHABITANTS PERHEALTH OF OF .......................... 96 FIGURE4.11.HEALTHEXPENDITUREPERCAPITA ANDPOVERTY INCIDENCE, BY PRO WORKERBY REGION,2003................................................................................ AND 2004 ...................................................................................................................................................... 97 FIGURE4.12. TOTALHOUSEHOLD EXPENDITURE SHARE OFHEALTHEXPENDITURE HEALTH AND INTOTALCASHEXPENDITURE, AREAOFRESIDENCE, 2003 BY .................................................................. 104 FIGURE4.13. CONCENTRATIONOFHEALTHSERVICEUSE, BY TYPEOFFACILITY,2003 FIGURE4.14. TOTALPROAGRIBUDGET SHAREALLOCATED PROVINCES, 1999-2005 ................105 ............................ AND TO 108 FIGURE4.15. ACCESS EXTENSIONTO ADVICE BYINCOMELEVEL EXTENSION AND BUDGET BY PROVINCE,2002 .......................................................................................................................................... 114 FIGURE4.16. PERCENTAGEWITHACCESS TO SAFEWATER, BYREGIONAND QUINTILE, 1997 AND 2003 .................................................................................................................................................... 117 FIGURE 5.1. STRUCTUREOF CONFLICTRESOLUTIONINERATI-NAMAPA, NAMPULA .................................. ........................... 140 FIGURE6.2.NUMBERmV/AIDSINFECTEDPEOPLE,BY GENDER,1998-2010 ...................................... FIGURE6.1. ADULTPREVALENCERATES(AGED 15-49), BYPROVINCE, 2002 AND 2004 152 OF 155 FIGURE6.3. POPULATIONPROJECTIONSFORSOUTHAFRICA, WITHAND WITHOUTmV/AIDS, 2000 AND 2025 .......................................................................................................................... Vlll ... FIGURE 6.4. EXPECTED AIDS DEATHS,AGE15 AND HIGHER,1999-2010 ................................................. 159 FIGURE6.5. NUMBEROF ORPHANS AGED0 TO 17, 1998-2010................................................................... 160 ix PREFACE TheIntegratedLens This assessment, reflecting poverty's many dimensions in Mozambique, combines multiple disciplines and diagnostic tools to explore poverty. It draws on a combination of approaches and tools from three separate analytical diagnostics developed by the World Bank-Poverty Assessment, Country Gender Assessment, and Country Social Analysis. It uses monetary, human, and social indicatorsand combinesquantitative and qualitative approachesto understandtrends in poverty and the dynamics that shape them. The objective is to support the development and implementation of propoor policies that really work by taking poverty's multiple dimensions into account. Because Mozambique has not collected nationally representative household survey data measuring poverty status and outcome indicators since 2003, the report focuses primarily on the changes in poverty and household community welfare through to that year. But when data are available after 2003, we use them, including data from a special nonrepresentative survey developedfor this report-the Poverty and Vulnerability Survey. The starting point for the analysis is chapter 1-using multiple quantitative and qualitative indicators on levels and changes in the opportunities and outcomes for households and communities inMozambique since 1997. The rest ofthe report explains these changes. Chapter 2 reviews the main economic developments, analyzes how changes at the macro and meso level affected household livelihoods, and how households, especially poor households, responded. Given the importance of the rural space, chapter 3 reviews these connections in more detail from the view point of the rural economy, the role of agricultural growth, and income diversification inpoverty reduction. One of the core functions of government is to finance and provide key social and infrastructure services, especially for the poor, who cannot purchase these services in the market. Chapter 4 reviews whether government spending in four areas-health, education, agricultural extension, and water supply-has reached the poor and the more remote areas and whether it has brought measurable improvements inwelfare. Another government function central to poverty reduction is to protect the rights of citizens, ensuring the poor and vulnerable access to resources to sustain and improve their communities. Chapter 5 analyzes a few of the efforts of government to increase the transparency and responsivenessof public institutions, to provide security to people, households, and communities, andto empower women within the householdand community. The HIV/AIDS pandemic in Africa is a threat to achievements in poverty reduction in all its dimensions.Chapter 6 uses the fragmentary data available to analyze the implications for propoor policy and consequences of the disease spreading in Mozambique. The government is implementing its poverty reduction strategy (PARPA 11). Chapter 7 summarizes the strategy and brings togetherthe implications ofthe analysis for the way forward. This study has been undertaken as a collaborative effort with the National Directorate of Studies and Analysis (DNEAP) in the Ministry of Planning of Mozambique. The World Bank team was X led by Louise Fox and included Rui Benfica, Malcolm Ehrenpreis, Hakon Nordang (consultant, Austral Consultoria e Projectos), Daniel Owen, Melissa S. Gaal, and Katleen van denBroek. Beth Anne Dabak contributed to the section on land andjustice issues in chapter 5. From the Ministry of Planning, Christina Matusse reviewed the concept note, and Maimuna Ibraimo and Claudio Massingarelaparticipated inpreparingthe analysis. This is the first combined Poverty, Gender, and Social Assessment led by the World Bank. It would not have been possible without the support from many individuals and institutions. Contributions and advice were received from Channing Arndt, Sergio Baleria, John Barnes, Carlos Casimiro, Humberto Cossa, Antonio Francisco, Konrad Matter, Ana Menezes, Daniel de Souza, Chris Tanner, Inge Tvedten, and Bina Valadon. The Poverty and Vunerability Survey for the report was performed by a team from the University of Eduardo Mondalne, led by Alexandre Mate. Cynthia Donovan and David Mather prepared a background paper for chapter 6 of the report, and MikkelBarslund a backgroundpaper for chapter 1. Joel Muzima provided assistance in the use of the education sector database. Bruce Ross-Larson, Amye Kenall, and Christopher Trott of CommunicationsDevelopment editedthe report. The team also consulted with Cruzeiro do Sul, Grupo Mopmbicano da Divida (GMD), and G20 members. Insightful comments on an earlier draft were received from Harold Alderman, Maria Correia, Caroline Robb, Sudhir Shetty, and Dominique van de Walle. Sudhir Shetty, EmmanuelAkpa, and Michael Baxter provided management support and guidance. The team is grateful for support from the joint government-donor sector working groups on Poverty Analysis and Monitoring Systems and on Gender Issues throughout the process. A special debt is owed to Bridget Walker-Muiambo for providing helpful comments and for securing financial support from IrishAid for the gender component. Financing was also received from the TFESSD trust fund, financed by the Norwegian and Finnish governments for the poverty and vulnerability survey, and the Belgian Poverty ReductionPartnershipProgram. xi ABBREVIATIONS AE adult equivalent ARI acute respiratory infection ART antiretroviral combination therapy CBN cost o f basic needs DHS Demographic and Health Survey EP1 primary grades 1-5 EP2 primary grades 6-7 ES1 secondary grades 8-10 ES2 secondary grades 11-12 GDP gross domestic product 'G o M Government o f Mozambique IAF National Household Surveys on Living Conditions (InquCrito aos Agregados Familiares) INE National Institute o f Statistics MADER Ministry for Agriculture and RuralDevelopment MDG MillenniumDevelopment Goal M E C MinistryofEducation and Culture M I S A U MinistryofHealthofMozambique MPF MinistryofPlanning andFinance MSE micro or small enterprise MTn Meticals NGO nongovernmental organization OIIL Local Initiative InvestmentBudget ORT oral rehydration therapy PARPA Plano de Accao para a Reducao da PobrezaAbsolouta (poverty reduction strategy) PETS Public Expenditure Tracking Surveys PO poverty observatory PPO provincial poverty observatory PRS poverty reduction strategy PVS Poverty and Vulnerability Survey SISTAFE Sistema de Administraggo de Fundos do Estado SWG sector working groups TIA National AgriculturalRural Income Surveys (Trabalho de InquCrito Agricola) technical and vocational training: xii OVERVIEW 1 Life for most Mozambicans improveddramatically in the decadefollowing the end of the civil war and thefirstfree elections in 1994. Household incomes and asset holdings increased, as did access to and the quality o f publicly provided services. Despite this improvement, many Mozambican households are still amongthe poorest inthe world. More than halfthe population lives inpoverty, with persistent gaps between farmers and city dwellers, women and men, and poor and rich. To extend the transformation that began inthe mid-1990s,policymakers can benefit from taking stock o f past successes while understanding more clearly the gaps and shortcomings. They need to find new ways to accelerate improvements inthe living standards o f all Mozambicans. 2 As specified in thegovernment'sfive-year plan (PARPAII),onegoal is to pursue propoor growth-by promoting agriculture and theprivate sector (box 1).A second i s to build human capital-by improving access to basic public services, especially for the poor. A third i s to improve governance and accountability-by getting government closer to its citizens. To do all this, the government will needto increase the value for money in its spending on public services. Itwill also need to target services on the rural poor and enlist poor communities in identifying needs and co-managing the delivery o f those services. Box 1. The Government's Poverty Reduction Strategies: PARPA Iand PARPAI1 The detailed poverty reduction strategy (PARPA-Plano de Accao paraa Reducao da Pobreza Absolouta) articulates the Mozambican Government's strategicvision for poverty reduction by promoting increased productivity and by improvingthe capacitiesand opportunities for all Mozambicans.The first PARPA was preparedin December 1999,and after public discussion, it was revised and presentedto donors as the basis for donor support in2001-05. PARPA Ielaboratedpriority policies and programsthat would foster humandevelopment and broad-basedeconomic growth. Emphasizingthe link betweensustained growth and poverty reduction, the strategy incorporatedpolicies and reforms to stimulate growth, identifyingphysical and humancapital and risingproductivity as fundamentalmacroeconomicdeterminantsof growth. The design of PARPA I1was completed in2006 and is now being implementedin 2006-1 0. It maintainedthis strategic focus, organized aroundthree main pillars: governance, humancapital, and economic development.The vision commonto both PARPAs includes consolidating national unity, developing each citizen's humanpotential, creatinga working institutional system, and increasingthe ability to create national wealth. 3 Topromote agriculture and theprivate sector, thegovernment should improve research and extensionfor commercialfarmers, large and small, andfor subsistence farmers, especially women. It should build infrastructureto increase farmers' access to markets and their ability to tap new technology. And it should strengthen producer associations to increase the benefits o f linking farmers to domestic food supply chains and to high-value, export-oriented supply chains. So that the private sector can create ... Xlll jobs and boost incomes inurbanareas, the government shouldencourage labor-intensive manufacturingexports ratherthanfocus only onhigh-profilemegaprojects, which do little for employment andbroad-basedincome growth. It should also developnew policiesandprogramsto reduce risk and increase incomes for householdbusinesses in the informalsector-businesses that will continueto be importantfor sustaininggrowth and reducingpoverty. 4 To build the human capital of Mozambicans, thegovernmentshouldfocus on the basics of education, health, water, and extension-greatly expanding access in rural areas and developingnew models to improve the quality of services. This will requireincreasingpublic spending on social and economic services, buildingonthe early successes ineducation, and gettingmore value for money fromthat spending. Part ofthis will involveshiftingresourceswithin ministriesto focus onthe basics and across sectors to reachthe underserved. Part will involvepublic-privatepartnershipsfor decentralized service provisionandhavingcommunitiesidentify needs and helpdeliver services to households.And partwill involvegood trackingsystems to link programoutputsto targets and outcomes, usingfrequent high-qualityhousehold surveys and other monitoringand evaluationinstruments.Only by establishingand refining such systems will it bepossibleto knowwhere resources are goingand what resultsthey are producing. 5 To improvegovernanceand accountability,thegovernmentshould empower communitiesas local agents of change. It shouldmake decentralizationwork for the poor by devolvingrealpowerto local communitiesso that they can advocate for more responsivepublic services, influenceservice delivery mechanisms, andhold local service providersaccountable for performance.It should also increase access tojustice for the poorby openingopportunitiesto engage with public institutionsand ensuringthat the judicial system increases its responsiveness to all Mozambicans.And it shouldmakepoor people's landrightsreal by increasingawareness of citizen rights to landand by strengtheningthe consultationsfor community land.All this demands reducingthe red tape andpetty corruptioninthe public administration. THERECORDOFACHIEVEMENTS 6 Mozambique was an extremelypoor country at the time of its electionsin 1994, with decimated infrastructure, a weak economy, andfragile institutions. Since then, it hasbeenastonishinglysuccessful at restoringgrowthand improvingwelfare. Sustained growth-driven primarilyby investmentsinphysicalcapital-reduced monetarypoverty from 69 percent ofthe populace in 1997to 54 percent in2003 andthe depth and severity of nonincomepoverty even more.Broad-based, labor-intensiveprivate-sectorgrowthwas efficient inreducingpovertyuntil 2003 because itwas equally distributed.At the same time, investments insocial and economic infrastructureextended access to public services, reducedwelfare inequalities,and supportedthe livelihoodsof the average Mozambican. xiv 7 Thesocial debt at the end of the civil war-a severe lack of social services in rural areas-required massive investments in buildings andphysical inputs and in trained staff to run the services. Donors were ready to fund good projects inunderserved areas, and the government was determined to get these funds and use them to reduce poverty. The greatest successes were inroads and in education, where highlevels o f spending(20 percent o f total government expenditures) producedthe greatest improvements inboth access and outcomes for poor and rural families. Inother sectors public programs were not developed or implemented as quickly to reach the underserved, so the gains were limited and some opportunities lost. 8 The ignitionfor growth camefrom new private investment inphysical capital and high levels of public spending. Public spending on rehabilitating the infrastructure needed to gain access to markets supported growth in rural incomes. Farmers had access to good quality land to expand and diversify production. Households increasedtheir integration with markets-locally and regionally. Usingcash gained from selling agricultural produce or working for cash either during harvest or in the off-season, they invested in new small and microbusiness ventures. They also invested inbetter housing and in sendingtheir children to school. But some households missedout and now face the prospect o f remaining trapped inpoverty. 9 The backbone ofpropoor growth in Mozambique was the response of family farmers andfamily-owned businesses-where more than 90percent of the laborforce in Mozambique works-to theprogrowth economicpolicies of thegovernment. These policies helddown inflation while reducing the cost o f doing business and lowering restrictions on competition, such as price controls and inefficient parastatal monopolies. 10 Since 2003 the blisteringpace of economicgrowth has continued. Mozambique's growth and poverty reduction episode i s already one o f the longest in low income countries. Research shows that growth cycles usually runout after about seven years in low income, non-oil producing countries. When growth starts after a crisis, it i s based (at least partly) on rehabilitation. Once this source o f growth i s exhausted, growth sputters out, or inequality increases, halting the poverty reduction impact, or both. Through good policies that providedpolitical and economic stability, Mozambique has so far beaten the odds on sustained economic growth. 11 But it is unclear whetherMozambique is continuing to beat the odds on shared growth. The latest comprehensive data on living standards are from 2003, when about halfthe population was classified as poor, and the evidence since then is sketchy. Government data show continuing increases in access to services, and rural income survey data show that average household incomes have increased. Based on extrapolations o f a few indicators, the Ministry o f Planning and Finance predicts a continuing fall inpoverty. But other indicators point to a slowdown inthat fall. Rural income inequality seems to be growing, and already highurban inequality persists, so fast growth may now have less o f a poverty-reducing effect. xv 12 The 2003 data show bewildering variation in welfare outcomes by province. In some areas infant mortality and nutrition are gettingworse while consumption i s going up.Recently collected perceptions data point to apopulationless content with what the government i s providing. I s this the result o f even higher expectations, or has the pace o f improvement changed? Certainly vulnerability remains high, threatening to undermine recent gains. And shocks to household income from weather, price fluctuations, and health continue to pushpeople back into poverty evenas other households climb out. Has the welfare improvement engine runout o f gas? Higher Incomes and Better Living Conditions through 2003 13 The latest national household incomeand expenditure survey-completed in 2003-showed evidenceof recovery and better living conditionsthroughout Mozambique. Eventhe poorest o f the poor had somewhat better living conditions (though most rural residents still lacked metal roofs on their houses). Most households had more livelihood choices, more diverse income sources, earned more money in cash, had more o f their children in school for longer periods, and could go to a health clinic more often. Rural communities-with two-thirds o f the population-had a better water supply, better market access, and were more likely to have at least a primary school nearby. Nearly two-thirds o f them reported that their welfare had improved over the previous three years despite a 10 percent fall inagricultural output in 2000 because o f severe flooding. 14 Poverty in Mozambiquefell significantly between 1997and 2003, with the percentage ofpoor falling from 69.4percent to 54.1percent (table 1).The national poverty gap index dropped from 29.2 percent to 19.9 percent. So, not only did 15 percent o f the population move above the poverty line, but the average household consumption o f those who had not yet escaped poverty increased. The squared poverty gap measure also fell-from 15.5 percent to 9.9 percent duringthe same period-a 36 percent drop inthe severity o f poverty. Table 1. All Three Standard Measures of Poverty Declined Significantly (percent) Headcount Poverty gap Squaredpoverty gap I997 2003 Change I997 2003 Change I997 2003 Change All 69.4 54.1 -22.0 29.2 19.9 -31.8 15.5 9.9 -36.1 Urban 63.9 51.6 -19.2 27.2 18.9 -30.5 14.8 9.0 -39.1 Rural 71.6 55.2 -22.9 30.0 20.4 -32.0 15.8 10.3 -34.8 Source: IAF data for 1997 and 2003. Note: Consistent with the 2003 urban andrural definitions. 15 Ruralpovertyfell slightlyfaster than urbanpoverty, though ruralpoverty overall remains higher than urbanpoverty Pgure 1).Poverty dropped by 16 points in rural areas and by 12 points inurban areas. But rural poverty i s still higher than urban poverty, with the percentage o f poor people inrural areas estimated at 55.2 percent, compared with 51.6 percent inurban areas. xvi Figure 1.Urban Poverty I s 52 Percent, Rural 55 Percent 80 -.... 7 ....... .~ 1997 1 . ...... -mi 2003 1 _i . . . . . . . - . Urban Rural Total Source:IAF data for 1997 and 2003. Note: Consistent with the 2003 urban and rural definitions. Poverty calculation based on consumption per adult equivalent. 16 Poverty has been reduced in mostparts of the country, though geographical variations are large figure 2). Poverty has falIen most inthe center, mostly inrural areas, driven by the large changes inthe populous provinces o f Sofala and Zambezia. Sofala, the country's poorest province in 1997, saw such significant reductions inpoverty that in 2003 it became the least income-poor province. Poverty also fell inthe north, though more in urban areas, and it rose inCab0 Delgado. The south experienced an increase inpoverty rates, especially inthe city o f Maputo and the surrounding province. Poverty in Inhambane, the second poorest province in 1997 after Sofala, fell only slightly, leaving the province as the poorest inMozambique in2003. The small poverty reduction inthe rural south was overwhelmed by the increase inurban poverty. Not only did poverty rise, but the depth and severity also increased inMaputo city and Maputo province. 17 Thesetrendsarefavorable, but incomepoverty in Mozambique remains high. As the national numbers suggest, 54.1 percent of the population still live inpoverty. Of the estimated 18 millionMozambicans in2003, nearly 10 millionwere below the poverty line. Eachprovince has at least 500,000 poor people, with more than 1 millionpeople in Zambezia, Nampula, and (disproportionately) Inhambane (see figure 2). xvii Figure 2. Poverty Rates Varied Considerably by Province in 2003 Province ~ Source: IAFdata for 2003. Access to ServicesIncreased 18 The different household access to services, goods, and assets helps explain the broader poverty outcomes. Mozambiquehas improvedaccess to public services, though important variations remain between income groups, regions, and urban and rural areas (figure 3). Boththe national household surveys and the quantitative and qualitative findings from the 2006 Poverty and Vulnerability Survey (PVS), conducted for this report, suggest improvements in access to public services, particularly for health and education. Although most households have benefited from these changes, improvements have sometimes been uneven, so access to public services continues to vary by income group and geographic location. The richest have benefited most inelectricity, health, and secondary education, while primary education has particularly benefited the poor. Urban areas have also benefited more than rural areas, with access to electricity, water, and sanitation benefiting the urban poor. 19 Physical access to lower primary education improved significantly across much of the country since 1997; improvements in access to upperprimary and secondary school were more uneven. Net primary school enrollments increased significantly between 1997 and 2003, particularly for the poorest households inboth urban and rural areas. Net secondary school enrollments also increased somewhat, but the gap between the poorest and richest householdscontinuedto widen, as did the urban-rural gap. For primary school the gap narrowed substantially betweenthe provinces inthe south and those inthe north and, particularly, those inthe center. For secondary school the gap in access betweenthe south and the rest o f the country i s increasing. xviii Figure 3. Access to Services Improved 100 90 80 70 60 50 40 30 20 10 0 01997 Source' IAF data for 1997 and 2003. 20 I n a major achievement, girls' netprimary school enrollment in grades 1-5 increased significantly, particularly for thepoorest quintile, rural areas, and central provinces. In2003 it almost equaled boys' enrollment rates inmost o f the country, even surpassing them in the south, overcoming longstanding prejudices about the value o f educating girls. The opposite is true in secondary education, however, where the gender gap is widening throughout the country, and particularly inurban areas and among the rich. Giventhe highcorrelation between education and poverty, these differences in access continue to pose challenges to future growth and poverty reduction. 21 Despite the improvements, constraints on access to education persist, particularlyfor secondary school. The poorest households continue to face greater barriers than the richest households. For primary education the real costs o f uniforms and school materials and the opportunity costs o f schooling pose real barriers to access for some o f the poorest households. For secondary school, where households still have to pay matriculation fees and cover all costs o f school materials, the constraints are even greater. Distance i s also a barrier, especially in rural areas, where only 1percent o f households live close to a secondary school, but also inurban areas, where the figure is still a mere 15 percent. 22 Socioculturaland economicfactors also continue topose a barrier to girls' enrollment, particularly in secondary education. Many households, when faced with the higher direct and indirect costs o f secondary education, prefer to send boys to secondary school rather than girls. Distance i s also a factor because families are less disposed to sending girls to live alone (in boarding arrangements) or with relatives or friends in areas with secondary schools. InZambezia most householdheads preferredto educate their xix sons-mainly because the girls might eventually be "married off' to other families and would spend most o f their time in (domestic) activities consideredto require less education. 23 Access to healthcare, measured by thepercentage of households seeking medical help when a household member is ill, increased by 5percentagepoints over the two survey periods. While access varies only slightly betweenthe bottom and middle income groups, a 13 point difference in access to healthcare separates households inthe poorest fifth o f the populations and and those inthe richest fifth (53 and 66 percent, respectively). The discrepancy i s also large between urban and rural areas, with as many as 74 percent o f urban residents usinghealthcare when ill,compared with only half o f rural residents. This seems to reflect the larger distances to health posts in rural areas. Of urban households 68 percent live within half an hour o f a health post. Only 21 percent o f rural households do. 24 Gains have been greaterfor water and sanitation thanfor electricity and other services-nd greater in urban areas. Across Mozambique access to water and sanitation improved, benefiting both urban and rural areas, with little variation among consumption quintiles. Although more than 60 percent o f the population remains without access to safe drinkingwater, the use o f safe water increased by 13 percentage points nationally, the distance to a water source (sometimes unsafe water) decreased for most households, and an additional 10 percentage points o f households benefited from improved sanitation, with the poorest quintilebenefiting the most. Evenso, households in rural areas have significantly less access to safe water and improved sanitation than those inurban areas. So Did Holdings of Assets and Durable Goods 25 Access toproductive assets-including land-may also have improved in recentyears, though differences continue across incomegroups and other social groups. Given the predominantly rural and agricultural economy, access to land i s an important part o f most households' livelihoods. The Poverty and Vulnerability Survey, confirming national trends noted elsewhere, found that access to land on average is not a problem for most households inthe communities studied. Most have access to land, especially inrural areas. Access varies significantly between urban and rural areas, however, reflecting mainly the relative importance o f nonagricultural activities inurban areas. There are still significant variations between richer and poorer households. More than a quarter o f the poorest third o f households have no access to land, compared with less than 15 percent o f households inthe richest third. 26 Households across all incomegroups improvedtheir holdingsof durablegoods between the two national surveys. For some o f the higher cost items (motorcycles, bicycles, televisions), the gap between the poorest and richest quintileswidened somewhat, though for most other goods the improvements were more evenly distributed. Significantly, houses were upgraded in all quintiles. Not only i s housing an important savings vehicle for most households, but many households see the quality (and security) o f housing as an especially important indicator o f poverty. So, the disproportionate increase inhouseholds with durable roofs between 1997'and2003 among poorer households i s important. Because upgrading i s usually a cash expense, it provides a good indicator o f improvements inwealth and welfare since 1997 for the poorest income groups. Another indicator o f better welfare is that the share o f income spent on food fell for all income groups, if slightly more for the richest quintile. Why Was Growth So Effective in Reducing Poverty? 27 Mozambique's rural poverty reduction is one of the most successful in the world. Several factors stand out. Farmers boosted the diversity o f their food crops, increasing food security and nutrition. And many farmers moved into small-scale commercial farming, increasing their incomes by selling surplus food crops or diversifying into such export crops as tobacco and cotton, mostly grown under contracting arrangements with agroprocessing/exportingfirms. Their higher income stimulated demand for goods and services in rural areas, and households responded by opening businesses, such as trading, brewing beer, and making bricks and charcoal. Larger farms, usingmore capital, sprang up and required wage labor, which some households supplied. Households with diversified sources o f income tended to be less poor in 2003 than those that did not diversify. And communities with access to markets through infrastructure and market information did better than those without. 28 Mozambique's poverty reduction is attributable to the character of its growth, including the higher crop incomefrom the rehabilitation of agriculture. Growth was also supported by donor aid, which appears to have been propoor as well, through backward linkages. Rural and urban economic growth raised incomes and stimulated demand for nonfarm goods and services. Growth remained propoor because changes in the structure o f productionwere labor intensive-pulling labor out of'agriculture into highervalue sectors, inboth rural and urban areas. This raised the productivity o f labor and incomes inagriculture, lifting incomes for the bottom half o f the distribution, and contributing 75 percent to the fall inthe poverty headcount. 29 I n urban areas strongprivate sector growth gave men new wageand salary jobs. Unexpectedly, many new wage and salary jobs were inthe service sector-where the demand for labor increased muchfaster than output. The manufacturing sector is normally the source o f new wage and salary jobs in low income countries, but it suffered from job destruction, sheddingjobs at more than 10percent a year on average as part o f the privatization and restructuring. The new small andmedium firms manufacturing for the domestic market did createjobs, butthe new megaprojectsproducinggoods for export did not. Those able to hold onto theirjobs or get new ones still earned a good living. Even after accounting for education and experience, wages inmanufacturing were higher than those inprivate services, and wages inconstruction and miningwere even higher. 30 The urban laborforce grew at 3percent ayear, and most of the urban labor force still had very little education, so not all comers couldfind wagejobs. Many ended xxi up inself-employment infamily business, inthe informal sector. Many people inurban areas reported themselves as poor even though they had regular self-employment income. Infact, by 2003 average earnings inthis sector were well above those inagriculture. 31 Not everyone left agriculture, andfew left completely-more than 80percent of the laborforce and 70percent of the households in 2003 still identified agriculture as their main economic activity. Women mostly stayed inagriculture, and mostly men left agriculture part time or full time to pursue other activities, which on average had higher labor productivity and thus higher earnings than agriculture. The demand for these goods and services produced by the nonagricultural sector was so strong that earnings still rose even as people kept starting new businesses. 32 Staying in agriculture, whileprovidingfood securityfor the household, was generally a lowerproductivity and income choice, leaving womenpotentially at a disadvantagein controlling household income. Because 90 percent o f women inthe labor force have at most two years o f education-the legacy o f exclusion from education-their opportunities are limited. Women are increasingly important as producers inrural agriculture, but usually only for subsistence. One reason is their huge burdeno f chores. Another is that householdsneed the security o f home-produced food, which women are incharge of. This can be disempowering-men control the cash. And growing higher value export crops requires access to technology, fertilizer, and credit, which women miss out on. 33 Most smallholderfarmers increased their crop income up to 2003 through more extensive use of land. This growth inoutput per person was the most important driver o f poverty reduction inrural areas, buthas resulted ina landproductivity that i s among the lowest inthe world. There are natural limits to this productionprocess, and smallholders in Mozambique are reaching them. Without changes, they will not see the growth o f total crop productionthat has been possible inthe past. Reliance on food crops i s especially risky because Mozambique's susceptibility to droughts and floods can leave subsistence farmers hungry inthe off-season and their children malnourished. Only those farmers who keep diversifying into higher value crops, usingyield-enhancing techniques, will realize the past growth o f crop income. Farmers now know that they have to diversify. Inall communities surveyed for this report, not having diversifiedwas identified as an important characteristic o f poverty. Smallholders who cannot achieve this transformation will have to rely on nonfarm sources o f income (wage labor or self- employment) to pull them out o f poverty. 34 The diversity and complexity of rural livelihood strategies, while importantfor raising incomes, have also increased risk. Household diversification into nonfarm self- employment in rural areas has increased income risk.Average earnings are high, but low earnings are not uncommon among those that diversify. Many households that entered the small enterprise sector or started selling labor for wages remainedpoor-often because they entered the very low-skill day labor market as a survival strategy. Households that lost nonfarm sources o f income usually plungedinto poverty. And an xxii income shock-such as poor harvest or illness inthe household-often causes households to draw down assets, making it more difficult to climb back out o f poverty. 35 Vulnerability wasperceived as related to the capacities and strategies of communities and households to deal with the risks of unexpectedshocks and risks. The most common coping strategy-in both urban and rural areas, with little variation across income groups-was to seek help from family and friends, an option more available in urban areas. Drawing down assets was more common inrural areas, despite the smaller asset portfolios o f rural households. Rural households were also much more likely to do nothing and merely try to get by. As many as 21 percent o f the poorest rural households reported having done nothing to cope. 36 Evidence is emerging that a subset of rural households, usually those in more remote areas, are becoming trapped in poverty. Households that cannot adopt more commercial, yield-enhancing production processes are losing ground. Many o f these low income farmers live incommunities with limitedmarket access-a geographical poverty trap. Without infrastructure and informationto provide market access, commercial buyers with new technology do not enter the area, and farmers have no market for their crops, so crop income stagnates. When overall income growth inthe community i s slow, small enterprise opportunities are limited, and demand for wage labor i s low. TheLooming Threat of AIDS 37 The spread of HIV/AIDS is a huge threat to households.Poor health is one o f their biggest risks, as identifiedinthe Poverty and Vulnerability Survey. And once an infected family member's status progresses to AIDS, a household has to pay for medicine, take time off production to care for the illperson, and risks the loss o f the main breadwinner, plungingthe household into poverty. 38 The spread of HIV/AIDS is also a huge threat tofuture growth andpoverty reduction. Mozambique's prevalence rate, risingrapidly, is now more than 16percent o f prime-age adults. Mozambiquewas slower than other countries to pick up the disease. As a result, the prevalence o f full-blown AIDS i s low. But rising infection rates indicate that prevention i s not working and risky behaviors are not changing. Infection i s driven by labor mobility, with the highestprevalence rates inprovinces with transport corridors connecting Mozambiqueto neighbors with even higher rates. Urban areas are more likely to be affected because risky sexual behaviors are more common. Women are more likely to be affected-in Mozambique women aged 20-24 are four times more likely than men inthe same age bracket to get infected. 39 Thespread and developmentof HIVIAIDS is expected to slow growth and increasepoverty. Unless treatment programs accelerate, Mozambique will see a loss o f educated and trained people, leaving a smaller labor force. Inhouseholds where both parents die, or communities with substantial losses among the parenting generation, the next generation could have lower human capital ifchildrenare neededat home for productive or caretaking tasks. Evidence shows that inMozambique household xxiii livelihoods change when a prime-age residentdies, regardless o f cause. The changes depend on the sex ofthe person that dies. When a male dies, women become less productive. PURSUING PROPOOR ECONOMIC GROWTH-IN AGRICULTUREANDTHEPRIVATE SECTOR 40 Because of goodfundamentals, Mozambique's growth episode is one of the longest on record in low income countries. Since the last household survey, Mozambique's economy continued its vigorous, broad-based growth, with GDP growing at about 7 percent a year since 2003 (5 percent per capita), and with no signs o f slowing down. Private consumption growth continues at about 3 4 percent per capita per year, slightly higher than during 1997-2003 (table 2). Agricultural growth has been strong, which should support continued improvements inwelfare in rural areas. Government spending, especially on service delivery, i s rising as a share o f GDP, thanks to high levels o f donor support. Research shows that growth cycles usually runout after about seven years in low income, non-oil producing countries. When growth starts after a crisis, it is based (at least partly) on rehabilitation. Once this source o f growth is exhausted, growth sputters or inequality increases to halt the poverty reduction impact, or both. Table 2. Annual Growth Ratesof GDP, Consumption, Investment, Exports,and Imports, 2000-08 (volume, percent change) 2000 2001 2002 2003 2004 2005 2006 2007 2008 Estimated Projected Projected GDP, total 1.9 13.1 8.2 7.9 7.5 7.8 8.5 6.8 7.6 Consumption 0.9 3.3 0.6 14.8 4.5 12.1 -3.7 11.1 8.3 Government 11.7 17.2 3.6 16.2 8.6 12.0 25.1 17.3 12.7 Nongovernment -0.3 1.5 0.2 14.6 3.9 14.6 -3.7 12.6 9.9 Gross domestic investment -8.7 -15.2 38.0 3.2 34.6 -6.4 16.6 27.0 18.3 Exports of goods and services Exports 31.9 51.6 21.0 30.2 13.1 9.4 13.7 8.O 11.2 Imports -2.4 -20.7 21.4 13.1 -7.5 29.1 31.7 29.4 10.7 Source: World Bank staff estimates. 41 Mozambique has natural assets that can support and sustain the high growth rates neededfor poverty reduction. Aside from a surplus o f fertile land suitable for rain- fed agriculture, the long coastline includes several underdeveloped coastal tourism areas. The central provinces have coal deposits, and this area is expected to rapidly develop as a large new coal mine and a rail link come on stream with private financing. Electricity is low cost thanks to hydroelectric development. The construction o f a new bridge over the Zambezi River i s expected to reduce north-south transportation costs dramatically. As a transport corridor for neighboring landlocked countries, the east-west transportation links could be an important economic asset ifthe Zimbabwe economy recovers. And the northernprovince o fNampula contains Nacala, one o f the deepest natural ports inthe xxiv world. All these advantages require substantial public and private investments to develop, however, and improvements inpublic management to maintain. 42 As the economy exploits these opportunities it will have to become much more nimble. Mozambique's drivers o f growth are changing, with education becomingmore important. The role o f aggregate factors (land, physical and human capital) in Mozambique's growth path has been changing since 2002. Analysis shows that the contribution o f human capital was 25 percent greater during 1997-2003 than inearlier periods, implyingan important role for more highly educated new labor entrants infuture growth. While this structural change will put more pressure on an education system still strugglingto meet basic access issues, it mightalso mean that Mozambique is capable o f usingphysical capital more efficiently-a development that could be consistent with propoor growth if it allows the absorption o f more labor out o f agriculture and other low productivity sectors. 43 Continuingpoverty reduction requires continued labor mobility from low productivity sectors to higher ones. Inurban areas private sector wage employment growth needs to accelerate in the industrial sector. As the rural sector sheds labor, the urban labor force i s expected to grow rapidly. This labor force needsjobs. While self- employment inthe service sector was an important driver o f poverty reduction during 1997-2003, ifthis sector has to absorb all o f the future urban labor force growth, productivity will decline to the point where incomes stagnate, especially among the poor and least skilled. And the self-employed sector i s highly dependent on overall urban income increases (figure 4). Figure 4. In UrbanAreas Employmentin AgricultureI s Down, Self-employment outside AgricultureUp I Annualgrowthinahareofomplaymentover1997~003 Share in urban amploymint in 2005 I 8.7 Source: IAF data for 1997 and2003. 44 To drive wage and salary-jobcreati w, Mozambique needs.foreign direct investment in labor-intensive industries. Confining foreign direct investment to capital and energy-intensive industrial megaprojects will not create the neededjobs. Mozambique's domestic manufacturing sector i s mostly very small, concentrated in low- tech food, beverages, and wood processing. There should be potential for diversification into labor-intensive manufacturing exports, such as textiles. Past performance indicates XXV that Mozambique should be able to attract the foreign direct investment required to create thesejobs. Maintainingand improvingthe investment climate are crucial, especially in areas most important for large manufacturing enterprises. 45 But even with a major increase in industrial sector wagejobs, Mozambique cannot createjobsfast enough to absorb all new entrants. The focus on the investment climate for large firms should be complemented by support for the poorly capitalized small and micro businesses. The highmean income for family business, especially inthe service sector inurban areas, i s good for poverty reduction. But the widening inequality within the sector is worrisome-it reduces the efficiency o f growth inreducing poverty. And it is an indication o f the growing heterogeneity of the service sector, with successful entrepreneurs working side by side with marginal ones. 46 Therural nonfarm sector ved by growing incomes in thefarm sector) will also be an importantpart of thepoverty reductionpicture. The market for many nonfarm activities is currently limited to the local community. Opening market opportunities beyond the community i s likely to yield a sizable impact and ensure the expansion and increased profitability o f these self-employment activities. 47 A key element of propoor growth is gender equity, as both a driver and a consequence of growth. Mozambique has already recognized and targeted increased educational attainments for all, including girls, as a driver o f growth. Less attention seems to have been paid to the implications for increasingproductivity inthe context o f the agricultural sector's growing feminization. This has particular implications for the policies requiredto sustain thetrend o fthe recent productivity increase in agriculture, the foundation o f Mozambique's propoor growth performance. Women are often overlooked by extension agents, so they may not get the informationthey need. Women have less access to the capital neededto invest inthese technologies. International experience has shown that successful agricultural projects must include women farmers early inthe designstage so that projects meet their needs. Strategies are neededto helpwomen diversify crops and raise output to a commercial scale without riskinghousehold food security. 48 Wideningthe employmentopportunitiesfor women, either in commercial agriculture or nonfarm activities-by increasingthe mobility of women out of agriculture-will also help reducepoverty. The fundamental measures may be outside agriculture. For example, improved infrastructure, such as rural water supply, can reduce the time women spend on household chores and free up more time for income-generating activities, including learning about better farming practices or buildinghuman capital. Supportinga Diverse andDiversified Agricultural Sector 49 Supporting a more diverse agricultural sector will require investments in research and extension to help smallholder subsistence farmers increase their productivity inbasic food crop production, commercial smallholders increase the scale and diversity o f their xxvi crop portfolio and their farm yields, and a large-scale farming sector generate employment and other linkswith rural smallholders. 50 Ruralpropoor growth will comefrom increasing agriculturalproductivity and diversifying cropproduction in thesmallholder sector. Because crop income remains important as a source o f income for the poor, sustainable increases inthis source will still be important to achieve propoor growth. And because the majority o f households are net food buyers, increases in crop income must originate from productivity gains infood crops and not only from price increases that reflect local scarcity. Productivity gains need to be reflected not only in increased yields per hectare but also inoutput per worker, from improved seeds and chemical inputs and from increasedmechanization. 51 Crop incomecan also be increased by having rural households diversify into profitable cash crops,particularly becauseproductivity-enhancing inputs aregenerally associatedwith such crops. Private investment incash crops for export brings technologies that can transform the agricultural sector. And the new technologies can spill over to other crops, raising incomes inthe whole community. 52 Encouraging large-scale, commercial agriculture can benefit smallholder farmers, especially if it strengthenslinks between the twogroups and does not crowd out smallholders.A successful agricultural strategy will rely on the private sector for needed investments and know-how. Commercial agriculture bringsnew technologies, which can spread to all farmers inthe area, large and small. Commercial farmers also subsidize infrastructure development, provide wage employment, and increase demand for products o f the local nonfarm sector. 53 Contractfarming brings improvedseeds andproduction techniquesto the farmers, provides inputfinance, and reduces thefarmer's riskfrom greater specializationand uncertain markets andprices. The tobacco sector inMozambiquehas shown that private extension services under these arrangements have a highrate o f return. Note, however, that contract farming requires an appropriate regulatory framework that strikes a balance between protecting the farmer while ensuring returns to the contracting company commensurate with the risk. Supporting Market DevelopmentandNew Institutional Arrangementsfor Stronger Market Participation 54 To enhance the efficiency o f agricultural markets, it will be important to invest in marketing infrastructure and rural services, improve access to information, and improve coordination mechanisms for key domestic food chains and export-oriented high-value supply chains. 55 Continuing investment in road infrastructure is important to open marketsto commercialsmallholders and to allow new investmentstoprosper, but complementary investmentsare also needed. With Mozambique's low population density, road infrastructure i s important for the success o f commercial agriculture because xxvii underdeveloped transport and rural infrastructure services increase marketing costs substantially and prevent domestic and export markets from developing. Inadditionto the main national roads, a well targeted programshould focus on maintainingsecondary and tertiary roads throughout the agricultural season. 56 Poor road access in many cases coincideswith other bottlenecks,such aspoor agroecology,low population density, weakservices,and a lack of water and electricity. So, a well targetedprogram needs to take into account the interactions with geographic, community, and household characteristics and consider complementary investments to ease those other constraints and maximize the impacts o f road construction and rehabilitation for market efficiency and household welfare. And funding and institutional arrangements (community-based maintenance contracts) needto be put inplace to ensure the sustainability o f road investments and maximize its long-term impact. 57 Access to market information and intelligence can improvethe competitiveness in the agricultural marketing systems, ensuring thatfarmers participate in a more leveledplayingfield in agricultural markets. Mozambique has had a public marketing information system inplace since early 1990.The system has evolved inresponse to user needs, but sustainability has beena constant challenge. Decentralizationto respond to the needs o f users inregional spots has beentaking place, and the use o f technology is progressing. The system should be supported to provide timely information on prices for differenttransaction levels, distribution channels, competitors, and buyer and producer trends. 58 Tohelpfarmers meet the required standards in high-value chains linked to domesticand regional and international markets, the market information systems should alsoprovide and disseminateinformation on regulations,grades and standards, post-harvest handling advice, and storageand transport recommendations.Public- private partnerships might be needed to accomplish this. The continuing exploration and use o f the Internet and the fast-growing cell phone network inthe country will aid inthis process. 59 Stronger coordination mechanisms,including thosefor contractfarming in key domesticfood chains and export-orientedhigh-value supply chains can improve smallholderparticipation in these markets. Their participation has so far been limited by their inability to meet quality standards, timeliness requirements, and economies of scale. Failures in credit and input markets complicate the problem. 60 Tosupport smallholder market access, mutually beneficial contractual links should be establishedbetween smallholdersand large tradersor processors in selected valuechains. Policy interventions inthe agribusiness sector needto strengthen links between smallholders and the emerging supermarkets (and other traders) in urban areas, and processors and traders in export-oriented value chains. To reduce the typically high transaction costs associated with contracting arrangements, farmer associations should be developed and supported to generate more balanced outcomes. xxviii SupportingIncome Diversification through Labor Mobility 61 More active labor markets and moreprofitable rural self-employment opportunitiescan help households improve and sustain their welfare levels. Although full-timeformal nonfarmwage employment opportunities may not be widely available to most households inthe near future (given the limitedavailability and growth o f formal sector jobs inrural areas), income diversification into seasonal or part-time wage labor will remain crucial for reducing ruralpoverty. Mozambique may also wish to consider rural public works schemes to expand opportunities for the poorest farmers who cannot break out o f subsistence. Past diversification o f household income came from pull factors-from a dynamic rural economy. Butthere will also be smallholders who cannot modernize or commercialize, and who cannot develop or stay profitable in off-farm activities. They have few assets and may lack adults inthe household. Providing income- earning opportunities through community-driven public works programs could be an effective safety net for them. 62 Employment linkages in thefarm sector are important in generatingfarm wage income to households, but its low returns and the seasonal nature limit the extent to which it can sustain ruralpoverty reduction. Opportunities for farm wage labor can be expanded as a resulto f the expanded activity o f larger farming households, such as commercial smallholders, or large private agricultural producers capable o f generating employment for their poorer neighbors. While this is important, especially for the asset poor, wages in agriculture tend to be low and seasonal, particularly for low-skilled labor working in small-scale farms. This points to the importance o f supporting a more diverse and diversified agricultural sector o f both commercial smallholders and well managed large-scale farms. 63 Alternative and moreproductive wage opportunities, on and off thefarm, are needed to maximize the effects of income diversification in rural areas. While the rates o f diversification have increased for virtually all groups, the graduation o f such activities to the major source o f income has been varied. This means that earnings from low- productivity, low-skilled work off the farm are also low. The ability to diversify i s positively correlated with the skill requirements o f the targeted activities. To be supported are alternative employment opportunities adequate for more skilled labor on and off the farm. But for the rural poor to benefit, continuing investments ineducation and skill development will be needed in rural areas. BUILDING HUMANAPITAkBY GETTINGPUBLICSERVICES CLOSER TO THEPOOR C 64 I n the mid 1990sthegovernment's challenge was to usepublic resources to reach millions of households neglected throughout theyears of conflict withpublic services in order to improve welfare, build human capital, and generate income. The strategy was to allocate a large share o f the budget (including donor funds) to public service delivery. The focus has been on improving quantity o f services, for the most part to build infrastructure. With Mozambique's lack o f any services inmany parts o f a vast country, this policy made sense. A highpercentage o f expenditures moved out o f Maputo xxix to rural areas for infrastructure construction (schools, roads, and water points), for equipmentand supplies, and for personnel costs. Giventhe low capacity o f the central government at the time, this was no mean feat. 65 I n quantitative terms the effort has been a success.Access and outcome indicators are up, especially in education.Health indicators for 2003 show higher service use among the poorest quintile.Nearly 60 percent o f children aged 6-1 1 inthe poorest quintile were inschool that year. Since then, enrollments have continued to climb. Gender parity has been nearly achieved at the primary level inmost areas. But at the postprimary level, neither girls nor the poor have access to education. 66 Since 2001 Mozambique has had a clear strategyfor allocating expenditures with geographically disaggregatedtargetsfor improvementsin access and outcomes (articulated in the PARPA). But the plan has only beenpartially implemented, and systemsto allocate expenditures directly toward these goals are not yet ready. Until recently, many donor-financed programs were completely off the books, so the government did not actually know the total spending inhealth. A systematic programto monitor outputs that reach the poor i s still lacking. And without good data on the incidence o f spendingand on the outcomes, it is impossible to increase the value for money. Mozambique needs to provide both clear data on whether the outputs were produced and whether the poor used the services, what quality services they received, and how the services improvedtheir welfare. 67 Butthe surgingdemand remains to be met.Primary schools are burstingwith students, and pupil-teacher ratios are climbing rapidly. Many poor households remain unable to use other services. Cost i s still the overriding access issue, especially for secondary education, because poor households simply do not have the cash to pay for the service. There are other costs, even for free services: travel time and money. Long distances to service facilities hurt girls and women more because they have less time and because safety duringthejourney is an issue. Girls inrural areas start school late for this reason. Inrural areas inthe north, for example, distance creates a vicious cycle. Long distances to fetch water keep girls at home to do chores, raisingthe opportunity cost to the family o f sending a girl to school. Long distances to school also demand time and effort to ensure girls' safety, again raising the opportunity cost to the family. And long distances to healthcare facilities, along with fees, raise the real and opportunity costs to the family, perhaps leaving girls inpoor health, another cost. This cycle disruptsthe education and skills o f women, lowers productivity, and devolves into chronic poverty. 68 The net outcome: wealthier households receive a much higher share of the government budget because they can use the services. Inurban areas it is easier for women to get to health clinics. But user fees play a role as well. Women who have to ask their husband for money to cover the transport cost and user fee for a clinic visit simply do not go. School fees cover a small portion o f the total cost o f secondary education, but they are highenough to keep the poor out and to allow the wealthier households to enjoy a larger chunk o f total expenditures. 69 It appears that not enough attention has beenpaid to expandingpublic services that directly increase the income of thepoor. Internationallyaccepted criteriafor allocating spending among sectors for povertyreductiondo notexist, so there is no benchmark for how muchmoney shouldbe spent on roads, for example. Country context i s an overridingcriterion, includingwhich area i s mostefficient at usingfunds. But agricultureandruraldevelopment havenever receivedmorethan 5 percentoftotal spending, and spending onruralroads has laggedas well. Analysis inMozambiqueand other countries shows that good extensionservices do raise productivityof small holders. Reachingand supportingsmallfarmers is particularly importantfor reducingpoverty, andpublicly financedinfrastructureis criticalto encouragingthe privatesector to invest, supportingruralgrowth. 70 At the national levelgovernment expenditurepolicy is moving in the right direction to change this reality-more than half of education spending isfor the expansion and maintenance of primary level schools,for example. Completingprimary educationhas a huge impacton householdwealth, bothinhigher earnings and inbetter householdmanagement of scarce resources. Buttoo many girls still cannot completefive years ofprimary school, and fewer complete the seven years requiredto progress to secondary school.This spills over into other areas-nutrition, health, andthe ability to make a living and runa household. Increasing Public Spending and GettingMore Valuefrom It 71 To maximizethe effectsofpublic spending onthe poor, the government needsto increase public spending on facilities and services usedby the poor andvulnerable, increase value for money inthe public sector throughhigherqualitypublic services and better delivery systems, and improvesystems to monitor government expendituresand evaluate how effectivelythey are allocatedto the poor. 72 I n education this translatesinto increasing the quantity and quality of the services delivered, with severalpriorities. Furtherexpandingthe school networkandreducingthe costs of educationto reduce distance and alleviate financialconstraintsfor poor households. Increasing access and completionrates among girls needto be a top priority, particularly in ruralareas. Continuingto pay special attentionto teacher training, to keepupwith surging demand, and investincurriculardevelopmentto improvequality at all levels. 0 Investingmore inpostprimaryeducationto keeppacewith progress inprimary access and completion,targetingrural areas andpoor and vulnerablegroups, especiallygirls. 73 To increase attendance amongpoorer households, conditional cash transfers could be considered,after a careful assessmentof theirfeasibility and overall xxxi sustainability. Scaling up existingsocial protection programs that have indirect effects on children could also be considered. 74 I n health thefocus should be on improving the coverageand quality of service delivery. Invest innew health posts and health centers to reach the underserved, and improve the distribution o f health personnel, includingwell remunerated and qualifiedprofessionals, to minimize shortages among rural and less favored areas and groups. 0 Develop outreach services that bringhealth services to the population, through a network o f community health workers (based inthe villages) working with existing health centers. Outreach activities should increase awareness o f the importance o f formal health services. 0 Manage the distribution o f medicines and diagnostics equipment to health .facilities, particularly in rural and remote areas. 75 Investmentsin education and information can improve access to health services and health outcomesandproduce much greater valuefor money than, say, new urban hospitals. Inthe short term disseminating informationcan improve childcare practices inrural areas lacking health facilities. 76 I n agricultural extension coverageshould be increased and actions well targetedto the needs of farmers. Adopt a holistic view that integrates the extension system with agricultural research to ensure that the materials developed by researchers are well adapted to the needs o f farmers, particularly poor farmers and groups with specific characteristics. That is, take into account the heterogeneity o f farmers and areas to be targeted. 0 Structure spending to allow for better geographical coverage and outreach to help farmers adopt technologies that will increase their production and productivity. 0 Promote infrastructure development, institutions (support associations in reducing transaction costs), and marketingto link farmers to markets. Promote public-private partnerships, including outsourcing, for better targeting and greater benefits to participating smallholders. 77 I n water investments in sustainable systems have economic value-they increase the time availablefor productive activities and reduce the incidence of waterborne diseases, thus reducing the health shocks to households and the needfor extra health expenditures.Access to clean water needs to be improved and the regional and rural-urban gaps reduced. Spending needs to keep up with population growth. xxxii Invest in water supply systems that are well distributed (spatially) to reduce the distance to water collectionpoints. This could reduce school dropouts and increase school completion, particularly for girls. Sustain existing public water supply systems, particularly inurban areas, through effective cost recovery, perhaps through user fees. Allocate the receipts to maintenance provision and system upgrading. Provide technical assistance at all levels-ranging from piped water systems in urban areas to public fountains and private wells inurban and rural areas-to ensure that existing systems work properly and are well maintained. Monitor and Evaluate Government Spending 78 Past weaknesses in monitoring and evaluation have allowedpoor practices to persist-at great human andfinancial cost. Without data it is impossible to know what planned strategies and programs will cost. Monitoring government expenditures and evaluating how effectively they are allocated to the poor will require monitoring systems that link spending to outcomes and targets. Evaluating resource flows through selected sector Public Expenditure Tracking Surveys can identify with leakages and increase understanding o f the effectiveness o f sector resource allocation. Regular-well designed and well planned-household surveys can monitor poverty trends and their determinants. They can also provide the informationneeded for costed sector strategies that can produce results. The timing o f those surveys should be aligned with the reporting needs for indicators inthe Poverty Reduction Strategy and other national strategies. IMPROVING GOVERNANCE ACCOUNTABILITY-BY GETTINGTHE STATE CLOSER AND TO ITS POOR CITIZENS 79 Mozambique's record on governance and accountability is mixed. The legacy o f colonialism, decades o f protracted conflict and centralized rule damaged state-society relations, trust, and downward accountability in Mozambique. The checks and balances developed during the postwar period have frequently beenunable to safeguardthe needs and interests o f citizens, particularly poor ones, leading to resentment, frustration, marginalization, and exclusion from the public sphere. Multiparty elections in 1994 were intended to support pluralism, but the winner-take-all system has resulted in a weak parliament and single-party dominance over all three branches o f government. The judiciary-largely absent from the lives o f the poor-is plagued by allegations o f corruption. The media and civil society organizations have limited social accountability mechanisms. 80 Sustainingpoverty reduction and increasing inclusion require social, economic,andpublic institutionsthat respond topoor people's needs. The government's poverty reduction strategy puts governance squarely within the poverty debate inMozambique. One o f the three pillars o f the PARPA I1poverty reduction xxxiii strategy focuses on governance, includingpolicies and programs to fight corruption, solidify the rule o f law, improve protection from crime, raise accountability o f the state, and support community and individual initiative. Three o f the main instrumentsare a new legal framework for decentralization and enhanced implementation o f the 1997 Land Law and the 2005 Family Law. The strategy also supports measures to implement other laws to improve thejustice system at the local level. Characterizing all these measures are attempts to bringthe state closer to the real lives o f Mozambique's poor. 81 Aforward-looking strategyfor Mozambique has to lend equal weight to empowering communities and their localities as agents of change and would be best positioned to do so by: Makingdecentralization workfor thepoor-by devolving real power to local communities so that they may advocate for more responsive public service delivery, influence service delivery mechanisms, and hold local service providers accountable for their performance. Increasingaccess tojusticefor the poor-by expanding poor people's opportunities to engage with institutions and ensuring that thejudicial system i s correspondingly responsive to necessary system changes to makejustice accessible to the poor. Makingpoorpeople's landrights real-by formalizing land rights and titling, improving overall awareness o f citizenship rightsto land, and strengtheningthe consultation process around community land. Making Decentralization Workfor the Poor 82 Decentralization and greater emphasis on citizen and community participation arepart of thegovernment's strategyfor improving service delivery. By bringing decisionmaking closer to the people and creating spaces for citizenparticipation in local planningand implementation, the government's decentralization reforms are intended to enhance the responsiveness o f public services to locally perceivedpriorities; increase their effectiveness, quality, andrelevance; improve downward accountability and local checks and balances; and raise the impact o f government policy on poverty reduction. 83 Toparticipate effectively in service delivery, people need opportunities to influence the mobilization and management of resources and the delivery of services through voice (consultative and oversight mechanisms channeling service user preferences to service managers andfrontline service deliverers) and choice (available options to select among multipleproviders) so thatpeople and communities are not forced to accept unresponsive service delivery. Critical parts o f an empowering agenda include access to information, capacity enhancement, and direct participation. xxxiv Informing communitiesof their role in decentralizedservice delivery. 84 Beyond legislativeadvances, consultativecouncils have been implemented in all districtsin Mozambique. Anecdotal evidence from the Poverty and Vulnerability Survey suggests that most consultative bodies were established from top-down government pressure-rather than from bottom-up demand-given their legal requirement as part o f the district deconcentration process. Indeed, inthe districts and communities visited (outside the pilotprovince o f Nampula), some consultative bodies were still not functioning properly. Evenina province (Niassa), the consultative forum was functioning in one o f the localities studied (Luelele, Mandimba) and not inanother (Lichinga). Strengtheningstate and local capacity. 85 Weaknessesin capacity-within both the state and civil society at local levels- threaten to undermine reform's potential benefits. Moving beyond legal institutionalization can ensure that participatory social accountability mechanisms work on the ground. Alongside the Local Initiative Investment Budget, large investments are needed inthe human, technical, managerial, and administrative capacity o f local governments to manage the district development process and associated budgets and to boost downward responsiveness and accountability. The decentralization reforms are based on radically different ways o f conducting government and deliveringsocial services. New skills and attitudes are required to facilitate participatory planning and encourage citizens' voices, initiatives, and engagement. Some promising activities are in evidence, such as Radio Moc and community radio broadcasts inthe central part o f the country, which have increased awareness o f the district budget and planning process. Improving representationandparticipation. 86 The extent to which local communitiescan meaningfullyparticipate in the local budget andplanning cycle is uncertain. This should not, however, lead to the conclusion that civil society i s incapable o f participating inthe development planning process. What is required i s building the capacity o f citizens, civil society organizations, and even the media to interact with local governments and hold them accountable (box 2). An important element o f such capacity buildingis improving citizen access to the informationand knowledge needed for accountability-on budgets, procurement, service delivery costs, and user rights (including citizenship rights more broadly). 87 To engagein social accountability mechanisms, the verypoorest in communitiesmay even need assistance in building their capacity to aspire-beyond the organizational, technical, and basic reading and math skills that community representativeswill need to master to effectively participate in these consultativebodies. Inshort, the education andtraining ofcitizens and civil society needto be more fully integrated inthe reform process to raise quality o f citizen participation inpreparing, implementing, and monitoringdevelopment plans. xxxv Box 2. Improving Participation in Planningand Monitoring Government Programs PARPA Iopened space for amore participatory process inthe development of planning instrumentsand promisedthat apermanent process would be establishedinvolving representatives ofthe sectors of government, the private sector, trade union confederations,civil society ingeneral, as well as funding agencies (including donors). To achievethis goal, the Poverty Observatory (PO) was formally set up by the Mozambican government with donor support on 19April, 2003. The PO has helpedbroadencivic engagement inthe monitoring and evaluation of public policymaking, with the aim of increasingpublic and downward accountability and responsiveness of governmentinstitutions and partnersinPARPA implementation.The PO has great potential for bringing together key stakeholders inpoverty planning, implementation, and monitoring. But it has remaineda purely consultativeforum, articulatedneither with any relevant sovereign organs (parliament) nor with the actualplanning and budgetingprocess ofthe state. And it is sidelinedby other monitoring and evaluationsystems, particularly by thejoint review mechanisms where the government's attentionis focused. Opportunitiesexist to build on the successes of the PO and the decentralizationprocess, and specifically to linkthe POs at national and provincial levelsto the bottom-up planning and monitoring ofthe consultativecouncils and forum at district levels and below. Giventhe many complexities, includingthe low capacity of civil society and citizens at the lower levels ofthe state, further resources and investmentswill be neededto develop a fully functioning and participatory monitoring system inMozambique. From apublic accountability perspective,much more clarity is required on expected results- recommendationsfrom the POs serve the poverty planning and programming cycle of the government.There is as yet no systematic feedback mechanism betweenthe government and the PO. A clear weakness inthe structure ofthe POs is the lack of broadrepresentationbeyondthe small and mostly urbanmiddle-class elite. At the provincial level, concerns about the role ofthe PO and its articulation inlocal strategies, plans, and budgets are even more pronounced. The work of local consultativecouncils shouldbe integratedinto the monitoring efforts of the provincial POs. Strengtheningthe incentivesfor participation and implementation. 88 Participation should be broadened beyond the districtplanning and budgeting stage to ensuregreater downward responsiveness and accountability. By limiting citizen participation to consultations during the district planning stages, the incentives to participate and the ability to hold district, province, and national governments accountable are weakened. Incentives to participate inthe district planning process are undermined.by the slow pace o f fiscal decentralization to districts. Perhaps more worrisome, however, i s that few incentive mechanisms are available to district consultative councils-and to citizens more generally-to hold district governments accountable for local decisions, plans, and investments.Not even the crude mechanism o f holding district officials accountable through the ballot box i s available to citizens in rural areas. So, while consultative councils may have addressed part o f the democratic deficit at the district level, they should not be seen as a substitute for democratic local government in the long term. xxxvi Learning frompilots. 89 Alongside communityparticipation inplanning is a need to institutionalize greater civic engagementin monitoring and evaluating districtplans and implementation. International experiences suggest several strategies to strengthen participatory monitoring and evaluation o f public service delivery, including citizen report cards, social audits, and community scorecards. These should be piloted-and scaled up when they prove effective. IncreasingAccess to Justicefor thePoor 90 Improving the understandingof how poor people can access the resources and assets to achieve a secure and sustainablelivelihood requiresfocusing on the systems of rights andjustice available to thepoor. Inrecent years the government has put in place a range o f legislative reforms that have begunto change the landscape o f rights, justice, and empowerment for the poor. The most significant i s the 2004 revision o f the Constitution. Other reforms-such as the 1997 Land Law and the 2005 Family Law-are also directly relevant to the lives o f poor people, regulating their access to key resources inside and outside the household. Decentralization is also changing the institutional landscape and the public policy approach aimed at realizingmany o f these rights. 91 Theformal justice system remains inaccessible to most of thepoor, with several constraintsto access. A recent survey o f national perceptions o f the state o f governance and corruption suggests that the three main obstacles are physical access and distance, costs, and corruption (table 3). General lack o f awareness and knowledge o f rights is also a factor. Table 3. HouseholdConstraints on Access to Justice Importance of Not Important Not Very Somewhat constraint at All Important Important Important Very Important costs 24.1 16.6 18.8 18.0 22.5 Bribes 26.6 16.3 27.6 13.3 16.2 Corruption 19.1 17.6 23.7 18.1 21.4 Incompetence 23.0 17.4 25.4 15.8 18.4 Waiting time/delays 20.2 17.1 22.9 20.1 19.6 Complexity 20.9 17.8 26.2 19.0 16.1 Execution 17.9 16.6 26.3 19.1 20.2 Distance 25.1 15.3 17.1 13.8 28.6 Source: Poverty and Vulnerability Survey. Disseminating information on rights and responsibilities. 92 The majority of poor men and women interviewed had very little insight on their specific rights and responsibilitiesas citizens.The few that recognized the existence o f rights highlightedthe right to vote, to live infreedom, to life, to work, and to health, food, and education. Some emphasizedcitizenship responsibilities-to pay taxes and to carry personal identification, a responsibility that carries its own set o f problems xxxvii for the poor. The majority o f poor households are even unaware o f recent laws directly relevant to their livelihoods and assets, such as the land and family laws. People's low awareness o f their rightsextends to land and inheritance rights. 93 Better disseminationof information on citizens' rights and responsibilitiesis thus critical. Given the substantial changes to the country's legal code since the early 1990s-particularly insmall-scale commerce, natural resources, and family matters-it is impohant to disseminate informationon these new laws and their implications for the rights o fthe poor. New laws are usually disseminatedthrough the Ofjcial Bulletin. But the sporadic and limited publishingo f the bulletinhas hindered widespread dissemination o f laws and other relevant legal information. Moreover, few laws have beentranslated into local languages and disseminated widely, which causes major difficulties ina country where only 8 percent speak Portuguese, the official language, as their native language and only 27 percent speak it as a second language. 94 The costs of accessingjudicial services are also a major obstacle. Court fees, costs o f legal representation, and transportation and accommodation costs for those living far from the courts-most o f the population-are usually prohibitive. Making thejustice systemfair. 95 Bribes and corruption areparticularly problematicfor thepoor in their access tojustice. As many as 35 percent o f the households interviewed inthe Poverty and Vulnerability Survey expected to have to pay a bribe to receivejudicial services. The police were rated as the fourth most corrupt public institution in Mozambique, with nearly a fifth o f responding households saying they always have to pay a bribe to the police inreturn for a service. Traffic police were rated as the second most corrupt after customs officials. While the proportion o f households considering the courts corrupt i s lower, it remains significant (11percent o f households claim that they always have to pay a bribe to receive assistance from the courts). 96 The representativenessand inclusivenessof communityjustice alsopresent concernsfor thepoor. More than 60 percent o fresponding households felt that the weak and the poor are unable to get their interests considered within thejustice system, and more than 40 percent felt that the system i s insufficiently independent from the government and manipulated by economic interests-and unfair overall as a result. The quality o fjustice services ranked as particularly poor. Indeed, 42 percent o f households rated the quality o f police services as either very bad or bad. The court system fared only slightly better, with 29 percent of households rating its services as very bad or bad. Establishingprocedures to link theformal and customary systems. 97 Marginalizedfrom theformal justice system, the majority of households- particularly in rural areas-rely on informal and traditional mechanismsof dispute resolution. Only 5 percent o f households in2005 resorted to the formal courts during the previous year. Inrural areas the proportionwas 3 percent. Because the state police have a xxxviii greater presence inrural areas, the proportion o f households that sought their help during the previous year was slightly higher but still very low (8 percent nationally and 4 percent inrural areas). 98 So, most citizensresort to community courts, traditional authorities, and local leaders. But despitebeing closer to the needs o f communities, community courts do not always overcome the problems associated with the formal system. Indeed, some households interviewed inthe Poverty and Vulnerability Survey questioned the fairness and integrity o f communityjustice systems. A male head o f household inPebane, Zambezia, for example, argued that "justice" could be bought-not only at the formal level but also from the community court. 99 Local authorities, by virtue of their roles as elected leaders with administrative mandates, often mediate and resolve community disputes. Buthow these mechanisms follow established legal procedures is unclear. The 2004 Constitution does not explicitly recognize the role o f traditional and local authorities within the justice system, despite their importance infacilitating accesstojustice for most o fthe poor. Butby confirming the principleo f legalpluralism, it allows for future legislationto better integrate such mechanisms into formal justice. Making Poor People'sLand Rights Real 100 The 1997Land Law and accompanyingregulations were toprotect the land- use rights of small-scalefarmers by recognizing the legitimacy of customary systems- and toprovide investors with secure long-term rights while maintainingstate ownershipof all land. The law broke ground by recognizing access rights to land through customary and good faith occupation as equivalent to state-allocated land-use and benefit rights. The law introduced legal pluralism and equivalence between traditional and formal justice. While the law encourages formalizing land-use rights, formalization is not required for the rightsto be legally recognized. And by emphasizing the rightsof individuals to take their land out of customary jurisdiction and affirming the primacy o f constitutional principles, the law aims to guarantee women's legal equality. 101 The law alsoprovidesprivate investors with secure rights through a renewable 50-year state leasehold. Before acquiring such land, however, a community consultation must set out the conditions for the takeover of the community right.Intheory, this consultation mechanism formalizes the participation o f local people in land allocation decisions, and so should protect local rights, guarantee local livelihoods, and generate new resources for local community development. Ownedby the state, the land cannot be bought or sold, but any private or community assets on the land can be sold, with the underlyingland-use and benefit rightsmerely transferred to the new owner. 102 Implementation of the law has been relatively slow, but several advancescan be identifed. The government-supported by donors and national nongovernmental organizations and community service organizations-has investedheavily inraising awareness o f the land law. Inaddition to making the law available inthe national press xxxix and OfJicial Bulletin, the law benefited from extensive consultations with civil society, translation into six languages, and-most important-a national campaign to take the basic messages o f the law to the local level. Nongovernmental organizations and community service organizations have also been very active throughout the process, educating the public on citizen rightsand raising awareness o f the land law. Building capacity and trainingpersonnel have also been important. The Centre for Legal and Judicial Training, part o f the Ministryo f Justice, has trained community and district judges and prosecutors, district administrators, and paralegals to assist local communities. Increasingpoor people 's awareness of the law. 103 Despite investments inpublic education,most households-particularly poor households and those headed by women-are unaware of the land law and its implications.The findings from the Poverty and Vulnerability Survey suggest that people's low awareness o f their citizenship rights extends to land rights. Indeed, both quantitative and qualitative fieldwork shows that nearly all respondents have little or no awareness o f the land law or its implementation. Only 15 percent o f urban households and 11percent o f rural have any knowledge o f how to obtain a land title under the current land law. 104 I n urban areas more than twice as many male-headed householdsasfemale- headedhouseholds know about titlingprocedures. Even inrural areas, the findings vary by gender, although somewhat less so. Householdwelfare also matters, especially in urban areas, where 30 percent o f the richest third o f households know about land titling requirements, compared with no households inthe poorest third. Even a decade after the law came into effect, very few households know their specific rights to the land they rely on for their daily livelihoods. 105 Unaware of their rights, thepoor are less likely to know and benefit from the real economic value of their land. Communities are also often unaware o f the value o f the landthat they surrender to investors-and the returns that it could generate through careful deals. Ifthe long-term benefits o f these transactions accrue mostly to investors, their impact on reducing poverty will be negative. Empoweringpoor households to register their land rights. 106 Voluntary land registration and low awarenessof rights reduce the demandfor land titles. Only a few households have formalized and registered their land rights.Even fewer households than those claiming to have knowledge about the law have titles to their land-less than I O percent inurban areas and only 1percent o f households in rural areas. Nationally representative data suggest that only 2.6 percent o f rural households have titles to their land. Because most people's rightsto land are still customarily and informally administered, they remain largely invisible to formal systems o f rights . registration and recording. xl Delimiting community land. 107 Delimitation initiatives are benejiting local communities. Inaddition to allowing community groups to negotiate better benefits from potential private investors, delimitationhas raised community awareness o f legal rights (beyond land rights alone) and empowered greater participation inmanaging local resources. And by providing a clearer definition o f community boundaries, it has improved social cohesion and reduced conflicts. Further evidence o f progress is shown by the many community consultations before private land investmentsand acquisitions. Community consultations take place in mostcases o f private claims on communal land-a great improvement on past practices, which often led to allegations o f private expropriations o f communal land. As a result, more o f the economic benefits o f private development are likely to be shared with local people and communities. Stoppingthe hijacking of the consultationprocess. 108 Although community consultationsfrequently accompanyprivate claims to community land, the quality of the consultation does not guarantee local rights and benefits. Inthe context o f rising demand for land from private investors+specially where community land-use and benefit rights are unregistered-the land law's community consultation mechanism provides a potentially important means o f formalizing land rights in a way that puts local rights firmly on the agenda. All private investors seeking land are required to conduct a consultation with the relevant community members to informthem o f the purpose o f the land use and to give them an opportunity to state their interests inthe land. The consultation must then stipulate the terms o f a partnership between the community and the investor, under which the land-use and benefit rights are transferred. 109 Although the requirement for community consultations has been fulfilled inmost private claims, the quality o f the consultation process i s usually very poor, either resulting inessentially unlawful occupation o f someone else's land or failing to produce real economic benefits at a level that can either maintain livelihoods or alleviate the poverty o f beneficiaries. 110 Therequirementfor communityparticipation hasput local rights squarely on the agenda. But unless the failure to generate the local benefits expected from community consultations i s addressed, the consultations may give land grabs by private investors a veneer o f respectability by demonstrating compliance with the law and apparently safeguarding local needs and interests. Making womenbetter off in claiming rights. 111 Mozambique is often said to have one of the mostprogressive andgender- sensitive land laws in SouthernAfrica. Butwomen continue to suffer discrimination in land rights, partly because o f a lack o f information and awareness about their rights, partly because o f continuing discriminatory customary practices, and partly because xli administrative andjudicial practices are still far from incorporating the norms and dynamics that the land law seeks to encourage. Women rarely actively participate inthe consultations-partly because dealing with the outside world is considered a man's role inmost rural areas. Women's marginalization is worsened bythe limited consultation process, whose brevity hinders the opportunities for women's participation behindthe scenes. Improving local land administration. 112 The weak, inefficient, and inaccessible land administration system charged with implementingthe land law compounds theseproblems. Successful implementation o f the land law requires efficient, accessible, and accountable cadastral services. But land administration inMozambiqueremains highly bureaucratic and centralized, institutionally and technically weak, and severely understaffed, particularly at lower administrative levels. Moreover, the costs o f delimiting and demarcating land holdings are prohibitive. The unresponsiveness o f local cadastral services-r even their absence-are ultimately why customary land tenure systems remain so pervasive. 113 A more decentralized, accessible, and responsive local land administration system is needed, as are efforts to enhance local communities' and individuals' awareness o f the rights stipulated inthe 1997 Land Law and their ability to realize them. * * * 114 Can Mozambique beat the odds and sustain inclusion in a growing economy? Can Mozambique realize its vision o f consolidating peace and democracy, social stability, public safety, and the guarantee o f individual freedoms as basic conditions for economic growth and a reduction inabsolute poverty? 115 Thisreport concludes that it can. Itcan pursue propoor growth by promoting agriculture and the private sector, especially labor-intensive activities, many o f them small and informal. It can build human capital by improving access to basic public services, especially for the poor, and by increasing the value for money inpublic spending. And it can improve governance and accountability by getting government closer to its citizens. To achieve these goals, the government will need to increase the value for money in its spending on public services. It will also needto target services for the rural poor and enlist poor communities inidentifying needs and delivering those services. And it will need to put inplace good tracking systems to link program outputs to targets and outcomes, usingfrequent high-quality household surveys. xlii 1.POVERTY'S MANY DIMENSIONS INMOZAMBIQUE 1.1 Poverty in Mozambiquefell significantly between 1997 and 2003, with the percentage of poorfalling from 69.4percent to 54.1percent. Rural poverty fell slightly faster than urban poverty, though rural poverty overall remains higher than urban poverty. These trends are favorable, but monetary poverty inMozambique remains high, and geographical variations are large. Poor households are found disproportionately in agriculture, where they are most vulnerable to climatic shocks and seasonal income shortfalls. They have larger families and fewer income earners per dependent, and these earners usually have no formal education. The poorest households even lack basic personal necessities, and most rural households live inprecarious shelter. Vulnerability remains highas well, as many households report lacking a stable and secure livelihood and lacking assets that could be sold to buffer shocks. 1.2 High vulnerability in Mozambique, both a cause and a consequence of poverty, is related to community and householdcapacitiesand strategies to deal with unexpected shocks and risks. The most common coping strategy-in both urban and rural areas, with little variation across income groups-is to seek help from family and friends. Drawing down assets was more common inrural areas, despite the smaller asset portfolios o f rural households. Reflectingtheir insecurity, 40 percent o f rural households inthe poorest communities saw only modest or nonexistent changes intheir overall welfare in the past five years. Community poverty is related to the absence o f solidarity, trust, and civic and moralvalues. 1.3 Poverty and vulnerability have a stronggender component.Female-headed households, especially those headed by widows, are more likely to be poor, especially in rural areas. They have fewer assets, and less diversified income structures. Overall in Mozambique, women have fewer income-earning opportunities. And female orphans in poor households are particularly vulnerable, more likely to be doing household chores and less likely to be in school. MONETARY DIMENSIONS POVERTYOF 1.4 Monetary poverty i s conventionally measured by total household consumption.' This report uses the Ministryo fPlanning and Finance's (MPF) methodology, which calculates a consumption aggregate by deflating food prices temporally and spatially. With a flexible food bundle, this amounts to computing different nominalpoverty lines by province, by urban or rural location, and by years2Complementary analysis was 'Both incomeand consumption can be usedto measure monetary dimensionsof welfare. Consumption-based approaches are preferred to income-basedapproaches for a variety of reasons. First, consumptiondata tendto be more reliablethan income data due to incomplete measurement, underreporting (because respondents are believedto be less willing to reveal income), and seasonality of income. Second, it fluctuates less than income (and is thus less influenced by seasonality), making it a better measure of welfare over time. Third, amajority of the labor force inMozambique spends part or all of its time on self-employed activities, making income difficult to measure. This methodology is explained inMPF (2004) and inappendix B. 1 conducted using the limited dataset collected for the rapid household livelihood survey conducted in 2006 as part o f the quantitative and qualitative fieldwork for this study: the Poverty and Vulnerability Survey, or PVS (box 1.2). Box 1.1.About Mozambique With 20 million people, two-thirds of them rural, Mozambique is the seventh most populous country in Sub-SaharanAfrica. It has a long coastline and diverse climate and is prone to natural disasters. Despite recent progress, Mozambique's gross national income per capita, at $3 10, ranks 191of 208 countries. A colony of Portugal for 470 years, Mozambique became independent in 1975. Soon after, it plunged into a civil war, which became regional as neighboring South Africa backed the antigovernment guerrillas, the MozambiqueNational Resistance (MNR, or Renamo). A ceasefire agreement was signed in 1992 between the government and the MNR, ending the war. A multiparty democracy under the 1990 constitution, Mozambique instituted multiparty elections and made a peacefultransition to new leadership in December 2004. By mid-1995 more than 1.7 million Mozambican refugees returnedfrom neighboring Malawi, Zimbabwe, Swaziland, Zambia, Tanzania, and South Africa, the largest repatriation in Sub-Saharan Africa. And an estimated 4 million internally displacedpeoplereturnedto their areas of origin. 1.5 Poverty in Mozambiquefell significantly between 1997 and 2003, with the percentage o f poor people falling from 69.4 percent to 54.1 percent (table 1.1). Indeed, all three standard measures o f poverty declined-the poverty headcount, the poverty gap (measuring the average distance to the poverty line), and the squared poverty gap (the measure most sensitive to gains at the b ~ t t o m )The poverty gap index dropped from . ~ 29.2 percent to 19.9 percent. So, not only did 15.3 percent o f the population move above the poverty line butthe average household consumption ofthose who have not yet escapedpoverty increased. The squared poverty gap measure also fell-from 15.5 percent to 9.9 percent during the same period-a 36 percent drop inthe severity o f poverty. Table 1.1. Poverty Measures, by Area, 1997 and 2003 (percent) Area Headcount Poverty Gap SquaredPoverty Gap 1997 2003 Change 1997 2003 Change 1997 2003 Change All 69.4 54.1 -22.0 29.2 19.9 -31.8 15.5 9.9 -36.1 Urban 63.9 51.6 -19.2 27.2 18.9 -30.5 14.8 9.0 -39.1 Rural 71.6 55.2 -22.9 30.0 20.4 -32.0 15.8 10.3 -34.8 Source: IAF data for 1997 and 2003. Note: Consistentwith the 2003 urbanand ruraldefinitions. ~~ These three measuresof poverty are all membersof the Foster-Greer-Thorbeckepoverty indexes.The poverty headcountindex showsthe share ofhouseholdsbelow the poverty line, but it does not measure how far below. The poverty gap shows how far below the poverty line householdsare on average, expressedas apercentageofthe poverty line.The squaredpoverty gaptakes into accountthe incomegap and inequality among the poor. A dollar ofincome gap for the extreme poor is given more weight than adollar of income gap for thosejust under the poverty line. As aresult, the index increases both with the income gap and with the existence of extreme poverty. 2 1.6 Rural poverty fell slightly faster than urban poverty, though rural poverty remains higher than urban poverty (figure 1.1). Poverty dropped by 16.4 percent inrural areas and 12.3 percent in urban areas. Rural poverty is, however, still higher than urban poverty, with the percentage o f poor peo le in rural areas estimated at 55.2 percent, compared with 51.6 percent inurban areas. 4p Figure 1.1.Poverty Rates, by Area, 1997 and 2003 I ao 70 60 50 Y 22 a, 40 30 20 10 0 I Urban Rural Total Source: IAF data for 1997 and2003. Note: Consistentwith 2003 urban and rural definitions. Poverty calculationbased on per adult equivalent consumption. 1.7 Poverty has been reduced in most parts of the country, though geographical variations are large (see table 1.1). Poverty has fallen most inthe s enter,^ mostly in rural areas, driven by the large changes in the populous provinces o f Sofala and Zambezia.6 Sofala, the country's poorest province in 1997, saw such significant reductions in poverty that in 2003 it became the least poor province in income poverty. Poverty also fell in the north, though more in urban areas, and it rose in Cab0 D e l g a d ~The south experienced . ~ an increase in poverty rates, especially in the city of Maputo and the surrounding province. Poverty in Inhambane, the second poorest province in 1997 after Sofala, fell only slightly, leaving the province as the poorest in Mozambique in 2003. The small poverty reduction in the rural south was overwhelmed by the increase in urban poverty. The National Statistics Institute changed the definitionof an urbanareato be more consistentandrobustafter the 1997 survey hadbeenconducted, so the definitionof an urban area is not consistent inthe two surveys. Inthe official tabulationsofthe data, the inconsistent data are used. Our analysis corrects for this deficiency (see appendix B).After adjustingfor this inconsistency, poverty inmost urbanlocationsfell in step with that in ruralareas, but slightly less in large urbanareas such as Maputo. Inthis reporttables are labeledas "consistent definition" where the corrected data has beenused. The center includesSofala, Tete, Manica, andZambezia; the northNiassa, Cab0 Delgado, andNampula; andthe southInhambane, Gaza, Maputo province, andMaputo city. The government believes that the change in Sofala is overstated, owing to an undermeasurementof consumptionin 1997. Government analysts attributethe increase in Cab0 Delgadoto poor sampling inboth years, but primarily inthe earlier survey, which led to an underestimation ofpoverty in 1997. Regionaltrends in povertychanges, however, show the same picture evenwhen povertyrates are calculated without the provinces with measurement problems. 3 Not only did poverty rise but the depth and severity also increased in Maputo city and Maputo province. 1.8 These trends arefavorable, but monetarypoverty in Mozambique remains high. As the national numbers suggest, 54.1 percent of the population still lives inpoverty. Of the estimated population of 18 million in 2003, nearly 10 million people were below the poverty line. Each province has at least 500,000 poor people, with more than 1 million poor people inZambezia, Nampula, and (disproportionately) Inhambane (figure 1.2). Figure 1.2. Population and Poverty, by Province, 2003 I Province Source: IAF data for 2003. HOUSEHOLD CHARACTERISTICSBY INCOMEGROUP 1.9 How are the geographic, social, and demographic characteristics o f Mozambican households correlated with poverty? 1.10 Thepoverty of households varies significantly by the gender of the head of the household. The income distribution is very equal among male-headed households, but female-headed households have a higher representation in the bottom quintiles, with a correspondingly lower representation in the top quintiles. The distribution improved slightly for male-headed households after 1997, but deteriorated quite significantly for female-headed households, which in 2003 had the largest share o f their population in the lower quintile. These results are largely confirmed by the recent light household survey o f the 2006 PVS.8Much as in 2003 the populations in both urban and rural areas were relatively equally distributed across tercile groups. But as underscored by the perceptions See appendix table A1.21. 4 data presented later in this chapter, female-headed households in urban and, particularly, inrural areas had the largest share of their population inthe poorest tercile. Male-headed households, by contrast, had the smallest share of their population in the poorest tercile, and inrural areas they had the largest share o f their population inthe uppertercile. Box 1.2. Data Used in the Report's Analysis Poverty is a multidimensional problem that goes beyond material deprivation-the lack of food, housing, land and other assets, and the means to attain them. It is also a problem of vulnerability-the inability to cope with adverse shocks and the lack of voice, power, and independence in relation to the institutions of society and the state. Reflecting poverty's many dimensions in Mozambique, this Poverty, Gender and Social Assessment combines multiple disciplines and diagnostic tools to explore poverty using monetary, human, and social indicators, and the perceptionsof poverty highlighted by poor individuals, households, and communities. Mozambique is not a data-rich country. The assessment is based primarily on survey data collectedbetween 1997 and 2003, supplemented by other sources. The maindatasets were: Two cross-sectional national multipurpose household surveys (Inqudrito aos Agregados Familiares, IAF 1997 and 2003). Their expenditure module is the base for the estimates ofthe population inpoverty. Two cross-sectionaldemographic and health surveys (Demographic and Health Survey, DHS 1997 and 2003), which provide the most recent survey data on social service delivery and healthand education outcomes. Cross-sectional rural household income surveys from 1996 and 2002 (Trabalho de Inqudrito Agricola, TIA) and panel survey data from 2002 and 2005. A small quantitative and qualitative Poverty and Vulnerability Survey (PVS) of households, groups, and individuals infour provincesconductedfor this report in2006.' Data from the Ministries of Health, Education, and Planning and Finance were used as well. The study draws on recently completed analyses prepared or sponsored by other donors, such as UNICEF's recent Childhood Poverty in Mozambique: a Situation and TrendAnalysis. The lack of nationally representative data on household consumption and poverty since 2003 limits the focus to developments up to 2003. Where possible, the analysis is updated using more recent data. Butthis means that the data sources might not produce a consistentpicture, since they refer to different periods. This is particularly true for the perceptions data, which were collected three years after the household survey data showed a major decline in poverty. These data are nevertheless used throughout the report. The report aims to better understand household and community organization, poor people's perceptions of their poverty, and the livelihood and coping strategies poor people use to improve their livesa2 1. See appendix B on methodologyfor details onthe quantitativeand qualitative surveys used. 2. See Kanbur (2001) on the importance o f combining quantitative andqualitative approaches to understandingpoverty trends anddynamics. 1.11 Poorer householdstend to be larger, have higher proportions of children, have a higher dependency ratio, and contain a higher proportion of disabled adults. Larger households have lower consumption (both per capita and per adult equivalent) and those 5 households with high proportions of children aged 6-1 4 also have lower consumption.' And while the economic dependency ratio rose in all quintile groups between 1997 and 2003, the bottomquintile has a muchhigher dependency ratio than the others. The bottom quintilealso has more disabledadults, addingto the burdenof dependency. 1.12 Poorer households are, moreover, predominantly rural and their heads work mainly in agriculture, have less education, and are older than those of richer households. The poor live in rural areas, where opportunities in sectors other than agriculture and paid employment are scarce, and their low human capital reduces the chances of taking advantage of opportunities in agriculture and outside it. There have been changes however, reflectinga general shift away from agriculture and toward other sectors dominatedby the upper quintile, the percentage o f household heads in the bottom quintile engaging in agriculture fell from 87 percent in 1997 to 80 percent in 2003. The education of householdheads in all quintiles also rose, though the bottom quintile still contains the largest proportionof householdheads without any education-estimatedat 38 percent. 1.13 Households across all incomegroups improved their holdings of durable goods between the two national survey periods. For some of the higher cost items(motorcycles, bicycles, televisions), the gap between the lowest and highest quintiles widened somewhat, though for most other goods the improvementswere more evenly distributed. Significantly, houses were upgraded in all quintile groups. Not only is housing an important savings vehicle for most households, but as highlighted by the PVS, many households see the quality (and security) of housingas an especially important indicator o f poverty." So, the disproportionateincrease in households with durable roofs between 1997 and 2003 among poorer households is important. Because upgrading is usually a cash expense, it provides a good indicator of improvements in wealth and welfare since 1997 for the poorest income groups. Another indicator o f better welfare is that the share o f income spent on fooddecreasedfor all income groups, if slightly more for the richest.," HUMAN DEVELOPMENT-LOOKINGBEYOND INCOME 1.14 Human well-being in Mozambique, as measured by the human development index, is fairly low. The human development index (HDI), developed by the United Nations Development Programme (UNDP), provides a composite indicator o f human capability and well-being (expressed as a number between 0 and 1, higher scores indicating better development). It is a weighted sum of three indices covering life expectancy, education, and per capita income.l2Reflectingthe poor regional results on This contrastsstarkly with the perceptionsof poverty andwealthbelow, which show that inmany partsof (particularly rural) Mozambiquehavingmany children is seen as a signofwealth. IoInthe areas studiedfor the PVS, most householdsownedtheir houses. Those that didnot-mainly poorer female- headedhouseholds andthe elderly-ofiencomplainedof the insecurityand precariousnessof their living arrangements. I' Decreasingsharesof food inthe consumption bundle usually suggests increasingwealth. Becauseper capitaincome is includedinthe HDI,this indexis a mixedindicator rather than a purely nonmonetary indicatorof well-being. 6 health described above, Mozambique currently ranks lowest among its neighbors on human development, just as it did 25 years ago.13This suggests that there could be a lag between changes in income measures of well-being and changes in nonmonetary dimensions, such as educationandhealth. 1.15 Gender-related development in Mozambique is also lagging behind neighboring countries, though the gap between women and men appears to be narrowing. The gender-relateddevelopment index (GDI) is a composite index of the same three indices captured inthe HDI,adjusted to reflect inequalitiesbetweenwomen and men inthe three dimensions of human development.Partly reflectingits low HDI, Mozambiquealso has a lower GDI than neighboring countries. As noted in the recent Mozambique National Human Development Report 2005, however, the improvement in gender-related development during the last four years has been faster than that in human development, suggesting a narrowinggap betweenwomen and men. 14 1.16 Overall improvements in human development, while still low, have been impressive. Reflectingthe dramatic reduction in incomepoverty, human development in Mozambiquehas also improvedsignificantly duringthe last decade. Since 1990 (the first year with data for all comparison countries), Mozambique's improvement in human development has outperformed other selected neighboring countries. In fact, only Mozambique, Uganda, and Malawi (much less so) have seen improvements in overall humandevelopment since 1990. 1.17 Regional variations in human development are great, however. There are significant differences between income and nonincome poverty (figure 1.3). There are also similarities in results, such as the low human development, particularly in the north (Cabo Delgado and Nampula), and the evidence of poor health outcomes in the center, despite reduced income poverty rates (Zambezia). The results in the south are somewhat contrary to the findings on income poverty, suggesting that relatively highlife expectancy and educational attainment may, from a human development perspective, outweigh the increase in income poverty in those provinces. These regional differences in outcomes reflect the differences in access to services, especially for healthcare and water (see chapter 4). l3See appendix table A1.13. l4UNDP 2006b. 7 Figure 1.3. Human Development in Mozambique,by Province, 2000 Human Development Index 205- 236 236 - 257 ,257 ,329 - .329- 422 Source: Earthtrends, World ResourcesInstitute, available at http://earthtrends.wri.org.Data from UNDP 2000. VULNERABILITY, SHOCKS, AND COPING STRATEGIES 1.18 Vulnerability is having a high probability of falling into poverty or into deeper poverty. It usually depends on the characteristics o f the household-on its asset and resource base (natural, physical, financial, human, and social capital)-and its capacity to respond to risks and shocks. The vulnerable are usually assumed to include the poor (who have a highprobability o f remaining poor) and the near-poor (who are likely to be pushed into poverty when faced with a shock or unforeseen change). Risks, Shocks, and Coping Strategies 1.19 The Poverty and Vulnerability Survey uncovered a range of shocks and risks that communities and householdsfrequently face.15 Vulnerability in the eyes o f the households was perceived to relate to the capacities and strategies o f communities and households to deal with the risks o f unexpected shocks and stresses. The most common covariate risks (those affecting all in a community) were shocks caused by climatic IsVulnerability can be analyzed inseveralways. Qualitative interview techniquesandparticipatory methodscan capturepeople's perceptionsof vulnerability, reasons for it, and strategiesto cope with it. Such data are the main source o f informationfor this report's discussionof vulnerability. Panel data-able to provide greater information on economic mobility, the relative sizes of chronic and transitory poverty, andthe characteristics ofthose who move out of povertyandthose who fall into it-are available inMozambique only for rural areas and livelihoods. 8 changes or natural disasters (such as droughts or floods), a downturn in the local economy, a sudden worsening o f the internal terms o f trade, or a health crisis (such as cholera). The main idiosyncratic shocks (those affecting individual households) included hunger, illness or death o f a family member, unemployment, and loss o f crops. The seasonality o f vulnerability affects both communities and individuals. 1.20 The most common shocks,particularly for thepoorest households in urban and rural areas, were illness of a family member and drought. As many as 54 percent o f households in the poorest tercile in rural areas, and 37 percent inurban areas, reported the household experiencing a shock due to family illness duringthe year preceding the PVS.16 And 34 percent in the poorest rural tercile reported drought, as did 17 percent in the poorest urban tercile. 1.21 The most common coping strategy in both urban and rural areas, with little variation across income groups, was to seek helpfrom family and riends. This option was more available to households in urban areas than in rural areas.' Drawing down on household goods and assets was more common in rural areas than in urban areas, despite rural households generally having fewer assets. Households in rural areas were also much more likely to do nothing and merely try to get by. As many as 21 percent o f the poorest households in rural areas reported having done nothing to cope with the shock. 1.22 Risk-coping strategies are usually shaped by personal, human, and social capabilities,and by the incomes and assets that an individual or household can deploy. Overall, the poor have few safety nets. Instead, they pursue a range o f risk mitigating and coping strategies, which usually vary by social group (in particular gender) and geographic area (table 1.2). In rural areas coping usually includes daily casual labor- often referred to locally as ganho-ganho18-on someone's farm in exchange for food or money. While ganho-ganho i s also practiced in normal times, it takes on particular importance as a coping strategy in times o f shocks and stress, when few regular activities are available to the poor. Ganho-ganho also takes place in urban areas, usually loading and carrying goods inmarkets. 1.23 Ganho-ganho is important for many, but the opportunitiesfor such work are normally scarce, partly because only a few people are well enough off to employ workers. Additional coping options in urban areas include noninstitutionalized social insurance and other strategies, usually based on precarious economic activities linked to informal trade in goods and services. Incomes from such activities are frequently low and irregular-and tend to go entirely to meet family consumption and subsistence needs. Other options are to draw down on existing assets by selling or pawning them, to reduce food consumption, or to borrow or beg for money. l6See appendix tables A1.6 and A1.7. l7See appendix table A1.8. Casual work, or gunho-gunho, is called oiimeiu inZambezia andNampula, andxicoropo in Gaza. 9 Table 1.2. Coping Strategies in CommunitiesVisited, 2006 Coping Strategy Social Group Casual day labor (on farms-of others, also includes nonagriculturalwork) All, but predominantly men and male youth Collection o f wild fruits, roots, and vegetables (mainly for consumption) Women and children Collection, preparation, and sale o f firewood and charcoal Male youth Relying on remittances All Selling, consuming, and drawing down existing household assets All Borrowing from friends and family All (those likely to reciprocate) Begging All (those with few assets left) Early marriage Young girls Prostitution Women, female youth, young girls Social networks All (limited to those with ability to reciprocate) Source: Authors' calculations basedon data from PVS 2006. Note: Sample not representative. 1.24 Copingstrategies and livelihood strategies vary by gender. Women also take part inganho-ganho,though they find fewer opportunities to do so. Other strategies ofwomen include collecting wild fruits, roots, and vegetables for consumption. And those divorced, abandoned, or single could be forced to develop sexual relations with better off men or even to engage in outright prostitution. Younger girls might choose to marry (or be married off). 1.25 Social networks and relationships are crucial coping strategies for poor households. Proximity o f extended families and links with them, good relations with neighbors and friends, and access to social, economic, and political elites were seen as crucial for a household's ability to weather shocks and risks. Such networks can open household access to financial assistance (such as remittances), material support, assets, and livelihood opportunities-outside the household and the community. They thus affect the dynamics o f ~ulnerability.'~ But these social networks are based on the capacity to reciprocate, something that many o f the poorest lack, leaving them with few means to cope.20 Especially Vulnerable Groups 1.26 Most Mozambicans, being poor, are vulnerable, but some are especially vulnerable, including individuals who are physically incapacitated and thus unable to Tvedten andothers 2006, 79. Insome areas informal networksof farmers and fishers helped members cope in difficult times. InAngoche, Nampula, small groups of 5-7 fishermen combinedtools andinstrumentsto enable all to go fishing (UEM 2006,64). And common inmany areas was Nkhumy,discussedpreviously,which helped farmers gather labor to work the land, especially during harvest. InGaza informal savings groups (such asxitique) were common ways for Community membersto help smoothconsumptionand better respondto risks and stress. 2oMozambiquehastwo government agencies to support vulnerable households: INGC (Instituto Nucionul du Gestcio dus Culurnidudes,at http://www.ingc.gov.mz/), which is responsible for disaster relief, and SETSAN (Secretariado Thcnicode SegurunGuAlimentur e Nutriqcio,at http://www.setsan.org.mz/Index.htm), which is Mozambique's TechnicalSecretariat for Food Security andNutrition, an intersectoralbody that aims at supporting the implementation ofthe nationalFoodSecurity andNutrition Strategy (ESAN-Esfruthgiude SegurucuAlimentur e Nutricionul). 10 work and those whose access to productive resources tends to be limited or who may otherwise be socially marginalized.2' The social groups most frequently noted as vulnerable were the elderly, the mentally or physically disabled, the chronically ill (includingthose living with HIV/AIDS), orphaned or abandoned children, widows and to less extentwidowers, and single, divorced, or abandonedwomen with children. 1.27 The elderly were invariably considered among the most vulnerable. This was particularly so for those living alone or in households with few young people. When people become older, they have to rely on the goodwill and the productive labor of others to survive-remittances or in-kind help from children, assistance from other relatives, neighbors or friends-and on drawing down their few goods and assets in exchange for food and labor. Ifthey have no one to help them and few assets of their own, they often end up as the poorest of the poor. Some, especially those who are disabledor ill,may be evenmorevulnerable. 1.28 Many of the physically and mentally disabled were often highlighted as especially vulnerable. A higher proportion of the poorest households have at least one disabledmember. Nhate and Simler, using small group estimates of poverty and drawing on the IAF and the 1997 census, found that households with a disabled member are considered poorer than nondisabled households. Such households had higher rates of poverty across all three poverty indices and thus are more vulnerable to shocks.22They have less access to services, particularly education, and are more likely to have nonworkinghouseholdmembers, further increasingthe household'svulnerability. 1.29 Although the chronically ill were often considered to be especially vulnerable, HIV/AIDS rarely came up as a reasonfor vulnerability in itself. However,HIV/AIDS is rapidly increasingthe number of chronically illpeople and households in Mozambique, among the most affected countries in the world. HIV/AIDS often renders individuals physicallyunableto work their land, and adult illnessesand deaths from the disease leave households with fewer able-bodied earners. And if the disease is publicly known or suspected, a householdmay be stigmatizedand socially marginalized,further addingto its vulnerability. In additionto causingvulnerability on its own, HIV/AIDS of course has led to increasingnumbers of orphans and single-headed household^.^^ 1.30 Orphans and children in lower income households are particularly vulnerable and increasingly prevalent. Orphans and vulnerable children lack social recognition. They have few resources and receive little assistance. And they may be neglected or abandonedby family members or neighbors.Someare forcedonto the streets, particularly in urban areas. Estimates of the orphan population in Mozambique range from 12-16 2'Sometimes(as inthe case of advancedAIDS) categoriesoverlap, and bothphysical incapacityand social stigma render individuals and householdsvulnerable. The stigma of HIV/AIDS meansthat those infectedwere rarely highlighted as more vulnerable. But those left behindby deaths-widows andorphans--were notedas potentially vulnerable. 22See Nhate and Simler 2006, 10. The findings were statistically significant nationally at the 99 percent level. 23For further discussionon the impacts ofHIV/AIDS on poverty and on the feminization ofAIDS inMozambique, see chapter6. 11 percent o f the child population (between 1.2 and 1.6 million children were orphaned in 2006).24 The proportion o f orphaned children i s slightly higher in urban areas (13.8 percent) than in rural areas (11.1 percent). Orphans are also more likely to live in households where the head has no education, suggesting that poorer households (those less likely to have education) may shoulder more o f the burden o f caring for the growing number o f orphans. Orphans also live disproportionately in households headed by women. While female-headed households account for less than a third o f the population, more than half o f orphans live in households headed by women. 1.31 Orphans are substantially worse off than nonorphans, though their vulnerability varies significantly by gender and wealth. In access to education-a key characteristic o f current and future poverty and vulnerability-orphans lag far behind nonorphans (table 1.3).25 In urban areas only 75 percent o f poor orphans o f school age attend school, compared with 96 percent o f nonpoor orphans and 95 percent o f poor nonorphans. Inrural areas the number o f poor orphans not attending school i sjust as high, but because attendance by all children in rural areas is lower, the difference is less striking.Nationally representative data from 1997 and 2003 DHS surveys, however, show the difference in school attendance declining between orphans and nonorphans, with the gap closing more for male than for female orphans.26 Table 1.3. School Attendance by Orphan Status, Area, and Wealth Tercile, 2006 (percentage of children aged 6-1 7) Urban Rural Poorest Richest Poorest Richest Orphan status All Tercile Tercile All Tercile Tercile All nonorphans 96.5 95.2 98.7 91.6 87.5 94.5 All orphans 89.7 75.0 96.0 79.2 75.0 83.3 Male nonorphans 97.7 96.3 98.4 91.5 86.2 93.8 Male orphans 91.4 88.9 92.9 78.3 75.0 80.0 Female nonorphans 95.2 94.4 99.0 91.7 89.5 95.3 Female orphans 87.8 63.6 100.0 80.0 75.0 85.7 Source: Authors' calculations basedon data from PVS 2006. Note: Sample not representative. 1.32 The large discrepancy in school attendance betweenpoor orphans and others in urban areas is mainly because of the strikingly low attendance of poor girl orphans in urban areas. O f poor orphaned boys, 89 percent attend school in urban areas. Only 64 percent o f poor girl orphans do. Compare that with 94 percent o f poor girl nonorphans and 100 percent o f girl orphans in richer households. In rural areas there is no gender difference for orphans living in the poorest households, and boy orphans are worse off 24The nationaldata on orphans andHIV/AIDS prevalence inthis report are from UNICEF(2006,202-4), unless otherwise stated. Orphansare children (under age 18) who have lost at leastone parent. 25Using IAF datafrom 1997 and 2003 onthe educationof nonbiological children, includingorphans, Nhate (2005) showedthat nonbiological children are especially discriminatedagainst indecisions on householdeducation andthat access for nonbiological children deterioratedbetweenthe survey periods.Datafrom DHSbetween 1997 and2003, however, show adeclining difference inschool attendance betweenorphansandnonorphans.Nhate (2005) also found that nonbiologicalchildren are discriminatedagainst inaccessto intrahousehold resources. 26See appendix table Al. 11. 12 than girl orphans on average. But both girl and boy orphans are substantially worse off than girl and boy nonorphans. 1.33 Orphans are also worse off than nonorphans in the amount of domestic activities and chores. While all children are regularly asked to take part in domestic activities and chores, orphans are substantially more likely to spend time on chores than nonorphans, both in urban and rural areas. Inurban areas orphans are more likely to take part in 8 o f the 12 activities or choresS2'In rural areas orphans are slightly worse off and are expected to engage more often than nonorphans in 9 o f 12 chores. 1.34 The sex of the child alsoplays a significant role in the likelihood of engaging in any domestic activity. Girls (both orphan and nonorphan) in both urban and rural areas are more likely than boys to participate in 8 o f 12 domestic activities, whereas boys (both orphan and nonorphan) are only more likely than girls to take part in 2 o f the 12 activities inurban areas and 1of the 12 inrural areas. Insome activities the sex of the child seems to be a stronger determinant (a nonorphan girl is more likely to be involved in cooking, washing, or the fetching o f water than an orphan boy), although orphan status i s also clearly an important determinant. 1.35 Paradoxically, despite the amount of chores that girls and orphans takepart in, both groups on average are reported (by the head of household) to enjoy the most leisure time. According to the household respondents in urban areas, girls have as much leisure time as boys, and orphans are 20 percent more likely to engage in leisure than nonorphans. In rural areas nonorphans are seen to have slightly more leisure time than orphans. But, again, girls and boys are seen as beingas likely to engage in leisure. Here at least male orphans are seen as having less leisure time than nonorphans, but girl orphans are thought to have the most leisure time o f all. 1.36 Female-headed households are more vulnerable. In some cultures in Mozambique widows do not have the right to inherit assets, which can mean that household assets are confiscated by the husband's family after his death. Female-headed households were also considered more vulnerable because they often have fewer assets, fewer productive resources, and hence less o f an asset base to deal with shocks. They are also predominantly engaged in agriculture, and their households are less diversified (especially in rural areas). Widowers were also often considered vulnerable. Without the support o f their wives, widowers may find it difficult to take care o f themselves. Having few assets, they may also find it hard to remarry or find another partner, though this is significantly easier for men than women.** Some, instead, resort to alcohol, further deepening their poverty. *'See appendix table A1.12. 28Mather and others 2004. 13 DRIVERS HOUSEHOLD OF CONSUMPTION 1.37 The main drivers of household consumption are education, demographic factors, and the employment sector of the household head (figure 1.4).29Multivariate regressions isolate the separate effects o f independent variables on household consumption. The analysis here compares the determinants o f the log o f per adult equivalent consumption in urban and rural areas. Separate regressions were run for each area, because the structures for urban and rural areas were found to be quite different. 1.38 Even a little education increases consumption, with rising returns, perhaps by making households more efficient. Among household heads, education is strongly correlated with higher consumption per adult equivalent. Reflecting the relative scarcity o f secondary and postsecondary education, the returns to education rise with each additional education level completed. But the impact o f education in urban areas is higher than in rural areas for all levels o f education (and especially for post-lower-secondary education). Compared with 1997, however, the payoff to education fell in urban areas for primax levels, remained the same for secondary levels, and increased for postsecondary leve1s. 1.39 The household head's sector is strongly related to consumption. Households working in agriculture are among the lowest consuming. Some sectoral differentials are 20-30 percent. The premiums in urban areas'are on the whole lower than in rural areas, except for households working in mines, trade, and public administration. In urban areas households working in education do not consume significantly more than those in agriculture, which may suggest an emerging teacher pay issue. Over time (since 1997) the premiums for working in trade and services fell in both urban and rural areas, potentially reflecting the influx o f labor into these sectors observed inrecent years. Inurban areas the premium for working in manufacturing fell significantly, perhaps associated with the privatization and restructuring o f parastatal enterprises o f the late 1990s. 29The full results of the regressionsfor bothyears are in appendix table A2.8. 30It is possible that part ofthis change is due to the reclassificationof urban andrural areas betweenthe two surveys. 14 Figure 1.4. RegressionResults:Determinants of Consumption, 2003 Number of children 0-5 Number of children 6-9 Number of children 10-14 Number ofmen 15-59 Number of women 15-59 Any disabled adults Age head Headwidowed* Educationof Head** Some education CompletedEP1 Urban CompletedEP2 Rural CompletedES1 Over ES1 Sector of Head*** Mining Manufacturing Transport Trade Services Education Health Public administration -0.5 0 0.5 1 1. wrce: World Bank staff estimates. Note: Bars indicate the percent change inconsumptionassociatedwith aone-unit change inthe explanatory variable. Coefficientsinsignificant at the 95 percent level have beenset to 0. Categoriesnot displayed inthe graph becausethey were not significant are: number of adults over 60, any disabled children, head female, headmarried, headpolygamous, headdivorced, head female*married, head female*polygamous, head female*divorced, headfemale*widowed, and headconstruction. EPI: Primary grades 1-5. EP2:Primary grades 6-7. ESl: Secondary grades 8-10. *Reference category: single male. **Reference category: no education. ***Reference category: agriculture. 15 PERCEPTIONSOF POVERTY IN2006 1.40 By seeing poverty through the eyes of the poor-as noted in the poverty report prepared annually by Mozambican civil society-people cease to be seen as the objects of poverty and are instead redefined as subjects and agents of their own de~elopment.~` This shifts the emphasis o f the poverty problem-as well as the public policies aimed at addressing it-onto the opportunities, constraints, and risks that poor individuals, households, and communities face, rather than merely being about the incapacity o f poor people.32 Defining Poverty in 2006 1.41 People's understanding of poverty is based on the distinct realities of different people's lives and experiences of poverty. Some characteristics o f poverty were deemed important for defining poverty at the community level (such as lack o f services and basic infrastructure), which were considered as external and structural factors tending to affect everyone in the community. And others were more internal, which help define poverty at the household and individual level (lack o f productive assets, social status). Many o f these defining characteristics o f poverty are also regarded as both causes and consequences o f poverty, highlighting its dynamic, complex, and interconnected nature. Attempting to accurately aggregate and summarize the different voices o f the poor and the struggles that they face on a daily basis is a near impossible task. 1.42 I n an attempt to conserve many "realities of poverty" encountered by the Poverty and Vulnerability Survey, seven dimensions of poverty are presented. First, the lack of income, money, employment, or livelihood figured frequently in people's perceptions o f poverty. The lack o f secure and stable livelihoods also received attention, emphasizing the cyclical and seasonal nature o f poverty. Second, the lack o f productive assets, inputs, and complements for livelihoods-such as land, labor, livestock, seeds, fertilizers, tools, and improved technologies-were also frequently cited. For example, access to (good quality) land was raised as an indicator o f poverty in areas with high population pressure (Quelimane, Zambezia). But as discussed in later chapters, what seems to distinguish the poorest from the less poor in most parts o f Mozambique are differences in access to labor to farm the land. Third, poverty invariably was defined as the lack o f basic household and personal necessities, including food, clothing, and shelter or housing. Fourth, a lack o f basic infrastructure (access to safe water, sanitation, irrigation, transport, energy, roads, and markets) and services (especially health, education, but also agricultural extension, credit, and social insurance) were also seen as an important aspect o f poverty. Fifth, poor health and low education were also frequently mentioned. A female focus group in Zambezia, for example, described poverty as the 31G20 2005, 13. 32This reference to the "incapacity ofpoor people" echoes the G20 (2005, 13) critique of the definition of poverty set out inthe first version of Mozambique'snationalpoverty strategy, the PAWA. This document (GoM 2001, 10) definedpoverty as "the incapacityof individuals to ensure for themselves and for their dependents aseries ofbasic minimum conditionsfor their subsistence andwell-being, according to the norms of society." 16 inability to keep children healthy, nourished, and clean. L o w levels o f education, lack o f knowledge about agricultural techniques and markets, even ignorance-suggesting culpability on the part o f the poor-were also described as characteristics o f poverty. 1.43 While these dimensions o f poverty are similar to those highlighted earlier using quantitative ("objective") indicators o f poverty, the two remaining dimensions o f poverty as highlighted by the poor differ significantly, and particularly help expand our understanding o f poverty in Mozambique: the lack o f social capital and poor governance and institutions. 1.44 The lack of social capital, the sixth dimension of poverty, was highlighted as a key element of poverty in some notable cases, as were the social, psychological, and spiritual dimensions of poverty. In community-level poverty, the absence o f solidarity, trust, and civic and moral values was noted. Such feelings were also perceived to undermine the ability to form associations and other collaborative ventures, which inturn could undermine the ability o f subsistence farmers to negotiate higher agricultural prices with buyers and traders. At the household or individual level, well-being depended on links to extended families, good relations with neighbors and friends, and having access to social, economic, and political elites. Such networks help households get financial assistance (such as remittances), material support, assets, and livelihood opportunities and may thus have important implications for the dynamics o f poverty. Social networks are based, however, on the capacity to reciprocate, something the poorest often lack. 1.45 Social characteristics were also considered important. The gender, age, and marital status o f a household head, for example, may influence a family's access to land and labor. Indeed, in parts o f Mozambique women can gain access to such resources only once married, through their husbands. For these reasons, women often noted the absence o f a husband as a defining,feature o f poverty. Having no children was also highlighted by women as a sign o f poverty-both because children were considered a source o f labor and future income and because having children was seen as a blessing in its own right. Other social factors indicating a poor household were unstable families, affected or even torn apart by a husband's alcoholism or by domestic abuse and violence (principally toward women and children). Some o f the discussions also brought up the psychological elements o f poverty-such as stigma, loss o f dignity, feelings o f powerlessness, and even culpability or being "cursed" by witchcraft for some. 1.46 Poor governance and institutions were also seen as a dimension of poverty. This dimension, referred to as political poverty in civil society's annual poverty report,33 i s related to the maintenance o f peace, stability, and public security, to the quality and breadth o f state-society relations and interactions, and to the participation in and influence over public decisionmaking. Communal and family conflicts and public insecurity- thefts, robberies, physical aggression, threats, and even murders-were highlighted as identifiers, causes, and consequences o f poverty, particularly at the community level. Many also spoke about neglect and a sense o f abandonment by the state and other 33G20 2005, 14. 17 external actors, such as nongovernmental organizations. But in some areas these feelings o f abandonment coexisted in seeming contradiction with perceptions o f a hostile and interfering state-poor people's sense o f powerlessness vis-&vis public officials, instances o f harassment by state police, tax collectors, and other officials, and corruption and bribe-taking in the delivery o f public services. There was also a sense o f political marginalization and exclusion from public institutions, processes, and mechanisms, leaving poor people unable to voice their needs and opinions and influence the decisions that affect their struggle with poverty. Perceptionsof Poverty and Inequality in 2006 1.47 Poverty is a relative concept, linked to inequality in social and power relations. Inall the villages, it became clear indiscussions of the dynamics of poverty and relations between "poor" and "rich" that perceptions o f poverty and inequality-xpressed through intracommunity differences in poverty and well-being-are often linked to social Pebane, Zambezia: "When you have shoes, you don't have to kick stones around....We, differentiation and marginalization. As noted by an elderly man in the fishing village o f on the other hand, do not know if we are sons o f the state or even sons o f God; for us a better future never arrives. We here are part o f the `Zero' and not part o f the `One."'34 In explaining this distinction, he noted that those who were part o f the One were either from the economic and political powers inthe village or connected with them. The Zeros were the poor and socially excluded, with few connections to the elite and powerful. In this village o f Ones and Zeros there are the privileged and the "sons o f a lesser God, deprived o f everything and forgotten by everyone."35 1.48 Inequality thus seemed central to the notion of poverty. While poverty at the community level would often be explained as due to external factors and constraints, intracommunal poverty would instead be related to unequal social and power relations- and to personal factors, such as misfortune and bad luck, to more culpable factors, such as ignorance and laziness, or to witchcraft.36 Poverty was blamed on unequal access to public resources and support. And the poverty o f elderly, disabled, or demobilized soldiers was blamed on unequal access to government- rovided social insurance (through the Instituto Nacional de Assistencia Social, INAS). It was often unclear why some 8 community members received support while the majority did not, though some alleged that favoritism, contacts in the district administration, or pure luck explained the unequal treatment. In an example from the municipio o f Angoche in Nampula, inequalities in ~~~ 34 UEM2006,75. 35 Ibid. 36 Allegationsof witchcraft serve two important social functions.First, people needto explain the unexplainable and painful. Death, accidents, andother troublingincidentsare often blamedon witches, frequently identified as socially marginalizedelderly menandwomen. Second, witchcraft accusationsenforce social conformity, targetingthose who are perceivedto step outside social norms. Accusedofusingwitchcraft, people who are well off or successful may be socially sanctioned(Tvedten andothers 2006,87). 37 See, for example, the male focus groupdiscussion inthe village of Mbaua, Mandimbadistrict, inNiassa (LJEM2006, 26). 18 political power and influence were highlighted as at the core o f the crisis o f unemployment and poverty.38 1.49 Almost all the people and households interviewed for the Poverty and Vulnerability Survey identified both themselves and the communities they lived in as poor. Very few recognized truly well off households (those with mopeds or cars and secure, salaried, formal employment) in their communities. Urban households considered themselves slightly better off than rural households. Nearly 40 percent o f rural households considered themselves as at the bottom o f the scale, and fewer than 5 percent placed themselves at the top (table 1.4). In urban areas only 31 percent o f households felt they belonged to the bottom o f the scale, and more than 12 percent saw themselves at the top. 1.50 Male-headed households in both urban and rural areas, moreover, also considered themselves much better off, on average, than female-headed households. About a quarter o f male heads o f household in urban areas, and a third in rural areas, placed themselves at the bottom o f the scale, compared with about half o f female heads o f household (slightly less inurban areas and slightly more inrural areas). About 16 percent o f male-headed households in urban areas, and 12 percent in rural areas, placed themselves in the top bracket, compared with about 6 percent and 1 percent o f female- headedhouseholds inurban and rural areas, respectively. Table 1.4. Self-AssessmentofWelfare Relativeto Others in the Community,2006 (percent) Urban Rural Female- Female- Male-Headed Headed Male-Headed Headed Assessment All Household Household All Household Household 1-Poorer 31.3 24.6 44.0 39.9 33.2 57.1 2 28.8 31.8 23.2 22.5 22.6 22.1 3 19.1 17.8 21.6 16.3 17.1 14.3 4 8.6 10.2 5.6 12.3 15.1 5.2 5 8.3 10.6 4.0 4.7 6.5 0.0 6 3.O 3.8 1.6 2.2 2.5 1.3 7 0.3 0.4 0.0 0.4 0.5 0.0 8 0.3 0.4 0.0 0.7 1.o 0.0 9 0.3 0.4 0.0 1.1 0.o 1.5 0.0 10-Less poor 0.0 0.0 0.0 0.0 0.0 Source:Authors' calculations basedon data from PVS 2006. Note: Sample not representative. 1.51 The overall perceptions of poverty are also reflected in food security indicators. In the 12 months before the interviews, as many as 43 percent of households in urban areas and 53 percent o f rural households complained o f having suffered from hunger.39 Not surprisingly, more o f the poorest households in both areas have suffered from hunger than in the wealthier tercile. Perhaps more surprisingly, more male-headed than female- 38UEM2006,42. 39See appendix table A1.15. 19 headed households in both urban and rural areas went hungry at one point during the previous year. The numbers are comparable when looking at the number o f meals consumed during the previous day. Of urban households 63 percent, and o f rural households 69 percent, noted that they consume only two meals or less a day (table 1S).40 While fewer rural households than urban households consume only one meal a day, urban areas have a higher proportiono f households that consume three meals a day. While more male-headed households seem to have suffered from hunger in the preceding 12 months, female-headed households on average consume fewer meals per day than male-headed ones, particularly in urban areas. Indeed, the overall difference in consumption o f meals per day between urban and rural areas is mainly explained by the difference between , male- and female-headed households. 1.52 Perceptions of trends in poverty are more ambiguous, however, and largely depend on the dimension of poverty being considered. Both qualitative and quantitative data from the PVS suggest that over the preceding five years some dimensions o f poverty have improved-particularly those relating to community-level poverty. Perceptions data from focus groups and semistructured individual interviews suggest few improvements in the delivery of agricultural extension and credit services, but more inthe access to public or community water sources and inthe delivery o f basic social services.41 Table 1.5. Number of MealsConsumedduringPreviousDay, 2006 (percent) Urban Rural Number Male- Female- Male- Female- of Headed Headed Headed Headed meals All Household Household All Household Household One 12.2 8.1 20.0 9.8 8.0 14.3 Two 51.0 54.2 44.8 58.7 57.3 62.3 Three 36.8 37.7 35.2 31.5 34.7 23.4 Total 100 100 100 100 100 100 Source: Authors' calculations basedon data from PVS 2006. Note: Sample not representative. 1.53 Even so, perceptions of trends in overall poverty suggest a worsening in the living standards and well-being of the poor. Findings from the PVS suggest that the improvements inwelfare at the community level have not been translated into perceptions o f profound and direct impacts on the poverty o f households. Indeed, despite the reduction in income poverty between 1997 and 2003, about 40 percent o f households in both urban and rural areas felt that their overall poverty situation either remained the same or worsened inthe last five years (tables 1.6 and 1.7). 40The data intable 1.5 shouldbe consideredwith caution, however, becausethe survey was conductedbetween May andJune, normally aperiodof relative food security. 41See appendix tables A4.16 andA4.30. 20 Table 1.6. Perceptionsof Change in HouseholdPoverty in RuralAreas over Last Five Years, 2006 (percent) Perception All Households Male-Headed Household Female-Headed Household Poorest Richest Poorest Richest Poorest Richest All Tercile Tercile All Tercile Tercile All Tercile Tercile Improved 26.8 11.0 45.6 32.2 16.7 48.1 13.0 4.7 30.8 Worse 40.6 61.5 24.4 37.2 54.2 54.2 49.4 69.8 30.8 Not changed 32.6 27.5 30.0 30.7 29.2 28.6 37.7 25.6 38.5 Total 100 100 100 100 100 100 100 100 100 Source:Authors' calculationsbasedon data from PVS 2006. Note: Sample not representative. Table 1.7. Perceptionsof Changein HouseholdPoverty in Urban Areas over Last Five Years, 2006 (percent) Perception All Householh Male-Headed Household Female-Headed Household Poorest Richest Poorest Richest Poorest Richest All Tercile Tercile All Tercile Tercile All Tercile Tercile Improved 25.2 11.7 42.7 28.4 17.7 43.4 19.2 4.1 41.2 Worse 37.7 51.4 23.9 37.3 45.2 25.3 38.4 59.2 20.6 Not changed 37.1 36.9 33.3 34.3 37.1 31.3 42.4 36.7 38.2 Total 100 100 100 100 100 100 100 100 100 Source:Authors' calculationsbasedon data from PVS 2006. Note: Sample not representative. 1.54 Household opinions differ significantly by gender. In rural areas as many as 87 percent o f female heads o f household felt that their poverty situation has either worsened or stayed the same, compared with 69 percent o f male-headed households. Moreover, nearly a third o f male heads o f rural households felt better off than five years before, nearly three times the proportion for female heads o f rural households. 1.55 Those who viewed themselves as poor in the subjectiveperceptions of change in well-being were indeed the poorest in the community. Households in the upper terciles viewed changes over the preceding five years significantly more positively than did the households in the poorest tercile. And the poorest households were much more likely to view the changes as having been for the worse than were the richest households. Overall, however, these perceptions contrast starkly with the findings on monetary poverty.42 42 Potentialexplanations for this apparentparadox are many.One is the nonrepresentative sample ofthe PVS. A second i s that perceptions of change in overall poverty stated here reflect poverty's many dimensions, not only changes in income poverty. A third is that it is not uncommonfor subjective measurementsofpovertyto differ substantially from suchobjective indicatorsas consumption. Subjective measurements are muchmore likely to fluctuate seasonally and be influencedby nonpoverty factors, such as public opinionsabout agovernmentpolicy. The discrepancy may also shed light on the potentialproblemsof using flexible consumption bundles to estimate poverty.The earlier analysis of monetarypoverty employedaflexible food bundle to estimate poverty to allow for substitutionamong goods due to changingrelative prices. Mainly looking at the costs of fulfilling aparticularenergy intake, this approachdoes not look at preferencesdue to culture, social norms, or habits. So, even ifpeople today do not face lower calorie intakes (inmost areas intakes are higher), they may still perceive themselvesas poor. Indeed,the qualitative assessment suggests that the inability to purchase or enjoy particular types of food (due to rising costs of livingandworsening internalterms of 21 1.56 Agricultural income data from rural areas document increasing inequality in recent years, which might provide some insight into this apparent paradox As discussed in greater detail in chapter 3, inequality appears to have risen in rural areas between 2002 and 2005. So, there i s a possibility-especially in rural areas-that growth after 2003 has not been trickling down to the poor to the same extent as in the earlier period. Increasing income divergence, coupled with frustrations from a potentially slowing pace o f poverty reduction, might underlie the recent perceptions data. Findings from the qualitative assessment moreover shed further light on this. Households frequently complained o f rising costs o f living and worsening internal terms o f trade43- particularly for subsistence farmers-which affect the poor more severely than others. Additional reasons noted were diminishing employment opportunities caused by a collapse in public-sector and parastatal jobs, a changing climate and increasing natural disasters, and greater vulnerability for households and individuals brought about by deteriorating health conditions, includingHIV/AIDS. 1.57 Even so, households remain hopeful about their future well-being. A higher percentage o f households in urban areas expected poverty to decrease or remain the same inthe near to medium term.44While the optimism of the urban areas was not necessarily shared inrural areas, a much lower percentage o f households inboth rural and urban areas was pessimistic about the future than was pessimistic about the past. This hope for the future may reflect an expectation that improvements in community well-being will eventually trickle down to the household. It may also suggest a lag between the growth in household consumption recorded earlier and household perceptions o f improved living standards. trade, for example) may itselfbe consideredatype of poverty.Finally, the PVS took place in2006, so it is possible that the poverty reductionuntil2003 may have slowed down. 43 Internal terms oftradei s definedhere as the agriculturalprice index divided by the price index for other goods and services. 44 See appendix table A1.19. 22 2. THE MECHANISMS OF SHARED GROWTH: STRUCTURAL CHANGE AND HOUSEHOLD LIVELIHOODS 2.1 High and sustainedpostwargrowth was the most importantfactor in Mozambique'spoverty reductionperformance between 1997and 2003. The ignitionfor growthcame from newprivateinvestmentinphysicalcapital and highlevelsofpublic spending. Privateinvestments increasedbecausethe government'sstabilizationpolicies reducedrisksand increasedthe rate ofreturn.These policieshelddown inflationwhile reducingthe cost of doingbusiness and loweringrestrictionson competition, such as price controlsand inefficientparastatalmonopolies.Growthwas also supported by donor aid, which financedpublic investmentininfrastructurerehabilitationandthe development o f a more efficientand effectivepublic sector. Donoraidallowedbothinvestment- neededfor growth-and consumptionper capitato increase, thus supportinghigher welfare. 2.2 But high growth only translatesintopoverty reduction ifit is broadly distributed-in other words, ifthe character of thegrowth ispropoor. In 1997, Mozambiquehada modestlyhighlevelof income inequality,and it stayedconstant through2003 (the lastmeasurement), indicatingthat the fruits of growthwere widely distributed.How didthis happen?Growthwas propoorbecausechanges inthe structure o f productionwere labor intensive-pulling labor out o f agricultureinto highervalue sectors, inbothruralandurbanareas.This raisedthe productivityof labor and incomes in agriculture, lifting incomes for the bottomhalfofthe distributionand contributing75 percentto the fall inthe povertyheadcount.The laborforce shiftedintothe service sector to meet the demand for nonfarmgoods and services that came from increasedruraland urbanincomes. Labor also shiftedintothe constructionsector, whichwas stimulatedby highrates of investment.These shiftscompensatedfor the loss ofjobs inthe manufacturingsector as overallwage employmentwent up between 1997and 2003. Finally,increased incomesplusbetter access allowedruralhouseholdsto sendtheir childrento school ratherthanwork inthe fields, reducingthe growthofthe rural labor force. 2.3 The backboneofpoverty reduction was the ability of family farmers and family-owned businesses-where more than 90percent of the laborforce in Mozambique works-to raise their incomes. Inthe ruralsector, public spending on rehabilitatingthe infrastructureneededto gainaccess to markets supported growth in ruralincomes. Usingcashgained from sellingagriculturalproduceor working for cash either duringharvestor inthe off-season, households investedinnew microbusinessor self-employmentventures. Inurbanareas, households also diversifiedas one member of the family-usually the male-was able to find a wagejob or earn income in nonagriculturalself-employmentor a small business.On average, nonfarmjobs generated higher earningsthan growingcrops. Butthere were very successful farminghouseholds as well that succeededby movinginto high-valuecrops. Overall,householdswith diversifiedsources of income tended to be less poor in2003 than those that didnot diversify. 23 2.4 Adults with more education were much more likely to be in high-earning occupations. Few people without at least some primary education were able to get a nonagricultural wagejob. Unfortunately, in2003, nearly 86 percent o f women inthe labor force did not finish EP1 (grades 1-5), as did 68 percent o f men. Lack o f education i s one reason why women were more likely to stay in agriculture; another i s the highburdeno f household chores women face, especially inrural areas, which constrains the number o f hours women can spend earning income. Not surprisingly,female headed households were less likely to be able to diversify their income, and thus some may be in a poverty trap. 2.5 Since 2003 the blisteringpace of economic growth has continued. Mozambique's growth and poverty reduction episode i s already one o f the longest in low income countries. Research shows that growth cycles usually run out after about seven years in low income, non-oil producing countries. When growth starts after a crisis, it i s based (at least partly) on rehabilitation. Once this source o f growth i s exhausted, growth sputters out, or inequality increases, haltingthe poverty reduction impact, or both. Through good policies that providedpolitical and economic stability, Mozambique has so far beaten the odds and sustained economic growth. 2.6 But it is unclear whether Mozambique is continuing to beat the odds on shared growth. The latest comprehensive data on living standards are from 2003, when about halfthe population was classified as poor, and the evidence since then i s sketchy. The agricultural sector overall has continued to grow, and this should be lifting more households out o f poverty. Government data show continuing increases in access to services, and rural income survey data show that average household incomes have increased. Based on extrapolations o f a few indicators, the Ministry o f Planning and Finance predicts a continuing fall inpoverty. But other indicators point to a slowdown in that fall. Rural income inequality seems to be growing, and already highurban inequality persists, so fast growth may now have less o f a poverty-reducing effect. Government strategies can increase the odds o f shared growth by focusing on the household firms and farms that are the livelihood the poor. THEENVIRONMENT POVERTYREDUCTION, 1997-2003 FOR 2.7 Mozambique has a typically low income African economy-rural, with 80 percent of the laborforce employed in agriculture, forestry, orfisheries. A vast country, with low population density, regional economic differences are important. In the south economic relations are dominated by the proximity to the South African economy, with Maputo less than four hours drive from the border. Road, rail, and sea travel links dating to colonial times facilitate the movements o f goods and people to and from South Africa and from there to the world. The south, more urban, has more than 60 percent o f the jobs in registered firms. There is a long tradition of Mozambican men from the south migratingto South Africa to work. 2.8 The central and northern provinces are the grain baskets, andproduce most of the traditional export crops (cotton, cashew, tobacco, sugar). The transportation network runs east to west, with trading relationships dominated by the proximity to Malawi and Zimbabwe. Mozambique's grain production can be the safety net for Malawi in years o f 24 poor production; the very poor quality o f the north-south transportation network makes shipping agricultural production from the north to internal markets in the south prohibitive. So, when droughts occur in the south, grain i s likely to be imported. The center, split by the Zambezi River, i s fertile (though subject to flooding). The largest city, Beira, on the southern side o f the river, i s the second main source o f manufacturing employment. 2.9 Mozambiquefaced serious macroeconomic diffulties in the mid-l990s, with annual inflation at 48 percent and the balance o f payments deficit at 17 percent o f GDP. Restoring macroeconomic balances was necessary to create a stable environment for investment. Emergency food aid-which had helped to finance the deficit-was being reduced, so imports had to be reduced as well. The macrostabilization policies (including gradual exchange rate depreciation) restored fiscal and monetary balances. Policy targets for this period included those for investment and poverty reduction. Investment rose to 14 percent a year over 1997-2003 financed by increased government and private savings (table 2.1). Table 2.1. GDP by Expenditure Category, Share, and Growth Rate, 1997-2003 (percent) Average Annual GDP Share of GDP Growth Rate 1997 2003 1997-2003 100.0 100.0 8.9 Consumption 97.0 89.4 7.6 Private 89.9 79.0 6.9 Public 7.1 10.3 14.9 Investment 21.8 27.4 12.5 Private 10.6 15.7 15.1 Public 11.2 11.7 9.6 Exports-imports -18.8 -16.8 7.2 Source: World Bank staff estimates. 2.10 The redirection of resources to investment reduced the growth of total consumption to 4 percent a year (2 percentper capita), a potential threat to thepoverty reduction target. In many countries such a slowdown in the growth o f private consumption would have increased poverty. But thanks to the character o f the growth process, aggregate consumption growth reduced the poverty headcount and improved nonmonetary welfare. The growth o f government consumption increased the quantity and quality o f public services, especially access to education, while that o f private consumption directly improved monetary well-beingas it reached most o f the population. 2.1 1 The main driver of growth was increasedpublic andprivate investment. At the most aggregate level, economic growth i s determined by the rate o f accumulation o f factors o f production (human and physical capital, labor) and the rate o f growth o f the productivity o f these factors. The main aggregate driver o f growth during this period was the growth o f physical capital, an unsurprising result given the growth rate of total investment(public and private) duringthe period. This investment, and efficiencies inthe use o f capital through rehabilitating critical infrastructure, fueled the high rate o f growth. The growth o f the labor force and human capital played smaller roles. The reasons were 25 that labor force growth was unusually low during this period. And the big government investments to increase enrollments had not yet reached the labor force in 2002, so increases in the labor force's human capital were also low. The lesson: if Mozambique had scaled up increases in human capital more, growth might have been even faster. 2.12 Mozambique's virtuous circle of investment,growth, and poverty reduction in 1997-2003 is even more impressive given the negative international and national price trends. Mozambique's international terms o f trade deteriorated mainly because o f falling export prices. Its import price index in2003 was about the same as in 1997, falling by 10 percent inthe intervening years and then climbing back up again. But its export prices fell by 15 percent, resulting in a 15 percent drop in national purchasing power. This price change was reflected, and magnified, in the internal terms o f trade (agricultural price index divided by the price index for other goods and services). Prices for agricultural goods fell 25 percent relative to prices for other goods, pushing down the purchasing power o f agricultural income^.^' 2.13 Growth reduced poverty thanks to the equitable distribution of its benefits among households. Consumption growth was about the same for all income groups over 1997-2003 (table 2.2). It was propoor in absolute terms-the consumption per person o f the poorest households increased in real terms throughout the period. In addition, urban consumption increased more slowly than rural, especially for the urban middle class, slightly reducing the urban-rural gap in welfare. This slower growth in urban household consumption i s related to the macrostabilization-specially the depreciation, which increased the cost o f imported goods and urban services such as transportation (higher fuel costs). Because these items are important in the consumption basket of those in the lower three quintiles in urban areas (imported rice i s a staple food in urban areas), real consumption growth was slower in those areas. So, the reduction o f poverty was also slower in urban areas. 2.14 National inequality, modest at the beginning of the period, stayed modest through 2003. Measured either by the Gini coefficient or the Theil index, national, urban, and rural inequality all were virtually unchanged between the two household surveys for 1997 and 2003.46 Inequality between rural and urban areas is also low-less than 3 percent o f total inequality is caused by the rural-urban income gap. While Mozambique's Gini-more than 0.40-is worrisome, it i s on a par with other low income countries. Of 81 countries in a World Bank database, 49 have a national inequality higher than Mozambique according to the most recent estimate. 47 Most o f the countries with lower inequality are inAsia, only a few inAfrica. 2.15 Inequality within urban areas increased, and this reduced the effect of growth on poverty reduction in urban areas. Ifthe highest quintile had not gotten a much larger share in the growth in consumption per adult equivalent inthe urban areas, the fall in the urban poverty headcount would have surpassed the rural one (table 2.3). For most population subgroups, the contribution to poverty reduction o f the change in the 45See appendix table A2.3. 46The confidence intervals for the nationalGini index ineachyear are close. See appendix table A2.5. 47See appendix table A2.6 and http://iresearch.worldbank,org/PovcalNet/jsp/index.jsp. 26 distribution of consumption within the group was less than 10 percent of the change in poverty so, the increase in consumption translated directly into poverty reduction. This was particularly true for rural households and for households whose primary earner worked in the agriculturalsector. Both these groups had very high consumptiongrowth. And both contain the majority of households in poverty and the majority of poor households-there is a highoverlapbetweenthese two groups. So, the equitable character of income growthfor these two groups made a huge difference inpovertyreduction. Table 2.2. GDP, Labor Force, Productivity, and Poverty by Sector, 1997 and 2003 (percent) Share of GDP Annual Average Growth Rate 1997 2003 1997-2003 GDP 100.0 100.0 9.0 Agriculture 34.4 27.3 5.4 Industry 16.0 26.9 17.4 Services (private) 40.9 37.2 7.5 Services (public) 8.7 8.6 8.8 Share of Labor Force 1997 2003 1997-2003 Laborforce 100.0 100.0 0.8 Agriculture 89.7 81.7 -0.8 Industry 3.4 3.1 -0.6 Services (private) 5.0 12.3 17.1 Services (public) 1.8 2.9 8.7 Average Productivig 1997 2003 1997-2003 All 4.0 6.5 8.4 Agriculture 1.5 2.1 5.6 Industry 18.2 54.1 19.9 Services(private) 31.6 20.2 -7.2 Services (public) 19.3 19.8 0.4 Poverty Headcount Annual Change in Poverty Rate 1997 2003 1997-2003 All 69.2 54.1 -4.0 Agriculture 72.6 58.2 -3.6 Industry 65.4 54.0 -3.1 Services (private) 54.6 44.4 -3.4 Services (public) 56.0 32.9 -8.5 Source: World Bank staffestimatesandIAF data for 1997 and 2003. Note: Service (private) includestrade, transport, and services. Service (public) includes health, education, andpublic administration. a. Average productivity equals sector GDP divided by sector employment. 27 Table 2.3. Decomposition of Change in Poverty by Location and Sector, 1997 and 2003 (percent) Decomposition Level and Population Share Change 1997 2003 Mozambique Poverty in 1997 69.4 Poverty in2003 54.1 Total change inpoverty 1997-2003 -15.3 Regional By Region Change inpoverty inthe north -3.7 North 32.3 32.3 Change inpoverty inthe central -11.7 Central 41.9 41.9 Change inpoverty inthe south 0.12 South 25.7 25.7 Total intraregionalcomponent -15.3 Populationshift (regionalmigration) 0.0 Interactioncomponent (residual) -0.0 Urban-rural By Area of Residence Change inurbanpoverty -3.6 Urban 29.0 32.0 Change inruralpoverty -11.6 Rural 71.0 68.0 Total intrasectoral component 15.2 Populationshift (urban-ruralmigration) -0.22 Interactioncomponent (residual) 0.11 Aggregatesectors (by head of household) By Sector of Employment of theHead Change inagriculturepoverty -11.1 Headinagriculture 78.7 71.3 Change inindustrypoverty -0.8 Headin industry 7.6 5.3 Change inservice (private) poverty -0.9 Headinservices (private) 9.4 17.5 Change in service (public) poverty -0.9 Headinservices (public) 4.1 5.8 Total intrasectoral component -13.9 PoDulationshift (sector shift) -1.5 Interactioncomponent (residual) 0.10 Source: Authors' calculations basedon IAF data from 1997 and 2003. Note: Consistentwith 2003 urbanand rural definitions. Individuals are assignedto the sector where the household head is employed. Ifthe head is not employed, they are assigned to the sector of employment of the oldest adult. If nobody works (less than 5 percent of all cases), they are assigned to agriculture. Service (private) includes trade, transport, and services. Service (public) includes health, education, and public administration. North includes Niassa, Cab0 Delgado, and Nampula; center includes Sofala, Tete, Manica, and Zambezia; and south includes Gaza,Inhambane, Maputo province, andMaputo city. 2.16 Migration, while important in the south, is insignificant nationally. Sectoral results were driven by consumption gains for households engaging in agriculture, explaining 11 percentage points o f the 15 percentage point national reduction in poverty since 1997. The contribution o f intrasectoral shifts (mainly from agriculture into private services), while reducing poverty, was not large. Despite these improvements, poverty continues to be mainly a rural and agricultural phenomenon, given the concentration o f the majority ofthe populationinthese areas. STRUCTURALCHANGES INTHE MOZAMBICAN ECONOMY 2.17 The economic shifi from postwar recovery to sustained growth brought major changes in the structure of output and employment-and in labor supply and demand. These structural changes were as important as the overall macroeconomic aggregates in helping explain why growth was sustained and equitably distributed. Shifts in output, shifts in employment, and shifts inproductivity combined to produce changes in poverty status by sector o f activity. 28 2.18 The decline in agriculture as a share of GDP and the growth of other sectors was unambiguouslypropoor (figure 2.1). Poverty is a characteristic o f the household, so in order to relate changes in the structure of output and employment to poverty, households are classified by the sector where the head o f the household works. Because the poverty headcount was higher in agriculture than in the other sectors in 1997, the growth pattern that opened opportunities for household heads to move out of agriculture and into other sectors was a key factor inreducing poverty. 2.19 Households working in agriculture increased their income. The reason incomes rose for households whose main earner worked in agriculture was that production increased even though the sector shed labor. The total labor force in agriculture fell, so that even with slower growth in output than other sectors, production per worker increased. This increase in labor productivity fueled the income growth for households whose main source o f income comes from agriculture (and for households who have only part o f their income from this sector). It was even strong enough to overcome the 25 percent fall in prices for agricultural products relative to other sectors.48 As a result, consumption growth for rural households whose primary income came from agriculture was as high as for those in other sectors, and income growth in these households accounted for 11.6 percentage points o f the total decline in poverty o f 15.3 percentage points. 2.20 Service growth was also propoor, despitefalling productivity. Primarily private sector services-such as retail trade, transport, food preparation, finance, and telecommunication-saw their output grow slightly slower than GDP. But employment in services more than doubled. Earnings growth was strong, despite falling productivity because o f the high income elasticy o f demand for services-as incomes increase, people want to consume more services.49Even after the increased employment, average earnings were higher in services than in agriculture, so this shift in the structure o f employment was a positive force for poverty reduction. Primarily public services also increased their share o f the labor force, but they remain small. But incomes grew rapidly in the public sector, so that even with a very small percent o f the labor force, the growth o f income for households inthis sector reduced poverty by nearly 23.1 percentage points (an average o f 8.5 percent a year), bringing the poverty rate for households headed by a public sector employee down to only 33 percent. 48While the fall in domestic terms oftrade reducedthe purchasingpower of incomesfrom agriculture, it increasedthe purchasing power ofconsumersof agricultural products. Because mostMozambicanhouseholdsare net food buyers, the effect ofthe trend indomestic terms oftrade is ambiguous on average. It would depend among other things on price trends in cash crops. For the poor inrural areas, where roughly 50 percent oftotal consumptionis home-produced, and many have income from other sectors throughnonfarmbusinesses,the effect could be small. See chapter 3. 49Fallingproductivity, often viewed negatively, is a mathematical certainty for a sector that experiencesrapid growth inthe labor force without asimilar growth inphysicalcapitalor technologicalchange. Inthe nonagriculturalsectors in acountry with incomesas low as Mozambique, it is normal andpositive from apoverty reductionstandpoint. 29 Figure 2.1. Agriculture's Share of Labor Force Remainsabove the Share of GDP, but not Industry or Private Services Agriculture: the large gap between production vs. employmentshare remains 41 1--e- -Share of1 Labor 20 Force 10 1997 2003 Industry: increasedshare of GDP and constant laborforce share 100.0 100 90.0 90 80.0 70.0 60.0 50.0 40.0 40 Labor 30.0 30 Force 20.0 20 10.0 10 0.0 0 I. 1997 2003 Private Services: decreasing share of GDP I and increased share of the laborforce 100.0 100 I 90.0 90 80.0 80 70.0 60.0 50.0 of GDP 40.0 30.0 20.0 20 of Lab0 10.0 0.0 0 10 Force I 1997 2003 I Source: World Bank staff estimates and IAF data for 1997 and 2003. 2.21 The industrial sector showed strong output growth, but changes in the structure of output destroyedjobs. The closing or downsizing o fparastatal manufacturing enterprises led to significant layoffs (figure 2.2). New private investment in manufacturing was substantial, dominated by three megaprojects-the Mozal aluminum smelter and two energy sector projects. While these projects contributed roughly 1.6 30 percent to GDP annually and boosted industrial production, they created little domestic employment.s0As a result, average productivity in the industrial sector grew overall. But there were some bright spots. Construction-a labor-intensive sector-boomed, doubling its share o f total employment to 2.0 percent. And businesses supplying the local market started up as well. The higher average productivity translated into higher incomes for those employed in industry, increasing the demand for services. So, overall, industrial growth supported poverty reduction as well. Figure 2.2. Average Annual Growth Rates of Output and Employment, by Sector, 1997 and 2003 25.0% - Senices 20.0% -E8gn Trade + 15.0% . Education 10.0% Public administration 5.0% -- i Health Mining -bl 0.0% - J 0.0% 5.0% 10.0% 15.0% 20.0% 25.0% 5 -5.0% J -10.0% Q, a Manufacturing -15.0% -20.0% Source: World Bank staff estimates and IAF data for 1997 and 2003. 2.22 Behind the rising average labor productivity, and thus labor incomes, was the withdrawal of subsections of theyouth population from the laborforce, primarily to be in school. Those aged 10-18 entered the labor force at a much slower rate than older cohorts and perhaps also left the labor force. It appears that most o f these workers or potential workers stayed out o f the labor force to go to school. The share o f those aged 10 and above who reported being in school as their primary activity increased from 17 percent to 22 percent, many o f them girls no longer required to stay home and help with domestic tasks." The ability o f households to support these dependents while in school is 50Inadditionto creating little employment for Mozambicannationals,these projects contributelittle to domestic absorption (domestic income), as muchof their value added is exported. As aresult, it is important to track the expenditure side of the economy as well-not only productionbut domestic income hasto rise, otherwise increases in supply of products producedprimarily for the domestic marketsuch as food crops and small scale services will drive down prices. InMozambique, despite low employment creation these large enterprises are an important part of the poverty reductionstory inthe past andthe future, as they are partially responsible for Mozambique's increasedexports and stable balance of payments, andthroughthe payment o ftaxes, they are beginningto support increasedgovernment, consumption andtherefore increasedservice delivery. 5'See appendix table A2.10. 31 another indication o f the strength o f greater household welfare. This schooling produced major improvements in human capital and should contribute to stronger economic growth inthe future. (It also increased employment inthe education sector.) 2.23 Mirroring the changes in the economy and the increased years of school in rural areas, the labor force in rural areas grew very slowly, while the urban labor force grew rapidly. The good performance o f the rural economy meant that this was not a source o f widening inequality. The movement o f households from rural to urban areas was low, and this migration played virtually no part in poverty reduction (see table 2.3). The entire increase inthe urban labor force relative to rural was caused by the decision to keep children inschool inrural areas. Box 2.1. Who is in the Labor Force in Mozambique? Mozambique household surveys classify household members over age six as "active"-in the labor force-or "inactive"-out of the labor force for a reason, including being a full-time student, ill or disabled, or "domestico(a), " meaning engaged in domestic household activity. While the number of people claiming to be out of the labor force as a domestico(u) was only 2.5 percent in 2002 (down from 7.6 in 1997), it is questionable whether these people should have beenexcluded from the labor force. People should be classified inthe labor force if they undertake a productive activity for at least one hour over a seven day period. I s this the case with domesticos(as)? In the PVS survey we included an additional set of questions on whether the householdmember was active inhousehold chores over the past seven days, including caring for household members, washing clothes, cleaning, growing food for the household, maintaining the home, collecting firewood, fetching water, and so on. In the survey almost all those who classified themselves as "inactive" domesticos reporteddoing at least one of these activities. Are these activities "economic," and should they count as in the labor force? Growing and preparing food is, and consumption of the production from these activities is included in the consumption aggregate. So, anyone who undertakes these activities should be considered employed. In addition, some household chores, such as home maintenance, collecting firewood, and fetching water, have been defined by the UN as productive ones, to be included in the national accounts. So, anyone doing these activities is also inthe labor force. We concludedthat most, ifnot all domesticos, should have beenclassified as inthe labor force. To avoid underestimatingthe labor force (as the official statistics certainly do), we included all those listed as domestico(a) as inthe labor force, inthe agriculture sector. 2.24 In sum: changes in the structure of output created opportunities for labor mobility, which brought households new income-earning opportunities. Most sectors with strong growth in output had a strong growth in employment-such as construction, trade, and education (see figure 2.2). The main outlier is manufacturing, where the decline in labor intensity was not for the most part propoor. The industrial sector overall had the slowest rate o f poverty reduction. In agriculture, where the majority o f the population works, the decline in employment was propoor. By the end o f the period, agriculture was only about a quarter o f Mozambique's GDP, even though 80 percent o f the labor force still works in this sector, highlightingthe need for even further changes in the structure o f employment to continue propoor growth. 32 2.25 Putting the pieces together, several factors, both aggregate and sectoral, explain the highly positive impact of growth onpoverty reduction. Strong output growth, fueled by a large increase in physical investment, translated into steady increases in household consumption. This increase was distributed equally among households for two reasons. Initial inequality was moderate. And the nature o f growth-specifically the high growth in labor productivity in agriculture, which translated into increased consumption for these households, plus the movement o f labor from agriculture into higher productivity sectors-kept equality roughly constant. Although the productivity increase was highest in manufacturing, job shedding was considerable, so the increase in productivity in this sector was not initially propoor. But in agriculture the job shedding and consequent increase in labor productivity was voluntary and strongly propoor. The complementary decline in average labor productivity in private services was also propoor because the average and marginal productivities in these sectors are still higher than agriculture. The slow growth o f the labor force, combined with investment-led output growth, increased overall labor productivity. This shift was probably a one-off, post- conflict event. In the future labor force growth is expected to return to over 2 percent a year, highlighting the need for even more investment and improvements in labor productivity to sustainpropoor growth. BEHINDTHE GROWTH STORY: CHANGES INHOUSEHOLD LIVELIHOODS 2.26 Aggregategrowth and structural change affect households through changes in livelihoods. As conventionally defined, livelihoods comprise "people, their capabilities, and their means o f living, including, food, income, and assets," where assets refer to both tangible assets (productive resources, household goods) and intangible assets (rights, claims, access to such resource^).^^ The livelihood strategies that individuals and households adopt often reflect their opportunities and assets (natural, physical, financial, human, and social), and remuneration (monetary or otherwise) from these activities determines the economic changes in their household welfare. Poorer households, with a lower asset base, tend to have fewer livelihood options. Changes in livelihood strategies represent the response o f households to the macro events analyzed earlier; and livelihood changes at the household level feedback into sectoral and aggregate economic performance. The causality i s not one way, but the results are changes in household income, wealth, and poverty. 52Chambers and Conway 1991. 33 Box 2.2. AnalyzingHouseholdLivelihoodDynamics The idea of livelihoods as household-level analysis-built from an agglomeration of individual events and decisions but taking into account the essentially communal nature of welfare-is attractive for poverty analysis becausepoverty is a household concept. Analyzing how livelihoods (the sum of economic activities of individuals) andhouseholdwelfare levels (returns to the factors of production owned and used by the household) change in response to events endogenous and exogenous to the household is necessary for understanding the impact of actions of nature, the government, and householdmembers on changes inwelfare status ofthe household over time. But implementing this type of analysis with standard cross-sectionalhousehold survey data, such as the IAF, is problematicfor three reasons. First, the cross-sectionalnational household data used here (IAF data for 1997 and 2003) provide snapshots of the economic activities of individuals inthe householdat two time periodsbut do not show changes in a given households' livelihood over the period. From these data, it is difficult to determine how a householdreactedto events, because we cannot identify that same household in the later dataset-we do not know its characteristics.We would needpaneldata inorder to be sure we hadthe same type ofhouseholdat two points intime. Second, our national household survey does not contain data on secondary activities of members, so we cannot observe the whole householdportfolio, only a subset. In particular, we do not know how livelihoods changed over the season-we have to assume that it is the same. We also do not have information onthe level of effort of individuals-how many hours each one works. Third, earnings per individual activity are observed only for those inwage jobs. For those who do not receive wages, we could observe income, but even there the income of the enterprise, involving several members of the household, is often attributed to one member. Therefore, an approximation oftheir individualearnings is neededto compare individuals. Lacking these data, the analysis makes four assumptions: There are no multiple activities or part-time workers-veryone who is employed works full time in one activity. Using this assumption, we can compare individuals by economic activity. There is one main economic activity inthe household, and this is proxied by the main activity of the household head. Using this assumption, we can compare households by livelihoodstatus. Total householdearnings equals total householdconsumption. Everyone in the household earns the same return on their (equal) effort, so individual earnings are total household earnings divided by the number of labor force participants inthe household. This methodbasically assumes that there is subsistence agriculture plus one main livelihood inthe household and that all economically active members participate in it. All members are paid the averageproduct of this activity. This assumptionis not too far off: 95 percent of households report household members working in two sectors or less, and almost all of the two sector households have one member in agriculture. Eight percent of households report some food consumption in kind. Two percent ofhouseholdsreport having all members with wage income, and we could have calculated the earnings for each of these household members directly. But to ensure comparability, we treated households with only wage labor in the same way as the others. This methodof approximating earnings of individuals is beingtested inthis report. 34 2.27 From an economicpoint of view, livelihoods can be describedprimarily by the structure of income sources. Economic livelihoods are the aggregate o f economic activities regularly undertaken by household members during a period (the multiple activities undertaken by members o f the household during a day, week, month, or year). They are influenced by boththe characteristics o f the individual, the other members, and a host o f other factors. Rigorous analysis o f the impacts o f structural change on livelihoods i s not possible with the data sources available (see box 2.1). In this section the following are analyzed at two points intime: The aggregate labor force trends to see how individual economic activities changed. Changes in the main economic activity of the household and in the sources o f household incomes as a proxy for the livelihood pattern. Changes in average earnings inthe household. Changes in the means o f these variables across subgroups o f households, especially for poor households. Through this analysis, the structural changes in Mozambique observed above are related to major changes in economic behavior at the household level. Due to the major differences in livelihood patterns, this analysis considers urban areas, where about a quarter o f the labor force lives and works and the economy is not dominated by agriculture, separately from rural areas, where agriculture still dominates. Trendsin the Labor Market 2.28 The dominant labor force trend in urban areas during 1997-2003 was the growth of the nonagricultural private sector: wage employment and self-employment. National wage employment grew modestly, thanks to restructuring in the public sector, where wage employment fell by an average o f 3 percent a year (tables 2.4,2.5, and figure 2.3). But in urban areas, the increase in private wage employment more than made up for the loss, with 20 percent annual growth. By 2002 the share o f the labor force employed in private sector wage jobs in urban areas was 18 percent, and private wage employment went from one-third to two-thirds o f the total wage jobs in urban areas and up to 65 percent inthe whole country. 35 Table 2.4. Type of Employment,by Area, 1997-2003 (percent) Share of All Workers Rural Urban Type of employment 1997 2003 Growth Rate 1997 2003 Growth Rate 1997 2003 GrowthRate Agriculture (all) 88.8 81.5 -0.6 94.8 93.1 -0.3 71.3 53.5 -1.9 Self-employed (nonagriculture) 3.8 8.1 14.3 2.0 3.6 10.6 9.0 19.0 16.5 Wage employment 7.3 7.5 1.2 3.2 2.5 -3.9 19.2 19.6 3.3 Private 2.5 3.7 7.8 1.1 1.1 0.6 6.4 9.9 10.8 Public 4.8 3.8 -3.2 2.1 1.3 -6.8 12.8 9.7 -1.7 Employers 0.1 2.9 66.1 0.0 0.9 65.0 0.4 7.9 66.6 All economically active 100.0 100.0 0.8 100.0 100.0 0.0 100.0 100.0 2.9 Source: IAF data for 1997 and 2003. Table 2.5. Distributionof the Labor Force by Sector, Type of Employment,and Area, 2003 (percent) Self-employed Wage Employment Employer Agriculture Nonagriculture Private Public Urban (29percent of the laborforce) Agriculture 100 0 0 1.5 0 Industry 0 5.2 33.1 3.1 30.6 Service (private) 0 94.3 63.8 29.2 69.0 Service (public) 0 0.6 3.1 66.8 0.4 Total 100 100 100 100 100 Rural (71 percent of the laborforce) Agriculture 100 0 0 4.9 0 Industry 0 15.8 49.5 4.4 42.4 Service (private) 0 80.5 48.0 13.7 57.6 Service (public) 0 3.7 2.5 77.0 0 Total 100 100 100 100 100 Source: IAF data for 2003. Note: Service (private) includes trade, transport, and services. Service (public) includeshealth, education, andpublic administration. 2.29 Growth in nonagricultural self-employment andfamily business is the second most important labor force and livelihood trend in urban areas.53Self-employment is the most important economic activity in urban areas. While self-employment in agriculture declined and is now limited mostly to subsistence agriculture (with some degree o f sale o f surplus), family business in nonagricultural sectors mushroomed, employing 20 percent o f the urban labor force. 53The classificationofemployment inthe householdsurvey gives an inadequatepicture of economic activity. The treatment of women's economic activity is unclear, becausewomen inruralareas often report beingout of the labor force when they are actually spending amajor part oftheir time doing householdchores defined as economic activity (such as fetchingwood) or insubsistence farming for the household.All women listed as domestics inthe surveysare classifiedhere as employedinagriculture.This is amild overstatement of the agricultural labor force, but at least a consistent one. 36 Figure2.3. In UrbanAreas Employmentin AgricultureI s Down, Self-employment outside AgricultureUp I Annualgrowthinrharc demplcymontwer19973003 I Source: IAF data for 1997 and2003. 2.30 Almost all nonagricultural family business employment and two-thirds of private wage employment is inprivate sector services. Most o f the growth inemployment inurban family businesses occurred intrade and retail and other small-scale services. This emplo ment creation i s clearly a response to increased demand for services in urban areas? Only 5 percent o f urban family businesses and self-employment are in manufacturing, mining, or other industry. Even wage employment growth i s concentrated in the service sectors (such as telecommunications, transport, larger retail and wholesale operations, and financial services). Only one-third o f total private urban wage employment was in the industrial sector by 2003, and this sector accounted for only 28 percent o f net wage job creation between 1997 and 2003. Most o f these new jobs were created by the construction sector. 2.31 The character of wage employment is heterogeneous. Wage employment as measured by the IAF can vary from employment in a large firm with a permanent contract, nearly complete job security, a regular salary, and all benefits provided for inthe labor code--or a job with no security and payment by the hour, day, or piece. The first type o f employment is the norm for public sector workers, and enterprise surveys indicate that about half o f workers in private manufacturing firms in urban areas have a similar arrangement." Other wage workers are on fixed-term contracts, where they have all these benefits except job security. Their contract lasts for a fixed period (up to four years) but can be terminated at any time with cause and with severance. And about 10 percent o f wage workers in2002 were casual workers, who have no regular job or contract and work only for cash or in kind.56These workers are found mainly in transport, construction, and personal services (for example, domestic service). 2.32 Anecdotal evidence suggests that for some labor force participants, employment in urban self-employment orfamily business is not voluntary. Often called informal, this sector i s very heterogeneous in urban areas in Africa, operating both as a 54 Without panel data on the labor force, it is impossible to identify where exactly thejob creation andjob destruction took place. Our discussion is limitedto net job creation. 55 See Fox andothers 2005, 34; Fox, DNEAP, and World Bank 2006. 56 See appendix table A2.13. 37 residual sector for people who seek wage employment but cannot procure it, and as an opportunity to those for whom self-employment is a positive choice.57Surveys in Latin America have shown that in some countries up to halfthe self-employed or those working in a family business would not leave this activity for a wage job paying as much as they earn in their informal sector activity. Reasons for this preference include flexibility of hours (especially important for women) and the intrinsic character o f the job- independence, and the like. Even so, the PVS survey of urban labor force participants (located in poorer neighborhoods and in the small and more remote urban areas) found people considering themselves "unemployed" even though they spend significant parts o f their day engaging in economic activities. This does suggest a preference for wage employment ifthey could get it (box 2.3). 2.33 One reason the self-employment sector has grown in Mozambique is the low education of the labor force.58 Only 17 percent of urban wage workers (including casual workers) in 2002 had less than five years of education, but 55 percent in the informal sector did. One-third o f wage workers in private wage jobs had completed 5 years o f primary education, and another third reported full completion o f primary education (7 years) or more. Even ifthe growth rate o f urban private sectorjobs were to double, few o f those now self-employed or working in family firms would qualify for them. And the education requirement for new private sector wagejobs has been increasing steadily. Box 2.3. Perceptionsof Unemploymentin Urban Communities The problem of `unemployment' was frequently cited as a key cause of poverty, particularly in the urban communities visited as part of the 2006 PVS. Unemployment was important in one community where several state-owned agroprocessing plants recently closed (the municipality of Angoche, Nampula), but the fieldwork uncovered that most of the participants interviewed in other urban communities were farmers or self-employed in agricultural or nonagricultural activities and had never had a regular salariedjob. Moreover, given that a significant percentage o f the people who engage in nonwage economic activities interviewed as part of the PVS cannot easily be defined as poor (which corresponds with the results of the nationally representative surveys of 1997 and 2003), this emphasis on the link between unemployment and poverty may instead reflect a desire or expectation on the part of respondents-even those with a regular economic activity in the self-employed service sector-to find formal or salaried employment (which would then, intheir minds, improvetheir well-being). The term unemployment may also indicate the lack of reliable, stable, and regular livelihood options-a situation of high vulnerability. Further discussions with interviewees suggested that perhaps both interpretationsmay be correct. Indeed, as concludedin one community perceptions report prepared for the PVS: Theproblem of employment (or unemployment) is not the principal question, because the people in this community work [in agriculture]. The central issue, instead, is to improve the profitability and productivity of the agricultural activities of thesefarmers to ensure that they get sufJicientfood and incomej?om these activities to enable them to...satis& their social needs. 57 See Fox and Gaal2007; Ruffer and Knight 2007. 58 See appendixtable A2.20. 38 2.34 Labor force shijits were less dramatic in rural areas-in 2002 more than 90 percent of the labor force still reported agriculture as their primary activity.59As in urban areas, the important structural change was from primary employment in agriculture to the nonfarm sector. Meanwhile, nonagricultural wage employment as a primary income source remains limited in rural areas. Growth in private wage employment as a primary activity was strong, but from such a low base that the growth barely registers in terms o f the overall labor force. Public sector jobs in rural areas also fell as the contraction in government jobs hit rural areas hardest when the cashew and other agroprocessing plants were shut down (box 2.4). Box 2.4. LocalEmploymentEffectsof ParastatalDownsizingin Angoche, Nampula Over the last decade both the urban and rural areas of Angoche, Nampula, have suffered significantly from the closings of several state-owned agroprocessing plants. The municipality and district of Angoche were previously home to three cashew processing factories and processing facilities for sisal, rice, shrimp, and other agricultural products. The area also benefitedfrom a small industrial fishingport, almost entirely closed down. While only a minority of the labor force in the area was employed in these industries, most households interviewed for the PVS had a family member who worked for the enterprises, which provided an important and regular source of income. With no alternative source of wage employment for those laid off, the communities inAngoche have been forced to rely on artisanalfishingand subsistence agriculture, as well as petty trading and other nonagricultural self-employment. While the closings of these plants may have had only a marginal impact on the national labor force, the effects in Angoche have been widespread and significant-not only on those who lost their jobs but also indirectly on the local economy as a whole. The effects have also taken on a gender dimension, because many of those laid off were women. Anecdotal evidence from focus groups with women in Angoche suggests that women lost a lot of intrahousehold influence once laid off. Some were also abandoned by their husbands and forced to turn to prostitution to make ends meet. Men also complained of abandonment: without employment or sources of regular income, they could not sustaintheir families and hence their marriages. 2.35 Self-employment in rural areas covers a very diverse set of activities. As in urban areas, most self-employment and family business employment and private wage jobs were in the private service sectors, although self-employment in manufacturing, mining,and other industry as a share of total nonagricultural employment is larger inrural areas than in urban. Many nonagricultural businesses involved natural resources, such as wood collection and sale, charcoal making, quarrying, hunting, and fishing. Others were similar to those in urban areas: carpentry, sewing, tailoring and jewelry making, bicycle or other repair, brickmaking or other manufacturing, and selling food and other household items.Some rely on farm production, such as beer manufacturing, livestock sales, or grain milling. For most activities, the customer is another local household. 59Ofthis, less than 2 percentwere not self-employedor in family business as their primaryactivity. 39 Changesin Household Livelihood Patterns 2.36 Household livelihoods became more diversified-which brought the potential for a reduction in vulnerability. In both rural and urban areas households supplied less labor to the market and sent more youth to school-the average number o f people over age 15 employed inthe household fell from 2.6 to 2.3 in rural areas and from 2.5 to 2.2 in urban areas. Yet the average number o f sectors o f activity found in the household increased.60 This is a striking result, because the data measure only primary economic activity, not all the economic activities o f household members. Urban households show much more diversity, mainly because they usually retain agriculture as a subsistence activity, providing mostly income in kind, while adding one or two nonagricultural ones. 2.37 The number and sources of income also increased, and more households had at least some cash income. On average, one person inrural households and two people in urban households reported earning a cash income, and few households depend on only one source o f income.61The increase in cash income i s an especially important change in purchasing power in the household because cash can be usedto meet many more needs in the household than income in kind (school fees, health expenses, transportation). Almost all households in rural areas and the majority o f households in urban areas produce food for their own consumption, but in urban areas it is only 12 percent o f total income on average. Surprisingly, fewer than 60 percent o f rural households report any income from selling agricultural crops, and on average it provides 14 percent o f total household income. Half o f urban households receive some employee income over the year, but only 20 percent o f rural households report any sort o f employee income.62 Rural households are slightly more likely than urban ones to report more than two income sources. This may be caused by seasonality or the ability o f rural households to combine farming with other types o f income-earning activities. 2.38 More extensive data on rural incomes show the extent of income diversification. Data on incomes in the rural sector are more complete and show that while crop income (in cash or in kind) is received by all households and remained the single most important source for all income quintiles over 1997-2003, its importance fell. Households in all income groups increasingly relied on noncrop sources such as livestock and nonfarm enterprise income (figure 2.4). While few rural households moved out o f farming as the main source o f income, 42 percent had some income from self- employment or family business, and 30 percent from sale o f livestock.63 6o See appendix tableA2.24. Dataon incomes inthe IAF surveys can only be tabulated for 2003, so anationalcomparison over time is not possible. Ina family business, cash income may be attributedto only one person, even if severalhouseholdmembers participateinthe business.See appendix table A2.25. Not all households with employment income have someone whose primary employment is as awage employee. This impliesthat for some households, wage employment is a secondary activity, perhaps seasonalor casual. Inthis datasource wage income includesagriculturalwage income, but inthe IAFtabulations employment inthe private wage sector does not. 40 Figure 2.4. More Rural Households Have Income from High-Value Sources Crop 1Wage labor I Livestock 1 enterprise Income sources Source: TIA data for 1996 and 2002. 2.39 Perceptionsdata on livelihoods andpoverty in 2006 confirm the importance of multiple income sources in poverty and vulnerability reduction. Inurbanareas the PVS data from 2006 show that the key determinants of perceived improvementin household well-beingwere more peopleengaged in gainfulwork, higher salaries or returns on labor, and a good harvest.So, strongagriculturalperformanceis still importantinexplainingthe well-being of poorer households, particularly for female-headed households, even in urban areas. This gender difference may result from the tradition of women being in charge of food security in the household.In rural areas richer households have benefited more from havinga higher proportionof members with gainful work outside agriculture and with higher salaries, but this factor was much less significant than diversity of sources.64 2.40 I n sum, the changes in the structure of the labor force have helped diversify national, regional, and local economies and the sources of household income. But for most households, overallwelfare still depends on agriculture. Earnings: Returns to EconomicActivity of HouseholdMembers 2.41 The effects on household welfarefrom shifts in the structure of employment and the composition of household livelihoods depend on the returns to these economic activities. For example, an increase in employment in a sector could decrease average wages. Or an increase in wages could decrease employment. For people who do not receivewages, an approximationoftheir earnings is needed(seebox 2.1). 2.42 The privatization and restructuring of the industrial sector increased the inequality of earningsfrom wage employment in the urban industrial sector. In urban areas industrial sector earnings grew a bit faster than earnings in the privately provided 64See appendix tables A2.16 and A2.17 41 service sector, and at about the same rate as agriculture. Note that if employer earnings had been included in the wage employee category, the index would have fallen substantially, indicating that new employers-most o f whom are running small firms- are not doing as well as employees (who are mostly inbig firms).65 Slow earnings growth in the privately provided service sector is to be expected given the fall in average productivity. It i s somewhat surprising that inequality fell a bit, but economic activity in the privately provided service sector, and the characteristics of those who undertake economic activities in it, became more heterogeneous, so it i s hard to know what was going on. Meanwhile, public sector wage earnings shot up faster than all other sectors. As shown below, this i s primarily because the level o f education increased in this sector, not because wages are particularly out o f line with the market. 2.43 Data on average earnings suggest that theperception of nonagricultural self- employment as a poor employment choice in urban areas compared with private wage employment is misguided. Despite the drop in average productivity, self-employment in the private sector-mostly the service sectors-is on average a well paid livelihood. By 2003, average earnings in the family business sector were 8 percent higher than the average for private sector wage earners. In both sectors earnings are less equally distributed than agriculture, indicating a potentially higher level o f risk (or at least more heterogeneity) for economic activities in this sector. The ratio o f inequality between sectors versus within sectors fell slightly, while that between types o f employment increased. Inany case, the main inequality i s within employment groups. 2.44 I n rural areas full-time, nonagricultural wage or self-employment pays well relative to agriculture, and this difference increased for self-employmeni/familyfamily business and the public sector. But the number o f earners in these groups i s still very low. Both for sector and for type o f employment, the ratio o f inequality between sectors (compared with within) went up, but total inequality in earnings in rural areas i s much lower than in urban areas. Inequality is lowest among households whose primary activity i s agriculture in both rural and urban areas. 2.45 I n addition to sector and type of employment,education alsoplays a major role in earnings. The above data do not control for personal attributes that normally influence earnings, most notably education. For example, one reason average earnings are higher in the public sector i s that education levels o f its workforce are much higher (table 2.6). Higher inequality of education also helps explain why earnings are more unequal inurban areas. How important i s education relative to sector o f activity? In chapter 1 the role o f the head o f household's education o f the head in raising household consumption was analyzed, and education was found to be very important, especially in urban areas, where completing even five grades o f primary school increased welfare an average o f 25 percent and completing secondary 70 percent over a household headed by a man with no education.66But the analysis o f the determinants o f consumption also showed that even controlling for education, location, and demographics, differences in household welfare 65Earnings data tend to follow alog-normal distribution, meaningthat there is along tail at the high endofthe distribution, and the meanis above the median, sometimes by a large amount. As aresult, some analysts use the median insteado f the mean. We stick with the meanbecauseofthe intuitive value of this summary measure. We also usethe Theil measure of inequality, which is designedto handle this type of distribution. 66See appendix table A2.28. 42 according to the sector o f activity o f the head remained and that the public services sector was well remunerated. Table 2.6. Highest Levelof Education Completed, by Type of Employment, 1997 and 2003 (percent) Type of Employment Level of education Agriculture Self employment Wage (private) Wage (public) 1997 2003 1997 2003 1997 2003 1997 2003 None, alphabetization 85.7 86.6 51.3 55.4 36.1 37.4 31.9 10.9 EP1 10.8 10.4 33.5 27.3 38.6 31.8 25.6 18.2 EP2 3.0 2.5 12.7 13.7 17.6 18.7 21.7 24.4 ES1 0.3 0.4 1.8 2.5 4.1 5.5 11.0 23.9 ES2 0.1 0.1 0.4 0.8 2.1 4.0 4.7 14.6 Technical school 0.1 0.1 0.3 0.3 1.6 2.0 4.3 6.4 Postsecondaryschool 0.0 0.0 0.0 0.0 0.0 0.6 0.8 1.6 Total 100.0 100.0 100.0 100.0 100.o 100.0 100.0 100.0 Source: IAF data for 1997 and 2003. Note: Education refersto highest level of educationcompleted. EPl: Primary grades 1-5. EP2: Primary grades 6-7. ES1: Secondary grades 8-10. ES2: Secondary grades 11-12. 2.46 For wage workers, education and experience are the most important in determining earnings. Regression analysis using 2003 data shows that structure o f returns to education i s linear and convex at higher levels o f education. The fact that the last years o f education are worth more than the first (the opposite o f what i s found in middle and high income countries) indicates major skills shortages at this level o f education (figure 2.5).67 There are few women who report wage earnings, but those who do earn even higher returns to education than menuntilpostsecondary education. As there are very few women in the survey who have either technical or postsecondary education (except teachers), it is not surprisingthat any differences are not significant. 2.47 Experience on thejob is also important-each year of experience (measured by age) adds about 6percent to wages. The observed returns to experience may actually represent the gains in productivity from learning on the job or they may represent something else about the workers that is not measured, in other words, a cohort (or selectivity) effect. This effect would come about because the high cost o f education a decade or more ago and the few education opportunities available at that time means that those who were able to get an education and a wage job 10-20 years before 2003 were most likely a few talented individuals, and these talents are partly what the experience variable i s measuring. With the expansion o f education, both the pure returns to education as well as the cohort effect should decline, as they have in other countries.68 2.48 After accountingfor education and experience,public sector employees do not earn more than others, but married men do earn a wagepremium. Wage jobs in both ''The dependent variable is the log of wages, so the coefficients can be interpreted as the percent change inearnings that can be expectedfrom a change inthe independentvariable. See Fox and Gaal2007, for examples. 43 the primarily public service sectors and the primarily private ones pay about as well as other nonagricultural wage jobs. Agricultural wage labor jobs suffer from a very negative sectoral effect, and casual (intermittent) wage work pays poorly regardless o f sector or education. This type o f employment i s only 14 percent o f all wage employment. Married meno f any type (even widowers) get a wage premium.Because there are very few single men in the dataset, it is not clear what this effect means, but it does account for most o f the difference in wages between men and women. Although education as a sector does not pay less than other sectors, the returns to teacher training are lower for women than those for any other type o f secondary education and are lower for men than those for technical education. It is possible that education salaries are biased downwards, but this requires further study. Labor Market, Livelihood Changes,Earnings, and the Poor 2.49 How have the livelihood shifls analyzed above affected thepoorest segments of the economy? Thus far, the analysis has focused on the whole labor force and on the average household. Given Mozambique's fairly equal income distribution and the high level o f poverty in 1997-nearly 70 percent o f the population-the average effect could closely approximate the effect on the average poor person. But it might not. Inthis section the household response to macroeconomic and structural changes i s disaggregated by quintile for a closer examination o f the relationship between changing livelihoods o f households in Mozambique and poverty reduction. This section also looks at changes in inequality within and between livelihood types, which indicates the riskiness o f livelihood choices. In this analysis the bottom 20 percent are considered the very poor, the second quintile the poor, the third and fourth quintiles the middle income group, and the fifth quintilethe highincome group. 44 Figure 2.5. For BothWomen and MenEducation I s the Most ImportantDeterminant of Wages, 2003 Percentage Change in Wage from a Unit Change in Explanatory Variable Age Female Referencecategory unmarried Married Polygamous Cohabiting Divorced Widowed Referencecategory: no education CompletedEPI Completed EP2 Women Completed ESI Completed ES2 Completed ET1 CompletedET2 Teacher Education Higher Education Referencecategory: manufacturing Agriculture Mining Construction Transport Casual Contract -1 -0.5 0 0.5 1 1.5 2 2.5 3 Source: Appendix table A2.22. Note: Regressionresults. Bars indicate the percent change in wages associated with aone unit change inthe explanatory variable. Coefficients insignificant at the 95 percent level havebeen set to 0. Categories, thus, not displayed inthe graph because they were not significant are: divorced, widowed, trade, services, education, health, and public administration. EPI: Primary grades 1-5. EP2: Primary grades 6-7. ESI: Secondary grades 8-10. ES2: Secondary grades 11-12. 45 2.50 In number and type of income source, the main difference is between the bottomfour quintiles and the top 20percent. The lack o f differentiation by income group ina number of sectors and types of income inthe household may reflect the weakness of the data. Only the upper income group shows a clear trend in the number o f income sources (more likely to have more than one) and the type o f income (more likely to have wage income and self-employment income, and less likely to have income in kind). All welfare classes have someone in agriculture, and some 40-50 percent o f households in every income group also earn income selling agricultural goods, reflecting the fundamentally agrarian and rural nature o f Mozambique's economy. But those whose primary activity i s wage employment, especially those in the public sector, are much more likely to be found in the wealthiest quintile. The wealthiest quintile i s also more likely to have household members engaged in self employment. The fourth richest quintile i s the one most likely to earn income by selling agricultural products; this seems to be the quintile where the more successful rural farmers are located. In terms o f shifts between years, only for public sector wage workers i s there a big shift in the quintile where they live; this is probably the result o f the major salary increase, which probably pushedtheir householdsup into the highestq~intile.~~ 2.5 1 Once again more detailed rural income data provide a clearer picture of the nature of ruralpoverty; income diversity increases with total income, suggesting higher vulnerability among thepoorest households. In rural areas all income groups except the highest one rely on crop income (including income in kind) as the main income source. High-income households have increased their reliance on wage labor (both on and off farm) and earn about a quarter o f their income from nonfarm enterprise. The probability o f a household engaging in wage labor as the main source o f income increases with income. All income groups have increased participation in the nonfarm enterprise sector, but success is correlated with higher income. It is unclear whether increased nonfarm income moved households out o f poverty over 1997- 2003 or whether higher income households were able to build larger and more successful businesses. Livestock is still an unimportant income source inrural M~zambique.~' 2.52 Wage labor income and nonfarm income is associated with higher overall income, and with education, infrastructure, market access, and gender of the household head. Households living in villages with more infrastructure, such as electricity and paved or all-weather roads near the village, are much more likely to have a nonfarm business generating wage labor income. Households with younger heads and male heads and members o f those households with education are more likely to find a wage job or work at a nonfarm business. Households with more members, more land, and those living in a village with more infrastructure are also more likely to be successful at a nonfarmbusiness. 2.53 One reason agricultural earnings remain low is the declining price of agricultural goods relative to nonagricultural goods. Evidence from the 2006 PVS suggests that price risk inagriculture and declining internal terms o f trade played stronger 69 See appendix table A2.26. 'O See appendix table A3.24. 46 roles in reducing living standards than the national household survey data indi~ate.~' Urban households cite increased costs o f living and loss o f employment or gainful work as key reasons-with poor harvests coming in a distant third. But the result varies, particularly by wealth group. 2.54 Increased costs of living (deteriorating internal terms of trade) seem to have disproportionately affected the poor. More than half the poorest households claim it as the main reason for their worsening well-being over the last five years. In rural areas the main reasons for worsened standards o f living were the same as in urban areas, but in a somewhat different order. Increased costs o f living were the main reason, followed by poor harvests and loss o f gainful work. More than 60 percent o f the poorest households consider the increased cost o f living as the main reason for their loss o f welfare-twice as many as the richest households. Poor harvests are also important for the poorest, and the loss o f gainful work (especially infemale-headed households) were also cited often. 2.55 For a riskier activity, thepossibility of a higher income has to be balanced with the possibility of a very low income, or even a negative one, a disastrous result for a household. Although agriculture has the lowest average earnings, how does the probability o f earning a higher income in another sector compare with the probability o f earning a secure income in subsistence agriculture? Perceptions data have confirmed the risk to rural livelihoods of a bad agricultural season (weather, prices, or otherwise). But what about the risk to a household o f deploying resources out o f agriculture? How i s this risk evaluated? Ifthe probability o f an income innonagricultural activity is lower than the average in agriculture, a household may not decide to undertake the activity because the downside risk i s too high. One way of comparing the risk o f activities i s to compare the variance inthe results. For earnings, which do not have a normal distribution, risk can be compared by examiningthe inequality in earnings.'* 2.56 Inequality in earnings has risen, and economic activities more remunerative than agriculture show a much higher difference in earnings, suggesting that diversification into these activities is not always a successful livelihood strategy. Although inequality o f income among households did not rise over 1997-2003, inequality o f earnings-as measured by the variance in results-went up substantially, as did the inequality o f earnings within certain types o f economic activity. In2003 average earnings from self-employment were still 50 percent higher than agriculture earnings, and the inequality o f income more than doubled, to twice that o f incomes from agriculture. Nonagriculturalwage inequality also increased, suggesting that it may not be good for an already poor household. Even though agriculture has lower average earnings, it does not have as much earnings variability within a given year. 2.57 Higher income rural households are more likely to engage in higher value nonfarm activities. In general, higher income rural households are more likely to have enterprises that require some capital, such as manufacturing enterprises, while lower income households that have a nonfarm activity are usually engaged in a very labor 7'PVS findings refer to a later period, which partly explains the difference. 72Alternatively, the variance ofthe log ofearnings canbe compared; ameasure of inequality is more intuitive. See appendix table A2.27. 47 intensive activity, such as firewood collection for sale. Likewise, only higher rural income households tend to have wage income from nonfarm sources. 2.58 The evidence on whether income diversification within the household reduces poverty is less clear. A higher proportion o f households relied primarily on wage labor as the main source o f income, and these households usually were not poor. As noted, it is impossible to know whether a household has a high income because it has a household business or whether it has a household business because they have a high income. The data here support both propositions. Because self-employment is risky, only a middle income household may be able to enter this market, as a household may need more land, assets, or capital, as well as more labor force participants inother sectors, to cover the risk o f starting a business. But having a business is associated with higher earnings, so starting a business may increase incomes, all other things being equal. 2.59 One way to analyze this question is to look at households with a head and a spouse and to see the relationship between beingpoor, and having both an agricultural income and a nonagricultural income, compared with having two agricultural incomes or two nonagricultural incomes. While the relationship is not entirely clear, ina majority o f households income diversification between the head and spouse i s not correlated with lower poverty rates. In fact, the opposite appears to be true. Where the head i s in agriculture, the poverty rate is higher if the spouse is in nonagri~ulture.~~ Where the head i s in a nonagricultural sector, poverty i s lower overall, but i s the lowest when both the head and the spouse are innonagricultural activities.74 2.60 While income diversification may not reducepoverty, qualitativefindings from the PVS suggest that households consciously diversify livelihood to cope with the dynamic and seasonal nature of poverty in Mozambique. Many o f the complementary activities adopted by the surveyed households take on additional importance during seasonal (primarily agricultural) scarcity and stress. Indeed, most households aim to diversify their livelihoods and income sources as far as their assets and resources allow to better weather the crises and seasonal changes that define the everyday lives o f poor Mozambicans. In a few cases, such as when primary activities have collapsed or disappeared, complementary strategies may take on a more primary role.75 The poorer segments o f the population, however, often have fewer livelihood strategies and opportunities to diversify because they have fewer assets and capabilities to draw on. 2.61 I n sum, all data on the relationship between shifts in the structure of the economy and employment and changes in household livelihoods and monetary werfare point to a virtuous relationship between macroeconomic adjustment, development and diversification of the structure of the economy, changes in patterns of employment, and changes at the household level in labor supply and livelihood structures, as all of these 73 See appendixtable A2.28. 74 This result may be picking up the fact that most couples who neither headnor the spouse i s inagriculture live in urbanareas and havehigher educationlevels. 75 Migration, for example, has become amuchmore important long-term, ifnotpermanent, strategyfor the households interviewedin Gaza, following the collapse ofthe agricultural industry.Also, inurban Angoche inthe provinceof Nampula, artisanalfishing has become amuchmore important livelihoodstrategy for householdsdirectly affected by the privatization andclosing ofthe state-ownedprocessingplants(cashew, shrimp) inthe area. 48 have contributed to poverty reduction in Mozambique. As Mozambique's economy moved from postwar stabilization into recovery and growth, it pursueda stabilization and growth strategy with high levels o f private and public investment in physical capital and declining but high levels o f foreign aid, for social services and infrastructure. The expansion in the economy was broad based, with much higher growth in the nonagricultural sectors. Even though the initial livelihoods o f the poor and the middle class were in agriculture, the growth pattern benefited most o f the population because the expanding sectors had higher average productivity and earnings, are highly labor- intensive, and have a high income elasticity o f demand. Thus, both an expansion in employment and a growth in earnings were possible in these sectors. One reason these two trends are positively correlated i s that the labor force was expanding slowly. Households reduced overall labor supply mostly by keeping children and youth, especially girls, in school longer. This decision did not seem to produce a drop in income, although without panel data it i s difficult to check. This development does suggest that the marginal product o f agricultural labor overall in Mozambique was in 1997 very low. While average productivity has improved, the productivity gap remains a threat to sustained propoor growth inthe future. GENDERDIMENSIONSOFPROPOORGROWTH MOZAMBIQUE IN 2.62 A growing body of microeconomic evidence supports the importance of considering gender issues in economic policymaking, not least for macroeconomic policy. This section explores the gender dimension o f Mozambique's propoor growth at the aggregate, sectoral, and household levels. To what extent did women contribute to economic growth? Did their contribution differ from men's and if so, why? Conversely, were the benefits o f growth gender-neutral? This section also looks at the effect o f growth on gender equality and the constraints to economic growth posed by gender inequality in Mozambique. Finally, it identifies gender-related policies that will help sustain Mozambique's propoor growth. The Contribution of Female Education to Growth 2.63 Mozambique's economicgrowth was held back by lowfemale education levels. While less central as a driver o f growth than the increase in physical capital, a growth- accounting exercise shows that increased education in the workforce accounted for 28 percent o f GDP growth between 1999 and 2004.76 Disaggregating this contribution by gender shows that the increase o f educated women in the labor force explains 11 percent o f the GDP growth, compared to 17 percent attributed to the increased supply o f educated men, suggesting.This suggests that ifwomen had been attending school at the rate o fmen and subsequently entered the labor force, growth could have been higher over the period. 2.64 The education level of the active labor force has increased since 1980, particularly after 1995, and the gender gap is slowly closing.77The share o f educated workers in total employment increased at all levels for both women and men, but no 76See appendix table A2.29. 77See appendix table A2.30. 49 group grew faster in the labor force than women with primary schooling: over five times since 1980 compared with three times for men. While the trend i s positive, the level is still very low: In 2003, 86 percent o f women in the labor force reported that they had not attended school or completed five years o f formal education, compared with 68 percent o f men. Because of significant investments ineducation made over the last decade, women's contribution to the education level o f the active workforce i s set to gradually increase as the larger cohorts enrolled from the mid-1990s graduate from the schooling system and enter the labor market. Not only can an aggregate macroeconomic effect be expected, but a poverty reduction effect i s expected as well. Results for sectoral and micro-level analysis below highlight the importance o f women's increased education and skills endowment for propoor growth. 2.65 Mozambique's aggregateprimary enrollment increase between 1997 and 2003 is due in largepart to an increase in girls' enrollment. Inthe younger age groups, girls' enrollment has caught up with boys': In fact, boys comprised the majority o f the 6 percent o f Mozambique's 7-- to 10-year-olds not attending school in 2003. And the trend was particularly accentuated among 11- to 15-year-olds, where the share o f girls enrolled increased from 34 percent to 39 percent, and the share o f boys attending school shrank slightly. This likely reflects the lagged effect o f an increase in enrollment, particularly o f girls, inprimary school inthe mid-90s, at the end o f the war, havingprogressed upward in the system. Gender-DisaggregatedTrendsin the Labor Market 2.66 Men have dominatedthe movement of adults out of agriculture into wage and self-employment in nonagricultural sectors, and as a result, the agricultural laborforce is increasingly made up of women. Between 1997 and 2003 employment in agriculture shrank as a share o f the economically active labor force from 90 percent to 82 percent. There were two key movements behind the shedding o f labor in agriculture: first, boys and girls delayed entry into the workforce to attend school; because most o f them would have worked in agriculture, the labor force fell. The second i s that men moved out o f agriculture to pursue higher earning activities in other sectors. An estimated 11 percent o f men in agriculture left to work inother sectors.78About half o f this movement took place in urban areas. As a result, by 2003 the share of economically active men employed in agriculture had fallen to 32 percent while the number o f women engaged in agriculture remained close to constant. Consequently, women working in agriculture made up a full half o f Mozambique's economically active population in 2003 and nearly two-thirds o f the agricultural labor forces7' 2.67 Accordingly, most of the diversification in household income sources accrued to men. Where did the men who left agriculture go? Countrywide agricultural and household surveys show households diversifying income sources, with male household members assuming new income-generating activities, mainly in the private sector. Estimates based on household surveys show some 283,000 additional men taking up wage employment between 1997 and 2003, primarily in transport, trade, and services, but also 78 Seeappendix table A2.31. 79 World Bank staff calculation, IAF data from 1997 and 2003. 50 in construction. In addition, men accounted for most of the increase in self-employed in nonagricultural sectors. As women's labor in Mozambique was less mobile, women's main income source still agriculture. The increased overall mobility has been positive for the economy and for rural household welfare, contributing to poverty reduction. But the strong gender bias has implications for gender equity within and outside the household. What determined this gender difference in labor mobility? 2.68 Evidence suggests that higher skill levels have enabled the labor mobility of men. The substitution o f nonagricultural labor for agricultural i s imperfect in Mozambique: Agricultural labor consists mainly o f smallholder farmers in rural areas, with skill levels often much lower than those needed to move.8oBut the substitution elasticity i s not zero: Men, with even a few years o f education, have managed to move into off-farm employment. But working women, especially those in rural areas, have education levels much lower than men. Among the rural workforce, the ratio o f men with primary education in the workforce remained more than double that o f women. And change is happening slowly: educated men's entry into the labor force continues to outpace women's every year." 2.69 I n urban areas more than 40 percent of households have both the head and the spouse out of agriculture. There is evidence to suggest that when household heads leave agriculture, spouses sometimes follow. When the main employment o f the household head-most often a man-isin agriculture, 98 percent of spouses also work in agriculture.82Households with the head employed in trade, services, education, and health are more likely to have spouses also in nonagricultural sectors. But overall, spouses are most likely to be in the agricultural sector, even in urban areas and even with household heads in nonagricultural employment. Livelihood Trends 2.70 Female-heads of households in rural areas are particularly constrained. Female-headed households have not been able to diversify their income porlfolios as quickly as their male counterparts.Between 1997 and 2003, diversification into nonfarm businesses has increased significantly (from 37 to 46 percent for male-headed households and 24 to 29 percent for female-headed households), but male-headed households were more successful in making nonfarm enterprises their major source o f income, increasing from 13 to 17 percent. Among female-headed households the proportion whose main sources o f income i s nonfarm remained unchanged at 11 percent over the period. As a result, while the share o f nonfarm microenterprise income increasedfrom 14 percent to 18 percent among female headed households, it remained ~onstant.'~Wage income i s a higher share o f total income for male-headedthan for female-headed households. NaaeraandNicholajsen 1998. See appendix table A2.29. 82See appendix table A2.32. 83See appendix tables A2.31 and A2.32. 51 2.71 The sources of rural and urban household income differ significantly, with women in rural households reporting very little cash earnings at all. Home production and the sale o f products are the two largest income earners in rural areas, with urban households earning incomes primarily from employment and ~elf-employment.~~Women in urban areas report earning 13 percent of the total urban income, through employment, self-employment, and sellingproducts, compared with 65 percent for mensg5Inrural areas the gender difference is even more pronounced, with women's reported earnings making up a mere 4 percent o f rural incomes, comparedwith 38 percent for men. The largest part o f rural household income by far, however, is in the form o f joint household home production, which includes subsistence agriculture. 2.72 Women consistently report cash earnings a fraction the size of men's. Women's and men's occupationalstatus is comparatively similar in Mozambique, with working women making up 37 percent of the population, compared with men's 32 percent in 2003. Yet women report earnings (including those in-kind) a fraction the size o f men's: 24 percent o f 2003 earnings were reported as men's employment income, compared with 3 percent for women.86And men's income from selling products was 14 percent, compared with a mere 2 percent for women. This i s a pattern at all income levels, with no great variation across quintiles. While the findings are consistent with expectations, a male bias in reporting cannot be rejected because there i s limited scope in the surveyto report joint earnings for activities such as sale of products, a major category inMozambique. A large share ofthe earnings from a family business, particularly inrural areas, may simply be reported as earned by the head o f the household (by default or cultural preference). Attributing income to men to the extent and with the consistency described above suggests, however, that women are not in control o f the income-that growth did not benefit women equally. Several studies show negative secondary effects on household welfare, nutritional status o f household members, and investment in household members education and health when women do not control household income, so this trend may have medium term growth impactseg7Moreover, it often harms women's bargaining power, with possible consequences for their mobility into other sectors. 2.73 Being a woman in wage employment carries a significant labor market penalty. For the 10 percent o f the labor force in Mozambique earning a salary, wage- regression analysis shows signs o f sex discrimination, with women earning 28 percent less income than men, even when controlling for differences in education levels, marital status and sector o f activity. Significant wage differences were found to stem mainly from marriage status, sector o f activity, and contract variables. While married men realize a premium in the labor market, married women do not. Working as a casual laborer rather than on a regular contract appears particularly disadvantageous for women, who face an additional negative factor o f 76 percent when paid hourly or daily on a casual contract. 84Home productionincludesin-kindincome from subsistence agriculture. 85The remaining22 percentof urban income is reportedas earnedjointly by householdsinthe form of home production, pensionsand alimonies, income from capital, and occasionalincome. 86See appendix table A2.35. "See, for instance, Khandker 1998;Kennedy andPeters 1992; Thomas 1990. 52 2.74 Does it matter if there are gender patterns in household production in Mozambique? First, gender roles in household production exist in Mozambique. Activities o f men and women in agriculture vary somewhat by region, but a stylized story would describe men as clearing land and tending livestock and being more involved in producing cash crops than food crops and women responsible for household food security and performing almost all household chores, including provision o f firewood, water and food, producing food crops, as well as responsible for a substantial part o f cash crop production (box 2.5).88 Second, women in rural areas usually work long hours-estimates from other countries find 14-1 6 hours a day.89This implies that labor saving technologies or infrastructure that reduces women's work load should reduce poverty and increase welfare, because it would allow women to be productive in both economic activities as well as inunpaid household work. Gender Equity and Poverty 2.75 Improving the productivity of women in agriculture would boost economic growth and reduce poverty. Using an economywide model, a 2005 study examines the links between the increasingly open economy in Mozambique, product prices and male and female earnings in the agricultural sector.90Including gender-aspects in a computable general equilibrium model and linking it to a microsimulation model o f households, the study analyzes poverty implications of different policy scenarios. Itfinds that the effect of opening the economy to trade has important growth effects, but only a small effect from a 88For further background, see Arndt andTarp (2000). 89World Bank 1997. 90Arndt andothers 2005; Arndt andTarp 2000. 53 Box 2.5. Who Does What in Mozambique? Women in Mozambique are in charge of fetching water, cleaning, cooking meals, processing food, collecting firewood and caring for children and the elderly. The PVS for this report in four provinces inMozambique questioned 1,49 1heads about how households' labor resources (people over the age of seven) are allocated. Results show that gender is the strongest determinant for households' division of labor, echoing similar surveys from other countries in Sub-Saharan Africa. The survey finds that women are almost exclusively charged with providing household's water, cooked meals, cleaning, food processing, energy, and care for elderly and children. For instance, in rural areas, households reported that 86 percent of women had collected water in the week prior to the survey, comparedwith 14 percent of men, and 65 percent of women were reportedto collect firewood, comparedwith 23 percent of men. Women are also engagingmore than men in agriculture,judging by the survey, although the difference is not as pronounced. Menare slightly more likely to work on house maintenance (7 percent more than women), and to tend farm animals (5 percent). The variation between rural and urban households and between households' income levels is, incomparisonto gender, very small (figure 1). Figure 1. Household Division of Labor, 2006 - ~ 100 Rural Women 90 ,!a Urban Women EE d 80 70 60 % 50 `El 5 40 f 30 5s 20 10 s o Source: Authors' calculations basedon data from PVS 2006. Note: Activities during last seven days. Sample not representative. While the survey is approximate, using a seven day recall and measuring neither the frequency nor the time burden of tasks, the results are strikingly uniform, suggesting that women are undertaking much of the basic low-productivity work in households that is essential for the survival of family members and for households' ability to supply labor. It also suggests that labor-saving investments in relatively basic infrastructure, such as water, sanitation, and energy, can help free up labor resources greatly. Only a handful of nationally representativetime-use studies exist for countries in Sub-Saharan Africa, and none ofthem are for Mozambiclue. 54 2.76 Macroeconomic, sectoral, and livelihood trends indicate that an increase in productivity in agriculture is essentialfor poverty reduction, especially in the work done by women. Labor will continue to move out o f agriculture, this mobility has been beneficial for both growth and poverty reduction, and policy measures should seek to bolster it. Women appear to face more binding mobility constraints, however, as they have not accessedthese sources o f income. Policy measures needto focus specifically on increasing the income earning potential o f women in agriculture. These measures should ones which save labor inthe nonagricultural tasks traditionally done by women. Box 2.6. When Women Select Them, CommunityInfrastructureInvestmentsRaise the Productivityof Women Villagers in Indonesia received grants through the World Bank-supported Kecamatan Development Project (KDP) for small-scale infrastructure investments selected by the community. The KDP developed several rules and processes to ensure that women's voices would be heard when selecting village projects. As a result, village investments substantially raised rural women's economic productivity by reducing the time women spent on traditionally female tasks, suchas water collection, andby helpingthem get to their fields faster. In one village a gravel road was built to the local rice paddies and created access for motorcycle taxis, making the transport of rice easier and reducing the time women spent traveling to and from the paddies. Ina nearby village a three kilometer-longwater pipe was built through a dense forest and up the steep sides o f avolcano to pipe water directly into the village, reducingthe time women spent collecting water. Other investments included community washing areas near the river, including concrete sinks andpoles for clotheslines. Over the course of three Kecamatan projects, women's status has progressively improved. Where previously women's roles were confined to the household, the KDP has promoted women's participation in community decisionmaking and used women facilitators in council meetings. It also initiated a specialplanning stream for women's groups, a competitive reward for promoting women's participation, and a women engineers program. Women submitted 6,170 proposals (or 55 percent) of the 11,275 proposals fundedacross23 provinces. CANMOZAMBIQUETHE ODDS?THEPROSPECTSFORPROPOORGROWTH BEAT 2.77 Because of good fundamentals, Mozambique's growth episode is already one of the longest on record in low income countries. Since the last household survey, Mozambique's economy has continued its vigorous, broad-based growth. GDP has grown at about 7 percent a year since 2003 (5 percent per capita), showing no signs o f slowing down. Private consumption growth continues at about 3-4 percent per capita per year, slightly higher than over 1997-2003 .91 Agricultural growth has been strong, which should support continued improvements in welfare in rural areas (figure 2.6). Government spending, especially on service delivery, is increasing as a share o f GDP, thanks to high levels o f donor support. Research shows that growth cycles usually run out after about sevenyears in low income, non-oil producing countries. When growth starts after a crisis, it is based (at least partly) on rehabilitation. Once this source of growth is exhausted, growth tends to sputter out, or inequality increases halting the poverty reduction impact, 91See appendix table A2.36. 55 or both. Through good policies which provided political and economic stability, Mozambique has already beaten the odds on sustained economic growth. 2.78 Mozambique has a lot of natural assets that can support and sustain the high growth rates neededfor poverty reduction. Aside from a surplus o f fertile land suitable for rainfed agriculture, the long coastline includes several underdeveloped coastal tourism areas. The central provinces have coal deposits, and this area expected to undergo rapid development as a large new coal mine plus rail link comes on stream with private financing. Electricity i s low cost thanks to hydroelectrical development. The construction o f a new bridge over the Zambezi River i s expected to reduce north-south transportation costs dramatically. As a transport corridor for neighboring landlocked countries, the east- west transportation links could be an important economic asset if the Zimbabwean economy recovers. And the northern province o f Nampula contains Nacala, one o f the deepest natural ports in the world. All these advantages require substantial public and private investments to develop, however, and improvements in public management to maintain. Mozambique's lack o f mineral resources may be a blessing, as the country has avoided the "curse" o f an economy dependent on this source o f growth. Figure 2.6. All Sectors Are Still Growing Rapidly in Mozambique,2000-08 I 25 15 g! E -4 riculture = +Industry 6 3 5 +Services -15 Year Source: World Bank staff estimates usingnational accounts data 2.79 It is unclear whether Mozambique is continuing to beat the odds on shared growth. Data are not yet available to analyze the poverty impact o f this growth. An integrated survey, capable o f measuring household welfare, has not been undertaken, and a new one i s not scheduled until2009, with data to be available in late 2010. So, evidence i s sketchy about the poverty reduction performance since 2003. Government data show continued increases in access to services, and rural income survey data show that average household income has increased. Based on extrapolations from a few indicators, the Ministry of Planning predicts a continued fall inpoverty. But other indicators point to a slowdown in poverty reduction. Rural household surveys show income inequality seems to be growing, and already high urban inequality persists, so the same high growth has 56 less o f a poverty-reducing effect. What are the prospects for continuing shared growth, and what i s requiredto achieve it? This section draws on the previous analysis to consider the prospects for Mozambique to continue on a growth path that i s equitable and poverty reducing. 2.80 Mozambique's drivers of growth are changing, with education becoming more important. The role o f aggregate factors (land, physical, and human capital) in Mozambique's growth path has been changing since 2003. A new growth accounting model shows a 25 percent greater contribution o f human capital during 1997-2003, implyinp2an important role for more highly educated new labor entrants in future growth. While this structural change will put more pressure on an education system still strugglingto meet basic access issues, it may also mean that Mozambique is capable o f using physical capital more efficiently-a development that could be consistent with propoor growth if it allows the absorption o f more labor out o f agriculture and other low productivity sectors. 2.81 Mozambique's growth was efficient at reducingpoverty because inequality did not change. However, the increasing education requirementsfor growth might increase inequality, especially within urban areas and betweenrural and urban areas. While overall inequality in Mozambique i s fairly low, inequality within the urban sector i s much more prevalent (meaning within cities and between richer urban areas such as Beira and Maputo and smaller, poorer urban areas). An upward shift in demand for educated labor not met by supply could increase inequality. The wage regressions show a very highrate o f return to education, which indicates a shortage at the postprimary level, and especially at the secondary and postsecondary levels. Data from a sample o f (primarily) manufacturing enterprises show that larger enterprises-ones that create a lot o f wage employment in urban areas-demand a much higher level o f education than the average urban labor force participant has. Mozambique is moving rapidly to address this skills shortage, but so far has been mostly at the primary level. Thus the gap in returns to education could widen further, increasing inequality, despite the government's push to increase the supply o f educated labor. 2.82 Poverty reduction requires continuing labor mobility from low productivity sectors to higher ones. In urban areas private sector wage employment growth needs to accelerate in the industrial sector. As the rural sector sheds labor, the urban labor force is expected to grow rapidly. This labor force needs jobs. While self-employment in the service sector was an important driver o f poverty reduction over 1997-2003, if this sector has to absorb all o f the future urban labor force growth, productivity will decline to the point where incomes stagnate, especially among the poor and least skilled. And the self- employment sector is highly dependent on overall urban income increases. 2.83 Because domestic capital is limited, Mozambique needs foreign direct investment in labor intensive industries to drive wage and salary job creation. Confining foreign direct investment to capital and energy intensive industrial "mega projects" won't yield the job creation performance needed. Mozambique's domestic manufacturing sector i s mostly very small, concentrated in low tech food and beverages and wood processing. 92 Jones 2007. 57 There should be potential for diversification into labor-intensive manufacturing exports such as textiles. Past performance indicates that Mozambique should be able to attract the foreign direct investment required to create these jobs. Maintaining and improving the investment climate are crucial, especially in areas that are most important for large manufacturing enterprises. 2.84 Even with a major increasein industrial sector wagejobs, Mozambique cannot create wagejobs fast enough to absorb all new entrants. The focus on the investment climate for large firms should be complemented by support for the poorly capitalized small and micro businesses. The high mean income for family business, especially in the service sector in urban areas, i s good for poverty reduction. But the widening inequality within the sector is worrisome-reducing the efficiency o f growth in reducing poverty. It i s an indication o f the growing heterogeneity o f the service sector, with successful entrepreneurs working side by side with marginal ones. More analysis will be needed to develop policy and programmatic recommendations to support this critical sector. The rural nonfarm sector (fed by growing incomes in the farm sector) will also be an important part o f the poverty reduction picture. The sustainability o f the natural resource- based activities i s limited. The market for many nonfarm activities is currently limited to the local community. Opening market opportunities beyond the community i s likely to yield a sizable impact and ensure the expansion and increased profitability o f these self- employment activities, if the supporting environment is in place, such as infrastructure and access to capital. 2.85 Continuing output growth and labor productivity improvementsin agriculture are critical for maintaining shared growth-and will require new strategies. Agricultural growth's effect on rural poverty i s particularly powerful because the rural economy is dominated almost entirely by smallholders. Sustained growth in per capita consumption for households in all income groups whose main occupation i s agriculture was the largest driver o f poverty reduction over 1997-2003. Inaddition to the fact that the rural poor are concentrated in agriculture, which allows growth in crop production and marketing to have a direct effect on rural poverty, positive externalities come from assuring food security and forward and backward links to the nonfarm sector. From this point o f view, the aggregate growth performance o f agriculture since 2003 i s a promising sign. But if it is not accompanied by a reduction of labor in the sector, the prospects for shared growth inrural areas are weaker. One factor that increased income per capita inthe agricultural sector-the withdrawal of low productivity young labor from the labor force to go to school-is one change unlikely to be repeated, at least not on the scale o f 1997- 2003. 2.86 Extensive agricultural practices-increasing output by working larger land area-are not sustainablefor smallholders. Rural farmers have been practicing "slash and burn" type techniques, which are now unsustainable. Land yields are too low-there i s no way to increase production. Adding more labor has very low returns. Smallholders have raised their incomes by adding nonfarm activities. But as in the urban sector, this will runout o f steam as well, because it depends on the growth o f agricultural incomes. 2.87 Increasing smallholder household income will require infrastructure investments to support rural nonfarm income growth, and new strategies to help 58 smallholders raise yields. Increased household income from both farm and nonfarm sources is associated with paved roads near the village. Mozambique needs to continue investing in transport infrastructure in rural areas. But as households shift labor out o f crop production into higher productivity nonfarm employment, the past strategy o f expanding the land under cultivation will become more difficult, as labor shortages appear. New farming methods, more productive per labor unit and provide higher yields per hectare, will be needed. They require improved seeds, drought resistant varieties, different farming practices, and the use o f more chemical inputs. Smallholders need to be supported through extension and credit to use these technologies. 2.88 Encouraging thegrowth of large-scale, commercial agriculture brings benefits to smallholderfarmers. A successful agricultural strategy will needto rely on the private sector for needed investments and know how-reliance only on publicly funded extension, credit, and investment i s not sustainable. Commercial agriculture brings new technologies to the area, which have spin-off to all farmers in the area, large and small. Commercial farms also subsidize infrastructure development, provide needed wage employment, and increase demand for products o f the nonfarm sector. Contract growing schemes, another form o f commercial farming supports smallholder farmers to modernize and earn higher incomes. These schemes bring improved seeds and production processes to the farmers, provide needed input finance, and reduce the farmer's risk from greater specialization. Analysis o f the tobacco sector in Mozambique has shown that private extension services provided by contract growers have a high rate o f return. Contract farming requires an appropriate regulatory framework that strikes a balance between protecting the farmer from a monopolist while ensuring returns to the contracting company commensurate with the risk. 2.89 Mozambique may wish to consider rural public works schemes to expand opportunitiesfor thepoorestfarmers who cannot break out of the subsistence trap. Past diversification o f household income has resulted from pull factors-in particular, a dynamic rural economy. A group o f smallholders cannot modernize or commercialize, and cannot develop or stay profitable in off-farm activities. They are usually low income, have few assets, and may lack adults in the household ("skip generation" households). Provision o f labor income earning opportunities through community-driven public works programs could be an effective safety net. * * * 2.90 A key element of propoor growth is gender equity, as both a driver and a consequence. Mozambique has already recognized and targeted increased education attainment for all, including girls, as a driver o f growth. Less attention seems to have been paid to the implications for increasing productivity with the growing feminization o f the agricultural sector. This has particular implications for the policies required to sustain the trend o f the recent productivity increase in agriculture, the foundation o f Mozambique's propoor growth. Women are often overlooked by extension agents so they may not get the information they need. Women have less access to the capital to invest in these technologies. International experience has shown that successful agricultural projects needto include women farmers early inthe design stage, so that projects meet their needs. Strategies are needed to help women diversify crops and raise output to a commercial scale without riskinghousehold food security. 59 2.91 Widening employment opportunities for women, either in commercial agriculture or nonfarm activities-by increasing the mobility of women out of agriculture-will also help reduce poverty. The fundamental measures may be outside the agricultural sector. For example, improved infrastructure to reduce the time women spend on household chores would free up more time for income-generating economic activities, including learning about better farming practices, or for building their own human capital. 60 3. AGRICULTURAL GROWTH,DIVERSIFICATION, AND MOBILITY-A CLOSER LOOKAT RURALPOVERTY 3.1 Agriculture has reduced rural poverty in Mozambique, particularly through land area expansion, crop diversification, and market participation. The scale o f production i s still small. Crop yields are generally low. And women, asset-poor, and uneducated farmers participate less in markets and grow less profitable crops. Vulnerability and the inability to cope with crop production shocks have proven severe, particularly among households already worse off. Income diversification into nonfarm activities and wage labor also helped households get out o f poverty or maintain their welfare, 3.2 Mozambique needs to promote a diverse and diversified agricultural sector, strengthen markets and coordination mechanisms, and support rural off-fm employment creation. A diverse and diversified agricultural sector can increase agricultural productivity and production efficiency to maximize the benefits to the rural poor. Agricultural markets and coordination mechanisms in value chains can reduce uncertainty and increase the benefits to smallholders and large producers alike. More employment opportunities can sustain a strong off-farm economy in rural areas to help households deal with the vulnerability associated with crop production. A more diverse and diversified agricultural sector will require research and extension to smallholder subsistence farmers to increase their productivity in basic food crop production-and to commercial smallholders to increase their scale o f operation, diversify their crop portfolio, and increase their farm yields. A large scale farming sector needs to be promoted in a way that it generates employment and other linkages (such as technology) to rural smallholders-and not crowd them out. Productive infrastructure, such as irrigation and a good investment climate, including flexibility in accessing and securing land tenure and improving access to finance, will be crucial inthis process. A more active and efficient agricultural marketing system will require marketing infrastructure and institutions and services that provide market information, establish grades and standards, manage risk, and enforce contracts. Contract farming in key domestic food chains (links to supermarkets) and export-oriented high-value supply chains will be fundamental. Developing and empowering farmer groups will be key to reducing transaction costs and promoting a more balanced and competitive system. A more diversiJied activity portfolio off thefarm will help households cope with shocks in crop production and prices. To maximize the benefits o f the off-farm economy, both farm and nonfarm wage labor and the self-employment microenterprise sector, public works programs can generate local wage employment inthe short run. A better investment climate can attract private sector investment in off-farm activities. And schooling and training can improve the access o f households to more profitable opportunities. Focused value chain studies 61 can identify opportunities to support agroprocessing industries and services with strong links to agriculture to maximize the effects on rural households. AGRICULTURALGROWTH POVERTYREDUCTION, 1996-2002 AND 3.3 Mozambican agriculture has been growing very rapidly, but it is vulnerable to climatic shocks. Starting from a very low base, agricultural growth was high in the immediate postwar period and throughout the 1990s. Mozambique has one of the highest agricultural GDP growth rates in Southern Africa.93 Since the peace accords, agriculture has grown rapidly because farmers returned to their land, and markets started to open. Average annual GDP growth for the agricultural, livestock, and forestry sectors was 6.2 percent between 1992 and 2003 (figure 3.1). Growth o f maize production, Mozambique's predominant food crop and a proxy for overall agricultural growth, confirms these production trends. By 1999 maize was cultivated by about 79 percent o f rural households and occupied about 35 percent o f total planted area.94 The rural economy has suffered heavily from droughts in 1991 and 1994 and floods in 2000, events followed by an immediate economic recovery. Figure 3.1. Growth of Agricultural GDP, 1992-2003 (percent) 30 1 20 10 : I 0 1992 1993 1994 1995 1996 1997 1998 -10 -20 -30 1 Source. World Bank 2006a, based on National Accounts data from NE. Note: Agriculture includesthe livestock and fishely sectors. 3.4 Production of basic food crops is the key driver for higher agricultural production, but export-oriented production of cash crops is also expanding. Basic food crop production grew at an average o f 3-4 percent a year. Maize and millet production showed the largest increases, followed by sorghum, beans, rice, and cassava. Production 93World Bank World DevelopmentIndicatorsfor various years. 94Agricultural Census 2000 data. 62 increases were highest in the central regions, particularly in Tete province.95The number o f households producing and total cash crop production for exports also increased. Particularly important are sugar cane (for larger enterprises), cashews, copra, cotton, and tobacco. International Fund for Agricultural Development estimates indicate that from 2000 to 2002 sugar cane production (predominantly a plantation crop) more than doubled.96 Tobacco production (predominantly under contract farming arrangements) increased, from less than 2,000 tons in 1996/97 to more than 50,000 tons in 2003/04, and the number o f farmers increased from fewer than 10,000 to more than 100,000. By contrast, cotton production, higher in 2002, has been cyclical, fluctuating with prices and demand in world markets and partly due to issues related to the management o f contract farming arrangements between agroindustrial exporting firms and smallholders. 3.5 Crop production was considerably more diversified in 2002 than in 1996-a key factor for growth in crop income and reductions in household vulnerability. The mean number o f crops grown per household increased from less than five in 1996 to almost eight in 2002, and the number o f crops grown by households inthe lowest quintile more than doubled." Richer households typically grow more crops o f all types than do poorer households. Food crops are the most numerous, followed by tree crops and horticultural crops. Households in the top two income quintiles are three times more likely to have a cash crop than those inthe lowest quintiles. Sustainingthat diversification and increasing access to more profitable crops for poor households are important for future poverty reduction driven by agricultural growth. Within the food crop group the largest increase in crops grown was in the pulses group (beans and groundnuts), followed by roots and tubers (cassava, sweet potato, and Irishpotato). Overall, the mean number o f crop groups grown increased from three to four per household, another indicator o f diversity. 3.6 Extensive cultivation has been the main source of agricultural growth. Much o f the growth in crop income has been through extensive agricultural practices, rather than through better seeds and chemical inputs. The area under cultivation grew by 3.3 percent a year between 1993 and 2001, and 3.9 percent a year between 2000 and 2003. The expansion is predominantly in the smallholder sector, with commercial farming shrinking, except in the southern region.98But between 1996 and 2002 there was only a modest increase in households adopting productivity-enhancing technologies. For instance, use o f fertilizer (manure or chemical) increased from 4 percent to 10 percent o f households (chemical inputs alone increased from 1 percent to 4 percent), animal traction from 7 percent to 11 percent, and small-scale irrigation from 4 percent to 11 percent. These technologies are used almost exclusively in cash crops, grown mostly under contracts with agroindustrial exporting firms. As a result, yields stagnated for basic food crops widely grown by smallholder farmers, predominantly for subsistence. In general, yields for many crops are far below potential (table 3.1), and Mozambique i s lagging behind most o f its neighbors. Extensive cultivation i s no longer a sustainable path. 95 See appendix table A3.3. 96 IFAD 2005. 97See appendix table A3.4. 98See appendix table A3.5. 63 Table 3.1. EstimatedActual and PotentialCrop Yields, 1998 Average Potential Yield Crop Average Actual Yield (t/ha) (t/ha) Maize 0.9 5.0-6.5 Sorghum 0.4 0.8-2 .O Rice 1.o 2.5-6.0 Beans 0.5 0.5-2.5 Cassava 6.0 5.0-10.0 Cotton 0.5 1.o-2.0 Source: Howard andothers 1998. 3.7 The area cultivated by smallholders is still too small to achieve adequate scale, but expansion is limited by practices and technologies. Average area sizes are still small, and plenty o f room exists to expand at the farm level to improve rural household welfare. Regression analyses on the determinants o f household income have noted the importance o f crop diversification ( enerally associated with land-area expansion) and productivity- enhancing technologie~!~Agricultural land's short-run elasticity on output growth is estimated at 0.8, and its long-run elasticity at 0.4.'0° Smallholder landholdings are positively correlated with net sales (and smallholder welfare) in key crops in rural Mozambique. The bottom line: while land area expansion i s still important to achieve the needed scale inproduction, household ability to expand that area may be limitedby scarce labor. But to deploy more agricultural labor is uneconomic for households, given the low returns. Gains in crop yields at the farm level, through productivity-enhancing technologies, can sustain agricultural growth and reduce rural poverty. Agricultural Markets and Household Welfare 3.8 Although still weak, market participation and spatial integration have improved, and crop commercialization hasgrown. Partly as a result o f improved security and road network conditions, domestic market integration has improved significantly, domestic trade is growing, and prices are converging across subregions.lo' The construction o f the bridge over the Zambezi River is expected to bring new dynamism. Agricultural growth in border provinces, such as Niassa and Tete, have benefited from increased crossborder trade with Malawi, Zambia, and Zimbabwe. Maize is the dominant crop inthat trade, with small quantities o f other food crops. 3.9 Participation in agricultural markets is importantfor rural household welfare, but the magnitude of the differences between market participants and nonparticipants varies by crop. Overall market participation increased from 1996 to 2002, and the magnitude o f the income gain depended on the type o f crop. Household income rises as households move from selling surplus food (usually maize) in the spot market to selling 99Walker and others 2004; Benfica 2006; Benfica, Tschirley, andBoughton2006. looWorld Bank 2006a. lo'See appendix table A3.7. 64 an undifferentiated contract cash crop (cotton), to selling quality-differentiated cash contract crop (tobacco).'02 3.10 Market participation and household success in key crop markets are correlated with education, gender, and the ownership of private assets. As expected, wealth and education are especially correlated with higher value contract and quality-differentiated crops. Female-headed households are significantly disadvantaged in both participation and earnings from sales in all markets because o f lower education levels and less labor available inthe household. Changesin the Structure of Rural HouseholdIncome 3.11 Rural household income grew for all regions and household groups, with top- and bottom-quintile households benefiting the most andfemale-headed households still lagging behind. During 1996-2002 real rural household incomes-the net value o f crop production and sales, earnings from nonfarm self-employment, and wage income-grew an average o f 8.7 percent a year. Growth rates were highly differentiated across groups and regions o f the country, but the strength o f income growth among lowest and highest quintile households shows some propoor rural income growth. Female-headed households remain concentrated in the lower quintiles, with an average growth rate significantly below that o f male-headed households. lo3 Table 3.2. Sourcesof Growth in RuralHouseholdIncomeby Quintile, 1996-2002 (percent) Change in Mean Total Income Attributable to Each Source Quintile of net Annual (I9962002) household Growth Rate incomeper in Income Crop Livestock Wage Nonfarm adult equivalent (I9962002) Income Sales Income Enterprise Total Lowest 6.5 77.2 5.O -3 .O 20.0 100.0 2nd 2.9 77.0 6.0 5.0~ 12.0 100.0 3rd 2.8 79.2 7.9 13.9 -1 .o 100.0 4th 4.1 39.4 9.1 38.4 13.1 100.0 Highest 12.4 -8.0 4.0 55.0 49.0 100.0 All households 8.7 9.9 5 .O 46.5 38.6 100.0 Source: TIA data for 2002 andBoughtonandothers (2006). Note: Change is the percentageoftotal change attributable to the source, calculated as a share ofeach source change in the total change. 3.12 The composition of income changed as crop income became less important, though remaining key for rural incomegrowth andpoverty reduction, especiallyfor the poorest groups. While crop income (drawn by virtually all households) remained the single most important source for all groups over the period, its importance is falling for lo'Boughtonandothers 2006. See appendix tables 3.8-3.10. IO3See appendix table A3.1. Data for this analysis comes from the National HouseholdIncome Survey (TIA) for 1996 and 2002. Details aboutthese surveys and key conceptsfor the analysis are presentedin appendix B. 65 households in all income groups and across all regions. Crop income's high share in total income (despite its slower growth), explains much o f the increase in mean income (nearly 80 percent) for the poorest groups (quintiles 1-3). For the highest quintile, crop income actually fell, and growth was motivated by a strong increase in wage income (accounting for 55 percent) and nonfarm enterprise income (accounting for 49 percent). Nonfarm enterprise income growth accounted for 20 percent of growth in the poorest income quintile, again illustrating the importance of nonfarm businesses in rural income growth and poverty reduction (table 3.2). RURAL POVERTY DYNAMICS AND THE ROLEOF DIVERSIFICATION,2002-05 3.13 Rural income growth has continued to be strong since 2002-but has not been propoor-and inequality increased. While mean rural incomes continued to grow, incomes in the lower quintiles did not improve, with incomes actually falling between 2002 and 2005. Over the same period, the third quintile experienced limited growth, but average growth inthe top two quintiles continued to be strong (figure 3.2).'04 Figure3.2. Average Annual Growthof Rural HouseholdIncome per Adult Equivalent,by Incomeper Adult EquivalentQuintile,199602 and 2002-05 1 0 10.0 c -E;5,2 5.0 J 0.0 S Q ) E 2 1996-2002 -5.0 0 rn E5 -100 4 -15.0 Periods and rural income per adult equivalent quintiles Source: TIA data for 1996, 2002, and 2005. 3.14 The poorest households lost out because the grain harvest was poor in 2005, especially in the central provinces, and they were unable to diversify out of food crops into other income sources. Diversification o f the household portfolio continued through 2005, with crop income remaining important but its share falling. While crop income remained the single most important source for all groups over the period, the process o f rural and agricultural transformation continued. Richer groups moved into nonfarm sources, such as livestock, wage labor, and nonfarm enterprise income. But households in the lowest quintile still earned 80 percent oftheir incomes from crop sales and subsistence Note that about two-thirds ofpeople are below the incomepoverty line. This implies that those inthe third and some inthe fourth quintilesare also poor butstill experiencedgrowth. Theendresultwas overall incomegrowth anda modestreductioninthe poverty headcountowing to those near the poverty line moving over the line (see top portionof table 3.3). 66 agriculture. This meant that with a relatively bad harvest in grains in2005 due to irregular rainfall, the households inthe lowest quintile, who depend much more heavily on income from crop production, were hurt most and the income gap increased. Chronic Poverty and Vulnerability in Rural Mozambique 3.15 Dynamic analysis, using datafrom a panel of households, helps understand the recent widening in rural inequality. This section uses data from a special 2002-05 rural household panel,lo5 and dynamic approaches to provide insights into household movements around a poverty line and take into account the effect o f time on household wealth, income sources, decisions, and strategies. Analysis o f individual household outcomes and their correlates provides more robust information on how mobility works in practice in rural Mozambique. Through this analysis, we can see who is actually left behind when a bad harvest occurs-and why.lo6 3.16 Vulnerability to shocks makes rural poverty persistent or chronic. The effects o f the fall in rural incomes among households show up strongly in the analysis o f rural poverty transitions. That has particularly affected households with less diversified incomes-that is, more dependent on rainfed agriculture and having fewer or less profitable off-farm sources. More than half o f all households (two-thirds o f the poor) remained poor between 2002 and 2005, and only a limited proportion o f households (15 percent) managed to stay out o f poverty in both periods (see lower part o f table 3.3). The proportion o f households that fell into and came out o f poverty i s almost the same. The odds o f households poor in 2002 becoming nonpoor in 2005 were 24 percent-against 49 percent falling into poverty (likelihood o f nonpoor in 2002 becoming poor in 2005). The high probability of people poor in 2002 remaining poor in 2005 indicates persistence in poverty in rural Mozambique, particularly in the face o f shocks, such as irregular rainfall as in 2005. Poverty transition status by region and province reveals a very common pattern o f tran~itivity."~An analysis o f reported days o f drought across the two datapoints indicates that the situation was relatively bad all over the country in 2005. The southern region, typically drought-prone, experienced more drought days than usual. In the north, where droughts are rare, the number o f drought days more than quadrupled over the period. The degree o f the drought in the central areas differed less across the two years, but its severity stood somewhere in between that o f the south and north. In spite o f this adversity, overall, the rural income poverty headcount rate in rural areas declined by three percentage points (see top part o f table 3.3). I O 5For details on the panelsurvey andthe designof poverty lines for the poverty dynamics analysis, see appendix B. IO6 Correlatesof poverty status have beenfound to be distinct from the dynamic processesthat cause householdsto fall into or escape poverty. lo' appendixtable See A3.13. 67 Table 3.3. Rural Income PovertyIncidenceand Transition Matrix, 2002 and 2005 (percent) Household Poverty Status Group Source: TIA data for 2002 and 2005. Note: The numbers inparenthesisare transition probabilitiesgiven the initial status. 3.17 The lack of upward mobility suggests highly persistent and severe rural poverty, but also significant upward and downward income movement within the group of poor households. To gain a better sense o f the relative stability o f households above and below the poverty line and how they move over time, an alternative transition matrix with four poverty status categories is used: extremely poor (incomes less than half the poverty line), poor (incomes above half the poverty line, but still below it), nonpoor (incomes above the poverty line, but below 1.5 times the poverty line), and wealthy (incomes over 1.5 times the poverty line). O f households considered extremely 'poor in 2002, most remained extremely poor in 2005 (56 percent); o f households considered poor in 2002, 41 percent became extremely poor in 2005 (table 3.4). Among the worst-off households, poverty i s persistent, with a high likelihood o f the situation worsening, particularly in 2005, when a weather shock occurred. A breakdown o f the analysis by region starts to reveal how bad the situation was in the north, where these events are less common. The north exhibits the highest rates o f immobility among the extremely poor- and downward mobility by the poor, nonpoor, and wealthy into extreme poverty."* '''Seeappendix table A3.14b. 68 Nonpoor 33.0 I 24.1 I 14.4 I 28.5 I 100.0 Wealthy 22.6 I 20.2 I 15.9 I 41.4 I 100.0 3.18 Although moving out of poverty was quite difficult under the circumstances facing households in thisperiod, those who escaped appear to do it relatively well. As expected, more "relatively less poor households" managed to escape poverty than the extremely poor. But most extremely poor and poor households that did escape poverty moved straight into the wealthy ~ategory.'~'While entering poverty was very likely among wealthy households, which reveals some instability and inability to sustain high welfare standards under adverse circumstances, an equal proportion o f households maintained their wealthy status in all regions. 3.19 Seasonality is an important part of the risks and shocks that ordinary households face in Mozambique. The seasonality o f vulnerability highlights important variations inrisks and shocks-and the ability of communities and households to respond. A recent quantitative assessment o f changes in food calorie intake over the agricultural year found significant regional differences in seasonality.' lo Households in the south seem to exhibit less seasonality in their calorie intake than those in the central and northern regions. Households on average are fairly able to smooth their calorie intake throughout the year. But the seasonality o f poverty and vulnerability imposes stress and hardship on many poor households. Participatory analysis, especially from the northern region, displayed particularly large variations in well-being throughout the year. It also showed fairly large variations among locales in the same region in the severity and timing o f seasonal fluctuations (figures 3.3 and 3.4 and box 3.1). logThe disaggregatedanalysis ofthe povertytransitions illustratedhere brings important insights. However,to better see the distribution and correlates inthe analysisof povertytransitions with other variables, the focus here will be on the simplest classification (poor and nonpoor) andthe resultingtransition status. Arndt andothers 2006, usingIAF data for 2003. 69 Figure 3.3. SeasonalVariation in Well-being and Vulnerability in Angoche, 2006 12 1 0 Months Source: Authors' calculations basedon data from PVS 2006. Note; Perceptionof well-being scale from 0-10: 0 i s very badly off, and 10 i s very well off. Sample not representative. Figure 3.4. SeasonalVariation in Well-being and Vulnerability in Ch6kw6,2006 lo 1 c0 4 - 2 - Months Source: Authors' calculations basedon data from PVS 2006. Note: Perceptionof well-being scale from 0-10: 0 is very badly off, and 10 is very well off. Sample not representative. 70 Box 3.1. The Seasonality of Vulnerability in Two Localitiesin Mozambique Bairro 1:Inguri B, Municipio ofAngoche, Nampula Angoche, a small town of previous grandeur, used to be an important fishing town with several processing plants (for cashews, rice, and so on). All but one were shut down in the 1990s. The closure of these plants led to a rapid rise inunemployment and underemploymentand to the near collapse of the local economy. Most households resorted to artisanal fishing, informal and petty trade, and subsistence agriculture. A few fortunate people remain employed in the formal sector-in public institutions, stores, and a small, recently opened shrimp-processingplant. The seasonality of vulnerability has been particularly strong inAngoche since the closures of the parastatal factories. Well-being normally begins to deteriorate in January and February. March and April are the most critical months. The supply of fish is at its lowest, and prices of staple goods-particularly dried cassava-are at their highest. Decemberthrough March coincides with the shrimp protection period, when shrimp trawling and fishing are not allowed. And the beginning of the school year strains household budgets with costs for matriculation (for secondary school), school materials, and uniforms. This is also when households normally suffer the most from illhealthdue to malaria, diarrhea, and even cholerainbadyears. InMay and June conditions improve. Fish, shrimp, and scallopsare abundant, both for eating and selling. And agricultural produce (fresh and dried cassava, maize, mapira, and rice) can be bought at fairly low prices. Well-being remains high into July and August, with a boost from good fishing conditions. Conditions start to decline again in November and December-the result of bad fishing conditions, the onset of shrimp protection, and higher prices for agricultural products.` Bairro 2: Xilembene, Chbkw2, Gaza Xilembene is a peri-urban town outside Ch6kw6. Central to the agricultural engine of Mozambique during colonial times, the town has a richhistory, knownacross Mozambique as the birthplace of Samora Machel, the country's first president. Until recently, it benefited from an extensive and sophisticated system of irrigation canals that stretched across much of southern Mozambique. But the region's infrastructure for agricultural support-which previously included several processingplants, an agricultural seedproduction factory, and several credit institutions- gradually fell apart due to a lack of maintenanceand investment. Irrigation has been in a state of permanent disrepair since the floods of 2000, leaving the fields of hundreds of households without easy and regular access to water. The gradual collapse of agricultural production and marketing devastated local subsistence farmers. Even so, households in the second bairro of Xilembene appear to have smoother and less severe changes in well-being during the year than those inurbanAngoche. September means the end of the really good times. The food situation becomes severe from late October until February or March, with most households eating fewer than two meals a day. The most vulnerable-orphans and female-headed households-rarely get more than one meal a day. This period also coincides with the warm and rainy season, bringing malaria and diarrhea. January burdens many households with schooling expenses. The winter months from April through July are times of abundance, with agricultural harvests bringing enough food for most people. Ingoodyears production is sufficient to sell the excessa2 1.UEM2006,467. 2. Ibid.. 118-9. 71 Rural Poverty Transitions Consistent with Perceptions of Welfare Change 3.20 Most households-even those who escaped poverty-view the period 2002-05 negatively. The TIA survey used here included questions on household perceptions o f welfare changes, similar to those included in the PVS study collected for the report. The results for 2005 are quite similar to our results for 2006-a negative perception o f change. The trend in household perception o f welfare changes is roughly consistent with income poverty transitions, but in each case households were much more likely to view their status negatively than the objective income data showed. While the majority of households in each group indicated that living conditions deteriorated over the period, this response was most frequent among the households that remained poor and that became poor. Likewise, the proportion o f households reporting improvements in living conditions was highest among those that escaped poverty and that maintained nonpoor status. About a third o f households in each group reported no change inwelfare (figure 3.5). 3.21 Households do seem to be correctlyperceiving the slowdown in income growth for the lower quintiles and the higher risk of poverty. In 2002, 30 percent o f rural households reported perceived improvements inwelfare over the last three years, but only 18 percent did in 2002, while the proportion reporting worsened conditions increased from 39 percent to 50 percent."' Figure3.5. Rural PovertyTransitionStatus and Welfare Perceptions in 2005 60 u) 50 r8 40 r r- 30 IC 0 er 20 10 n 0 1 Stayedpoor Becamepoor Escaped poverty Stayed nonpoor Povertytransition status (2002-05) lrrproved rn Did not change 0 WorsenedI Source: TIA data for 2002 and 2005. TheRole of Agricultural Markets and Off-farm Diversification 3-22 How successfully households gained and sustained diversified income sources determined their poverty transition status in the period. Among households that were initially poor (and mostly hit with the weather shock), those able to participate in food markets and gain access to nonfarm enterprise income were more likely to escape poverty. ' ' ISee appendix table A3.15. 72 Failing to sustain nonfarm enterprise activity, and more importantly wage income, reduces the probability o f escaping poverty. Among households initially nonpoor, most o f which reached that status following successful income diversification in the past, closing down nonfarm enterprises and losing wage income sources increased significantly their probability o f falling into poverty.' l2 3-23 Households that escaped poverty or maintained nonpoor status were more likely to enter ruralfood and cash crop markets and by 2005 had a higher probability of owning key assets and having access to information extension and market information. Overall, households that escaped poverty and those that remained nonpoor over the period are more likely than those that stayed poor or entered poverty to have nonagricultural assets, such as radios and bicycles, and access to extension, market information, and improved storage.' l3Increasing participation in food and cash crop markets through more profitable marketing operations is important for households to escape poverty. More households maintained or increased their participation in food and cash crop markets among those that escaped poverty than among those that remained poor.' l4And growth in returns in those activities were also significantly higher for these households, though in absolute terms households that retained nonpoor status had the highest returns in both food and cash crop markets at the end o f the period.'" 3.24 A change in household income sources had a big effect on poverty status. Finding and maintaining off-farm sources of income, such as wage and nonfarm self- employment, help households escapepoverty, and notfall into it. Among the households that escaped poverty the proportion with wage income and nonfarm businesses increased significant1 and households with those as the major sources o f income almost tripled (table 3.QZ6 Many households reached nonpoor status due to successful involvement in nonfarm self-employment and wage jobs. Nonpoor households that keep those sources of income are much less likely to fall into poverty. Among households that kept their nonpoor status, 21 percent maintained their wage labor income sources and 37 percent maintained their nonfarm self-employment sources. A smaller proportion lost them. Likewise, the highest proportion o f households that remained nonpoor never had access to these sources o f income, and among those exitin poverty, a great deal had access to wage or markets or maintained a nonfarm enterprise. 118 'I2Here we estimated separatelytwo probit models for two subsets of the panel sample.(1) Probit estimationfor escapingpoverty that uses householdsthat were poor in 2002. The dependentvariable is 1 ifthe householdescaped poverty in2005,O otherwise. (2) Probit estimation for entering povertythat uses householdsthat were nonpoor in 2002. The dependentvariable is 1 ifthe householdfell into poverty in2005,O otherwise. For detailedregression results, see appendix tab1A3.16. 'I3See appendix table A3.17. 'I4See appendix table A3.18. 'I5See appendix table A3.19. 'I6See appendix tables A3.21 and A3.22. 'I7See appendix table A3.20. 13 Table 3.5. RuralPovertyDynamics and Off-farm Income Diversification,2002-05 Poverty Transition Status Off-farm income diversification Stayed Became Escaped Stayed All dynamics Poor Poor Poverty Nonpoor Households Wage income status Never hadwage income 66.1 58.9 58.6 57.3 62.4 Lostwage income 7.4 13.8 3.4 6.5 7.5 Enteredwage market 21.0 17.5 30.3 15.5 21.3 Maintainedwage income 5.5 9.7 7.7 20.8 8.8 100.0 100.0 100.0 100.0 100.0 Nonfarm MSE income status NeverhadMSE income 39.9 30.0 27.6 25.3 34.1 LostMSE income 14.8 27.2 8.6 16.0 15.7 EnteredMSE activity 23.2 18.1 34.1 21.1 24.1 MaintainedMSEincome 23.2 24.6 29.7 37.5 26.1 Source: TIA data for 2002-05. MSE:Micro or small enterprise. 3.25 Returns to nonagricultural wage labor are significantly higher than those to agricultural wage labor. Gaining and sustaining access to those opportunities is crucial in determining household poverty status. Most poor households did not diversify into nonagricultural wage jobs, and those that did moved into unspecialized categories. Average returns to households that remained in poverty were much lower, while the returns to households that escaped poverty increased, especially for specialized nonfarm wage labor. Sustained participation and aggregate wage levels in nonfarm wage- employment opportunities determined how well nonpoor households fared. In fact, a slightly higher proportiono f households that did not fall into poverty engaged in nonfarm wage labor opportunities, particularly in the specialized categories. And participation among households that fell into poverty decreased from 17 percent to 10 percent. While average returns among households that fell into poverty dropped, those that maintained their status saw their average returns stagnate-but at a higher level than other groups.' l8 3-26 The proportion of households diversijying into agricultural wage jobs grew, but success in escaping poverty or remaining nonpoor was defined by the rates of diversification into, and the profitability of, specific types of agricultural wage opportunities. The proportion o f households with agricultural wage jobs more than doubled, but aggregate average returns fell, which suggests that some households did not benefit as much as they expected and that poverty outcomes differed greatly. In both periods households that did not escape poverty were more likely to have an agricultural wage job than were those that escaped poverty. The fundamental difference, however, was that those households that did not escape poverty were predominantly engaged in unskilled low-paying wage jobs on small farms, while those that did were engaged more inequally unskilledjobs (buton larger farms) and inskilledhigherpayingjobs. Likewise, householdsthat fell into poverty were mostly engaged in small-scale unskilled low-paying farmjobs."' See appendix table A3.23. ' I 9Ibid. 74 3.27 Diversification into nonfarm enterprises in natural resources extraction Cforestryproducts, wildlijie, and fshing) was important in helping households escape poverty. Overall, nonfarm businesses in natural resources extraction require little capital and skills. Poor households that escaped poverty showed the highest increase in these types o f activities, particularly in firewood collection and grass and cane products. But diversification rates into natural resources extraction activities among nonpoor households that did not fall into poverty hardly differ from those o f other groups and did not change much. This suggests that these enterprises were not as important in reducing the likelihood of households falling into poverty. Diversification into manufacturing and trading appears to have been effective in helping better-off households sustain their level o f welfare. 3.28 Nonfarm self-employment in manufacturing and services was importantfor all groups, but success depended greatly on the types of businesses targeted. For instance, trade in alcoholic beverages and in nonfood products was important in taking households out o f poverty and in preventing others from falling into it.'20The proportion o f households that became poor engaged in beverage trading fell from 6.8 percent to 4.8 percent, while the proportion o f households that escaped poverty rose from 5.2 percent to 13 percent. Likewise, for households that maintained nonpoor status (the group that already had the highest proportion in this business in 2002), the proportion rose from 8 percent to 15 percent. Households that remained poor had the lowest diversification. The same patterns, though smaller in magnitude, are observed for trade in nonfood products and some services (repairing bicycles and the like) and inconstruction. THEWAY FORWARD SUSTAININGRURAL FOR POVERTY REDUCTION 3.29 The evidence outlined in previous sectionspoints to the key role of agriculture in rural poverty reduction in Mozambique in the past decade. In the light o f its continued predominance as a livelihood for rural households, progress in poverty reduction in Mozambique will continue to depend on the performance in this sector. Evidence has also indicated that income diversification into off-farm self-employment activities and wage labor was important in increasing income levels and ensuring that households stay out o f poverty. So, to sustain reduction in rural poverty, it will be important to: Raise a diverse agricultural sector with higher productivity and production efficiency to maximize the benefits to rural households in terms o f better food security and higher crop incomes. Enhance agricultural markets and coordination mechanisms in value chains to benefitfrom economic linkages. I2'See appendix table A3.24. 75 Support the generation o f wage-employment opportunities to sustain a strong off- farm economy in rural areas to help households deal with the vulnerability associated with productionagriculture. Supporting a Diverse and Diversified Agricultural Sector 3.30 Supporting a more diverse agricultural sector will require investments inresearch and extension to help smallholder subsistence farmers increase their productivity in basic food crop production, commercial smallholders increase and diversify their crop portfolio and farm yields, and a large-scale farming sector generate employment and other links to rural smallholders. 3.3 1 Rural propoor growth will comefrom increasing agricultural productivity and diversifying crop production in the smallholder sector. Because crop income remains important as a source o f income for the poor, sustainable increases in this source will still be important to achieve propoor growth. And because the majority o f households are net food buyers, increases in crop income must originate from productivity gains in food crops and not only from price increases that reflect local scarcity. Productivity gains need to be reflected not only in increased yields per hectare but also in output per worker, from improved seeds and chemical inputs and from increased mechanization. 3.32 Crop income can also be increased by having rural households diversify into profitable cash crops, particularly because productivity-enhancing inputs are generally associated with such crops. Households growing those crops tend to apply these inputs in food crops or fallow techniques. Private investment in cash cropping for exports brings technologies that can transform the agricultural sector. And the new technologies can spill over to noncash crop growers, raising incomes inthe whole community. 3.33 Encouraging large-scale, commercialized agriculture can bring benefits to smallholder farmers, especially if it strengthens links between the two groups and does not crowd-out smallholders. A successful agricultural strategy will rely on the private sector for needed investments and know-how. Commercial agriculture brings new technologies, which can spread to all farmers in the area, large and small. Commercial farmers also subsidize infrastructure development, provide wage employment, and increase demand for products o f the local non-farm sector. 3.34 Contract farming brings improved seeds and production techniques to the farmers, provides input finance, and reduces the farmer's risk from greater specialization and uncertain markets andprices. The tobacco sector inMozambique has shown that private extension services under these arrangements have a highrate o f return. Note, however, that contract farming requires an appropriate regulatory framework that strikes a balance between protecting the farmer from a monopsonist while ensuring returnsto the contracting company commensurate with the risk. 76 Supporting Market DevelopmentandNew InstitutionalArrangementsfor Stronger Market Participation 3.35 To enhance the efficiency o f agricultural markets, it will be important to invest in marketing infrastructure and rural services, improve access to information, and improve coordination mechanisms for key domestic food chains and export-oriented high-value supply chains. 3.36 Continuing investment in road infrastructure is important to open markets to commercial smallholders and to allow new investments to prosper, but complementary investments are also needed. With Mozambique's low population density, road infrastructure i s important to the success o f commercial agriculture because underdeveloped transport and rural infrastructure services increase marketing costs substantially and prevent domestic and export markets from developing. Inaddition to the main national roads, a well targeted program should focus on maintaining secondary and tertiary roads throughout the agricultural season. 3.37 Poor road access in many cases coincides with other bottlenecks, such as poor agroecology, low population density, weak services, and a lack o f water and electricity. So, a well targeted program needs to take into account the interactions with geographic, community, and household characteristics and consider complementary investments to ease those other constraints and maximize the impacts o f road construction and rehabilitation for market efficiency and household welfare. And funding and institutional arrangements (community-based maintenance contracts) need to be put in place to ensure the sustainability o f road investments and maximize its long-termimpact. 3.3 8 Access to market information and intelligence can improve the competitiveness in the agricultural marketing systems, ensuringthat farmers participate in a more leveled playing field inagricultural markets. Mozambique has had a public marketing information system in place since early 1990. The system has evolved in response to user needs, but sustainability has been a constant challenge. Decentralization to respond to the needs o f users in regional spots has been taking place and the use o f technology i s progressing. The system should be supported to provide timely information on prices for different transaction levels, distribution channels, competitors, and buyer and producer trends. 3.39 To help farmers meet the required standards in high-value chains linked to domestic and regional and international markets, the market information systems should alsoprovide and disseminate information on regulations, grades and standards, post-harvest handling advice, and storage and transport recommendations. Public- private partnerships may be needed to accomplish this. The continuing exploration and use o f the Internet and the fast-growing cell phone network in the country will aid in this process. 3.40 Stronger coordination mechanisms, including those for contract farming in key domesticfood chains and export-oriented high-value supply chains, can improve smallholderparticipation in thesemarkets. Their participation has so far been limitedby their inability to meet quality standards, timeliness requirements, and economies o f scale. Failures in credit and inputmarkets complicate the problem. 77 3.4 1 To support smallholder market participation, mutually beneficial contractual links should be established between smallholders and large traders or processors in selected value chains. Policy interventions in the agribusiness sector need to strengthen links between smallholders and the emerging supermarkets (and other traders) in urban areas, and processors and traders in export-oriented value chains. The effects will be maximized if policies are oriented towards improved efficiency and productivity o f the entire value chain as opposed to focusing only on the crop production node o f the system. To reduce the typically high transaction costs associated with contracting arrangements (and potentially rent-seeking behavior by agroindustrial firms and traders), it i s recommended that farmer associations are developed and supported to generate more balanced outcomes. SupportingIncome Diversification through Wage-Employment Opportunities 3.42 Increased and more profiiable rural wage and self-employment opportunities can help households improve and sustain their welfare levels. Although full-time formal nonfarmwage-employment opportunities may not be widely available to most households inthe near future (given the limitedavailability and growth of formal sector jobs in rural areas), income diversification into seasonal or part-time wage labor will remain crucial for reducing rural poverty. Mozambique may also wish to consider rural public works schemes to expand opportunities for the poorest farmers who cannot break out o f subsistence. Past diversification of household income came from pull factors-from a dynamic rural economy. But there will also be smallholders who cannot modernize or commercialize, and who cannot develop or stay profitable in off-farm activities. They have few assets and may lack adults in the household. Providing income-earning opportunities through community-driven public works programs could be an effective safety net for them. 3.43 Employment linkages in thefarm sector are important in generatingfarm wage income to households, but its low returns and the seasonal nature limit the extent to which it can sustain ruralpoverty reduction. Opportunities for farm wage labor can be expanded as a result o f the expanded activity o f larger farming households, such as commercial smallholders, or large private agricultural producers capable o f generating employment for their poorer neighbors. While this i s important, especially for the asset poor, wages inagriculture tend to be low and seasonal, particularly for low-skilled labor working on small-scale farms. This points to the importance o f supporting a more diverse and diversified agricultural sector o f both commercial smallholders and well managed large-scale farms. 78 4. GOVERNMENT EXPENDITUREPOLICIESAND THE POOR 4.1 A key public policy goal in Mozambique is improving access to socialservices for thepoor and excluded,especially those in remote areas. Many o f these services are publicly provided, and all are at least part publicly financed. So, the major government policy action toward this goal has been to increase public expenditures on service delivery. Donors have supported this, through direct financing o f investment projects and through budget support for current expenditures. The result has been increasedresources for key publicly provided services, such as health and education, in both real terms and in the share ofpublic spending. More recently, public policy has focused onthe efficiency o f resource use and the quality o f service provision. This expenditure policy underlay the first national poverty reduction strategy (PARPA I), identified four priority areas which for public expenditure: education, health, basic infrastructure, and agricultural and rural development. The 2006 revised strategy (PARPA 11) increasedthe scope o f targeted expenditure to add HIV/AIDS,governance andjudicial systems, and other such priorities as social action and labor and employment. 4.2 Public spending on service delivery in Mozambique has been growing rapidly, and thegovernment has been effective in allocating resources topriority areas, and ofen to rural and underservedareas. But there have been problems o f weak capacity and poor infrastructure. And results have differed across sectors, with education showing outstanding achievements, water fairly good progress, and health and extension the least. 4.3 I n educationgovernment has devotedsubstantial expendituresto increasing education accessfor underservedareas and excludedpopulations. Progress has been impressivefor the first five grades o fprimary education (EP1) and less so for the next two grades (EP2). There has been a large increase inthe number o f schools and in enrollment, with the enrollment o f girls increasing significantly. Spending on EP1was propoor. But with demand growing faster than infrastructure, and the proportion o f unqualified teachers and pupil-teacher ratios rising, quality is becoming an issue. The secondary education system suffers from poor supply, particularly inrural areas, and large gender gaps inaccess and outcomes. 4.4 I n health the distribution of healthfacilities is skewed toward urban areas, and though there has been an increase in staff, incentivesremain poor. Outreach efforts have expanded preventive care, but poor and rural areas still fall behindin access. In general, poor women and rural residents seek less curative care. The greatest constraint to healthcare use i s distance, especially for poor and rural households. Cash expenditures for health are generally low in absolute terms, but, when compared with the relatively well off, the poor spend a greater share o f their incomes on it. Education is also an important determinant o f healthcare use and child healthcarepractices, especially inrural areas. Rural people, especially women, are less satisfied with the health system. 79 4.5 I n extension, agricultural spending ispoorly oriented toward thepoor. Muchof the spendingthat canbe trackedis targeted at nonpoorbudget itemsand away fromthe provinces,where most ofthe poor live.Agriculturalextensionhas beengivenlittle attention and receivedlimitedfunding, resultinginpoor geographicalcoverage ina countrywith a lowpopulationdensity. 4.6 I n water coveragehas increased,but regional and rural-urban gaps remain wide. Overall, there is roomfor improvementin meetingthe needs o fthe populationby reducingdistance, increasingservice provision, and improvingquality. 4.7 To maximizethe effects ofpublic spendingonthe poor, the government needsto increasepublic spendingon facilities and services usedby the poor andvulnerable, increasevalue for money inthe public sector throughimprovedqualityofpublic services andbetter delivery systems, and improvesystemsto bettermonitor government expenditures and evaluatehow effectivelythey are allocatedto the poor. Increasing public spending infacilities and services used by thepoor requires focusingon spendingonthe basics, such as education, health, and water supply infrastructure,particularlyinruralareas to reduce the distance and financial constraintsfor households.Forhealthand extension, developingand running outreach serviceswill ease those constraints. Increasing the valuefor money in thepublic sector requiresimprovingthe quality of public services throughbettertrainedteachers, healthworkers, and extension personnel.It requirespromotingpublic-privatepartnershipsfor better targeting andmore effectiveresults.Itrequiresensuringthat decentralizationcontributesto effectivetargetingofthe poor.And it requiresidentifyinghowpublic money can improveaccess to services andthe quality of life ofthe poor andvulnerable. Improving systems to better monitor government expenditures and evaluate how effectively they are allocated to thepoor requiresincreasingthe accountabilityfor public spendingthroughstrengthened public financialmanagement systems. It requiresevaluatingresource flows (PublicExpenditureTracking) and spending allocations (Public ExpenditureReviews),and conductingregularandwell plannedhouseholdsurveys. To evaluate strategies and monitorresults,monitoring systems shouldbe basedinprogrambudgetingto link spendingto targets and outcomes. THESTRUCTURE OFPUBLICEXPENDITURE 4.8 Before 2000 Mozambique'spublicfinancial management was characterizedby poor practices across every dimension. Accountingusedantiquatedsystems. Budget coveragewas poor.Expendituresoutcomes were almost impossibleto track. Cash managementwas fragmentedandweak, and auditingineffective.There was little transparency and accountabilityfor the use of resources. One majorreasonwas the 80 government's weak capacity after the country emerged from years o f civil war. Another reason was the inadequate legal basis for financial management, corrected with a new financial management law in 2001 and the phased introduction in2004 o f a new integrated financial management system, SISTAFE. 4.9 Data on expenditures by sector before 1999are unavailable, and data since then have severe deficiencies. Data have been available by functional classification since 1999 (such as wages or inputs), but data by spending unit (such as primary, secondary, or tertiary education) or by program (such as livestock services) are only now emerging. Much donor financing, outside the government financial management system, was thus not tracked or includedin sector spending estimates. This situation is beingcorrected through a concerted effort to include donor programs in government accounts and have donors deliver their fundingthrough the government budget, financing recurrent and capital costs. 4.10 Linkingpolicy goals in the PARPA topublic expenditures,both at the budgeting and executionstages, is criticalfor assessingthe efficiency and effectiveness of government resource allocation and management, but Mozambique is notyet able to do this in most sectors. The main exception is education, where data on expenditures by level o f education have beenavailable since 2002. The severe lack o f data limits the scope o f this analysis, especially for the poor's access. However, the government, along with the World Bank, made a major effort inthe early 2000s to track overall expenditures inthe PARPA areas of emphasis. So, budget transparency has improvedinthese areas, and some analysis is now possible. The government's maintools for tracking the consistency o f expenditures with PARPA promises are annual reviews o f budget outturns by sector. The government has committed ineach PARPA to make sure that expenditures inall PARPA priority sectors are at least 65 percent ofthe budget. 4.11 Government spending has increased at least asfast as GDPsince 1999, implyingpositive real growth. National accounts show government consumption per capita (current expenditures) growing 12 percent a year during 1997-2003. Although some o f this increase included a large increase ingovernment salaries, it also reflects a large increase in inputsdevoted to providingpublicly supplied goods and services. 4.12 Governmentspending in PARPApriority areashas been increasing as a share of totalgovernment expenditures, indicating that a large share of increased governmentfunding over thepast decade hasgone to these sectors. The expenditure in priority areas as a share o f GDP averaged 16 percent between 1999 and 2006, and the share intotal government expendituresaveraged 65.4 percent (table 4.1 and figure 4.1). The sectors most fundedwere education (averaging 20 percent o f total expendituresover the period), infrastructure (16 percent), and health (12 percent). 81 Table 4.1. Sectoral Expenditures as a Percentage of Total Expenditures, 1999-2006 I999 2000 2001 2002 2003 2004 2005 2006 General administration 16.9 15.3 7.2 8.8 13.7 9.2 10.2 19.2 Education 14.4 21.7 21.7 15.9 21.5 21.8 20.4 19.9 Primary/secondary/technical 11.3 18.8 17.2 13.9 18.4 18.6 17.8 16.8 Higher education 3.1 2.9 4.4 2.0 3.2 3.2 2.6 3.1 Health 12.0 14.1 9.7 10.3 11.5 11.9 13.1 15.8 Healtwgeneral 12.0 14.1 9.2 9.5 11.2 11.5 12.1 14.3 HIV/AIDS 0.0 0.0 0.5 0.8 0.4 0.4 1.o 1.4 Inpastructure development 11.9 17.2 10.6 11.9 13.5 11.5 17.2 15.7 Roads 0.0 0.0 2.7 8.8 10.3 9.2 12.4 9.8 Water, sanitation, andpublic works 0.0 0.0 7.8 3.1 3.2 2.3 4.8 5.9 Agriculture and rural development 4.7 6.9 3.1 4.6 4.1 4.5 4.8 3.5 Governance andjudicial system 7.9 8.6 7.2 8.0 9.4 10.1 8.8 9.2 Security and public order 5.8 5.8 4.7 4.9 5.6 5.9 4.9 4.2 Governance 0.5 1.o 1.1 1.4 1.4 1.5 1.5 2.8 Judicial system 1.7 1.8 1.4 1.7 2.4 2.7 2.4 2.3 Energy and mineral resources 3.4 4.4 2.1 2.4 2.9 2.4 2.3 0.4 Social action and labor and employment 1.o 1.7 1.2 1.2 1.2 1.1 0.9 1.1 Social actions 0.6 1.3 0.9 0.9 0.8 0.7 0.6 0.7 Labor and employment 0.4 0.4 0.3 0.3 0.4 0.4 0.4 0.4 Other sectorsa 27.9 10.2 37.2 36.9 22.2 27.5 22.3 15.2 Total expenditureb 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Total expenditure as apercent of GDP 24.1 25.5 31.7 27.5 23.6 22.2 24.6 28.8 Expenditure in PARPA priority areas aspercent of total expenditure 55.2 74.5 61.2 65.3 63.9 65.0 63.0 66.3 Source: MPF 1999-2006. a. This is a group of sectors that includes communications; recreation, culture, and religion; housing and community development; tourism; construction; and nonspecifiedexpenditures. Given the difficulty of systematically reporting on these items individually, they are lumpedinto a single category. b. Total expenditure excluding bank restructuringcosts, net lending, and interest payments. 4.13 Although thegovernment has mostly kept thePARPA total expenditurepledge, focusing on broadpriority areas does not necessarily mean that spending has been propoor. For example, ifmost education spending is allocated to tertiary education, the poor will not benefit because they tend not to be enrolled intertiary education. The next sections attempt a more detailed spending analysis, usinghousehold survey data to map expenditures to the use o f services by the poor and comparing spendingwith the poor's perception o f the effectiveness o f service delivery. The higher quality o f data for education means that the analysis is most complete for this sector. Expenditure data for health are not separated by type o f healthcareprovided and are thus less complete. Other sectors suffer from similar problems. 82 Figure 4.1. Total Expenditure in PARPA PriorityAreas-as a Percent of GDP and Total Expenditure, 1999-2006 *$ c 30.0 1 - 1 20.0 - A v A A - A - A 10.0 7 - 0.0 -1 1999 2000 2001 2002 2003 2004 2005 2006 2007 Years &inpercentofGDP-m- in percent of total expenditure Source: MPF 1999-2006 EDUCATION 4.14 General education in Mozambique hasfour levels, twofor primary and twofor secondary. EP1 covers the first five years o f primary education (ages 6-1 1, or grades 1- 5) and i s the level most commonly attended. EP2 covers the next two years (ages 12-13, or grades 6-7). EP1 and EP2 schools may be located together, but inrural areas the largest number o f schools covers only EP1. ES1 covers ages 14-16 (grades 8-10), and ES2 ages 17-18 (grades 11and 12). Secondary school may take a variety o f forms: general, which is the most common, and technical, which includes agriculture, teacher training, and the like. 4.15 I n 1997 thegovernmentfaced an enormouseducation deficit. Very high absolute poverty and difficulties in access caused enrollments to plummet during the civil war, rebounding only since 1997. Gross enrollment inprimary school was only 50 percent, and net enrollment was below 40 percent. Infrastructure was invery poor shape, and schools were absent inmany rural areas. Schools often lacked inputs(teachers, books supplies, andthe like). Many teachers were not qualified. GovernmentPolicy and SpendingPatterns 4.16 Governmentpolicy since the end of the civil war has been toprovide quality educationfor all, but until now it hasfocused mostly onprimary education. The government emphasizes the goal o f universal primary education (EP1 and EP2) by expanding the infrastructure network and improvingthe efficiency o f resource use. This 83 goal was spelled out inboth PARPAsand inthe later Education and Culture Strategic Plans, the most recent covering 2006-1 1. The strategy has increasingly focused on reducing regional and gender differences in access and outcomes. The 2006-1 1 Education Sector Strategic Plan, approved in2006, seeks to expand access to basic education throughout Mozambique, improve the quality o f education, and strengthen the institutions and the framework to achieve those goals. It intends to develop a balanced system that will sustain growth inthe sector. 4.17 Since 2000 the government has attempted to meet this goal through high expenditure onprimary education, devoting an average of 20percent of revenue- about 5percent of GDP-to education. Spendingon education as a share o f total public expenditure has been fairly constant, but it fell to 4.4 percent o f GDP in 2002 because o f an overall budget reduction. In2006 expenditures climbed back to the trend level, reaching the average for Sub-Saharan Africa. Since 2002 primary education has received more than 50 percent o f the education budget, and secondary roughly 14 percent, with a slighttrendtoward a larger share o fthe budget for secondary educationto meet higher demand.'21 Although total primary education expenditure (current and capital) is high, Mozambique i s still unable to reach the Education for All benchmark o f 3.5 percent o f GDP.'~~ 4.18 Since 2003 government policy hasfocused on lowering the cost of primary education and ensuring that a higher proportion of expenditures reach children. Three major primary education financing and curriculum measures were introduced: Abolition of primary school fees. In 2004 the government issued a ministerial decree abolishing the ASE (Ac@o Social Escolar) and all other fees and levies in primary education, beginning in the 2005 school year. This decree still allows parents, communities, and others to contribute cash or in kind on a strictly voluntary basis. Direct support to schools. Following the abolition o f primary school fees, direct support to schools was used to compensate for the loss o f discretionary budget at the school level. The objective o f such a transfer i s twofold: to procure basic teaching and learning materials to enhance the quality o f education and to strengthen the school management capacity as part o f the decentralization process. The program provided approximately $0.50 per term per primary student in 2004 directly to all primary schools in Mozambique. The funds were used for nonsalary expenses, including purchasing essential teaching and learning materials and repairing school buildings. The allocation per student has increased each year. ''I appendixtableA4.2. See ''' EducationforAll Fast Track InitiativeinWorld Bank (2005). Mozambiquehas reachedthe Educationfor All goal of spending20 percent oftotal budget revenue oneducation. By 2004 primary education spendingas aproportion of GDP was about 2.5 percent. 84 New primary school curriculum andfree primary textbooks. A new primary curriculum was implemented in 2004. It required that 20 percent o f the curriculum be decided at the provincial or district level, organized the curriculum into three main blocks (grades 1-2, grades 3-5, and grades 6-7). It created a semiautomatic promotion evaluation system with automatic promotion within each block but promotion between blocks based on school exams. And it allowed the possibility o f teaching in the native tongue for grades 1 and 2. Textbooks for all primary subjects are also supposed to be free for every student. The Ministry o f Education and Culture (MEC) and donors are assessing the effects of these policy changes on school quality, school expenditure management, and enrollments among poor and underserved groups. 4.19 Expenditure increaseshavegreatly expanded infrastructure and stafft especially at the EPl level and in rural areas. Between 1996 and 2005 the number o f public schools at the EP1 level almost doubled. The focus was on rural areas with an increase inthe number o f schools o f about 88 percent over the period; urban areas saw only a 32 percent increase (figure 4.2). The number o f EP2 schools also increased significantly-from 94 to 383 inurban areas and from 218 to 1,211 inrural areas (figure 4.3). The populous central region o f the country received a very large increase.'23 The number o f teachers hiredover the period also increased significantly, but the proportion o f qualifiedteachers has been decreasing (figure 4.4). 124 Figure 4.2. Number of EP1Schools in Urban and Rural Areas, 1996-2005 10000 - 9000 w 0 8000 r0 0 7000 cn r a 6000 5000 Lc 5 4000 SI ; 3000 2000 1000 0 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Urban Rural National I Source: MEC database. 123See appendix table A4.3. 124See appendix table A4.5. 85 Figure 4.3. Number of EP2 Schools in Urban and Rural Areas, 1996-2005 5 1200 v, L0 0 1000 2u) 800 w 3 600 n 400 5 z 200 5 0 Source: MEC database. Figure 4.4. Total Number of Teachers and Proportion of Trained Teachers in EP1, 1996-2005 s 50 - -80.0% 0 -- 70.0% 5a, 40 -- m v -- -- 60.0% 2 30 -- -- 50.0% m 0 2 za,0 -- 40.0% 20 -- -- 30.0% 5f - I O - - -- 20.0% z -- 10.0% +xe 0 7 0.0% a 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Total W teachers +proportion of trained teachers Source: MEC database. 4.20 Infrastructure improvementshave not kept up with surging demand, especially in underserved areas. Overall, pupil-teacher ratios are on the rise, particularly inrural areas and inthe northern and central provinces. Pu il-teacher ratios inE P l increased from 54 pupils per teacher in 1992 to 74 in2005.'2PAlthough more teachers have been hired, the ratio is still very highinthe northernprovinces-where the poverty rate is highest and average spending per pupil is low-but falling inthe southern provinces.'26 Ratios in EP2 started lower than those inEP1 but are trending upward as well.'27 Pupil- teacher ratios are generally higher in rural areas (figure 4.5). This pattern indicates that in the face of major bottlenecks, richer areas (the south and urban areas) and the schools lZ5See appendix table A4.5. See appendix table A4.7. UNICEF 2006. I2'See appendix table A4.5. 86 usedmore by the middleclass (EP2) receivedmore benefitsfromthe infrastructure expansion. Figure 4.5. Pupil-Teacher Ratiosin EP1Rural and Urban Areas, 1996-2005 1996 1997 1998 1999 2000- 2001 2002 2003 2004 2005 ----cRural Urban Source: MEC database. Access to Education Services 4.21 Enormousprogress has been made in increasing enrollment rates and closing gapsfor EPl, but the results arefar less impressivefor EP2 and secondary school, with large differences based on gender,province, rural-urban location, wealth, and age. Ministry o f Educationdata show that gross andnet enrollmentrates for EP1have beengrowingvery rapidly.NetenrollmentinEP1went from 40 percent to 83 percentin 10 years.I2*Net enrollmentrates for EP2 are abysmal. Only 7 percent o f students who shouldbe goingto EP2 at age 11do so. One reason is that students start primaryschool late and may haveto repeat. (Note in figure 4.6 the low enrollmentrate ofchildrenat age six compared with age eight). But ifthis was the mainexplanation,gross enrollment shouldbe over 100percent.Itwas only 42 percentin2004,jumping to 47 percentin 2005 (the first year of the free primary educationprogram,which does seem to be raising enrollment). This meansthat fewer than half of Mozambique's childrenare finishing seven grades of school(table 4.2).'29 "*These data are from the MEC administrative database. They may be a little bit higher than what would have been measuredby household surveys. '''Seeappendix tables A4.7 andA4.8. 87 Table 4.2. Gross andNet EnrollmentRates, by EducationLevel, 1996-2005 (percent) GrossEnrollment Rates Schoolyear EPI EP2 ESI e52 (age 6 1I) (age 12-1 3) (age 14-1 6) (age i7-1 8) 1996 67.5 18.5 5.9 1.6 1997 73.8 19.6 5.8 0.7 1998 78.6 21.0 6.0 1.6 1999 85.3 22.4 6.3 1.4 2000 92.1 25.3 7.6 1.5 2001 101.2 29.2 8.5 2 2002 106.5 32.9 10.6 2.4 2003 112.7 36.9 12 2.9 2004 121.2 42.7 13.8 3.4 2005 131.3 47.0 17.0 3.7 Net Enrolment Rates EPI EP2 E N e52 1996 39.3 2.2 1.3 0.3 1997 42.5 2.2 1.4 0.2 1998 45.1 2.5 1.3 0.5 1999 50.1 2.5 1.4 0.2 2000 54.7 2.7 1.7 0.2 2001 61.1 3.3 1.9 0.3 2002 64.1 3.6 2.3 0.4 2003 69.4 4.5 2.7 0.5 2004 75.6 5.6 3 0.5 2005 83.4 6.7 3.9 0.6 Source: MEC database. EP1: Primary grades 1-5. EP2 Primary grades 6-7. ES1:Secondary grades 8-10. ES2:Secondary grades 11-12 Figure4.6. Enrollmentof Boys and Girls, by Age, 2003 100.0 -s 80.0 60.0 a"2 40.0 20.0 0.0 7 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Age I +Boys -m- GirlsI Source: IAF data for 2003. 88 4.22 Despite someprogress, differences in enrollment betweengirls and boys beenachieved inthe southern province^.'^^ Differences are greater inrural areas. Girls remain-particularly in the northernprovinces. Gender parity inenrollment has already start school late, especially inrural areas where the distances may be longer. But after age seven, girls have a slightly higher enrollment probability than boys inrural areas, but it decreases fast with age, starting at age The combination o f the late start and the fact that girls drop out o f school earlier than boys, especially inrural areas, leaves girls with less schooling.132 4.23 Enrollment patterns show that household income is strongly correlated with whether children attend school. Inall wealth quintiles both girls' and boys' enrollments are increasing untilage 10-1 1,when the percentage o f all children enrolled sharply decrease^.'^^ Enrollment rates are the highest inthe top quintile, and the smallest difference between boys and girls i s found inthe richest quintile. The pattern suggests that children are leaving school to work. 4.24 Besides gender and age, having twoparents mattersfor enrollment rates. Orphans have a lower probability of being enrolled in school, and children with better educated and wealthier parents have a higherpr~bability.'~~ The probability o f being enrolled decreases ifthe child is not the child o f the household head or spouse.*35School fees or other costs were not found to affect the chances of a child being enrolled, but this variable was very poorly measured in the household survey, so no firm conclusions can be drawn. 4.25 Despite infrastructure improvements, long distances to school seem to be most important in reducing parents' willingness to send their girls to school. The probability that a child goes to school sharply decreases as the time it takes to get there increases. The presence o f a school inthe village increasedthe enrollment o f girls from poorer households. 136 One reason i s parental concern for the child's safety, especially younger children. Another is the time that the commute takes away from household tasks. Distance to water also has a negative effect on the probability o f girls being enrolled in primary school in rural areas. This indicates sensitivity among poor parents to the high opportunity cost o f sendingtheir children to scho01.'~ ~~ See appendix table A4.9. World Bank 2005a. "'See See appendix table A4.10. appendix table A4.11. 134World Bank 2005a; Handaandothers 2004. Nhate 2005; WorldBank 2005a. Handaand others 2004. 13'The analysis on distanceto the water source was inaddition to the analysis based on the enrollmentregressionsused inWorld Bank(2005, PovertyandSocialImpactAnalysis inEducation). The negativeeffect suggeststhat girls are the prime water collectorsinthat distance. But when the water source i s more than two hours away, it does not have any effect, suggestingthat carrying water from sources more than two hours away may not be amongyoung girls' duties because it is too far from the home. 89 Efficiency in the Education Sector 4.26 Quality appears to be getting worse, again especially in underservedareas. Higherpupil-teacher ratios andmore unqualifiedteachers usually indicate that quality is dropping. Another indicator i s repetition and dropout rates-also on the rise. In2003 only 37 o f 100 students who started grade 1 reached grade 5. Again, the northern provinces had the lowest completion rates and the widest gender gaps. 4.27 Despite improvementsin educational outcomes in recentyears, actual completion and dropout rates do not display the same success. As shown earlier, the level o f accessto primary school has increased impressively inrecent years, but the share o f pupils actually completing first-cycle primary school (EP1) remains low (table 4.3). Dropout rates have fallen but remain high, making the relatively good access rate to primary school much less significant. The poor rates o f survival of pupils through the successive grades mean that, though initial access i s fairly good, the EP1 completion rate i s low. Table 4.3. Primary School (EP1) Completionand DropoutRates, by Province, 1997and 2003 Primary School @PI) Primary School (EPI) Completiona Dropout Province 1997 2003 1997 2003 Niassa 62.0 61.7 17.9 19.5 Cab0 Delgado 57.9 64.5 19.6 19.1 Nampula 59.1 64.9 18.7 17.0 Zambezia 51.0 57.3 26.8 20.5 Tete 61.0 68.7 24.1 14.8 Manica 62.7 67.9 14.3 11.8 Sofala 67.8 73.2 11.0 10.7 Inhambane 63.5 68.0 14.2 10.7 Gaza 60.1 67.7 16.6 13.2 Maputo province 60.6 75.3 15.5 8.8 Maputo city 64.2 76.3 11.3 5.9 Total 60.6 67.5 I7.1 13.8 Source: MEC database. a. Primary school (primary grades 1-5) completion:the data presentmodified completion rates, meaning the enrollment inthe final year of the cycle, regardlessof age, expressed as apercentageofthe segment ofthe populationthat is at the age correspondingto the official age for graduation. 4.28 Thereare large variations in completion and dropout rates among regions and by gender. While improvements can be seen across the country, Maputo province and city continue to show the best outcomes, whereas central and northern regions lag behind. Nearly universal access and completion infirst-cycle primary education characterizes Maputo city, although universal access and completion remains elusive particularly inthe north, but also inthe center-regions that performedthe best interms o f monetary poverty reduction. By 2003 completion rates in every province were lower among girls. Nationwide completion rates were only 32 percent for girls compared with 48 percent for boys. 90 Implicationsfor thePoor 4.29 Household education expendituresincrease with overall consumption, butfall as a share of total cash expenditures,indicating thatpoor households are having harder timespayingfor kids' education. Inall provinces and across urban and rural areas, total education expenditures increase with wealth levels-richer households spend more in education (bars infigure 4.7). However, education expenditure as a share o f total cash expenditure-a measure o f the (potential) difficulty o f households to support education costs-decreases with levels o f income.'38The level o f hardship i s much stronger inrural areas where poverty i s more severe, indicatingthat highopportunity costs o f household labor can lead to reduced enrollment and to highdropout and low completion rates.13' Figure 4.7. Total HouseholdEducationExpenditure and Share of Education Expenditure in Total Cash Expenditure, by Area of Residence,2003 I -4.5% I4O0 T -- 4.0% -- 3.5% -- 3.0% -- 2.5% -- 2.0% -- 1.5% -- 1.0% -- 0.5% 1 2 Household consumption quintiles - 7 0.0% 3 4 5 0Rural lkban -A- Rural+&ban Source: IAF data for 2003. 4.30 Thepoor use the educationsystem less than the rich, and thusget lessfrom government expenditures.140 By calculating a cost per student usingenrollment per quintile ineach level, the benefit that each quintile receives from education expenditures 13' See lines in figure4.7. '39One exception is among the highest income households, where urbanhouseholdsspend ahigher share oftheir cash income ineducation. This may haveto do with awider range of education servicesavailable for urbanhouseholds, particularly in secondary and highereducation for which higher prices are charged, andricher households struggle to pay for in urbanareas. Richer ruralhouseholds do not havethat choice. I4OUsingthe method from Demery (2003) government expenditure data from 2002 and 2003 for different subsections of education(primary, secondary, tertiary, andtechnical) are mappedto the users of the services (retrievedfrom the 2003 householdsurvey). This allows for the quantification of who benefits from government expenditures by characteristicsof the users, typically gender and consumption or income quintiles. 91 can be calculated. Across the education sector, 35 percent goes to pupils inthe richest quintile, while the remaining 65 percent is distributedmore or less equally among the four lower consumption quintiles (table 4.4).141 Table 4.4. BenefitIncidence:EducationExpenditureAllocationby Quintile, 2003 (percent) Distribution Across Quintiles Cfromhouseholdsurvey) Education Lowest 2nd 3rd 4th Highest Primary education 21.8 20.6 19.6 19.5 18.5 Secondary education 12.8 13.6 17.2 17.8 38.6 TVE 5.2 6.0 14.7 13.6 60.4 Adult literacy 18.1 17.8 26.3 21.1 16.8 Teacher training 5.3 16.2 10.3 5.4 62.8 Higher education 5.0 3.1 4.2 2.2 85.5 Total education2003 16.7 16.0 16.1 15.6 35.6 Source: IAF data for 2003. 4.3 1 Primary education expenditures-half of government education expenditures- are reaching thepoor, but the rest of government expenditures on education are not. All government expenditures are reachingmore boys than girls. The increase in enrollments has translated into increased use o f the education system by poor households. As a result, primary education expenditures (and adult literacy) are fairly equally distributedacross the quintiles.But secondary education increasingly benefits richer quintiles since they can access it-especially the richest one, which receives nearly 40 percent of secondary education spending. Cost risingwith education level means that good results o f poorer students at the primary level are overshadowed by the low access to higher levels. Technical and vocational education, which costs two to three times as much as general secondary education, i s even more concentrated among richer households and on boys. More than 60 percent o f teacher training benefits the richest quintile.And for higher education 85 percent o fexpenditures go to the richest quintile. Expenditure equity can be improvedby better efforts to increase access for the poor, such as bringingschools closer to communities, lowering costs, and improving quality and efficiency-in terms o f both retention and completion rates inprimary and secondary education. Table 4.5 shows how all government expendituresare reaching more boys than girls, with the differences increasing at higher education levels. 14'Actual dollar figures for 2002 and2003 are in appendix table A4.14. 92 Table 4.5. BenefitIncidence: ExpenditureAllocationby Quintile and Gender, 2003 Primary School Secondary School All Public Schools (EPl + EP2) (ESGl + ESG2) Boys Girls Di$ Boys Girls Diff Boys Girls Diff Quintile (1) (2) (3) (4) (5) (6) (7) (8) (9) Lowest 11.3 9.6 1.7 11.9 9.9 2.0 7.7 5.1 2.6 2nd 11.0 8.8 2.2 11.6 9.1 2.5 8.3 5.2 3.1 3rd 10.6 9.0 1.6 10.8 8.8 2.0 10.6 6.6 4.0 4th 10.3 9.1 1.2 10.5 9.1 1.4 10.7 7.0 3.7 Highest 10.7 9.6 1.1 9.6 8.8 0.8 22.5 16.1 6.4 All 53.9 46.1 7.8 54.4 45.7 8.7 59.8 40.0 19.8 Source: IAF data for 2003. EPl: Primary grades 1-5. EP2: Primary grades 6-7. ESl: Secondary grades 8-10. ES2: Secondary grades 11-12. Perceptionson Changes in the Education Sector 4.32 Satisfaction with educationis generally higher among the urbanpopulation. About two-thirds of urban households-but only half thosein rural areas-perceived an improvement in education during 2001-06. Actually, in rural areas almost a quarter o f households perceive that conditions have worsened. Inurban areas only 7 percent perceive it that way (figure 4.8). Female-headedhouseholds inrural areas express the most dissati~faction.'~~ While better-offhouseholds inurban areas express more developments. satisfaction, the oorest tercile households inrural areas complain the least about recent 1 2 4.33 The chief reasonfor perceived improvement in the educationsector in both urban and rural areas was the expansion of the school network. About 67 percent o f urban households and 62 percent o f rural ones indicate that expansion o f the network i s the main reason they perceive improvements inthe education sector. The increase in teachers was listed as important only inurban areas, with improved curricula much more important in rural areas.144This may reflect policy changes that allow more room for locally relevant topics in school curricula inthose areas. 142See appendix table A4.16. 143See appendix table A4.17. '44Ibid. 93 Figure 4.8. Perceptions of Change in the Education Sector, by Area, 2006 71.2 n 1 I I 53.6 22.2 22.5 23.9 Urban Rural 10 hrproved Did not change Worsened 1 Source: PVS data for 2006. Note: Sample not representative. 4.34 As expected, the main reasonsfor theperceived worsening in the education system were distance to schools (reflecting low school density) among rural respondents and corruption and bribes among urban respondents. Distance (lack o f schools) was explicitly referred to by 38 percent o f rural respondents. Inurban areas lack o f school books was also a major factor (29 percent o f respondents). While this problem was not cited inrural areas, other evidence suggests that rural areas also face serious constraints inthe availability o f teaching materials, despite those areas beingdominated by primary schools that benefit from the free book policy. Lack o fteachers i s more o f a concern inrural areas (12 percent) than inurban areas (8 percent), consistent with the higherpupil-teacher ratios inrural areas.145 4.35 Poor households also cited local level nuisance regulations as obstaclesto attending school. For example, many municipalities impose a bicycle tax. Parents complain that students are harassed by police while bikingto school, demanding the payment o f the tax. They are especially concerned about girls. This tax actually yields very little revenue. With distance being a major impediment to access, it seems counterproductive to enforce a tax that reduces access. HEALTH 4.36 The system of healthcare service delivery in Mozambique has threeparts: hospitals (central, provincial, and district), health centers, and health posts. By 2005, there were 46 hospitals (1Io f which were central), about 775 health centers, and 435 health posts.'46 The public system i s complemented by a very small urbanprivate system and a system o f traditional village healers. 145See appendix table A4.18. 146See appendix table A4.19. 94 GovernmentPolicy and SpendingPatterns 4.37 Thegovernment's aim is a healthypopulation through improved and equitable access to basic healthcare, especiallyfor thepoor. Government policy statements, including both PARPAs, point to the health sector as an important contributor to human development and to poverty reduction-both directly and indirectly. The PARPA I1core objective inthe health sector i s to improve the health status o f the population, especially the most vulnerable groups, through an equitable allocation o f resources that prioritizes poor and vulnerable groups and expands the health network to target areas that have been less favored. It also stressesthe needto improve the quality o f health services.'47 4.38 Expenditures in health as a share of total expenditureshave been uneven but rising recently. Overall, the trend inhealth expendituresis hard to follow. A significant share o f expenditures is funded by donors, including vertical programs (figure 4.9). In 2000 poor financial management resulted inthe Ministry o f Health spending only 80 percent o f its budget allocation. Better expenditure management, more donor funding, and greater allocations for HIV/AIDS have boosted expenditures in real terms and as a share o f the total government budget. But Mozambique's expenditure per capita i s lower than that o f neighboring c o u n t r i e ~ .Mozambique has yet to reach the World Health ' ~ ~ Organization target o f $10 per capita per year. Figure 4.9. Health Expenditure Shares, by Source, 1997-2002 70 - 60 - 50 - 1997 1998 1999 2000 2001 2002 --tStateBudget--eDonorFunding--tHouseholdsIUsers Source: MISAU data for 1997-2002. 4.39 I n 2002 there were leakagesfrom theprovincial to the district level and persistent regional inequalities in resourceallocation. In2002 the government conducted an ExpenditureTracking and Service Delivery Survey, revealing poorly kept financial records-incomplete for nearly half the sampled districts. Moreover, allocation and execution data at the provincial level were inconsistent. Leakages were evident inthe distribution o f drugs from provinces to districts. Also apparent were considerable 14'GoM 2005 14'See appendix table A4.20. 95 confusion, poor transparency, and little control inuser payments, cost recovery in consultation, and drug charges.I4' Since this survey major efforts have been undertaken to improve financial management. A follow-up survey i s warranted once implementation o f the new budget system i s initiated. 4.40 Healthfacilities and health staff have increased, but the sector remains understaffed, and distribution is skewed toward urban areas. In2003 there were about 100health workers per 100,000 people and 3.8 physicians (national and expatriate) per 100,000 people. Mozambique has one doctor for every 44,000 inhabitants, similar to Malawi, but worse than Zambia (one doctor per 11,000 inhabitants) and the Sub-Saharan average o f one doctor per 22,000 inhabitants in2003. There were about 20 nurses per 100,000 people in2003. Inparallel with the expansion o f the health facility network, the last three years saw a 4.2 percent increase inthe number o f staff. Human resources in health are concentrated inthe urban areas. Only 3 percent o f the physicians work in rural areas, and more than halfthe country's 712 doctors work inMaputo city. Auxiliary nurses and midwivesare well distributed regionally: 60 percent o f auxiliary nurses and 68 percent o f auxiliary midwives work inthe rural areas.*50Figure 4.10 illustrates the regional differences with respect to the number o f inhabitants per physician and per health worker. Bringingthe health facilities closer to the people requires increasing the number o f facilities. Addressing this issue goes to the heart o f structural issues inthe sector, includingthe affordability and cost-effectiveness o fthe model o fprimary healthcare beingimplemented inMozambique, the preferences o f health professionals, and the incentives they are offered. Figure 4.10. Number of Inhabitants per Physician and Number of Inhabitants per Health Worker by Region, 2003 Mozambique 44,058 I 4 Maputo city p;SO3,907 South 32,413 Center 44,742 North 58,673 0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 Inhabitant per physician Inhabitant per healthworker Source: MISAU2004. 149Data fromthe Ministry of Healthare acknowledgedandgreatly appreciated.PublicExpenditureTrackingSurvey, MPF andMISAU2003, andWorld Bank2004. lSo Bank2006c. World 96 4.4 1 Given the inequalities infacilities, health expendituresare distributed unequally acrossprovinces. Ingeneral, health expendituresper capita are lowest inthe northern provinces and highestinthe southern and central ones (figure 4.1 1). Although equity i s a guidingprinciple inthe health sector, resources per capita vary substantially even between districts. Inequities exist for all types o f resources, including drugs, equipment, and staff. One reason i s that there are more facilities per person inthe southern provinces. The north has almost twice as many people per health center than does the south. Figure 4.11. Health Expenditureper Capita and Poverty Incidence, by Province, 2003and2004 .-3$70 t 60 50 .-3 40 l30 20 10 0 - 0Ewpenditurepercapitain2003 +-Expenditureper2003 capitain 2004 Rrcent poor in Source: Expenditure data from MISAU 2003 and 2004 (DespesaJustificada);percent poor from IAF data for 2003. Access to Health Services 4.42 Physical access to healthcare service is still a largeproblem in rural areas. Poor people and rural dwellers who live closer to a health facility and who have more education are more likely to use healthcare. Rural residents are 50 percent less likely to seek care when they are illthan are urban dweller^.'^' More than half o f rural women, but only 23 percent o f urban ones, who had been sick report distance as the main reason for not seeking care.152Evenhigher income households inrural areas report distance as a Is'Forhealthcare service use, household surveys (IAF) provide information on curative care (seekingtreatment incase of illness or injuries), while for healthoutcomes much information comes from the MozambiqueDemographicand HealthSurveysof 1997 and2003, which offer comparable data on arange of healthanddemographic issues and allow analysis by wealth quintile. "* See appendix table A4.24. 97 major constraint. The importance o f distance is confirmed by multivariate analysis with the 1997 and 2003 Household Survey (IAF) data. There are strong positiveeffects from primary education and strong negative effects from distance on usingcurative healthcare. The positive effects o f primary education are strongest for rural and poor groups. The negative effects o f distance are strongest for rural and poor areas, where there i s less transportation or roads are inworse ~ 0 n d i t i o n . l ~ ~ 4.43 Bothpreventive care and curative care have seen improvementsin the use of healthcare, but improvementswere less impressivefor curative care.154 And although inequality inthe use o f preventivecare is falling, it is increasing inthe use o f curative care, with households that are poor, rural, and inthe northern provinces still showing much lower use.155Access to healthcare i s clearly a problem, with only 57 percent o f people who reported being sick in 2003 usingformal healthcare facilities for treatment. The most substantial increase intreatment seekingbehavior between 1997 and 2003 was inthe richest quintile, indicatinglingeringinequality.156By contrast, inpreventive and home care the poorest quintile saw the largest changes.lS7A regional gap remains for both types o f healthcare, with better health service use inthe south than in the north. Use o f curative health services is worsening in some provinces (for example, Niassa and Tete show a drop inmedical treatment for diarrhea and acute respiratory infections between 1997 and 2003). This i s distressing, because many common diseases, especially among children, can be cured by simple health treatments, such as prescription drugs. 4.44 Efforts have been especially strong to improve maternal and child healthcare in the northern and centralprovinces, where health outcomes are lower. Overall, live births attended by skilled healthpersonnel and vaccination coverage are increasing especially inthe northern and central provinces (table 4.6). 4.45 Higher incomes increasetheprobability thatpeople use healthfacilities. Income has apositive effect on the likelihood of seeking care in thepoor group, while it has no effect in the nonpoor group. But it is less important when the magnitude o f its effect is compared with the effects o f education and distance. User fees are also a greater constraint for rural residents, perhaps resulting from a seasonal cash shortage. Rural residents still rely heavily on traditional healers, while urban residents rely more on the formal health system. 153 Distance i smeasuredintime to reachthe facility by any means usedto get there: walking, biking, or any other meansof transport. 154Preventive care includesvaccinations for children and antenatal care. Curative care includes treatment sought when showingsymptoms of an illness. See appendix tables A4.24 and A4.25. See appendix table A4.27. I5'See appendix table A4.25. 98 Table 4.6. Number of Live Births Attended at Health Facility and Vaccination Coverage, by Province, 1997 and 2003 (percent) Live Births Attended at Public Children 12-23 Months Fully Provinces Health Facility Vaccinated I997 2003 Change I997 2003 Change Niassa 44.4 45.9 1.5 48.2 46.6 -1.6 Cab0 Delgado 31.O 29.6 -1.4 25.4 57.9 32.5 Nampula 28.5 36.8 8.3 34.4 53.9 19.5 Zambezia 23.5 32.6 9.1 23.2 44.7 21.5 Tete 41.0 47.4 6.4 48.0 55.0 7.0 Manica 43.0 55.7 12.7 46.5 61.6 15.1 Sofala 35.4 51.4 16 49.6 63.9 14.3 Inhambane 56.3 49.6 -6.7 71.7 90.6 18.9 Gam 65.7 62.6 -3.1 63.0 82.3 19.3 Maputo province 75.7 84.8 9.1 61.9 92.5 30.6 Maputo city 86.5 88.2 1.7 82.0 91.3 9.3 Rural 33.3 33.8 0.5 36.4 56.0 19.6 Urban 81.3 80.6 -0.7 85.0 80.5 -4.5 Total 43.7 47.4 3.7 47.3 63.3 16.0 Source: DHS 1997 and 2003. Health Outcomes 4.46 Health outcomesshow large variations,geographically and by incomegroup. The infant and child mortality rates remainamong the highest inthe world and higher than most neighboring countries (table 4.7), though they have been declining steadily in recent years, by 15 percent and 18 percent respectively. Urban areas and the south generally perform much better. Indeed, while rural areas experienced greater declines in both infant and child mortality, the urban-rural gap remains large. Moreover, Cab0 Delgado and Manica saw increases in both infant and child mortality, while Maputo city and Niassa experienced an increase in infant mortality (table 4.8). InCab0 Delgado and Maputo city income poverty increased by more than 10percent, which could help explain the increase inmortality rates, but inboth Manica and Niassa poverty fell by about 29 percent.'58 Is*Maternal mortality rates are also very high nationally, althoughthe DHSfor 2003 registeredasignificant reduction inrecentyears. World HealthOrganizationestimates were 1,062 deaths per 100,000 live births in2000 (MISAU 2000), andthe DHS for 2003 showedadecreaseto 408 deaths per 100,000 births. It is debatable, however, whether the fall in the rate is as large as these figures suggest, becausethey were obtained using different methodologies.The ratesremain very high. 99 Table 4.7. Regional Comparison of Welfare Indicators (percent) Infant Severe Female HIVAIDS Region Mortality Stunting Malnutrition Prevalence Malawi (2002) Average 112.5 24.4 8.8 Rural-urban ratio 1.4 1.9 1.7 Low-high ratio 1.5 2.9 1.7 Mozambique (2003) Average 123.6 17.9 8.1 Rural-urban ratio 1.4 2.2 1.6 Low-high ratio 2.0 4.2 2.0 Tanzania (2004) Average 82.5 12.3 9.4 Rural-urban ratio 1.2 1.84 1.3 Low-high ratio 1.4 4.70 1.6 Uganda(2002) Average 89.4 23.8 10.4 Rural-urban ratio 1.7 2.2 2.5 Low-high ratio 1.8 1.4 3.1 Zambia (2002) Average 93.9 22.1 15.0 Rural-urban ratio 1.3 1.7 1.54 Low-high ratio 2.0 2.2 2.1 - Source: World Bank 2007, Socio-Economic Differences in Health, Nutrition, and Population, Country Reports; HIV/AIDS data are from UNAIDS/WHO (2006b) andare for 2005. Note: Infant mortality is the number of deaths to children under 12 months per 1,000 live births, based on experience duringthe 10years precedingthe survey. Severe stunting (height-for-age) i s the percentageof children aged 3 years (varies by survey), with a height-for-ageZ- score ofbelow -3 standard deviations ofthe medianreference standardfor their age. Female malnutrition is the percentageof women aged 15-29 years with abody mass index of less than 18.5, definedas weight inkilograms divided by square ofheight inmeters.The low-highratio is highest quintile/lowest quintile. -isnotavailable. 4.47 While health status improvedfor all incomegroups; thegap between the health status of the richest andpoorest remains significant. Despite reductions in infant and child mortality rates inthe poorest quintile outpacing improvements inthe richest quintile, infant mortality rates remainmorethan twice as highfor the poorest than for the richest. 100 Table 4.8. Selected Social Outcome Indicators (percent) Wasting Stunting (under Life HIV/AIDS Infant Under 5 (under age Expectancy Prevalence Mortality Mortality agejve) five) 1997 2005 2000 2004 1997 2003 1997 2003 2003 2003 All 42.3 47.1 11.0 16.2 147 124 219 178 41.0 4.0 Area Urban 48.8 50.8 - - 101 95 150 143 29.2 3.1 Rural 40.2 45.7 - - 160 135 237 192 45.7 4.3 Province Niassa 42.2 45.1 6.2 11.1 134 140 213 206 47.0 1.3 Cab0 Delgado 37.9 42.4 7.5 8.6 123 177 165 240 55.6 4.1 Nampula 39.9 44.4 4.8 9.2 216 164 319 220 42.1 6.0 Zambezia 37.0 48.6 10.0 18.4 129 89 183 123 47.3 5.2 Tete 43.8 44.3 16.3 16.6 160 125 283 206 45.6 1.6 Manica 42.7 46.5 17.3 19.7 91 128 159 184 39.0 2.8 Sofala 42.2 44.8 20.6 26.5 173 149 242 206 42.3 7.6 Inhambane 46.0 49.6 7.8 11.7 151 91 193 149 33.1 1.3 Gaza 45.6 49.0 12.6 19.9 135 92 208 156 33.6 6.7 Maputo province 50.6 54.0 14.4 20.7 92 61 147 108 23.9 0.5 Maputo city 58.3 58.6 13.5 20.7 49 51 97 89 20.6 0.8 Quintile Poorest - - - - 188 143 278 196 49.3 5.6 Poorer - - - - 136 147 214 200 46.7 4.3 Middle - - - - 144 128 216 203 46.2 3.0 Richer - - - - 134 106 187 155 35.2 3.9 Richest - - - - 95 71 145 108 20.0 2.5 Source: Healthdata are taken from Demographic and Health Survey, 1997 and2003; andIAF data for 1997 and 2003. Note: HIV/AIDS prevalenceis the percentageof adults, ages 1549, with HIV/AIDS. Stunting (height-for-age) is the percentage of children under age 5 who are below -2 standard deviations from the medianof the InternationalReference Population(not comparableto 1997). Wasting (weight-for-height) is the percentagebelow -2 standard deviations (not comparableto 1997). -isnotavailable. 4.48 Even with strong consumptiongrowth in recent years, malnutrition rates remain very high. Malnutritionrates inMozambiqueare high inabsolute terms, though notas highas in some neighboringcountries.Evenso, the persistence of high malnutritioninMozambiqueremains puzzlinggiventhe highconsumptiongrowth in recentyears. The rates for stunting (long-termchronicmalnutrition)are especially troubling, particularly inruralareas and innorthernand centralprovinces.Wasting (short-term,or acute, malnutrition) is also present, thoughto a muchsmaller extentthan stunting.Butfor wasting, regionaldisparitiescanbe quite bewildering.For example, Sofala hadthe lowestpoverty rates in 2003 butthe highestprevalenceof wastingandthe fifth highestpercentageof stunting.Muchlikethe other healthoutcomes, differencesin children's nutritionalstatus betweenthe poorest andrichestquintiles are striking, and significantlyhigherthan inneighboringcountries.Rates for bothstunting andwastingare morethantwice as highfor children inthe poorest householdsthanfor children inthe richestquintile. 101 4.49 Despite thegrowth in real consumption across most of the country, chronic malnutrition did not improve accordingly, and it even worsened in rural areas and someprovinces. Sofala and Zambezia-two provinces with the highestconsumption growth and reduction in income poverty-both saw increases in stunting.'59The south- with low reductions or even increases inpoverty-also had stuntingrates go up, except Maputo city, which despite its significant increase inpoverty experienced a slight decrease in stunting. Overall, rates fell somewhat. 4.50 Most noteworthy is how little children's nutritional status changed, suggesting that the elasticity of householdconsumption to anthropometric outcomes is low (at least in the short to medium term). Nutritional status affects primary school enrollment, learning ability, productivity, and future earning potential, and when low itjeopardizes the children's chances o fescapingpoverty. This raises an important policy question: how to translate the country's economic gains into real improvements innutrition and health and into long-term reductions inpoverty. International evidence suggests that income growth alone is not sufficient to solve the malnutrition problem'60-because households do not necessarily spend additional resources on acquiring more calories or on improving their diet.Even so, economic growth is normally expected to contribute more than what has been observed inMozambique,161suggestingthat other factors may be significant in explaining this nutritionalparadox (box 4.1). Box 4.1. The NutritionalParadox: Consumption Growth and Malnutrition in Mozambique What explains this nutritionalparadox?Why has substantialconsumption growth inMozambique not led to similar reductions in malnutrition among children? Four factors have been shown to have some explanatory power inMozambique: The seasonality of consumption and cropping patterns. The composition of householddiets. Women's knowledge and educationlevels. Women's ability to use their knowledge when makingfood consumption, expenditure, and other decisions. Seasonality highlights how consumption varies throughout the seasons, often depending on cropping patterns, seasonal variations in food and crop prices, and the timing of harvests. Seasonalitywas identified inPVS data as a key dimension of householdvulnerability. Arndt and others (2006) find that seasonality in calorie consumption in Mozambique varies dramatically by 15'The stuntingrateshere are for children under age 3, which contrastswith the stuntingratesin appendix table A4.23, which were for children under age 5. The rates are not fully comparable. Barslund(2007), usinga slightly different methodology to calculate changes betweenthe 1997 and2003 time periods, confirms these general findings. See appendix table A4.28. I6OAldermanand others 2005; Haddadandothers 2003. 16'Simler andIbrahimo 200, 18. 102 region and is related to maize cultivation (a seasonal crop). In the central provinces-the maize basket, where seasonality is the most significant-in the post-harvest season (June-July) calorie consumption is up to 40 percent higher than in the "hunger" season (January-February). The central provinces experienced both some of the highest consumption growth and poverty reduction as well as the worst anthropometric outcomes between 1997 and 2003. But seasonality i s less pronounced in the southern provinces, yet they also saw deteriorations in stunting rates, suggestingother factors might also be at work. Households in Mozambique are found to have a poor diet. So, increased consumption may not necessarily translate into improved nutrition. Barslund (2007), using DHS 1997 and 2003 data, finds that dietary diversity is correlated with improved nutritional outcomes (reduced stunting) at provincial levels, especially inrural areas. Despitethe generalpreferencefor dietary diversity, the links between consumption growth, changes in diet, and child health are weak, and other variables at the household level intervene, including the knowledge of household care providers (and particularly women's knowledge), and women's relative bargaining power and ability to realizetheir rights within the household. Barslund notes that mothers' education was highly correlated with improved nutritional outcomes, both in terms of reduced stunting and wasting rates. And mothers' knowledge about `good practices' in raising children is correlated with the nutritional status of children in the household.' Mothers' knowledge about practices that prevent or treat their children's illnesses, however, varies significantly by area (urban or rural) and by the wealth level of the household. Using an index of childcare practices, Barslund finds that knowledge about child care practices is highly correlated with reduced stunting rates in rural areas. In urban areas mothers' education more generally remains strongly significant. Bothyears of schooling and more immediate (access to) information for mothers, moreover, have a strongpositive effect on childcare practices, setting out clear and actionable policy recommendations. Successful translation of knowledge into behavior often depends on women's relative bargaining power within the household. While few representative studies on women's intrahousehold bargaining power have been conducted for Mozambique, evidence from the quantitative elements of the PVS finds consistent discrepancies inhouseholddecisionmakingpower betweenmen and women. Women have very little influence over key household expenditures, including education, health, and food. As seen from studies in other countries,there are potentially important nutritional consequencesfrom this discrepancy. There is clearly a needto devise specific means for addressing malnutrition inMozambique. Followingits commitment at the World Food Summit in 1996 to reduce malnutritionby 50 percent, the government has developeda specific Food Security and Nutrition Strategy, approved by the Council of Ministers in 1998. The strategy focuses on improving knowledge and attitudes inrelationto feeding andprimary healthcare.Morerecently aSecretariat for Food Security and Nutritionhas beencreatedto monitor and evaluatethe food and nutritionsituation of vulnerable populations.But as is clear from the studies cited above, more is urgently neededto address this nutritional paradox. 1. van den Broeck 2007. Implicationsfor the Poor 4.5 1 Overall, thepoor spend a greater share of their income on health expenditures. While rural residents spend less inabsolute terms, they generally spend a larger share of their income on health services. Inboth rural and urban areas the richest households 103 spend more on health, but the poorest spend approximately the same (rural areas) or even more (urban areas) than middle income households. This may be related to the patterns o f demand for health shaped by the incidence o f diseases among households. As a percentage o f total cash expenditures among those that paid for health services, poorer households tend to spend a greater share, but there i s little difference between middle and highincome households (figure 4.12). Figure4.12. Total HouseholdHealth Expenditureand Share of Health Expenditurein Total Cash Expenditure, by Area of Residence,2003 1 11 ';%I"4500 T 12.0% 4000 -- 2: i1 6.0% I !8'oo/o 4.0% E I000 -- -- 2.0% 500 -- - - 0 , r 0.0% 1 2 3 4 5 Household Consum ptlon Qulntlles Rural Urban t- Rural +Urban Source: IAF data for 2003. 4.52 All health services are badly targetedtoward thepoor-with hospitalsbeing the worst. Health data are not clearly divided by type o f healthcare (primary healthcare or hospital care), so it i s not possible to conduct a benefit incidence analysis.'62 But the household survey data can shed some light on the proportion o f health services used by different consumption quintiles, ignoringwhether it was preventive or curative healthcare. Usingthe 1997household survey, the survey found that only vaccination services for children were well targeted toward the poor. The 2003 household survey does not distinguishbetween preventive and curative hea1th~are.l~~ So, figure 4.13 examines health-service use ingeneral, plottingthe cumulative proportion o f benefits to the cumulative proportion o f the population (poorest to richest). It shows that all health services are still badly targeted toward the poor, with hospitals clearly showing the worst targeting, where the highest consumption quintile (the top 20 percent) constitutes close to 40 percent of the total users o f hospitals, and the bottom 20 percentjust over 10 percent. Several data problems ledto this conclusion. Personnelexpenditure is not separatedby type of healthcare.In some districts information on ruralhospitalsis missingwhile inother districts it is combined with expenditure on primary healthcarefacilities, and data do not includeexpenditure inthe three central hospitals. (The question inthe household survey groups all hospitaltypes inone category, so it is impossible to separate people who visited central hospitals.) 163The questionwas askedas follows: "Did you visit any healthcareprovider duringthe last two weeks?Ifyes, which type?" No informationis available onthe type of service received from the healthprovider, andno detailedinformation is collectedon maternity or childcare. 104 Figure 4.13. Concentration of Health Service Use, by Type of Facility, 2003 100 90 80 70 60 50 40 30 20 10 Lowest 2nd 3rd 4th Highest CumulatiE proportionof population (poorest to richest) 1--t Health posts -a- Health centres -A- Hospital -45degreeI Source: IAF data for 2003. Perceptions of Changes in Health Services 4.53 Although most householdsperceive improvements,more than one-third still see no changes,particularly in rural areas amongfemale-headed households. As with education, satisfaction(during2001-06) is higher among urbanhouseholds(67 percent) than amongruralhouseholds (63 percent), with men inbothareas more likely to indicate improvementsthan women (table 4.9). While there are no major differencesinthe perceptionfor improvementbetweenrelativelywell off households inruralandurban areas, relatively well off householdsinrural areas aretwice as likely as their urban counterpartsto perceiveconditionsinthe healthsystemas ~0rsening.l~~These figures reflectthe inequality inthe distributionand quality of healthservices. '64See appendix table A4.29. 105 Table 4.9. Perceptions of Change in Health Services, by Area of Residence and Gender of the Household Head, 2006 (percent) Gender of the Household Head Changes inperception of health Female All services Male Head Head Households Urban areas Improved 68.6 63.2 66.8 Did not change 20.8 22.4 21.3 Worsened 10.6 14.4 11.9 Total 100.0 100.0 100.0 Rural Areas Improved 65.3 57.1 63.0 Didnot change 15.1 15.6 15.5 Worsened 19.6 27.3 21.7 Total 100.0 100.0 100.0 Source: PVS 2006. Note: Sample not representative. 4.54 Better service and growth in the health network were cited as major reasonsfor perceived improvement in both rural and urban areas. More than two-thirds o f households indicate that better services are the main reason for satisfaction with changes inthe health system. As with education, expansion inthe health network is more valued inrural areas, increased staffsize, inurban areas.165 4.55 Distance to healthfacilities and lack of medicinesare the key reasonsfor perceiving a worsening health system in rural areas, while thepoor quality of service is most important in urban areas. Other complaints from rural households include distance to health facilities and lack o f medicines. Urban households complain about the poor quality o f service inhealth facilities and waiting time.'66 4.56 I n sum, the richest quintile, the urban dwellers, and the southernprovinces show the highestpositive health outcomes and the highest use of healthcare services. The most important constraint to healthcareuse is distance, especially for poor and rural households. Regression analysis indicates that education i s also an important determinant o f healthcare use and o f child healthcare practices, especially inrural areas.16' Investment inprimary educationwill also have positiveeffects on the health sector, but inthe long- term. Inthe short term, investment in information distribution can improve childcare, practices in rural areas where health facilities are lacking. Butthere i s no substitute for increasing the number o f health posts, especially inrural areas. See appendix table A4.29. See appendix table A4.3 1. "'van den Broeck2005 106 AGRICULTURALEXTENSION-FORTHEPOOR, NOTTHEAGENTS 4.57 Governmentspending in agriculture has been guided since 1999 by the sector developmentprogram, one of thefirst sectorwideprograms in Mozambique, developed in response to thefinancing requirementsof the Ministry of Agriculture'sambitious restructuringprogram, which accompanied theprivatization of state-owned agricultural enterprises.The ministry neededto reduce direct interventioninproduction andmarketingandto support the transformationofthe ruraleconomy. Set up in 1999by donors as a pooledfundto support the ministry's budget, the sector strategy (PROAGRT) has three mainareas o f intervention,eachwith specific objectives: Institutional reform and modernization-to transform the Ministry of Agriculture into a modern institution that can formulate, regulate, and implement policy to enable favorable development for smallholder farmers and consolidationof large-scalefarmers. Building the public sector's capacity in service provision-to increase production and productivity and to raise family incomes and improve food security. Sustainable management of natural resources-to protect, preserve, develop, and ensure access to naturalresourcesandtheir sustainable use. 4.58 Despite the boost to expenditurefrom PROAGRI, government spending on agriculture has been more variable and significantly lower than on other sectors, such as education, health, and infrastructure, averaging less than 5percent of expenditure since 2001. Most o fthe variationin spending can be tracedto the slow start o f PROAGRIandthe subsequent hugejump inexpenditures inthe lastyears ofthe program.Government sources do notprovidea detailedbreakdownoftotal spending on agriculture and rural development, so most government expendituredata come from PROAGRI'sprojectmonitoringrecords.The programcanbe creditedwith the increased decentralizationof spending inagriculturalandruraldevelopment.In 1999, at the beginningofthe program, only 20 percent ofthe resourceswere transferredto the provinces,with the rest spent at the centrallevel.Inlateryears the share allocatedto the provincesincreasedto about 50 percent(figure4.14). Butwith the focus on institutional development, further decentralizationcouldnotbe expectedfrom this program.As a share oftotal expenditureson agriculture and ruraldevelopment, fundingfrom PROAGRIrangedfrom 15 percent to 83 percent during 1999-2005 .168 PROAGRT funded predominantlyrecurrentexpenditures,such as goods and services, butalso covered capitalexpenditures.Institutionaldevelopment absorbedthe largestshare of expenditures, peakingat 70 percentin2003.169 See appendix table A4.32. 16'Muchofthe institutional development expenditurecoveredconsultingservices andtraining programsoverseas and in Mozambique-required for implementingthe Ministry of Agriculture's new budgeting, monitoring, andevaluation systems, upgraded skills for technical staff, andtechnicalassistanceinagricultural research. See appendix table A4.33. 107 Figure4.14. TotalPROAGRI Budget and Share Allocated to Provinces,1999-2005 60 800 -- 50 .-C> 1 40 ig 30 20 C 200 -- E8 10 0 1999 2000 2001 2002 2003 2004 2005 1DTotalbudget(billionMTs)--ePercentallocatedtoprovinces 1 Source: Ministry o f Agriculture, PROAGRI database, 1999-2005. GovernmentPolicy and Spendingin Extension 4.59 Governmentpolicy in agricultural extension is aimed at improvingproduction, productivity, and incomes of rural smallholders in an environmentally sustainable way. Itfocuses on assistingproducers ina demand-driven manner and on rationalizing and strengthening extension services. The official government policy spelled out inthe national Extension Master Plans (for 1997-2006 and 2007-16) is in line with PARPA (I and 11) and the Agricultural Sector Development Programs (PROAGRI Iand II), which establish the needfor an agricultural extension system that can transform smallholder agriculture through innovation. The goal i s an extension system that raises productivity in agriculture, animal production, forestry, and wildlife, increases incomes, improves food security, and prevents natural resource degradation. 4.60 Agricultural extensionfor smallholders requires innovativestrategies. The recently released Agricultural Extension Master Planhighlightstwo key challenges: complex and diverse socioeconomic and farming systems combined with low population density, and low effective demand for extension by smallholders due to low education and weak market orientation. To address these challenges, new approaches emphasize three principles: deconcentrating services to the district level, enhancing participation of the target groups inproviding services, and engaging other actors ina coordinated effort, including outsourcing.170 The extension strategy promotes active participation o f producers to ensure that extension efforts respondto their needs. And to face logistical and financial challenges, it promotes a publicly financed extension system open to ''OIncountrieswith limitedmarket participation,weak institutions,poor roads, andalimited privatesector, involvementininput delivery and marketingoutsourcing may be premature(Finney 2003). 108 Box 4.2. PROAGRI and Smallholders PROAGRI's national coverage has eight components: Institutional development. Supportto agricultural production. Livestock development. Extensionservices. Research. Landmanagement. Irrigation. Forestry and wildlife. Each had specific objectives. Although smallholders were central, the program focused on capacity building-and thus financed fees for consultantsand salaries and benefits of staff in the Ministry of Agriculture. So, the primary target group became, in essence, ministry staff. As beneficiaries of the agricultural support services, the smallholders became a secondary target group, with much smaller budgets for components that directly serve them. Outcomesand lessons in agricultural extension Evaluations of PROAGRI credit the program with success in institutional capacity building and in decentralized planning and budgeting. In extension the program also yielded good results in institutional development by strengthening a unified extension system and improving the qualifications of extension agents. But it did poorly in services to smallholders. Extension is a staff-intensive activity, requiring much travel and many field visits to disseminate messages. The absence of adequate funding limits the ability of the public extension system to serve farmers in rural areas. Evaluation reports indicate, however, that PROAGRI and the National Directorate for Rural Extension increased yields of major crops in intervention areas, though the higher yields still represent about half the genetic potential. The reports also note that the number of farmers assisted by public extension remains small. Ifthe increased yields were replicated by the majority of Mozambican farmers, the country could achieve self-sufficiency in maize even without increasingthe area currently cultivated. One lesson from the PROAGRI Iextensioncomponent is the need to recognize the heterogeneity of farmers and thus to take into account the different inputs and technologies that eachregionrequires. Another key issue is the weak link with research, preventing successes from reaching smallholders. Why this lack of coordination? Research and extension had their own agendas and priorities, both faced funding and humancapital constraints, and no formal structures strengthened the links. These aspects must be taken into account in future interventions. Both research and extension have recently released their master plans, and the need is recognized for better coordination to maximize the impact on smallholder production and productivity, with potential benefits for welfare. multiple financial and delivery arrangements, including partnerships with other extension providers (such as nongovernmental organizations and private firms) through 109 outsourcing, cost sharin and cost recovery from farmers and farmer groups and associations (box 4.2). 177, 4.61 Extension was supposed to be afocus, but it absorbed only 7.6percent of the budget in 2002, dropping to 2 percent in 2004, even though PROAGRI was still funding 49percent of totalgovernment expenditures in the sector. Average spending on extension duringthe PROAGRIperiod corresponds to an annual expenditure o f about $4,500-$5,000 per extension officer and roughly $20 per assisted ho~sehold.Research, ' ~ ~ a key activity to improve the yields o f small farmers, received the bulk o f its funding in the last two years (table 4.10). Table 4.10. PROAGRIBudget by ExpenditureComponent, 1999-2005 Component 1999 2000 2001 2002 2003 2004 2005 Institutional development 37.3 43.8 46.2 64.5 70.4 17.6 20.2 Livestock 11.0 3.7 3.8 3.7 3.0 1.0 1.2 Rural extension 14.4 13.3 5.0 7.6 4.2 2.2 1.5 Supportto agricultural production 7.1 4.3 10.1 8.6 3.5 4.4 6.8 Forestry andwildlife 24.1 7.5 7.5 2.6 2.1 1.4 1.5 Landmanagement 3.8 10.2 8.8 5.7 4.0 2.8 2.8 Research 2.3 10.6 10.9 5.5 6.9 11.6 27.7 Irrigation 0.0 6.6 2.6 1.9 1.7 1.1 0.8 Other 0.0 0.0 5 .O 0.0 4.2 58.0 37.4 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Source: Ministry o f Agriculture, PROAGRI database, 1999-2005. Access to Extension Services 4.62 When accountingfor public,private, and nongovernmental organization extension, many districts have some kind of extension coverage, but important differences exist acrossprovinces. About 117 of 128 districts o f the country receive at least some extension services, half from the government.173 4.63 Density of extension workers in Mozambique is still low, making access to extension services difflcult. Again, important differences emerge across regions. Extension worker density averages 1.3 extension workers per 10,000 people, with 0.7 workers in Sofala inthe center and 2.7 inNiassa inthe north (table 4.11). The southern region gets the highest extension worker density at 1.4 per 10,000 inhabitants. According to the IAF (2003), about three-quarters o f the population live more than 50 kilometers from extension facilities-and about half o f those live more than 200 kilometers away.*74 But inreality people do not always haveto walk to extension facilities because extension workers travel to meet farmers. According to the 2002 National Agricultural Survey, '" Extension Master Plan2007. 17'ExtensionMaster Plan2007. 173See appendix table A4.34. 174World Bank 2005c. 110 about 32 percent o f rural households live in a village with access to extension officers. The survey also reports that about 14percent ofthe ruralpopulation received some extension advice in2002. By 2005, 16 percent had. The biggest increases were inthe south, while the central region saw a de~1ine.l~~ Table 4.11. ExtensionWorker Densityin Mozambique, 2004 (percent) Extension Workers Region Province Population Extension Workers per 10,000Inhabitants Government NGOs Private Province Zone Niassa 966,579 58 135 68 2.7 C tiCab0 Delgado 1,588,741 91 183 - 1.7 z Nampula 3,563,224 121 291 - 1.2 1.3 Zambezia 3,645,630 55 213 19 0.8 Tete 1,461,650 56 54 172 1.9 3 L Manica 1,280,829 62 71 - 1 .o u L5 Sofala 1,582,256 86 26 - 0.7 1.2 Inhambane 1,401,216 54 95 - 1.1 53 Gaza 1,333,540 73 177 - 1.9 53 Maputo 1,074,793 52 64 - 1.1 1.4 Total 17,898,458 708 1,309 259 1.3 Source: Coughlin (2006) using data from Ministry of Agriculture. -is not available. 4.64 Most extension assistanceis in agriculture, but livestock assistance is also important in someprovinces. While 27 percent o f communities surveyed by the National Agricultural Survey received agricultural assistance, only 17 percent received livestock advice. Livestock extension assistance is strongly correlated with livestock ownership rankings by province.*76 4.65 Recipients of extension are more likely than nonrecipients to use specific techniques, except in the south, where extension services have grown significantly in recent years and both recipients and nonrecipients are equally likely to use them.177 Specific techniques include intercropping, soil conservation, crop rotation, and line planting. The nearly equal use o f techniques across recipients and nonrecipients inthe south is not necessarily bad, as it may reflect important spillovers from extension through demonstration effects-nonrecipients learning and successfully applying techniques from their recipient neighb01-s.'~~ 17'See appendix table A4.36. 176TIA 2002. 177TIA 2002, ECON2005, and TIA 2005. 17*See appendix table A4.37. 111 Implicationsfor the Poor 4.66 Evidence is mixed on the impact of extension,particularly on thepoor. Despite highspendingon agriculture inthe decade after the war, there has beenno clear upward trend inmajor crops since 1997, except for the expansion of sugar and tobacco-almost entirely a result o f private sector investment.179 Institutional capacity inagriculture has been strengthened, but production growth and poverty reduction have been only modest, as have intermediate outputs, such as research and extension services. Other factors, such as poor roads, marketing constraints, and the limitedtechnologies for dissemination to farmers, could contribute to the poor impact. Thejustification for heavy public spending on agriculture is questionable. Untilfood system constraints are eased and adequate technology i s available, it i s not worth spending heavily on extension services.18' 4.67 Regression results do not show any effect of access to, and receipt of, extension on household income orperceptions of improved economic conditions. Access, but not receipt o f extension, has a nearly significant effect on poverty incidence, suggesting regional differences inpoverty rather than effects on individuals.181 There i s a mild association between extension contact and crop income. Both access to and receipt of extension have a low level o f statistical significance and are consistent with about a 5 percent gain in net crop income. But interms o f total income those households were as poor as their neighbors. Walker and others (2004) argued that the absence o f statistically significant positive results on community and householdvariables suggests that extension messages are not making a difference inrural income. The lack o f impact may be explained by constraints on access to improved inputs and to technologies more adapted to location. But the absence o f effects does not substantiate the claim that nongovernmental and public sector extension specialists are working mainly with richer farmers. This i s consistent with the results in appendix table A4.36 showing that there i s no substantial difference among extension recipients inthe use o f specific techniques across income quintiles. 4.68 But in another study, rural extension significantly affects rural crop production, andpublic extension is better oriented towardpoor households. Despite the very limited coverage-only 14 percent o f rural households receive extension informationfrom public or private firms or from nongovernmental or anizations- having access to extension increases crop incomes about 8.4 percent. lf2One explanation o f the difference betweenthese results and the regression results i s that Walker and others (2004) controlled for a much wider range o f factors-reaching an adjusted R2o f 0.50. Meanwhile, World Bank (200%) controlled for a more limited set o f factors-reaching an adjusted R2o f only 0.28. Important variables may have been omitted from the model, and extension may be capturing their effect. Usingquintile analysis World Bank (2005~) found that the impact o f extension diminishes as crop income increases; the effect o f "'Finney 2003 Is'Ibid. Is'Walker andothers 2004. Is*World Bank2005c. 112 extension i s highest for lower income quintiles. The study concluded that agricultural extension in Mozambique i s primarily benefitingthe rural poor. 4.69 The lack of data on overallfunding in agricultural and rural development prohibits a general conclusion about how well expendituresreach thepoor. General agricultural funding does not appear to be propoor, because a large part o f the PROAGRI budget is allocated away from the provinces andto nonpoor spending. 4.70 Extension, whilepoorlyfunded, does reach thepoor, particularly public extension. Although all household income quintiles increased their access to extension services, the lowest amount i s consumed by the poor. Butthe gap is narrowing-by three percentage points between 2002 and 2005. So, at least interms o f access, there are clear indications that the poor can benefit-if the quality of extension provided continues to generate positive returns.lg3 4.71 Public extension spending and the use of extensionservices seem to be correlated. The distribution o f farmers who report receiving extension across provinces in 2002 appears to closely follow public spending on extensionby province. Access to extension by poor and nonpoor i s somewhat correlated to extension budgets by province (figure 4.15). InNampula, where most o f the country's people live, the absolute number o f people assisted is high, accounting for a large share o f the country's total. But a large share o f the province's populationi s not reached. InTete the relatively highproportion o f people targeted results from the large private extension systems associated with cash crop production. And in Sofala private extension relatedto agroindustrial crops, such as cotton, tobacco, and sugarcane, may be increasing the access o f both poor and nonpoor households. 4.72 The effects of extension on consumptionper capita indicate that nongovernmental organization extensionyields a larger impact thanpublic extension. The poor do, however, have more access to public extension than to other types o f extension. On average N G O extension has a slightly higher and more significant impact on welfare than public extension services.'84This is not surprisingbecause NGO and private extension normally have more resources per farmer and are more market oriented. Among the poor with access to extension most have access to public extension. But public extension generally i s the most important source for all income groups, and conclusions about its propoor nature need further investigation. See appendix table A4.37. World Bank 2005~. 113 Figure 4.15. Access to Extension Advice by Income Level and Extension Budget by Province, 2002 35 T A [b" - - aH N 5 -- - E - 6 5.-e * - 5 : - - 4 .Q c \ - - 3 2 - - 2i -- 1 C O - 0PoorestQuintileReceivedExtension2002 Highest Quintile ReceivedExtension2002 -A- ExtensionBudget2002 (Billion MZM) Source: Ministry of Agriculture,PROAGRIdatabase, and TIA 2002. WATER SUPPLY-A KEYOBJECTIVE OF GOVERNMENT POLICY 4.73 Under PARPA 11improving access to water and sanitation in order to meet the MDGs is a key objective. Government policy recognizes that access to potable water significantly affects health, education outcomes, and productivity, as well as gender equity, since women usually collect householdwater. The principles and policies for achieving these goals are set forth inthe 1995 National Water Policy and the 2005 revision, which emphasizedwater's role ineconomic development. Mozambique i s committed to meeting the MillenniumDevelopment Goals on water and sanitation, which entails providing access to water and sanitation to more than 7 millionpeople inrural areas and 3 million inurban areas by 2015. The sector policy framework, built on decentralized management and demand-driven approaches, incorporates lessons from experience. 4.74 Government expenditures on water are apoor indicator of total spending on water. They vary but have recently increased. Government expenditures are a small fraction o f total expenditures for water, however. Investments inwater supply-the bulk o f the cost o f improving access to water-are usually financed by donors. Donor projects have been mostly off budget, leaving them out o f expenditure statistics, but this has been changing with the implementation o f SISTAFE (Sistema de Administraqiio de Fundos do Estado). The water sector suffers from poor project implementation. Funds for donor 114 projects are disbursedmore slowly than inother sectors, perhaps contributing to the uneven expenditures. 4.75 Userfees are an important source of operationsand maintenancefunding in the water sector. Urban householdsmay pay some upfront costs for water connections. Once the investments are made, user fees are normally collected from households to pay the costs o f operation and maintenance. Inpractice, the fees may not cover the costs o f a brokenpump, for example, so public funds may also be used for maintenance. Funding for maintenance is critical. Studies show that a large percentage o f borehole wells inrural areas are out o f order. Despite Progress,Access to Safe Quality Water in Poor and RuralAreas I s Still Limited 4.76 Gains have been greaterfor water and sanitation thanfor electricity and other services, and greater in urban areas. Across Mozambiqueaccess to water and sanitation improved, benefiting both urban and rural areas, with little variation among consumption quintiles.Although more than 60 percent o fthe population remains without access to safe drinkingwater, the use of safe water increased by 13percentnationally, the distanceto a water source (sometimes unsafe water) fell for most households, and an additional 10 percent o f households benefited from improved sanitation, with the poorest quintile benefiting most.'85Evenso, households inrural areas have significantly less access to safe water and improvedsanitation than those in urban areas (box 4.3). Focus group discussions confirmedthese general trends, with the largest improvements inurban areas. 4.77 Although the inequality in access hasfallen, differences are still large across provinces and between rural and urban dwellers. Data from the Labor Force Survey (2004/05) indicates wide differences in access to safe water across provinces, ranging from less than 22 percent inZambCzia Nampula and Niassa to over 74 percent inMaputo province and 96 percent inMaputo city. Access to safe water in rural areas i s estimated at just over 20 percent, reachingclose to 70 percent o fthe population inurbanareas nationwide. These differences are more important than differences across consumption quintiles, where the gap betweenthe richest quintile and the poorer quintiles is closing. Despite improvements, these figures omit the seasonality of the water supply, which has importantconsequences for householdaccess to water. Some rural communities notedthat they generally had sufficient access to water during the rainy season (from Decemberthrough March) but frequently encounteredsevere shortages duringthe dry season (from June or July through September). Women and childrenoften suffermost from this shortage. Responsible for finding and carryingwater, they have to wait in lines at public fountains andwells or must walk even farther to find water. 115 Box 4.3. Quality of Water in Erati-Namapa, Nampula In many communities the quality of water-access to safe drinking water-is a severe and continuing problem, often noted as a key cause of illness and health problems.Typical sources of water in the study areas were tap water (inside or outside private homes), public fountains, traditional wells (some private), and rivers, small lakes, and streams. Of these, communities in rural areas were most likely to have a mix of traditional wells and rivers, lakes, and streams, with households in urban and peri-urban areas often having a broader mix, including public and private fountains. Those with several sources of water normally used water from wells and rivers for washing and cooking and water from public fountains for drinking. But some households, particularly inruralareas, had only one source of water, used for washing, cleaning, drinking, and cooking. In the district of Erati-Namapa in Nampula the local population was in the midst of a cholera epidemic during the fieldwork. Only 6 percent of the district's more than 200,000 people have access to safe water. Even in the district capital of Namapa, with more than half the district's population, the main sources of water for the majority are nearby wells and the Lurio River. The water from the river, however, is frequently contaminated. Bathing, washing, and ,cleaning of clothes often take place adjacent to where most people collect their drinking water. In rainy periods drinking water is further contaminated by human waste because the village lacks a functioning sanitation and drainage system. The water situation in Namapa is not unlike that in many communitiesinrural and periurbanMozambique. I Source: UEM 2006, 53. 4.78 Distance to water,particularly to safe water, varies by quintile and urban-rural location. The poor live farther away from their water sources, especially for users o f safe water, and rural households live farther from any type o f water than urban households (figure 4.16). Some provinces are not well covered by water sources. Water coverage has increased and poor-rich gaps are closing, but regional and urban-rural gaps remain wide. Mozambique will need to make major efforts to meet the Millennium Development Target for water. WHATI S NEEDED AN EFFECTIVE FOR GOVERNMENT PROPOORSPENDING? 4.79 Public spending on service delivery in Mozambique has been growing rapidly, and success has been variable acrosssectors. The social deficit was reduced but not eliminated. Many challenges remain. To maximize the effects o f public spending on the poor, the government needs to make efforts in ensuring: Increased public spending in facilities and services used by the poor and vulnerable areas and groups. 0 Increased value for money in the public sector through improved quality of public services and better delivery systems. 116 Figure 4.16. Percentage with Access to Safe Water, by Region and Quintile, 1997 and 2003 100 A 90 ao 70 .-ti 60 7 50 iij 8 a 40 30 20 10 0 Lowest 2nd 3rd 4th Highest Rural Urban Total nAccessin1997 -Access in 2003 -A- Less than 30 minutes from safewater (2003) +Less than 30 minutes from unsafe water (2003) Source: IAF data for 1997 and 2003. Improved systems to better monitor government expenditures and evaluate how effectively they are allocated to the poor. Increased Spending and Valuefor Money in the Public Sector 4.80 I n education this translatesinto increasing the quantity and quality of the services delivered, with severalpriorities. Further expand the school network and reduce the costs o f education to reduce distance and alleviate financial constraints for poor households. Increasing access and completion rates among girls need to be a top priority, particularly in rural areas. Continue to pay special attention to teacher training to keep up with surging demand, and invest in curricular development to improve quality at all levels. 117 Investmore inpostprimary educationto keep pace with progress inprimary access and completion, targeting rural areas and poor and vulnerable groups, especially girls. 4.81 To increase attendance amongpoorer households,conditional cash transfers could be considered,after a careful assessment of theirfeasibility and overall sustainability. Scaling up existingsocial protection programs that have indirect effects on children could also be considered. 4.82 I n health thefocus should be on improving the coverageand quality of service delivery. Improve the distribution o f hospitals, health posts, and health centers and o f health personnel, including well remunerated qualified professionals, to minimize shortages among rural and less-favored areas and groups. Develop outreach services that bring health services to the population through a network o f community health workers (based in the villages) working with existing health centers. Outreach activities should increase awareness o f the importance o f formal health services. Manage the distribution o f medicines and diagnostics equipment to health facilities, particularly in rural and remote areas. 4.83 Investmentsin education and information will be important to improve access to health services and health outcomes. Since education i s an important determinant o f healthcare and childcare practices, investments inprimary education are expected to have positive effects inthe health sector. Inthe short term disseminating information distribution can improve childcare practices in rural areas lacking health facilities. 4.84 I n agricultural extension coverageshould be increased and actions well targetedto the needs of farmers. 0 Adopt a holistic view that integrates the extension system with agricultural research to ensure that the materials developed by researchers are well adapted to the needs of farmers, particularly poor farmers and groups with specific characteristics. That is, take into account the heterogeneity o f farmers and areas to be targeted. 0 Structure spending to allow for better geographical coverage and outreach to help farmers adopt technologies that will increase their productionand productivity. 0 Promote infrastructure development, institutions (support associations in reducing transaction costs), and marketing to link farmers to markets. 0 Promote public-private partnerships, including outsourcing, for better targeting and greater benefits to participating smallholders. 4.85 I n water access to clean water needs to be improved and the regional and rural- urbangaps reduced. Spendingneeds to keep up withpopulation growth. 118 Invest in water supply systems that are well distributed (spatially) to reduce the distance to water collection points. This may increase the time used in productive activities, reduce school dropouts, and increase school completion, particularly for girls. Set up or help establish adequate water supply systems to minimize the incidence o f water borne diseases. Sustain existing public water supply systems, particularly in urban areas, through effective cost recovery, perhaps through user fees. Allocate the receipts to maintenance provision and system upgrading. Provide technical assistance at all levels-ranging from piped water systems in urban areas, to public fountains and private wells in urban and rural areas-to ensure that existing systems work properly and are well maintained. 4.86 Monitoring government expenditures and evaluating how effectively they are allocatedto thepoor will require monitoring systems that link spending to outcomes and targets. Evaluatingresource flows through selected sector Public Expenditure Tracking Surveys (PETS) can identify with leakages and increasedunderstanding o f the effectiveness o f sector resource allocation. Regular-well designed and well planned- household surveys can monitor poverty trends and their determinants. The timing o f those surveys should be aligned with the reporting needs for indicators inthe Poverty Reduction Strategy and other national programs. 119 5. GETTING THE STATE CLOSER TO ITSPOOR CITIZENS 5.1 Mozambique's record on governance and accountability is mixed. The legacy o f colonialism, decades o f protracted conflict, and centralized rule damaged trust, downward accountability, and state-society relations inMozambique. And the checks and balances developed during the postwar period have frequently been unable to safeguard the needs and interests o f citizens, particularly poor ones, leading to resentment, frustration, marginalization, and exclusion from the public sphere. Multiparty elections in 1994 were intended to support pluralism, but the winner-take-all system has left parliament weak and one political party dominating all three branches o f government. The judiciary-largely absent from the lives o f the poor-is plagued by allegations o f corruption. In addition to the state's poor technical and administrative capacity to respond to the needs o f the poor, the relatively weak capacity o f the media and civil society organizations has also limited social accountability.lg6 5.2 Sustaining poverty reduction and increasing inclusion require social, economic, and public institutions that respond to poor people's needs. The government's poverty reduction strategy puts governance squarely in the poverty debate inMozambique. One o fthe three pillars ofthe PARPA I1strategy focuses on governance, including policies and programs to fight corruption, solidify the rule o f law, improve protection from crime, raise the accountability o f the state, and support community and individual initiative.lg7 Three major instruments to pursue these are a new legal framework for decentralization, enhanced implementation o f the 1997 Land Law, and the 2005 Family Law. The PARPA strategy also supports measures to implement other laws to improve the localjustice system. What characterizes all these measures are attempts to bringthe state as a development actor closer to the lives o f Mozambique's poor (box 5.1). ~~~~ Box 5.1. Institutional Reform I s a Long Process, and Results Are Often Slow in Coming In the 2006 Poverty and Vulnerability Survey (PVS), commissioned for this report, poor communities and householdswere asked about localaccountability, about their accessto land and protection of rights in their lives and livelihoods, and about their experience with the legal reforms already enacted. This chapter uses this qualitative data to analyze poor people's experience of these reforms and the issues identified as important for implementation. Overall, the picture is one of a state intransition, with a longroad still to travel. The reforms have not yet addressed the disempowerment poor people feel toward the state. But signs of progress are emerging insome areas. 5.3 A forward-looking strategy for Mozambique has to lend equal weight to empowering communities and their localities as agents of change. It would best do so by: The findings here, summarizedin World Bank (2007, 16-17), are drawn from the 2005 USAID Corruption Assessment (USAID 2005) andthe 2005 World Bank InstituteAnti-Corruption Survey (Austral 2005). According to the latter survey, Mozambique'sscore has improvedonly slightly since 1998, and it remainsweaker than its Southern African neighbors. GoM2005. 120 e Makingdecentralization workfor thepoor-by devolving real power to local communities so that they can advocate for more responsive public service delivery, influence service delivery mechanisms, and hold local service providers accountable for their performance. Increasingaccess tojusticefor thepoor-by expanding poor people's opportunities to engage with institutions and ensuringthat the judicial system i s correspondingly responsive to systemchanges to makejustice accessible to the poor. e Makingpoorpeople 's landrights real-by formalizing land rights and titling, improving overall awareness o f citizenship rights about land, and strengthening the consultation process for community land. MAKINGDECENTRALIZATION WORK FORTHE POOR 5.4 The quality and responsiveness of governance reflect large geographical variations, particularly for service delivery. While service delivery has improved recently, access and outcome indicators continue to vary by region, with poorer results in the center and particularly in the north, as well as between urban and rural areas. The quality o f the state and its responsiveness to poor citizens' needs for service delivery underlie these geographical variations. Such inequities may undermine the recent poverty reduction and perpetuate marginalization if left unchecked, particularly for the rural poor. 5.5 Decentralization, including greater emphasis on citizen and community participation, is an important element of thegovernment's strategyfor improvedservice delivery.I8' By bringing decisionmaking closer to the people and creating spaces for citizen participation in local planning and implementation, the government's decentralization reforms are intended to enhance the responsiveness o f public services to locally perceived priorities; increase their effectiveness, quality, and relevance; improve downward accountability and local checks and balances; and ultimately raise the impact o f government policy on poverty reduction. 5.6 Decentralization is a key part of Mozambique's strategy to reducepoverty. To ensure greater responsiveness o f public policy and service delivery to citizen needs, community consultation and civic engagement have been institutionalized and mainstreamed in district development planning and budgeting. Despite this important and historic advancement, much remains to be done to strengthen decentralization's demand- side and social accountability dimensions. The risk is that neglecting these dimensions will cause state responsiveness to the needs of the poor to continue to falter. Given the high expectations, a decentralized but unresponsive state would lead to further popular Decentralizationis usedhere to refer to thejurisdictionaltransfer of functions andresponsibilities-administrative, fiscal, or political-to lower tiers of government. Conventionally,these transfers of responsibilitiesrange on a continuumof autonomy from deconcentration (low degree of autonomy) throughdelegation (slightly more autonomy) to devolution (high degree ofautonomy). Deconcentrution refers to the transfer of authority from the center to the periphery inacontext of territorial hierarchy. This is often seen as a way to strengthen and enhance the capacity of local governmentsto better implementthe policies of the central state andto improve state delivery of services. Devolution occurs where the central government cedes power andauthority to autonomouslocalgovernance structures and providesthem with sufficient resources (or the ability to generate and keepthem) for independentmanagement. Delegation falls somewhere in between. 121 frustrations, underminingpublic support for the reforms and the government's antipoverty agenda more generally. 5.7 While assessing the impact o f the government's decentralization strategy on poverty reduction would be premature, this section provides insights about the process o f implementation, particularly in areas that have received less attention, such as the demand-side dimensions o f decentralization reform, includingthe social mechanisms that citizens and civil society organizations use to hold governments and service providers ac~ountable.'~~ Key Features of Decentralization So Far 5.8 Decentralization in Mozambique uses a dualist approach of municipal devolution in urban areas (municipalization) and district deconcentration in rural areas. Although decentralization o f state functions has been a subject o f debate since the 1992 peace settlement, the state remained highly centralized untilthe last ten years. With the 1997 municipal legislation, the government adopted a two-pronged approach to decentralization, differentiating between urban and rural areas. Law 2/97 established 33 municipalities, comprising all 23 cities in Mozambique and 10 o f the country's 68 rural towns, to be administered by autonomous and elected local governments (auturquias), with devolved powers to manage the municipalities with their own budgets (funded through own-source revenues and intergovernmental transfers). 5.9 With the 2003 approval of the Law for Local State Bodies (L0LE),l9Othe rest of the country-the rural areas where more than 70 percent o f the population lives and that account for about 90 percent o f national territory-is governed by provincial and district governments as deconcentrated units of the central government. Rather than being elected by the local population, as the mayors and city councils of the municipalities are, provincial and district governments are appointed representatives of the central government. 5.10 while the Law for Local State Bodies can best be understood as a gradualist approach to decentralization in rural areas, the re orms have brought about four important changes,particularly at the district level." First, the LOLE establishes the district government as a legal entity and the basic unit for planning and development. Second, it makes the district a planning and budgeting entity for the first time, with the responsibility to prepare budget and expenditure proposals and the ability to receive budget allocations. Third, it recognizes district development plans as the principal instrument for planning and budgeting. Fourth, it regulates community participation- through formalized district consultative councils-in preparing, implementing and World Bank 2007, 19. I9OLaw 8/2003, Lei dos drg6os Locuis do Estudo,is often referredto as LOLE. Detailedorientationsregardingthe practicalimplementationofthe law are set out inthe associatedregulations approvedinApril 2005 (Decree 11/2005). The approach set out inthe law and associatedregulationsdraws on a range ofdistrict decentralization pilots, jointly supported by donors andthe government, and inparticular on the district developmentplanningprocesspilotedjointly by the governmentandthe UnitedNationsCapital DevelopmentFund(UNCDF), amongothers, indistricts across the provinceofNampula. 19'Barnes 2005, 8. 122 monitoring district development plans.'92 Intheory these plans would also draw on inputs from consultative councils and forums at lower administrative levels, ultimately comprising representatives from every community. 5.11 Substantial work remains to complete the legalframework for implementation. I n some areas regulations are missing, while in others, implementing regulations are contradictory. Specific attention should be concentrated on sectoral legislation that defines how the sectors are organized, operate, and deliver public services. To ensure congruence between sectoral legislation and that o f district deconcentration and municipal decentralization, a comprehensive review o f the sectoral laws should be carried out to address harmonization o f legislation. In addition, the legal review should also look into legislation that provides the general framework for public sector function, including the Public Finance Act and the Public Service Act. For fiscal decentralization reforms to be successful, these laws should be fully consistent and compliant with LOLE and the broader decentralization policy framework. 5.12 I n 2006 the central government experimented with fiscal decentralization by providing direct funding to district governments-in the form of the Local Initiative Investment Budget (OIIL). In 2006 the government made approximately $300,000 (7 million meticals) available to each o f the 128 districts for district and community investments identified with the help o f the consultative councils and forums.'93 The government further increased the funds available to district governments for 2007.194 However, the use to which these funds could be directed was narrowed in 2007 to focus uniquely on job creation and income generation and disengaged from any capital investment for social or economic infrastructure. In2008 OIIL is to be increasedwith the same restrictions and $90,000 per district added for infrastructure investments. The earmarking o f most o f the OIIL for job creation and income generation both restricts local choice and positions the district as a lending institution without frameworks or support. Combined with the lack o f sector deconcentration to districts in most areas, the room for local choice in allocating resources i s still highly limited. This underscores the importance o f capacity building efforts in order to pave the way for effective decentralization. RespondingBetter to Consumersand Citizens 5.13 Empowerment gives people and communities the opportunity and the capability to participate in social, economic, and political spheres-most immediately manifest through participation in local governance and local service delivery. To participate in local governance, citizens need institutionalized opportunities to influence local planning and policy processes, local decisionmaking systems, and accountability mechanisms linking decisionmakers and citizens. To participate in service delivery, people need opportunities to influence the mobilization and management o f resources and the delivery o f services through voice (consultative and oversight mechanisms channeling service user preferences to service managers and frontline service deliverers) and through Provincial Assemblies, not directly addressedby the law, are explainedingreater detail inthe 2004 Constitution. Working alongside the existing and appointedprovincialgovernments, they are designedto play arole inpreparingand approvingprovincialplanning instruments. The first set of elections were held inlate 2007. 193Barnes2005, 11. 194World Bank 2007, 19. 123 choice (the availability o f options to select among multipleproviders)-so that people and communities are not forced to accept unresponsive service delivery. Critical constituents o f an empowering agenda include access to information, capacity enhancement, and direct participation, with the following actions: 0 Informing communities. Strengthening capacity, Improving representation. Strengthening the incentives for participation and implementation. Learning from pilots. 5.14 Decentralization has laid the groundwork for improved downward accountability and opportunitiesfor popular participation in developmentplanning. At the municipal level, citizens can elect and hold to account their local representatives. Mayors and city councils in all 33 municipalities are directly accountable to their constituents for service provision and local revenue generation and use. In practice, however, the ability-and sometimes the willingness-of municipal governments to respond to their citizens' needs varies greatly (box 5.2). Given municipalities' continuing reliance on intergovernmental transfers rather than own-revenue collection, the autonomy that they exercise in practice remains questionable and potential1 undermines downward responsiveness and accountability to local taxpayers and citizens.795 The recently approved Municipal Finance Law provides some small positive developmentsfor municipalities but contains mostly negativeprovisions that essentially impose constraints on the municipalities. It does not allow for the reassessment of properties (though the full extent o f this will not be clear until the new fiscal code i s approved). It caps the rate for property taxes and provides an exemption for new properties for five years. And it limitsthe allocation o f investmenttransfers to 1.5 percent o f the national fiscal revenues. The potential for Maputo and other municipalities to raise revenues is significantly reduced, with consequences for institutional reform and service delivery. Transfers to municipalities currently constitute only 1 percent o f the national budget and are capped at 1.5 percent. In addition, Decreto 33/2006 provides for the transfer o f additional competencias (functions) to municipalities inthe social sectors. `95Barnes 2005, 3. 124 Box 5.2 MunicipalizationExperienceon the Ground Decentralization in urban areas has taken a different route from that in rural districts. With the municipal elections in 1998, the transformation of urban towns into locally elected auturquias began inearnest. The fieldwork sampledlocalities belonging to 4 of the 33 municipalities created since 1998, in Gaza (Ch6kwb and Manjacaze), Nampula (Angoche), and Zambezia (Quelimane). The results from two of the municipalities, Gaza and Angoche, are summarizedhere. Municipality of Chdkwt? Gaza ) Ingeneral, respondents inboth municipalities inGaza felt that the introduction of the autarquius had brought about improvements, especially in participation and accountability in decisionmaking. With local elections the communities have put in place local leaders who know how to addresstheir main concerns and difficulties. The municipalities have also been fairly good at involving local communities in decisionmaking, which has built trust in local institutions of governance. In Chdkwb the energy and water situation has improved somewhat, particularly for the minority able to pay for utility connections. The main roads through town have also improved. Even so, many respondents felt that more should have been done and that progress thus far mainly benefitedthe better off. Common was the impressionthat the government failed to live up to the promises of the elections and that much still had to be done to extendthe improvements in water and energy to the rest of the populace. A particular criticism was for the higher taxes levied by the municipal government, which-combined with the limited progress inthe delivery of public services-were seen as unjustified. But some respondents felt that the main reason for the lack of progress was the municipality's continuing low human, infrastructure,and financial capacity. Municipality of Angoche,Nampula InAngoche, Nampula, the situation is further complicatedby its dual process of decentralization. Since 1998 the town of Angoche has had an elected municipal council, and since the early 2000s the district of Angoche has taken part in the recently formalized district deconcentration pilot. Since 2002, however, the municipal council has been dominated by the national opposition party, Renamo (which boycotted the 1998 elections), while the appointed district government continues to be dominated by the national government party, Frelimo. Although the borders of the municipality and the district of Angoche are distinct, local political coexistence has confused administrative reporting structures and the lines of accountability between the local people and the state. Especially within the municipality, there are two administrative structures (including secretarios de bairro and their deputies and staff), one representingthe national government and another representing the new municipal government. While these structures may have enhanced the political checks and balances in Angoche by increasing local political competition and opportunities for community participation, the dual power structures may be confusing state responsibilities and lines of accountability. Despite these complications, respondents (from focus groups and semistructured and key informant interviews) agreed that the municipal situation has improved over the last few years. Even so, the area's strong political divisions and overlapping structures of district and municipal governance led respondents' perceptions of achievements to vary by the time and government in question: the national and district governments (Frelimo), the municipal government between 1998 and 2002 (governed by Frelimo), or the municipal government since 2002 (governed by Renamo). In this heavily Renamo area perceptions about achievements by Frelimo-dominated bodies were mostly quite negative. Many blamedthe government for the closures of factories and mass unemployment. One respondent noted, however, that the new municipal government has not improved things either because it has not beenable to replacethe lostjobs. 125 5.15 I nforming communities of their role in decentralized service delivery, the legislated consultative councils and forums at district levels and below (including administrative posts, localities, and villages) are unprecedented in opening spaces for citizen participation and civic engagement in rural development planning-despite misgivings about decentralization due to the exclusion o f rural areas (thus far) from the political decentralization and suffrage provided to urban areas.'96 The institutionalization o f such mechanisms o f social ac~ountability,'~~ with participatory mechanisms inscribed in laws or regulations, is extremely rare, even internati~nally.'~~Some gender equity is also institutionalized: women's participation inthe local councils i s guaranteed through a 30 percent quota, established in the 2003 Guidelines on Participation and Community Consultation and the 2005 regulations. Moreover, the reforms legally recognize community authorities as interlocutors between rural communities and the district administration, reversing decades o f legal marginalization o f traditional institutions and leaders in local g~vernance.'~~ 5.16 Beyond legislative advances, consultative councils have been implemented in all districts in Mozambique. In preparation for the disbursements o f the Local Initiative Investment Budget in 2006, a study by the National Directorate o f Planning verified the existence o f consultative councils in all districts and in 95 percent o f administrative posts.200 The study also suggested that most district councils had been involved in developing the district plans presented for Local Initiative Investment Budget fundingthat year. The PVS survey confirmed these results. Consultative councils also existed in all 16 districts visited for the PVS. While only recently created in the provinces o f Niassa, Zambezia, and Gaza, pilot councils have existed for quite some time in the districts o f Angoche and Erati-Namapa, among others, inthe province o fNampula. 5.17 With the accelerated implementation, however, the quality of citizen participation in development planning risks falling. At this early stage o f reform, anecdotal evidence from the PVS suggests that most consultative bodies were established from top-down government pressure-rather than from bottom-up demand-given their legal requirement as part o f the district deconcentration process. Indeed, in the districts and communities visited (outside the pilot province o f Nampula), some consultative bodies were still not functioning properly. Even within a single province (Niassa), the consultative forum was functioning in one o f the localities studied (Luelele, Mandimba) and not inanother (Lichinga). Strengthening State and Local Capacity 5.18 Weaknesses in capacity-in both the state and civil society at local levels- threaten to undermine reform's potential benefits. Moving beyond legal institutionalization is needed to ensure that participatory social accountability mechanisms 196See, for example, Serrano (2002), Braathen(2003), and Sitoe andHunguana(2005). 197Social accountabilityis definedas "an approachtoward buildingaccountabilitythat relies on civic engagement, i.e. in which it is ordinary citizensand/or civil society organizations who participatedirectly or indirectly inexacting accountability" (Malena 2004, 1). 19*Ackerman2005, 17. 199Kyed andBurr 2006. Cited inGoM and PAP 2007, 50-1. 126 work on the ground. Alongside the Local Initiative Investment Budget, large investments are needed in the human, technical, managerial, and administrative capacity o f local governments to manage the district development process and associated budgets and to boost downward responsiveness and accountability. The decentralization reforms are based on radically different ways o f conducting government and delivering social services. New skills and attitudes are required to facilitate participatory planning and to encourage citizens' voices, initiatives, and engagement. Some promising activities in evidence, such as Radio Moc and community radio broadcasts in the central part o f the country, have increased awareness on the district budget and planningprocess. Improving Representation and Participation 5.19 Beyond formalizing participation mechanisms in local government, participation must also be institutionalized in society itseIJ2" The extent to which local communities can meaningfully participate in the local budget and planning cycle i s uncertain. This should not, however, lead to the conclusion that civil society i s not capable o f participating in the development planning process. Required are parallel investments in building the capacity o f citizens, civil society organizations, and even the media to interact with local governments and hold them accountable. An important element o f such capacity building i s to improve citizen access to the information and knowledge needed for accountability-on budgets, procurement, service delivery costs, and user rights (including citizenship rightsmore broadly). 5.20 To engage in social accountability mechanisms, the very poorest in the communities may even need assistance in building their capacity to aspire202-beyond the organizational, technical, and basic reading and math skills that community representatives will need to master to effectively participate in these consultative bodies. The education and training o f citizens and civil society need to be more fully integrated in the reform process to raise quality o f citizen participation inpreparing, implementing, and monitoring development plans. 5.21 The emphasis on rapid implementation of reforms may also risk undermining the inclusiveness and representativeness of consultative bodies. International experience suggests a tendency for participatory initiatives to be subject to elite capture, often by the most articulate, well off, well connected males and leaders in a community.203Without mechanisms to ensure greater transparency and inclusiveness, the benefits o f decentralization thus risk capture by local elites, reinforcing existing local inequalities rather than improving the lives o f the poor.204 5.22 Anecdotal evidencefrom the PVS suggests that this may indeed be a problem in some communities. Representatives o f the consultative bodies in the communities visited usually included traditional and religious leaders, political party members, and representatives o f other organizations (economic associations and social organizations, including female and youth organizations). The opinions o f religious and traditional 20'Ackerman 2005, 18. 202Appadurai 2004. 203Ackerman 2005,22. 204Ackerman 2005. 32. 127 authorities-coupled with education level and expressive ability-were often determinants in selecting representatives for the consultative bodies. Despite the gender quota, women were highly underrepresented insome communities. 5.23 The overall impressionfrom the PVS is that opportunities to participate had not changed significantly despite the recent introduction of consultative bodies. Consultation and participation protocols continue as before for most o f the people in the areas studied, and the impact o f deconcentration appears not yet to have trickled down to ordinary community members. Building meaningful participation into the process takes time and needs credible and continual monitoring and assessment. 5.24 Impact assessmentsfrom the districts that took part in the decentralization pilot initiatives in Nampula province suggest, however, that opportunities for participation have improved since the introduction of participatory district development planning.205This may suggest that with time the inclusiveness o f the more recently constituted consultative bodies may also be enhanced. Even so, the assessments from Nampula also showed that the consultative bodies-despite these improvements-still tend to be dominated by more educated, older, male community members. 5.25 Greater emphasis is needed during reform implementation to improve the inclusiveness and representativenessof the consultativebodies. First, efforts are needed to overcome the many obstacles to participation facing most community members. These obstacles range from cultural factors (with age and gender often important) to constraints on time and availability to skills and comfort with participating. Overcoming such obstacles will, however, require substantial long-term investments. Second, the demand for citizen participation needs to be strengthened. This again requires investments in the capacity o f local communities and civil society organizations to participate and in the capacity o f local government officials to mobilize and integrate citizens in district development planning, implementation, and monitoring. Third, to overcome problems o f elite capture, efforts should mobilize informal social organizations and community groups (such as water user groups, agricultural producer organizations, and the like) and better link these potentially more representative groups with the formal consultative committees. Fourth, the ability to represent a community and the effectiveness o f community representation rely on trust. To build and sustain community trust, transparency is requiredinthe flow of informationabout decisions and debates inthe consultative bodies. StrengtheningtheIncentivesfor Participation andImplementation 5.26 Beyond the implementation challenges, participation should be broadened beyond the district planning and budgeting stage to ensure greater downward responsiveness and accountability. By limiting citizen participation to consultations duringthe district planning stages, their incentives to participate and their ability to hold district, province, and national governments accountable are weakened. 5.27 District governments are legally mandated only to consult citizens (through district consultative councils), not to integrate their priorities into district development plans. And despite the recognition o f district development plans as the principal 205See UNCDF (2004), IlalandChisvo (2005), andAllen andDupont (2006). 128 instruments for planning and budgeting, provincial and national governments are not requiredto integrate these plans into their development plans. 5.28 Incentives toparticipate in the districtplanningprocess are undermined by the slow pace of fiscal decentralization to districts. Except for limited discretionary funds available to district governments, most budgetary decisions on district investments are made nationally. So, budgets proposed locally by the district government4rawing on the input o f district consultative councils-are merely proposals that may be "aggregated into a unitary budget at a central level, which is then submitted to parliament for approval."206 District planning and budgeting remains more o f a mapping and priority- setting exercise-at least in the short run-where citizens and communities are asked merely to make suggestions. In 2006 the central government initiated a district-level development grant (or allocation), which provides financial resources to the districts for financing projects that are locally defined as priorities through a participatory process. In 2006 and 2007 each district was given MTn 7 million (approximately $300,000) for this purpose, though the amount i s expected to increase slightly for the 2008 budget. There i s no clear methodology for allocating these financial resources among districts, and there is confusion about the scope o f activities that the MTn 7 million can be directed toward. 5.29 Perhaps more worrisome, however, is that few incentive mechanisms are available to district consultative councils--and to citizens more generally-to hold district governments accountablefor local decisions,plans, and investments. Not even the crude mechanism o f holding district officials accountable through the ballot box is available to citizens in rural areas. So, while consultative councils may have addressed part o f the democratic deficit at the district level, they should not be seen as "a substitute for democratic local government" inthe long term.207 Learningfrom Pilots 5.30 Limited opportunities for participation in district planning and budgeting hinders theprocess, outcomes, and impacts of districtplans. At the very least improved communication and more transparent information flows are needed "to ensure that those who participate in the planning process are adequately informed o f the outcome o f their involvement."208 The exact role o f consultative councils and forums in monitoring and evaluation is unclear in the legislation and accompanying regulations-suggesting a top- down and state-driven approach. A risk i s that the concerns o f citizens and service users will not be appropriately taken into consideration in district planning and budgeting. Alongside community participation in planning, therefore, is a need to institutionalize greater civic engagement in monitoring and evaluating district plans and implementation. International experiences suggest several strategies to strengthen participatory monitoring and evaluation o f public service delivery, including citizen report cards, social audits, and community scorecards (box 5.3). 5.31 At the municipal level it is imperative to stick to the good practice rule that finance should follow function. According to Decree 33/2006, municipalities will be 206Barnes 2005, 10. 207Barnes 2005, 13. 208Ibid. 129 responsible for social services. Legislation also mentions that the financing for the decentralized provision o f these services will be secured. But the decentralization o f expenditure responsibilities should not move too far without the corresponding allocation o f resources. Thus, it is imperative to establish a stable, predictable, and transparent transfer system with a clear allocation o f local revenue sources (taxes and fees) by level o f government. Box 5.3. Community Scorecards in Malawiand Maputo The community scorecard methodology was first developed in Malawi by CARE International through its Local Initiatives for Health Project in 2002.' Malawi has a long history of deficient healthcare services for the poor. The main goal of the project is improving the provision of health services to the rural poor by empoweringusers. The model applied in Malawi includes four elements.2 First, facilitators organize community meetings with villages near the health center to be evaluated. Participantsare asked to talk about their health problems, their access and use of health services, and their opinions of the health center. The facilitator then helps the participants design a list of indicators to evaluate the health center. Finally, the participants are asked to rank the health center's performance on each indicator. Second, the staff at the health clinic go through a similar process. They are asked to discuss the present situation at the clinic, develop a series of indicators, and rank their performance on these indicators.Third, an "interface meeting" is organized, where community members and clinic staff present their respective scorecards, compare the outcomes, and work together to design solutions to the common problems identified. Fourth, the action plans are implemented, including follow- up visits. There is evidenceof significant improvement inthe health center's services betweenthe interface meetings.Most of the improvement can be attributed to the community scorecard. Almost all the indicators received higher scores in the second scorecard, with significant improvement in "respect for patients," "listening to patients' problems," "honest and transparent staff," "giving priority to serious cases," "no discrimination in providing supplementary nutrition," and "no preferential treatment."' In 2006 the Maputo municipal government experimented with this tool as part of the participant process to develop the urbanreform strategy. The process was well received andjudged a success by policymakers. INCREASINGACCESS JUSTICE TO FORTHE POOR 5.32 Improving the understandingof how poor people can access the resources and assetsfor a secure and sustainable livelihood requiresfocusing on the systems of rights and justice available to the poor.209In recent years the government has put in place a range o f legislative reforms that have begunto change the landscape o f rights, justice, and empowerment for the poor. The most significant i s the 2004 revision o f the constitution. Other reforms-such as the 1997 Land Law and the 2005 Family Law-are also directly *09Moser and Norton2001, 21. 130 relevant to the lives o f poor people, regulating their access to key resources inside and outside the household. Decentralization is also changing the institutional landscape and the public policy approach aimed at realizingmany o f these rights.210 5.33 Theformaljustice system remains inaccessible to most of thepoor.21'There are several constraints to access. A recent survey o f national perceptions o f the state o f governance and corruption212suggests that the four main obstacles are physical access, distance, cost, and c~rruption.~'~General lack o f awareness and knowledge o f rights is also a factor.214This was recognized even by the Chief Justice o f the Supreme Court in his speech opening the 2005 Judicial Year: "There exist...factors which limit the degree o f access to justice., .like ignorance or deficient awareness of the law on the part o f citizens, the lack o f ajuridical culture, as well as corruption more generally. Access to justice presupposes a general awareness o f the rights and freedoms accordedto citizens which we lack inthis country."215 DISSEMINATING INFORMATIONONRIGHTSAND RESPONSIBILITIES 5.34 ThePVS brought to thefore thefact that the majority of poor men and women interviewed had very little insight about their specific rights and responsibilities as citizens.216The few that recognized the existence o f rights highlightedthe rights to vote, to live in freedom, to life, to work, and to health, food, and education. Some emphasized citizenship responsibilities-to pay taxes and to carry personal identifi~ation,~"a responsibility that carries its own set o f problems for the poor (box 5.4). 5.35 The majority of poor households are even unaware of recent laws directly relevant to their livelihoods and assets, such as the land and family laws (box 5.5). Findings from the PVS suggest that people's low awareness o f their rights extends to land and inheritance rights. 5.36 Better dissemination of information on citizens' rights and responsibilities is critical. Giventhe substantial changes to the country's legal code since the early 1990s- 2'o The discussion in this sectionreliesprimarily on the results from the 2006 PVS and arecent study ofthejustice sector inMozambique(AfriMAP 2006). 211 The officialjustice system includes courts, prosecutors' offices, state-sponsoredlegal aid, the state police, andjails. 212Austral 2005. The study, commissionedas a World BankInstitute Anti-Corruption Assessment, surveyednearly 4,000 individuals, including representatives from 2,475 households, 486 firms, and 992 public sector functionaries. 213 See appendix table A5.1. 214Delays are also akey constraint to access.According to AfriMAP (2006, log), civil actions at the primary instance may take up to four years. Those who enforce the laws-judges, prosecutors, andpolice-also lack basic information on the laws they are meantto enforce. A study of sittingjudges Mozambique'sNationalJudicial Training Center found that improvededucation andbetter access to legislation,jurisprudence, andjudicial doctrine were cited as the most importantactionsthat could be taken to improvethe judiciary. See Centro de FormapLoJuridicae Judicihia (2005,62). 215 Speechof Supreme Court President, Mario Mangazeon the openingofthe 2005 Judicial Year, pp. 12 and 13 as cited inAfriMAP 2006. 216For the purposes ofthe survey these were definedas the constitution,the 2005 Family Law, andthe 1997LandLaw and its associated regulations. 217 See specifically the results from the communities studied in Gaza (UEM 2006, 134). 131 particularly in small-scale commerce, natural resources, and family matters-a sharper focus i s needed on disseminating these new laws and their implications for the rights o f the poor and on improving access to legal information more generally. New laws are usually disseminated through the Official Bulletin. But the sporadic and limited publishing o f the bulletin has hindered widespread dissemination o f laws and other relevant legal information. Moreover, few laws have been translated into local languages and disseminated widely,218which causes major difficulties in a country where only 8 percent speak Portuguese, the official language, as their native language and 27 percent speak it as a second languagea219 Box 5.4. NationalIdentificationCards and ClaimingCitizenshipRights Being able to claim the most basic citizenship rights in Mozambique requires a state-issued personal identification. An identification card is required to receive social insurance support. Those without an identification card are subject to harassment and possible detention by police. Even so, many respondents notedthat they do not have an identification card (because they never obtainedone or becausethey have lost one) and face many difficulties obtaining one. To obtain an identification card in the locality of Macuacua in Manjacaze district, Gaza, one needs first to receive a pass from the president of the locality, then a birth certificate certified by the registrar and notary inthe locality where the individual was born. Next, an application has to be submitted to the municipio of Manjacaze to apply for the identificationcard. After six months waiting time, the card is then issued. The costs are prohibitive for many, leaving them unable to claimtheir rightsand benefits as citizens.' 1. UEM 2006. 5.37 Without clear and unbiased legal information, the poor are particularly vulnerable to misinformation or even elite capture, especially in rural areas. Indeed, the PVS fieldwork found that access to information remains a great challenge to enhancing people's ability to realize their rights and to improving governance more generally. The main sources o f information for most people are local meetings with community authorities and local government officials and informal oral updates from family members, neighbors, and friends. While these methods may be appropriate given the widespread illiteracy in rural areas, they are also more prone to capture by vested interests.220 5.38 The costs of accessingjudicial services are also a major obstacle, cited as the second most important constraint among households. Court fees, costs o f legal representation, and transportation and accommodation for those living far from the courts-most o f the population-are usually prohibitive.221 'I8 The 1997 LandLaw was, inthis regard, an exception. * I 9M E 1996/97.Besidesthe official language(Portuguese), Emakhuwais spokenas aprimary languageby 26 percent of the population, Xichangana by 11percent, Elomwe by 8 percent, Cisenaby 7percent, Echuwabo by 6 percent, other Mozambican languages (32 percent), andunspecified(1 percent). 220 UEM2006,9-10, 134, and 153. There have beensome attemptsto provide legal informationthroughcommunity radio, though this canbe beyondthe reachofmany poor families due to lack ofpower or inability to buy radios or batteries. '*'AfriMAP 2006. 132 Box 5.5. The Family Law and IntrahouseholdRelations,2006 Parliament passedthe new Family Law in2003 protecting a range of women's rights and for the first time legally recognizing customary marriage, thus offering official sanction to property claims and custody rights for women. The Family Law, signed into effect by the president in March 2005, asserts that both spouses are responsible for the family and can represent the household. Inthe past a woman required her husband`s consent before taking a paidjob. The law stipulates marriage as a monogamous union, with the minimum marriageage for both sexes at 18 years. Domestic violence and adultery constitutegrounds for divorce, as does abandonment ofthe conjugal home for a period exceedinga year. The Family Law addresses intrahouseholdrelations directly by establishingas principle that the household is headed by both spouses, administering the couple's property equally. If widely implemented, it could have significant effects for the rights of women, both in terms of property inheritance rights and decisionmaking status in the household. Results from the qualitative part of the PVS suggest that there is very limited awareness of the existence of the new Family Law and its ramifications. The survey looked at questions such as perceptions of ownership and inheritance rights and the knowledge of where and how to appeal decisions for land concessions, with a particular focus on perceptions of women's rights to own and inheritproperty. So, what is the current decisionmaking status of men and women in the household?Results from the quantitative part of the PVS indicate that gender continues to be a significant determinant of household decisionmaking power. There are important variations, however, by religious affiliation, geographical location, and wealth ofthe household. The sex of the decisionmaker is a strong determinant of intrahousehold decisionmaking responsibility. Both in rural and urban areas, men are predominantly making the households' expenditure decisions on education, health, and food. Women have a relatively smaller, though still considerable, disadvantage indecisions over food expenditures.When asked separately, male heads answer marginally different than female spouses when queried on decisionmaking power, with women reporting a higher degree of both spousal and joint decisionmaking than men. But the overall picture remainsconsistent. Urban women have a slightly higher say over expenditure decisions than rural women, but the difference is not large. When asked, both male heads and spouses report relatively morejoint and spousal decisionmaking in urban than in rural areas, particularly on education and health expenses. There is also a difference in rural-urban decisionmaking in the relative wealth of households: women in relatively wealthier households are perceived to have more say over expenditure decisions than women in relatively poorer households, but only in urban areas. Religious affiliation appears to be a determinant in the decisionmaking power between men and women in households. Male Muslim household heads report making expenditure decisions to a larger degree than male heads in non-Muslim households. The pattern is consistent over all categories of expenditure, and particularly so inrural areas. There are potentially important welfare consequences from this discrepancy in decisionmaking power between men and women. Gender-based decisionmaking is not only a problem from an equity or rights perspective: there can also be potentially important development consequences. Evidence from other countries suggests that mothers' control of the household income has a significant effect on consumption expenditure, children's schooling and health, and households' labor supply (Khandker 1998; Kennedy and Peters 1992; and Thomas 1990). The intrahousehold decisionmaking discrepancy betweenmen and women can be one explanation behind the finding that nutrition inequality increased inMozambiquewhile income inequality decreased. 133 5.39 Distance and geographical marginalization remain key constraints to the poor's access tojustice. Inthe 2005 governance and corruption survey, distance was cited as the most significant constraint to accessingjustice (table 5.1). Courts and other justice institutions are based mainly in urban areas and provincial capitals. In theory, courts should exist ineach province and district-achieved inprovinces butnot indistricts. Only 93 district courts exist for 128 districts.222 Physical distances within districts are also substantial. In Chilembene in the province o f Gaza the nearest district court i s in Ch6kw6, a city more than 50 kilometers away-which presents substantial difficulties, particularly for the poor, who may not be able to afford transport and accommodation. Inthe north the distances are greater, infrastructure i s more limited, and services are fewer, so the constraints facing the poor may be even greater. Table 5.1. HouseholdConstraints in Access to Justice (percent) Not Important Not Very Somewhat Very Constraint a t i l l Important Important Important Important costs 24.1 16.6 18.3 18.0 22.5 Bribes 26.6 16.3 27.6 13.3 16.2 Corruption 19.1 17.6 23.7 18.1 21.4 Incompetence 23.0 17.4 25.4 15.8 18.4 Waiting time and delays 20.2 17.1 22.9 20.1 19.6 Complexity 20.9 17.8 26.2 19.0 16.1 Execution 17.9 16.6 26.3 19.1 20.2 Distance 25.1 15.3 17.1 13.8 28.6 Source: Austral 2005, table 18.3. Making the Justice System Fair 5.40 Bribes and corruption areparticularly problematic in accessingjustice for the poor. As many as 35 percent o f households interviewed expectedto have to pay a bribe to receive judicial services. The police were rated as the fourth most corrupt public institution in Mozambique, with as many as 19 percent o f households saying they always have to pay a bribe to the police in returnfor a service.223Traffic police were rated as the second most corrupt after customs officials. While the proportion o f households considering the courts corrupt i s lower, it remains significant.224 Even the public prosecutor in Mozambique recognizes corruption injustice and the problems it creates: "Justice sector officials have easy access to money through bribes; this i s why the citizens inthis country don't approach thejustice system to complain about violations o f the law, given that the perpetrators o f the crime have a financial advanta e over them. Inthis situation, the rule o f law doesn't exist and cannot exist.,322 8 222This figure understatesthe problem, as some ofthe more urbanized districtshavemore than one district court. 223Austral 2005, table 22.3. 224Elevenpercent of householdsclaim that they always have to pay abribeto receive assistance from the courts. 225Cited inAfriMAP (2006); translationby the author. 134 5.4 1 Beyond these constraints to access, however, the representativeness and inclusiveness of thejustice sector at the community level also present concernsfor the poor. Both the 2005 governance and corruption survey and the recent U.S.Agency for International Development (USAID) corruption assessment identified the justice system as particularly problematic.226The 2005 survey showed a general perception o f serious problems with the justice system, particularly in how much it represents or defends the interests o f the poor.227More than 60 percent o f households felt that the weak and the poor are unable to get their interests considered within the justice system, and more than 40 percent felt that the system is insufficiently independent from the government, manipulated by economic interests, and unfair overall as a result.228The quality o fjustice services ranked as particularly poor.229Indeed, 42 percent o f households rated the quality o f police services as either very bad or bad.230The court system fared only slightly better, with 29 percent o f households rating its services as very bad or bad.231 5.42 These perceptions were largely echoed in the qualitative PVSfieldwork. The state police, for example, were uniformly accused o f corruption, and many respondents complained o f having to pay fines or bribes to the police ifthey failed to present personal identification or ifthey needed the police to investigate a theft or robbery. Petty traders in Quelimane, Zambezia, also complained o f having to regularly pay fines and bribes to the municipal police when they did not have the correct licenses for their operations, often wiping out their daily incomes. The fieldwork in N i a ~ s a ,G~a~ ~ a and N ~ m p ~ l a ~ ~ ~ , ~ ~ a confirmedthese trends. Establishing Proceduresto Link the Formal and Customary Systems 5.43 Marginalized from the formal justice system, the majority of households- particularly in rural areas-rely on informal and traditional mechanisms of dispute resolution. Findings from the 2005 survey suggest that only 5 percent o f households 226 USAID 2005, 14. 227 See appendix table A5.2. 228 The majority ofrespondentsinall cases-exceptfor when askedabout the ability of the weak andthe poor to get their interestsdefendedby thejustice system-claimed no strong opinion either way. 229 Austral 2005, table 13.4. 230 As many as 64 percent of householdsconsideredthe quality oftraffic policing either very bador bad. 23' This compares with 13.6 percent of householdsrating education servicesas bador very bad(the lowest for any service). 232UEM2006, 35. 233UEM2006, 137. The Gazafieldworkhighlightedthe randomness of injustices targetedat youth. The country's highly formal andcomplex civil andcriminal codes-dating back to the colonialperiod-largely remaininforce. In urban areas the lack of an identification card was acause for corruption and ledto the detention ofyouth unable or unwilling to pay. These youth then awaited formal charges injail, throughthe "preventative prison" process, intermingled with hardenedcriminals. A 2003 Save the Children Norway study (cited inUNICEF 2006, 176) showed that the average age among the children andyouth detainedwas 16-17. The Mozambicancriminal code (article 42) states that those under age 16 are not criminally liable. Ofthejailed childrenandyouth interviewed inthe study, 18 percent were younger than 16, including children as young as 12. The study also showedthat the principles of due process had not beenfollowed inmany cases. For example, 45 percent ofthe children reportedhaving beendetained illegally for longer than the official 48-hour limit, with an average detention ofthree months. Ofthe 38 percent ofthe children who hadbeensentenced-to sevenmonths inprisonon average-many hadbeengiven no information on their trial. 234UEM2006,63. 135 resorted to the formal courts during the previous year. In rural areas the proportion was even lower (3 percent). Because the state police have a greater presence in rural areas, the proportion o f households that sought their help during the previous year was slightly higher but still very low (8 percent nationally and 4 percent in rural areas).235Instead, most citizens resort to informal and traditional mechanisms o f dispute resolution, such as community courts, traditional authorities, and local leaders. 5.44 Community courts are ofen the most visibleface of justice for households. Perhaps the biggest gap in the system o f rights and justice has been the failure to include community courts in formal The more than 1,650 community courts across the country are "the most widespread officially recognized judicial forum in Mozambique."237But they have no formal links with the formal courts, and they receive no material, financial, or human resources support from the government. And most community judges have little or no legal training, and many have only a very basic understanding o f the laws that they are supposed to apply. Table 5.2. Resolution of Problems by Entity in Urban Areas, by Gender of Household Head and Tercile, 2006 (percent) Male Heads of Household Female Heads of Household Item All Lowest Highest All Lowest Highest Family 16.5 19.4 9.6 11.2 16.3 11.8 Traditional Authorities 7.6 6.5 4.8 7.2 12.2 0.0 Local authorities 57.2 58.1 60.2 65.6 51.0 70.6 Community police 4.7 3.2 3.6 4.0 10.2 0.0 State police 12.3 11.3 19.3 11.2 8.2 17.6 court 0.8 0.0 2.4 0.0 0.o 0.0 Church 0.4 0.0 0.o 0.8 2.0 0.0 Do not seek help 0.4 1.6 0.o 0.0 0.0 0.0 Total 100.0 100.0 100.0 100.0 100.0 100.0 Source; Authors' calculations basedon data from PVS 2006. Note: Sample not representative. 5.45 Most laws applied are based on customary practice and, as one community judge from Nampula noted, "common sense.'238As a result, there is no judicial supervision or formal control. The 2004 constitution, however, recognized the existence o f community courts, which bodes well for future legislation to provide a better framework for integrating them into the formal judicial system. 235Austral 2005, tables 13.1 and 13.3. 236AfriMAP 2006. 237AfriMAP 2006. *38AfriMAP 2006. 136 Table 5.3. Resolutionof Problemsby Entity in RuralAreas, by Gender of HouseholdHead and Tercile, 2006 (percent) Male Heads of Household Female Heads of Household Item All Lowest Highest All Lowest Highest Family 16.6 18.8 14.3 9.1 11.6 15.4 Traditional Authorities 32.2 29.2 33.8 33.8 53.5 7.7 Local authorities 44.2 47.9 41.6 54.5 32.6 76.9 Community police 1.5 0.0 3.9 2.6 2.3 0.0 State police 4.0 4.2 5.2 0.0 0.o 0.0 0.0 court 1.5 0.0 1.3 0.0 0.0 Church 0.0 0.0 0.0 0.0 0.0 0.0 Did not seek help 0.0 0.0 0.0 0.0 0.0 0.0 Total 100.0 100.0 100.0 100.0 100.0 100.0 Source: Authors' calculations based on data from PVS 2006. Note: Sample not representative. 5.46 Despite being closer to the needs of communities, however, community courts do not always overcome the problems associated with theformal system. Indeed, some respondents questioned the fairness and integrity o f community justice systems. A male head o f household in Pebane, Zambezia, for example, argued that "justice" could be bought-not only at the formal level, but also from the community court: "In the community court you need to pay to be `right' and get justice. Even if a person i s in the wrong he'll win the trial if he pays. On top o f that, the person who then loses the trial still has to ay an indemnity fee, including a percentage which the judge demands for himself."23 B 5.47 Traditional and local authorities often alsoplay an important role in resolving community conflicts and disputes. In rural areas traditional authorities-especially the traditional chief, but also religious leaders, traditional doctors, and other informal agents-continue to be important in facilitating access to justice at the community level (box 5.6). Particularly in communities in the center and north where Islamic law plays a greater role, religious leaders also tend to exert substantial influence. Exactly how these traditional agents work varies by community, but they tend to rule through community councils (with representatives o f different community leaders and elders) and to be based on customary practices and law. And no mechanisms exist to ensure compliance with national laws (including human and women's rights) and legal procedures (including procedures o f appeal, and redress). 239UEM2006. 103. 137 Box 5.6. TraditionalAuthorities and Their HistoricalRole in Dispute Resolution Portuguese colonial rule allowed for a variety of different social, economic, and political models at the local level. Where issues were not relatedto colonial rule, per se, the state referred back to customary laws and authorities. Beginning in 1920 under the Indigenato regime, the colonial government adopted the customary systems already in place. The colonial rulers used these regimes as tools to control rural society but allowed the local authorities a great deal of leeway. This led to a wide range of legal practices and did little to create a unified body of customary law.' The traditional chiefs functioned as the administrative auxiliaries of the colonial government, presiding over the community councils. These councils, in turn, were linked to Portuguese rule and embodied legislative, administrative, executive, and judicial functions. The chiefs were of noble lineage, but when they showed resistance to the colonial power or disobeyed orders, they were subject to removal and replaced by individuals more favorable to the colonial regime. The new authorities were often brought from outside the ruling hereditary system. These politically motivated ruptures in the chief system created problems of legitimacy, eroded cultural identity, and eliminatedthe basis for resolving local disputes. With Mozambique's independence, colonial institutions, including justice and customary traditions, were rejectedas backward by the ruling party, Frelimo. Traditional authorities aligned with the colonial system, discredited by the Marxist government, were replaced by dynamizing groups and the popular courts-with the objective of creating a fairer and more equal Mozambique under the aegis of the executive branch. The popular courts remaineduntilthe 1990 constitution, which embraced democratic principles and an independent judiciary. The new constitution created a three-tieredjustice system similar to the judicial structure at the end of the colonial period, with the district courts as the lowest level ofjustice. The delinking of the popular courts from the official justice system meant that some communities no longer have official courts to adjudicate disputes. Traditional or community courts may still function either within the dynamizing groups or the local authorities. While legitimate in the eyes of the local populations, ties between the community courts and the local authorities may perpetuate inequalities in communities, running counter to constitutional guarantees and national laws. 1.Meneses andothers 2003. 5.48 Local authorities (including secreta'rios de bairro in urban areas and secreta'rios dapovoapZo in rural areas), by virtue of their roles as elected leaders with administrative mandates, of en mediate justice and resolve community disputes. But much like the traditional councils, how much these mechanisms follow established legal procedures is unclear. The 2004 constitution does not explicitly recognize the role o f traditional and local authorities within the justice system, despite the importance o f these actors in facilitating access to justice to most o f the poor. But by confirming the principle o f legal pluralism, it allows for future legislation to better integrate such mechanisms into formal justice.240 5.49 Both urban and rural households are more likely to trust local authorities and informal and traditional mechanisms of dispute resolution than formal structures of justice.241The state police and the courts play a greater role in urban areas and among 240AfriMAP 2006. The land law precededthe 2004 constitutioninthis approach, recognizingthe role oftraditional authorities andcustomary laws inregulatingaccess to land. 241See appendix tables A5.3 andA5.4. 138 male-headed households and richer households. Poorer households, by contrast, are more likely to revert to family, traditional authorities, and community police. While concerns are often raised about the treatment o f women by traditional authorities and customary practices, a clear majority o f female heads o f households prefer traditional and informal structures o f authority to those o f the formal system. 5.50 Most households and communities usually rely on a complex combination of mechanisms for dispute resolution. The most common problems in the PVS were physical aggression caused by alcohol abuse, marital conflicts (including adultery) and domestic violence, curses from witchcraft, robberies, and disputes over water and land. Which institution community members approached varied by conflict type, severity, and urban-rural location. 5.5 1 Internal family problems are most ofen dealt with in the family. Most other issues are dealt with by local leaders, includingtraditional authorities, religious leaders, or administrative officials (especially the secretbrio de bairro and secretbrio de povoaqfio), depending on the problem. If a problem cannot be resolved, it is often passed to the councils o f elders or community courts, considered the highestinstances o fjustice inmost communities. Only if an issue is deemed severe (such as a large theft or physical aggression) will it be passed on to the state police or the formal courts (figure 5.1). 5.52 Mozambique's constitution is in itself enlightened and generous in terms of providing rights to its citizens, but the government is limited in its capacity to ensure these rights are enjoyedfreely. Mozambique's justice sector faces a range o f constraints in meeting its constitutional mandate to enforce the law, protect the rights of Mozambique's citizens, and provide checks and balances against the excesses o f the executive and legislative branches. Priorities for judicial reform include: 0 Attention to criminal justice. From the traffic fines imposed by traffic police- reportedly the most corrupt institution in Mozambique's justice ~ e c t o r ~ ~ ~the- t o apparent upswinginpetty thefts and violence inurban and rural areas, security i s a pressing issue. Injustice is perpetuated through petty corruption, poor legal aid and prison conditions, as well as a fundamental lack o f knowledge regarding one's basic rights. Stronger sector accountability. There has been considerable donor interest and financing o f Mozambique's justice sector for a number o f years. But changes and concrete results have been slow in coming. Stronger sector accountability is needed, above all, by those who use the system-be it through improvements in the collection and reporting o f data, prison conditions, dissemination o f legal information, and the rightto legal aid, among others. 242UTRESP 2004. PesquisaNacional sobre Governa@io e ConuppiTo. 139 Figure5.1. Structure of ConflictResolution in Erati-Namapa,Nampula I District court I t I State Dolice I t Community court r- Religious leaders Secrethrio de bairro Legal information in local languages. The government and the donors have put considerable energy into bringing the country's legal framework up-to-date. But much less concerted effort has gone to training and disseminating these laws inthe official language, Portuguese. Unbiased legal information-presented in a simple format and in local languages-would improve both individual and collective economic and social decisionmaking ineven the remotest o f villages. MAKING POORPEOPLE'S LAND RIGHTSREAL 5.53 Land isperhaps the most important resource available to thepoor, although its relative abundancein most of Mozambique means that it is not a binding constraintfor continuing growth and poverty reduction. Mozambique has an estimated 36 million hectares o f cultivable land, but only about 4.9 million hectares were cultivated in 2003. Moreover, the total land used for agriculture i s expected to remain at less than 40 percent through 2020.243Even so, improving the security o f land tenure may help ensure access to land for the poor, particularly poor women. Secure tenure i s an important vehicle for increasing the productivity o f land and enabling the poor to use their land for investment and collateral. 5.54 At the same time, land is an important resourcefor national investment and economic growth, especially that led by the private sector. Simple, fair, and transparent procedures for acquiring land-use rights are thus a requirement for "serious long-term private sector investment."244 Balancing these two concerns has to be at the forefront o f an agricultural and rural development focused growth and poverty reduction strategy for Mozambique, and is also at the center o f the government's second PARPA. 243Chemonics(2006, 21), citingFA0 and World Bankdata. 244Chemonics2006. 140 5.55 The legal instruments defining land rights in Mozambique have evolved rapidly. After the 1992 peace accords a new land policy was adopted in 1995, and the new land law was passed in 1997. New rural land regulations appeared in 1998 and a technical annex was added in 1999. Both the first and the second PARPAs also established principles for natural resource use, including land, with the explicit aim of improvingthe livelihoodsofthe ruralpoor. 5.56 The 1997 Land Law and accompanying regulations were to protect the land- use rights of small-scale farmers by recognizing the legitimacy of customary systems and toprovide investors with secure long-term rights while maintaining state ownership of all land. The law broke groundby recognizingaccess rightsto landthrough customary and good faith occupationas equivalentto state-allocatedland-use and benefitrights.The law introduced legal pluralism and equivalence between traditional and formal justice. While the law encourages formalizationofuse rights, formalizationis not requiredfor the rights to be legally recognized.245And by emphasizingthe rights of individuals to take their land out of customary jurisdiction and by affirming the primacy of constitutional principles,the law aims to guarantee women's legalequality. 5.57 The law alsoprovidesprivate investors with secure rights through a renewable SO-year state leasehold. Before acquiring such land, however, a community consultation must set out the conditions for the takeover of the community land-useand benefit rights. Intheory, this consultationmechanismformalizes the participationo f local peoplein land allocation decisions, and hence "should protect local rights, guarantee local livelihoods, and generate new resources for local community Owned by the state, the landcannot be boughtor sold, but any privateor community assets on the landcan be sold, with the underlying land-use and benefit right merely transferred to the new owner.247 5.58 An increasingly active informal marketfor land complicatesthe distribution of land rights. In recent years an increasingly vibrant informal market for land has developed, particularly in urban areas, suggesting rising demand for land in parts of the country. The amount of purchased land steadily increased between 2002 and 2005, particularly in the north and south. As expected, land purchases are significantly more common in richer households than inpoorer households, exponentiallyso over the period in question, perhaps suggesting a rising concentration of land ownership. Recent studies confirm this trend.248Interestingly, however, while the proportionof purchased landthat i s titled was higher in 2005 than in earlier surveys, most land still i s not titled before purchase. 5.59 Rising demand for land may ultimately, however, undermine legally recognized but invisible rights to land. Although the new land law ensures that most households with access to land have legally recognized rights to it, with increasing 245This aspect o f the law, particularly progressive, sets the 1997 Land Law apart from most other land laws, including the recently adoptedAngolan land law, which also recognizes customary law but makes formalization mandatory. 246Norfolk andTanner 2006, 2. 247Ibid., 3. 248Tanner 2005, 11-13; Tanner andBaleira 2006. 141 demand for land-particularly inurban areas-those who do not have formally registered rights risk missing out on the potential benefits from land markets and losing the tenure security accordedto them by law: "Unregistered community and good faith land-use and benefit rights may be legally recognized, but invisible to anyone but local people and their neighbors. Faced by rising demand for land, local people with unregistered rightsmay then be exposedto de facto expropriation ... [by private] investors."249 5.60 Implementation of the law has been relatively slow, but several advances can be identified. The government-supported by donors and national nongovernmental organizations and community service organizations-has invested heavily in raising awareness of the land law. In addition to making the law available in the national press and the OfJicial Bulletin, the law also benefited from extensive consultations with civil society, translation into six languages,250and-most important-a 1998 national land campaign to take the basic messages of the law to the local Nongovernmental organizations and community service organizations have also been very active throughout the process, educating the public on citizen rights and raising awareness o f the land law. Buildingcapacity and training personnel have also been important. The Centre for Legal and Judicial Training, part o f the Ministry o f Justice, has trained community and district judges and prosecutors, district administrators, and paralegals to assist local communities. Increasing Poor People'sAwareness of the LawZS2 5.61 Despite investments in public education, most households-particularly poor households and those headed by women-are unaware of the land law and its implications. The findings from the PVS suggest that people's low awareness o f their citizenship rights extends to land rights. Indeed, both quantitative and qualitative fieldwork shows that nearly all respondents have little or no awareness o f the land law or its implementation. Only 15 percent o f urban households and 11 percent of rural ones have any knowledge o f how to obtain a land title under the current land law (table 5.4). 249 Tanner 2005, 5-6. 250 Inadditionto Portuguese,translations exist in Changane, Sen%Ndau, Xitswe, Maconde, andMacua. Unofficial translations have also beenmade into many other languages. 25' Tanner 2005,6. Six basic messagesfrom the law were taken to the locallevel: consultations (between local communities andprivate investors) are obligatory, communities can sign contracts (with investors andthe state), women have equal rights, rights ofway must be respected, rights shouldbe registered, andwhat to do inthe case of conflicts. 252 For a broader discussionof the conceptual andoperational implicationsof "making rights real," see Gaventa and others (2002). Inthe contextof Mozambique, see Tanner andothers (2006). 142 Table 5.4. Awarenessof How to ObtainLandTitle, by Area, Gender of HouseholdHead, and Tercile, 2006 (percent) Urban Rural Head of household All Terciles Lowest Highest All Terciles Lowest Highest All 15.2 0.0 30.0 10.5 10.4 11.5 Male head 18.2 0.0 32.5 11.9 10.8 12.1 Female head 8.8 0.0 20.0 6.7 10.0 8.3 Source: Authors' calculations basedon data from PVS 2006. Note: Sample not representative 5.62 In urban areas more than twice as many male-headed householdsasfemale- headed households have knowledge of titling procedures. Even in rural areas the findings vary by gender, though somewhat less so. Household welfare also matters, especially in urban areas, where 30 percent o f the richest households know about land titling requirements, compared with no households in the poorest tercile. Even a decade after the law came into effect, very few households know their specific rights to the land they rely on for their daily livelihoods. 5.63 Unaware of their rights, the poor are less likely to know and benefit from the real economic value of their land. Communities are also often unaware o f the value o f the land that they surrender to investors-and the returns that it could generate through careful deals. If the long-term benefits o f these transactions accrue mostly to investors, their impact on reducing poverty will be considered negative. Encouraging Poor Households to Register Their Land Rights 5.64 Voluntary land registration and low awareness of rights reduce the demand for land titles. Only a few households have formalized and registered their land rights. The PVS found that even fewer households than those claiming to have knowledge about the law have titles to their land-fewer than 10 percent in urban areas and only 1 percent o f households in rural areas.253Nationally representative data from rural areas suggest that only 2.6 percent o f rural households have title to their land, despite regional variations.254 5.65 Even so, the TU data show that the number of households with land titles doubled between 2002 and 2005. Voluntary land registration may explain the low demand for land titling, but the PVS suggests that low awareness o f the process and benefits o f formalizing land rights may be a more likely explanation. Indeed, lack o f awareness o f how to obtain a title was the most important reason cited in both urban and rural areas (table 5.5). Female heads o f household are particularly unaware o f rights and procedures for land titles-close to 70 percent in urban areas and up to 80 percent in rural areas. 253See appendix table A5.11. 254See appendix tables A5.12 and A5.13. TIA 2002 and 2005. The TIA data show that landtitling i s more common in the northandthe center, possiblyas aconsequenceofthe greater presence ofnationalnongovernmental organizations working on the land law inthese areas. Although householdsbelongingto higherwealth groups have greater awareness of the land law, the TIA survey suggests no clear relationshipbetween income and landtitling. 143 Table 5.5. Reasons for Lack of Title, by Area and Gender of Household Head, 2006 (percent) Urban ~ ~~ ~~~~~~ Rural Reason All Male H e a h Female Heads All MaleHeads Female Heads Lack of interest 22.5 26.4 14.0 27.9 32.7 15.0 I t is complicated 6.2 5.0 8.8 5.0 6.3 1.7 Lack of knowledge 57.9 52.9 68.4 62.1 55.3 80.0 Lack of money 3.9 4.1 3.5 4.6 5.O 3.3 Land is borrowed 9.0 10.7 5.3 0.5 0.6 0.0 Other 0.6 0.8 0.0 0.0 100.o 0.0 0.0 Total 100.0 100.0 100.0 100.0 100.0 Source: Authors' calculations basedon data from PVS 2006. Note: Sample not representative. 5.66 Low formalization leaves the land rights of most of the population invisible and outside official records. The current cadastral database in Mozambique relies on informationfrom land-use rights applications that have already been processed. Given the limited demand for formalization of land rights and the low proportion o f land that has been delimited or titled, cadastral records cover only a tiny part o f the country. A land-use and occupation survey was recently conducted, but it was geographically limited and relied on conventional surveyingmethods, which failed to capture most community land- use rights. Because customary and good faith land accounts for at least 80 percent o f land rights in Mozambi ue, the land rights o f most o f the population are not captured in the national ~adastre.2~'As noted in a recent study, because most people's rights to land are still customarily and informally administered, they remain "largely invisible to formal systems o f rights registration and recording."256 5.67 The government has done little to promote communal land registration, and public resources allocated to this activity have dwindled. While public resources for community land delimitation were low in 2001, they have declined The voluntary registration o f land rights has also meant, accordin to one observer, that "there i s no pressure on public services to record these r i g h t ~ . ~ ~ ~ ~ ` Wthel eexemption from h i registration remains important for the poor, who do not have the knowledge or capacity to formalize their rights, efforts are needed to make land rights for the poor more visible in the formal cadastral systems.259 Delimiting CommunityLand 5.68 Community land delimitations are also under way, ifslowly. Because most o f the rural poor still obtain their land rightsthrough community and good faith occupation, community land delimitation provides insights into land rights registration in Mozambique.260 The most comprehensive survey o f community delimitations was 255 Chemonics 2006, 4. 256 Norfolk and Tanner 2006, 5. "'See appendix table A5.16. Tanner 2005, 5. z59 Norfolk and Tanner 2006, 2. z60 Chemonics 2006, 33. 144 conducted in 2003,261 but a 2005 update (with more recent figures for the northern provinces) suggests that only 185 community delimitations have been conducted and fewer than half (88) have received certificates o f delimitation.262While this may seem low, given that an estimated 2,000-3,000 communities exist in Mozambique (comprising more than 10,000 villages); since land registration i s entirely voluntary according to the land law, these figures may represent some progress. Moreover, these delimitations have occurred despite the lack o f national focus on community land delimitations. Infact, most all o f those completed have been supported by national nongovernmental organizations, with donor assistance.263 5.69 Studies also suggest that the delimitation initiatives benefit local communities.264In addition to allowing community groups to negotiate better benefits from potential private investors, delimitation has raised community awareness o f legal rights (beyond land rights alone) and empowered greater participation in managing local resources. And by providing a clearer definition o f community boundaries, it has improved social cohesion and reduced conflicts.265 5.70 Further evidence of progress is shown by the many community consultations before private land investments and acquisitions.Community consultations take place in most cases o f private claims on communal land266-a great improvement on past practices, which often led to allegations o f private expropriations o f communal land. As a result, more o f the economic benefits o f private development are likely to be shared with local people and communities. Improving the ConsultationProcess 5.71 Although community consultationsfrequently accompany private claims to community land, a growing body of evidence suggests that the quality of the consultation does not sufficiently guarantee local rights and benefits. In the context o f rising demand for land from private investors--especially where community land-use and benefit rights are unregistered-the land law's community consultation mechanism provides a potentially important means o f formalizing land rights in a way that puts local rights firmly on the agenda. All private investors seeking land are required to conduct a consultation with the relevant community members to inform them o f the purpose o f the land use and to give them an opportunity to state their interests in the land. The consultation must then stipulate the terms o f a partnership between the community and the investor, under which the land-use and benefit right is transferred.267 5-72 Although the requirementfor community consultations has been fulfilled in most private claims, Tanner and Baleira (2006) found that the quality of the 26'CTC 2003. 262See table 5.1. IIED2005. Norfolk and Tanner 2006, 12. 264Chemonics (2006,369, citing Knight(2002) andNorfolk (2004). 265Chemonics 2006, 34-5. 266Tanner andBaleira 2006. 267Norfolk andTanner 2006, 5. 145 consultation process is usually very poor,268either resulting in "essentially unlawful occupation o f someone else's land" or failing "to produce real economic benefits at a level that can either maintain livelihoods or alleviate the poverty o f `beneficiarie~."'~~~ 5.73 Whether community consultations are representative is often questionable. The consultations rarely involve more than one meeting, and only very few community representatives participate, usually community leaders.270 Such an approach raises questions about the depth and seriousness o f the consultation and about who can legitimately claim to represent the interests o f constituents-an important definitional question. Although community authorities-identified in the definition of legitimate actors in state administration in Decree 15/200O-play an important role representing the community to the state, they should not be relied on to "necessarily represent the group o f private individuals who collectively hold a Local Community DUAT [Direito de Us0 e Aproveitamento da Terra] The study found that community consultations included only conversations with these community authorities in many cases, rather than the broader community, or specific individuals within the community, whose land rights were at stake. 5.74 Local land administration officials often consider the community consultation requirement to be merely "one of several administrative ste s for securing a new DUAT," rather than a "means of protecting local vi hts."= One cadastral service officer even described the process as a "waste o f time." 73 This procedural view o f the 4 $ community consultation process i s further reinforced by a Ministry o f Agriculture and Rural Development directive requiring all new land-use and benefit rightsto be processed and approved within 90 days. While this much time should in theory be sufficient to ensure proper consultations with the community, evidence from tracking land-claim cases in northern Mozambique suggests that the time for community consultations is only one or two days, while most o f the time goes to bureaucratic procedures.274 So, the land administration needs to streamline the process to ensure that enough time i s available to properly consult communities, raise their awareness o f their legal rights and alternatives, and provide them with real opportunities to share inthe benefits o f their land.275 5.75 Many o f the challenges posed by community consultations are summarized in a recent report on the land reform in Mozambique, developed as part o f the Southern African Development Community (SADC) Land and Agrarian Reform Initiative: 268 Tanner andBaleira 2006. 269 Norfolk andTanner 2006, 5. 270 Tanner andBaleira 2006, 5-6. 271 Norfolk andTanner 2006, 7. 272 Norfolk and Tanner 2006, 8. 273 Tanner andBaleira 2006, 11. 274 Chris Tanner, personalcommunication, November9,2006. 275 The role ofthe state may sometimes further complicate matters.According to the landlaw, landadministration officials shouldbe facilitators o fthe land claim andtitling process. Inpractice, however, this neutralstate role is complicatedby the history of antagonistic relationships betweenthe state and some communities, particularly inrural areas. Such relationshipsmay also be further underminedby perceptionsof collusion betweenthe state andprivate investors. 146 "there i s a lot o f evidence that the majority o f consultations are badly done, resulting inturn in complex conflicts between locals and investors who later come to occupy the lands. Rarely i s more than one meeting held, and in the majority o f cases decisions are taken by a small group o f the community [which] do not always include the legitimate leaders o f the community. Inaddition, the official minuteso f the meeting are frequently inadequately compiled and do not include all the information on the agreements made."276 5.76 The requirementfor community participation put local rights squarely on the agenda. But unless the failure to generate the local benefits expected from community consultations i s addressed, the consultations may `lead to "enclosures with a human face,"277 giving land grabs by private investors a "veneer o f respectability by demonstrating compliance with the law, and apparently safeguarding local needs and interests.,y278 Making WomenBetter off in Claiming Rights 5.77 While women have gained unprecedented legal rights under the land law, judicial, cultural, and structural factors undermine their enjoyment of these rights. Women are the majority o f smallholder farmers in Mozambique, making up 62 percent of those who claim agriculture as their primary occupation. Securing women's land tenure may hence have a range o f positive primary effects (on yields) and secondary effects (on household well-being).279 But their tenure security i s the most fragile, particularly for rural women, often considered dependent on their marriage or kinshiprelations. However, with the 1997 Land Law and the 2005 Family Law, "the legal framework for women's equal rights,to land in Mozambique i s basically in place." 280 Indeed, Mozambique is often said to have one o f the most progressive and gender-sensitive land laws in Southern Africa.281But women continue to suffer discrimination in land rights, partly because of a lack o f information and awareness about their rights, partly because o f continued discriminatory customary practices, and partly because administrative and judicial practices are still far from incorporating the norms and dynamics that the land law seeks to encourage.282 5.78 From a judicial perspective the emphasis on legal equivalence and pluralism introducestension between statutory and customary law, withpotential implicationsfor gender equality. As the recent Swedish International Development Agency (SIDA) Gender Equality update for Mozambique states: 276SADC (2006), cited in Chemonics (2006, 35). 277Chris Tanner, personalcommunication, November9, 2006. 278Tanner 2005, 19. 279Arndt and others 2005. Chemonics 2006, 36. The proposeddomestic violence law provides further protectionto women. Section 5, subsection(i)of the draft law prohibits "property violence"-all violence causingthe deteriorationof objects, animals, or materialgoods of women or their family nucleus.Moreover, section8, subsection(3) (0 states that the government may take measuresto "guarantee the safe return of the woman who hadbeen forcedto abandonher residencefor safety reasons, with the immediate applicationof [a restraining]order." "' Collier 2006, 36. 282Ibid. 147 "The new legislation recognizes women's equal land rights to men and to this extent i s based on the assumption o f gender equality. However, it also formally recognizes what are widely held to be patriarchal, customary systems o f land tenure wherein rights and duties are allocated differently, according to gender and, therefore, based on gender difference. This tension creates uncertainty about the correct interpretation and application o f the formal law, which inturn may result in insecure tenure for women." 283 5.79 The new land law tries to counter these fears by emphasizing women's recourse to theformal legal system and the equality between women and men under the constitution. Moreover, the law includes two additional elements to secure women's access to land. First, women have full co-title rights to participate in all decisions about how community land i s used. Second, they have the right to take their land outside the communal title-hold system and claim their individual right to the land from the state. 5.80 While these steps are crucial advances over the previous law and set Mozambique apart from most other developing countries, the issue remains "how women's interests are represented in the local reconstruction of customary rules and practices, and the actual participation of women in the implementation of the land law. jGS4 The survey o f community consultations found that women rarely participate actively in the consultations-partly because dealing with the outside world is considered men's role in most rural areas. Women's marginalization is made worse by the limited consultation process, whose brevity hinders the opportunities for women's participation behind the scenes.285 5.81 The law could overemphasize community rights at the expense of individual rights-especiallyfor women. The law tends to treat the community as a black box when describing customary practices' implications for land rights:86 without clearly specifying how individual land rights are guaranteed. This i s a problem for women and other roups whose access to land may be compromised, including orphaned children and youth.!87 5.82 Women have low awareness of their citizenship rights, including land rights, increasing their exclusion. Women's low awareness o f their land rights has been raised as a key reason for discrimination against women before the new land law.288But as seen earlier, women continue to be less aware than male heads o f households o f their rights to land. So, despite legislative improvements, women continue to face a range o f obstacles inclaiming their land: 283Collier 2006, 37. 284Chemonics 2006, 36. 285Tanner and Baleira 2006, 20-1. 286 "'Access Sergio Baleira personal communication, November 9, 2006. to land in many rural parts of Mozambique is often gainedthrough family and kinship relationships, as well as marriage. The tenure security of single, divorced, or abandoned women, youth and orphanedchildren may therefore be especially fragile. 288Chemonics 2006, 36. Constraints of awareness and capacity also extend to the institutions and organizations that are supposedto enable women to claim and realize their rights, including thejudicial system, the land administration system, nongovernmental organizations, and even donors. 148 "Being in a situation o f ignorance about their rights under civil laws in Mozambique, allied to aspects o f their socialization that impede them from taking decisive positions in certain situations, women do not have the necessary tools to exercise their rights as owner or co-proprietor o f specific assets, and are subject to arbitrary expulsions and expropriation [of their land] inside their communities."289 5.83 The weak, inefficient, and inaccessible land administration system charged with implementing the land law compoundstheseproblems. Successful implementation o f the land law requires efficient, accessible, and accountable cadastral services. But land administration in Mozambique remains highly bureaucratic and centralized, institutionally and technically weak, and severely understaffed, particularly at lower administrative levels, "leading to shortcuts and nonobservance o f procedures and regulations" o f the land law.290Moreover, the costs o f cadastral services (delimitation and demarcation o f land holdings) are prohibitive, which together with the geographically centralized land administration system make access to services a real challenge for the majority. The unresponsiveness o f local cadastral services-or even their absence-is ultimately why customary land tenure systems remain so pervasive. A more decentralized, accessible, and responsive local land administration system i s needed, along with efforts to enhance local communities' and individuals' awareness o f the rights stipulated in the 1997 Land Law and their ability to realize them. RecommendedActions 5.84 Decentralization will increase the likelihood of effective implementation and increased accessibility at the local level but needs to be complemented with a push for broad land registration and improved monitoring. The benefits o f community registration and titling in increasing awareness and strengthening negotiating positions are significant. Monitoring o f implementation should include quality-based performance measures. 5.85 More emphasis needs to be attached to the protection of individual rights, especially those of women. Dissemination efforts will need to emphasize explanation o f the land law and how individual rights relate to it. 5.86 Incentives for the effective use of the land and inclusion of smallholders can ensure that gains to society are maximized over the long term. The land tenure system has been susceptible to opportunistic behavior by powerful private investors to the detriment o f the disadvantaged and poor. An incentive system i s required that would strengthen the inclusion o f smallholder farmers and prioritize agricultural investment projects likely to maximize links to smallholder livelihood. 289Seuane andRivers-Moore2005, 17. 290Chemonics2006, 3. 149 6. THE IMPACT OF HIV/AIDS INMOZAMBIQUE 6.1 Mozambique's estimated HIVprevalence rate amongprime-age adults, at 16.2 percent, was the eighth highest in the world in 2006.291In 2007 an estimated 441,000 children in Mozambique had lost one or both parents to AIDS, and women-the primary caregivers-were four times more likely to be infected as men in the age group o f 20 to 24. Moreover, prevalence i s increasing at an alarming pace. At the end o f 2004, 1.4 million Mozambicans were estimated to be infected, up from 1.1 million in 2001. Some 2 million Mozambicans are projected to be infected by 2010. Infection i s correlated with labor mobility, with the highest prevalence rates in urban areas and in provinces with transport corridors connecting Mozambique to neighboring countries even worse afflicted. The disease has a devastating effect for most patients and for family members. With high prevalence levels and infection rates, the disease will make its impact felt in Mozambique for decades. 6.2 The rapidly rising prevalence rate calls for significantly increased efforts to prevent infection by changing sexual behavior. Country-specific analysis suggests that it i s particularly important to target older men's risk-taking behavior. A main driver behind Mozambique's high incidence appears to be older men having multiple sexual relations, while not always using condoms, with women both from younger age categories (particularly ages 15 to 24) and with women their own age. Analysis shows the spread o f HIVIAIDS to be a considerable threat to future growth and poverty reduction in Mozambique. Experience from Uganda suggests that efforts to persuade men to have fewer sexual relations andiuse condoms can slow incidence. Box 6.1. Data Used in the Analysis This chapter examines the impact of HIV/AIDS on people's lives inMozambique, on society, and on economic growth andpoverty. Informationfrom the Ministry of Health's (MISAU) sentinel site surveillancedata, the Ministry of Agriculture's surveys, and the Demographic and Health Surveys(DHS) is usedto investigatethe interrelations betweenlivelihoods and adult mortality in Mozambique, with a focus on how the feminization of HIV/AIDS affects poverty gaps between households.A growing body of specific HIV/AIDS research for Mozambique exists, but it remainsscarce comparedwith research materials from other countries inthe region. Above all, no recent DHS data exists. To fill some of the gaps, the chapter draws also on evidence from neighboring countriesfor comparison. HIV/AIDSPREVALENCEINMOZAMBIQUE 6.3 Mozambique's estimated 16.2 percent HIV/AIDS prevalence in the prime-age population is more than twice the Sub-Saharan average of 7.2 percent. The global 291UNAIDS 2006. The prevalencefigure for prime-age adults refersto people aged 15 to 49. Mozambiqueis ranked after Swaziland (33.4 percent), Botswana(24.1 percent), Lesotho (23.2 percent), Zimbabwe (20.1 percent), Namibia (19.6 percent), SouthAfrica (18.8 percent), andZambia (17.0 percent). 150 prevalence rate in 2005 was 1.1 percent, representing around 40 million adults and children living with HIV/AIDS, nearly 26 million living in Sub-Saharan Africa and 1.4 million in Mozambique.292 The data underlying the Mozambique prevalence estimate were collected during the latest epidemiological survey round in September 2004 to January 2005. 6.4 The disease took hold later in Mozambique than in neighboring countries, but prevalence continues to grow rapidly. The first case o f HIVIAIDS was recorded in Mozambiquein 1986.293 1988 HIV prevalence was less than 1percentamong antenatal In clinic attendeestested inMaputocity. The disease spreadrapidly after that, however,with prevalence among antenatal attendees tested in Maputo city rising to 22 percent in 2002.294HIV prevalence among male sexually transmitted disease patients tested in Maputo increased from 3 percent to 19 percent between 1987 and 1997.295The returnto Mozambique of an estimated 2 million refugees from neighboring countries with high prevalence rates following on the 1992peace accords is often ascribedto be an important factor explainingthe highincidenceinthe mid-1990s, particularly inthe centralregion.296 The incidence continuesto be high long after the return of the refugees, suggesting that the most important factor in the spread of the disease may be related to high-risk behavior, such as engaging in concurrent sexual relations and practicing unsafe sex, in combinationwith the absence ofmale circumcision.297 6.5 Mozambique's infection rate remains alarmingly high. While some countries have begun reporting slowing incidence and even decreasing prevalence of the disease, such as in Uganda and Kenya, Mozambique's infection rate has yet to show signs of slowing. Prevalence is estimated to have increased by 2.6 percent between 2002 and 2004, with the most rapid incidence in the south and center. The latest data show a dramatically worsening epidemic overall, with rising infection levels in all regions. The MISAU projects that the number of people with HIV/AIDS will be nearly 2 million in 2010.298 Life expectancy is estimated to drop to less than 37 years by 2010, compared with the 50 years projected without HIV/AIDS.299Heterosexualcontact is the primary mode of HIV transmission in Mozambique, with mother-to-child transmission, blood transfusions,and injectingdruguse beingminor contributorsto the epiderni~.~" 292UNAIDS 2005 and2006. 293MISAU 2005. 294Ibid. 295UNAIDS and WHO 2006. 296Ibid. 297That returningrefugeeswere a factor in introducing the disease into the population comes easily because Mozambique borders five countries with some ofthe highest prevalence inthe world. Yet it is atheory that may call for critical investigation:the high levels of incidencethat continue over 10years after the return of the refugeessuggest that behavioralcharacteristics are more importantdeterminants. 298MISAU 2005 . 299Ibid. 300WHO andUNAIDS 2005. 151 6.6 Prevalence rates vary considerably by region, with the southern (20.4 percent) and central (18.1 percent) provinces worse afflicted than the northern ones (9.3 percent).3o1And prevalence is increasing everywhere in Mozambique, except for the northern province o f Niassa (figure 6.1). Sofala province, which includes the city o f Beira and the country's main transport corridor to Zimbabwe, faces the highest rate (26.5 percent), more than three times that o f the northernmost Cab0 Delgado province (8.6 percent). This i s no anomaly: The highest prevalence in Mozambique appears along transportation and commerce routes connected to neighboring Malawi, South Africa, and Zimbabwe, disproportionately affecting mobile populations, such as miners, migrant workers, drivers, uniformed services, and their partners. There are geographic differences inhousehold structure that could explain some ofthe regional variance. Households inthe northern provinces tend to center around a nuclear family, with monogamous parents. Households in the central and southern provinces are more likely to configure differently, including large numbers o f female-headed households due partly to higher labor migration rates and partly to more polygamous households. Northern provinces have higher levels of Muslim households, where male circumcision, found to lower infection risk significantly, is a cultural norm. Figure 6.1. Adult Prevalence Rates (aged 15-49), by Province, 2002 and 2004 (percent) 30.0 80 3% 25.0 70 4 20.0 7 6o 19e 15.0 50 p8 ' u m 40 P5 10.0 30 sP 20 s 5.0 10 0.0 0 Source: Author's calculations basedon data from GoM (2005) andMPF (2004). 301Northern provinces include Niassa, Cab0 Delgado andNampula; central provinces include Zambezia, Tete, Manica and Sofala; southern provinces include Inhambane, Gaza, Maputo province, and Maputo city. 152 Box 6.2. Monitoring the HIVIAIDSEpidemicin Mozambique Obtaining accurate datais an important challenge inthe combat ofthe HIV/AIDS epidemic. Severalmonitoringefforts havebeenundertakeninMozambiquefollowingthe first recorded HIV/AIDS case in 1986.In 1988the MISAU's national STD/HIV/AIDS program establishedthe first four sentinel sites inselected antenatalclinics, measuringprevalenceandmonitoringtrends inHIVIAIDS infection.Sentinel surveillance expandedduringthe nexttenyears to three more sites inthe central region. Today, throughout all 11provinces, there are a total of 36 sites, 17 ruraland 19 urban. Dataare collectedinsamples ofthe first 300 pregnantwomen at their first antenatalconsultation per sentinel site (over a maximum periodo fthree months) andare then extrapolatedto create provincial,regional, andnational HIV prevalenceestimatesandto projectthe national demographicimpact ofHIV/AIDS. Only a quarter ofthe 144 administrative districtscontain sentinel sites, so the weightingexercise assumes that nonsentineldistricts havethe same prevalencerate as the district most similar.This way of estimatingHIVIAIDS prevalence is commonto many countries inSub-SaharanAfrica, yet it has many problems.For instance, women who presentthemselves at clinics are a select population. Monitoringdifficultiesare compoundedby the stigma of HIV/AIDS. Qualitativeevidence suggests that AIDS deaths may be underreportedinMozambiqueas elsewherebecauseof stigma. Nevertheless, sentinel site data-in combinationwith surveys suchas the TIA, the DHS, andthe IAF-canDrovide a basisto estimatethe imDactofthe enidemic. STRUCTURALDETERMINANTS HIV/AIDSINMOZAMBIQUE OF 6.7 The correlation between HIV/AIDS prevalence and poverty in Mozambique appears to be negative. The province with the lowest poverty rate, Sofala, has the highest HIV/AIDS prevalence rate. The province with the highest poverty, Inhambane, meanwhile, reports among the lowest HIV/AIDS rates. The pattern is consistent also in the provinces of Manica and Zambezia, both with low poverty and high rates. With better communication infrastructure and higher mobility, a common factor in the less poor provinces, the evidence appears to suggest that the same mechanism helping to decrease poverty is causing higher HIV prevalence rates. In a second phase, after a lag between infection and symptoms, health levels can be expectedto start deteriorating more rapidly, with households losing labor to the disease and childrenbeingpulledfrom school because of time and financial constraints. As a result, negative effects on aggregate poverty levels may follow. 6.8 Cities have higherprevalence than rural areas. Judging by 2004 prevalencedata from sentinel sites, and 1997 poverty data for the corresponding sites from a poverty map based on the 1997 IAF, prevalence rates vary significantly within regions, with cities displaying particularly high rates.302This finding is consistent with other data but should be treated with care: the district poverty rates are calculated based on the 1997 household 302MPF 2002. See figure 6.2. 153 survey, reflecting a situation eight years before the HIV/AIDS surveillance year, when prevalence rates were significantly lower. HIV prevalence among antenatal clinic attendees tested in Maputo city in 2002 was 18 percent, compared with 12.3 percent outside Maputo city. Among women aged 15 to 24 attending antenatal clinics in2002, 15 percent inMaputo city were HIV positive compared with 12percent inother sites. 6.9 Data for analyzing the impact of HIV/AIDS on individual wealth do not exist yet for Mozambique. Evidence from other countries in Sub-Saharan Africa show no significant results linking HIV/AIDS status and poverty on the individual Using the first Demographic and Health Surveys that collected information on HIV status, de Walque finds a robust correlation between women's HIV/AIDS seropositive status and wealth in Cameroon, Kenya, and Tanzania.'04 For men, the correlation is significantly positive in Cameroon and marginally negative in Burkina Faso. While the evidence cannot be generalized and the findings are not wholly consistent, there are no robust indications showing that being seropositive is associated with higher poverty in individual asset wealth. THEFEMINIZATIONOFHIV/AIDSINMOZAMBIQUE 6.10 Women accountfor nearly half the 40 million adults living with HIV in the world, and 57 percent in Sub-Saharan Africa. Social determinants o f women's vulnerability to HIV include gender gaps in resource access, economic empowerment, bargaining power, poverty, education levels, violence, and cultural and sexual norms. Hormonal changes, vaginal microbial ecology and physiology, as well as a higher prevalence o f sexually transmitted diseases also contribute to women's higher susceptibility. Successful prevention strategies for women address the full spectrum o f gender inequalities that promote HIV infection. 6.11 The gender gap in infection rates is large and widening in Mozambique. Women make up an estimated 58 percent o f those infected with HIV/AIDS.Women aged 20 to 24 are four times more likely than men in the same age group to become infected with HIV/AIDS, and girls and women aged 15 to 19 are three times more likely. Moreover, this trend i s projected to continue across all age groups, with the gender gap among 20 to 24 year olds growing most (figure 6.2).305HIV prevalence among pregnant women rose at 23 o f the 36 clinics included in the 2004 surveillance round. Among pregnant women in Caia, along the railway link to Malawi, HIV prevalence rose from 7 percent in 2001 to 19 percent in 2004.306 No study exists to pinpoint the exact determinants behindthis increasing gender gap, but social factors are often assumed to be behind it. The existence o f polygamous relationships and Mozambique's very high '03de Walque 2006. '04de Walque 2006. The DHS surveys used inde Walque (2006) were from Burkina Faso (2003), Cameroon (2004), Ghana (2003), Kenya (2003), and Tanzania (2003-2004). '05UNAIDS 2005; INE and others 2002. 'OsMISAU2005. 154 prevalence o f labor migration (and return migration) o f prime-age men may play an important role in combination with gender gaps in access to resources and rights. Moreover, while the prevalence rate for women is three to four times that o f men in age group 15 to 29, older age groups do not show the same gender-discrepancy in infection rates. Infact, most men infected with HIV are 25 years or older. Figure 6.2. Number ofHIVIAIDS InfectedPeople, by Gender, 1998-2010 iousands) All w a s (~1000) 1000 200 800 150 600 100 400 50 200 0 0 Source; Authors calculations basedon data from ME and others 2002. 6.12 Low female bargaining power, in combination with the HIV/AIDS stigma, makes women more vulnerable to infection and hampers prevention work in Mozambique. Young women in Maputo who were involved in multiple transactional or cross-generational relationships report that the use o f condoms, or even raising the prospect o f using condoms in discussions with sexual partners, was considered to jeopardize income.307The possible long-term effect o f contracting HIV/AIDSwas seen as relatively less important than the short-term effect o f losing income. 6.13 These age and gender-relatedpatterns of prevalence, as well as the difference in young women's bargaining power relative to older men, suggest the main drivers of the epidemic. First, relatively older men having sex with multiple partners, without always using a condom, appear to infect women both their own age and women inthe 15 to 24 age bracket (who in turn infect younger men). Second, cross-generational sexual interactions skew power relations to women's disadvantage, limiting their ability to negotiate for safe sex. Experience from Uganda suggests that efforts to persuade adult men to have fewer sexual partners, complemented by condom prevention, can be successful successful. 6.14 Women's education and wealth determine their awareness of HIV/AIDS and theperceived possibility to discuss it openly with a sexualpartner. Women's awareness o f HIV/AIDS, knowledge o f prevention methods, and the perceived possibility o f 307Hawkins, Mussa, and Abuxahama2005. 155 discussing prevention with partners differ considerably by area o f residence, education level, and wealth (table 6.1).308Less than one-third o f women without education could name "condom use" and "limiting sexual relationships to one uninfected partner" as methods to avoid infection. Meanwhile, half o f all women with primary education named these methods, as did nearly 80 percent o f women with secondary education. This i s not surprising: The positive effect o f education on protective behavior-including condom use, counseling, testing, and discussion among spouses-is evident in all countries that conduct regular surveys on HIV/AIDS through the DHSS.~'~ At the same time, however, education has also been shown to predict a higher level o f infidelity and a lower level o f abstinence, which may offset the positive effect. Table 6.1. Awareness of HIVIAIDS andPreventionMethods, 2003 Knows Using Condoms and Limiting Sex to One Has Discussed Uninfected Partner HIV/AIDS Prevention Characteristic Has Heard of HIV/AIDS Reduces Risk with Partner Area Rural 93.7 38.0 45.1 Urban 99.1 57.6 60.3 Education No education 91.7 31.1 39.3 Primary 98.2 51.4 56.8 Secondary 100.0 78.5 82.1 Wealth quintile Lowest 90.5 29.5 38.5 2nd 93.7 34.3 41.9 3rd 95.9 43.1 50.6 4th 98.7 51.4 53.5 Highest 99.7 65.8 69.2 To61 95.7 45.2 49.5 Source: DHS data for 2003. 6.15 The richer a woman is in Mozambique, the more knowledge she has o prevention methods and the more she is able to broach prevention with her ~ a r t n e r . ~ ! Women with little or no education and low income might be at greater risk o f infection. Although lack o f overall awareness o f the disease adds to the risk o f infection, more important factors appear to be not knowing how to avoid the disease and not being able to discuss it with sexual partners. Evidence shows, however, that knowledge o f the disease and o f prevention behavior alone do not lead to results.311In countries that have reversed incidence trends changed sexual behavior-such as fewer partners and increased condom use-has been the most important reason for success. Mozambique's government has started a female condom distribution program to complement its male condom 308DHS 2003. 309de Walque 2006. 310See table 6.1. 311de Walque 2006. 156 distribution effort. To date, far fewer condoms are distributed per man annually than in neighboring countries, including South Africa, Zimbabwe, and Botswana.312 6.16 Since HIV/AIDS tends to strike people in their most productive years, older women in Sub-Saharan Africa are assuming an increasingly important care burden, especiallyfor orphans. One survey indicates that two-thirds o f caregivers in households affected by HIV/AIDS in South Africa were women, with a third o f them older than 60.3*3 The Mozambique IAF contains no time-use module allowing reporting on care for patients or for other family members, and the data do not show particularly large numbers o f households headed by women over the age o f 60 in any region-so Mozambique's situation may differ from South Africa's. The quantitative survey conducted for this assessment in four provinces, however, indicates that girls and women are almost exclusively in charge o f the care burden for sick household members.314 DEMOGRAPHICEFFECTS HIV/AIDS OF 6.17 Lqe expectancy, infant mortality rates, and population structures in Mozambique have deteriorated rapidly due to HIV/AIDS. The demographic impact o f HIV/AIDS in Mozambique i s thorough: life expectancy i s declining fast and population structures are being distorted. Life expectancy in Mozambique is pro'ected to be 37 years in 2010 rather than the projected 50 years without HIV/AIDS.312Infant mortality in Mozambique is expected to be at least 25 percent higher in 2010 with HIV/AIDS than it would have been Current population projections for Mozambique show the demographic pyramid changing from a concave shape, with a relatively young population, to a convex Ifthe current HIV/AIDS incidence trend continues, the demographic structure o f the population could transform radically from a pyramid to a chimney, approaching the projections for neighboring South Africa, with fewer people in their 50s than in their 60s and fewer women than men in the prime-age categories (figure 6.3). In countries with prevalence rates higherthan 20 percent, the population i s estimated to be more than one-third smaller than without HIV/AIDS.318 312The MISAU reports that 43 million male condoms were distributedin2006, equaling8.9 per man, compared with 17 per man inhigh-performing countries. Studiesby PopulationServicesInternational indicate a demandconstraint to condom use (World Bank Aide-Memoire, MozambiqueHIViAids ResponseMission, March 12-23,2007). 3'3Steinberg and others 2002, cited inUNICEF 2004. 314See box 2.3 in chapter2. 315INE andothers 2002. 316WHO 2005. Inthe whole of Africa the share o f under-five mortality due to AIDS increasedfrom 2 percent in 1990 to 6.5 percent ten years later. 317We use the mediumvariant of the UnitedNations PopulationDivision (UNPD 2004). These projectionsare lower than INE's projections, but following Klasen and Woltermann(2004), the UNprojectionsare probably the more realistic becausetheir fertility and AIDS assumptions lie closer to the current trends inMozambique. 318UNAIDS 2004. 157 Figure6.3. PopulationProjectionsfor SouthAfrica, with and without HIV/AIDS, 2000 and 2025 6.18 HIV/AIDS tends to be most concentrated among prime-age adults, with considerably negative effects for the active labor force size. The female labor force in Mozambique risks a larger decline: prevalence and mortality rates for women are substantially higher than for men, and women are the prime caregivers in Mozambique's households (figure 6.4). Projections o f AIDS deaths suggest that two current key demographic characteristics o f the pandemic will continue: an increasing number o f adult deaths and a widening gender gap, with more women dying from AIDS than men every year. The consequence will be higher household dependency ratios, with potential poverty effects on households that divert resources away from productive investments, ultimately dampening economic growth. 158 Figure 6.4. Expected AIDS Deaths, Age 15 and Higher, 1999-2010 Source: Authors' calculations based on data from INE2002. Orphans 6.19 The increasing number of children who have lost one or both parents is a grim effect of the HIV/AIDS epidemic. The well-being o f orphaned children almost always suffers, with potentially negative consequences for their socioeconomic status in adulthood. Lower psychological well-being, lower education levels, food shortages, and exposure to violence are possible consequences. Mozambique had many orphans before the HIV/AIDS epidemic, and large numbers o f non-AIDS orphans will continue to exist inthe future. 6.20 An estimated441,000 children (under 18) had lost one or both parents to AIDS in 2007, twice the number in 2003. The MISAU estimates that since 2004 children having lost both parents to AIDS outnumber those having lost parents by other causes (figure 6.5). The trend i s projected to continue. By any measure, Mozambique's orphaned population i s among the larger in the There are numerous consequences to this many orphans. One is potentially lower aggregate education, because orphaned children face a lower probability o f attending school. DHS 1997 and 2003, however, show the difference in school attendance declining between orphans and nonorphans, with the gap closing more for male than for female orphans (table 6.2). 319Using a different estimate than the Ministry o fHealth, UNICEF rankedMozambique's orphaned population in2003 as the 1lthmost severe inthe world inabsolute terms (UNICEF 2005). In2003, UNICEF estimated the number o f AIDS-orphans inMozambique to 470,000. Using the same method, and not accounting for relative size o f populations, UNICEF found larger numbers o fAIDS orphans only inNigeria(l,SOO.OOO), South Africa (1,100,000), Tanzania (980,000), Zimbabwe (980,000), Uganda (940,000), Congo (770,000), Ethiopia(720,000), Kenya (650,000), Zambia (630,000) and Malawi (500,000). 159 Figure 6.5. Number of OrphansAged 0 to 17,1998-2010 Having lost one parent Having lost both parents 1,400 ,-- - 400 1,200 350 r: 1,000 300 800 250 7J $ 600 f3 400 x)O 200 50 0 1898 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Source: Authors' calculations basedon data from MISAU 2007 Table 6.2. Ratio of Current School Attendance of Orphansversus Nonorphans, by Type of Orphan, 2003 (percent) Typeof orphan 2003 Maternal 0.81 Paternal 0.97 Bothparentsdied 0.80 Source: GoM 2006, basedon DHS data HIV/AIDS and the Impact on the Livelihoods of Poor Households 6.21 Increasing mortality rates due to the HIV/AIDS epidemic can be expected to have significant negative effects on agriculture and rural development. Many studies report a multifaceted loss to the livelihood o f a rural household due to HIV-related prime- age adult illness and death. Predicted impacts on households include loss o f income, on- and off-farm labor, and technical knowledge o f agricultural production and marketing- and the liquidation o f assets, including farm equipment and livestock, to cover medical and funeral expenses. Evidence from across the developing world o f the impacts on rural households' livelihoods i s heterogeneous, with factors such as the sex o f the deceased prime-age adult and the prior poverty level o f the household shown to be significant. 6.22 Rural households in Mozambique with a prime-age adult death do not on averageface worsepoverty than households without a prime-age adult death, but there are important gender-related differences. A study based on a three-year panel o f over 4,000 households surveyed in 2002 and 2005 showed that the impact o f prime-age mortality due to illness varies considerably by the sex and household position o f the 160 deceased individual as well as by region.320A prime-age male death is more likely to significantlyreducehouseholdsize, income, and assetsthana prime-agefemale death is. 6.23 Householdsfaced with a prime-age female death are considerably more likely to attract a new prime-age female than households with a prime-age male death are to attract a new prime-age male. When a prime-age male dies in rural Mozambique, evidence indicatesthat no new manjoins the household.321Instead,the householdtends to become female-headed. Infact, when a prime-agemale dies, estimates show that it results in a significant loss to the household of 1.23 members, indicatingthat some children leave the household when men die (table 6.3). Households that lose a prime-age woman to HIV/AIDS, meanwhile, face a size reduction of only 0.17 adults, indicating that of five households, more than four with a prime-age female death were able to attract another prime-age adult to the household.This finding also holds when looking at difference-in- differencesbetweenhouseholdswith prime-agedeath andhouseholds without it. 6.24 A prime-age male death results in a significant loss of 25percent of household income, while householdsfacing aprime-agefemale death report no significant income loss. Households losing a prime-age woman face no income loss because her labor is replacedby recruitinganother prime-age But consumptionrequirementsin male- death households also fall by 25 percent, with the net effect that there is no significant income loss per adult equivalentmember inthe household. 6.25 There are gender-related differences in asset ownership effects of a prime-age death. A prime-age male death results in a 21 percent reduction in land holding, compared with an 18 percentreductionfollowing a prime-agefemale death. The regional variation is large, however,with reductionson total landholdingssignificantand larger in the north when prime-agewomen die (results are not significant inthe south). This may be a result of matrilineal systems north of the Zambezi, where the deceased woman's brother may inherit the land.323There are sizeable reductionsin livestock due to a prime- age male death, butno significantreductioninresponseto a prime-agefemale death. 320Donovanand Mather2007. 32'Ibid. 322Ibid. 323Ibid. 161 6.26 Widow-headedhouseholds experiencesignificant reductions in size and in assets that make them vulnerable tofuture shocks. The gender-relateddifference in household coping mechanisms bodes illfor widow-headed households' longer term welfare. Significant reductions in land holdings and in livestock assets mean that they will have to rely on lower livestock income, makingthem more vulnerable to future income The TIA survey that underlies this research does not providemuch informationon actual time-use of adults or children, and the analysis therefore implicitly assumes that children are not playing a role in household income generation.325With many consumption studies showing that money in the hands of female adults is more likely to be spent on children's food, education, and health, it may bepossiblethat childrenare more likely to be pulledfrom school by a surviving widower in Mozambiquethan by a surviving widow when a new prime- age woman is notrecruitedto the household. 6.27 Studiesfrom other countries in Sub-Saharan Africa show that children in households with a prime-age death are less likely to go to school, with variations depending on the sex of the child and of theparent that died. Whether HIVIAIDS- related or not, adult morbidity and mortality reduces children's school enrolment and achievement. In Kenya school participation declines substantially after a parentaldeath, es ecially when the motherdies andwhen the child is a young girl or a weak student?' In South Africa a maternal death reduces not only the child's probability of being enrolled in school but also the average years of schooling and the average money spent on education.327These findings appear to hold for Mozambique, where both the sex of the deceased parent and the sex of the child affect enrollment likelihood, with maternal deaths and female children being the combination with the most adverse effect on education.328This could reopen the gender gap in enrollmentand completionrates, which has beenclosing. 6.28 Neighboring countries also show effects of a prime-age adult death on household income, cropping patterns, and labor availability+ ffects that depend on the sex and householdposition o f the deceased. Despitethe difficulty in making meaningful generalizations from country-specific studies, evidence from neighboringcountries corroborates the findings for Mozambique. Poor rural farm households in Zambia have more difficulty coping with the death of a male householdhead than nonpoor There was no general switch to labor- 324Ibid. 325No comprehensive time-use study exists for Mozambique. 326Evans and Miguel2005. 327Case and Ardington 2004. 328See table 6.2. 329Chapoto andJayne 2005. When amale householdhead died, poorer householdsexperienceda 13 percent decline in gross value of output while nonpoor householdsmaintainedtheir gross output value. The deathof a prime-age male resulted in an 11 percent decline intotal land cultivatedwhile the deathof aprime-age female resulted in a3 percentdecline andthe death of male headsor spousesresulted ina20 percent decline. 163 saving crops, such as roots and tubers, among households that lost a prime-age member.33o Malawi women from households that experienced prime-age death In reallocate their time from labor-intensive farming and heavy chores to cash- generating tasks, while men's time appears to be insensitive to the Household size inKenya declined significantly in the case o f a male death but not a female one. Moreover, the death o f a head or spouse affected family size negatively, while the death o f a nonhead or nonspouse did not affect family size.332In Rakai, Uganda, household size declined by one person after any type o f prime-age death.333 So, while the evidence from other countries in the region is heterogeneous, in most cases households appear not to recruit new members following the death o f a male prime-age adult, incontrast to when a female prime-age adult dies. MACROECONOMICOFHIV/AIDS IMPACT 6.29 The negative effects of HIV/AIDS on the economic growth prospectsfor Mozambique are estimated to be severe. The effects o f HIV/AIDS on Mozambique's growth prospects were first estimated in 2003 in a computable general equilibrium model that explored the constraints on growth o f the productivity growth impact; the population, labor, and human capital accumulation impact; and the physical capital accumulation impact.334The model estimated that Mozambique's economy will be between 14 percent and 20 percent smaller in 2010 because o f HIV/AIDS.335Depending on the scenario modeled, the prevalence o f HIV/AIDS restricts Mozambique's economic growth by an estimated 0.3-1 percentage point annually due to negative effects on human capital accumulation, decreases in productivity because o f lower health and education and loss o f knowledge, diminished labor force due to higher mortality, and diversions o f government and household resources from investments to higher spending on health. The most significant growth slowdown effect for Mozambique-a full 41 percent-was from lower productivity. The physical capital accumulation effect accounted for 35 percent o f the slowdown, and demographic and labor force effects accounted for 24 percent.336 6.30 Subsequent modeling exercises corroborate the order of magnitude of reduced growth due to the HIV/AIDS epidemic. A growth accounting model from 330Chapotoand Jayne 2005. InZambia the death o f a nonspouse female actually results in a 5 percent decline o f the root andtuber area. The authors conclude that the poorer households headed by HIV/AIDS widows are in especially precarious positions. 33' Anglewicz and others 2005. 332 Yamano andJayne 2004. 333Menon 1998. 334Arndt 2003. 335 These percentages are calculated for the author's base AIDS scenario. 336 Three different scenarios are used: base AIDS scenario; less effect scenario, which softens the effects assumed inthe base AIDS scenario; and the education scenario, with increased education funding. 164 2007 indicates considerable potential o f the epidemic to weaken future economic growth.337 The base case employs official government HIVIAIDS prevalence estimates (16 percent o f the working population) and indicates a significant cost to the economy. A hypothetical 0 percent HIV/AIDS scenario estimates annual growth to be 0.8 percentage points greater than in the base case scenario, consistent with earlier estimates for both Mozambique and other high-prevalence countries. The model does not account for some o f the more complex dynamics associated with the disease, such as the relationship between general health conditions and infection, as well as the lag between infectionand symptoms. ANTIRETROVIRALTREATMENT 6.3 1 Antiretroviral therapy coverage in Mozambique increased rapidly in 2006, with higher coverage in the southernprovinces than northern ones. The first-line drug regimen in Mozambique is estimated to cost $140 per person per year.338In October 2005 there were only 32 sites administering antiretroviral therapy, and Mozambique's coverage rate was in the bottom tier in the region (table 6.4). 2006 saw a great improvement, and in December there were 150 sites in Mozambique administering antiretroviral treatment to 44,100 patients, 16 percent o f those estimated to need treatment (table 6.5). By comparison, across Africa an overall estimated 4.7 million people needed antiretroviral therapy in 2005, while an 810,000 o f them-1 7 percent-received it (table 6.6).'" Table 6.4. PercentageReceivingAntiretroviral CombinationTherapy (HAART) in Mozambique,by Gender and Region, October 2005 Region Male Female Total South 17.9 17.4 17.6 Central 4.6 4.4 4.5 North 1.6 1.3 1.4 Total 7.4 7.4 7.4 Source: GoM 2006, basedon DHS data. Table 6.5. Number of PeopleReceivingAntiretroviral CombinationTherapy (HAART) in Mozambique,2006 September October November December 2006 2006 2006 2006 Number receiving HAART 34.184 37.133 40.475 44.100 Estimated number inneedo f HAART 270,3 17 Percentagereceiving HAART 16 Source: MISAU and GoM 2007. 337Jones 2007. UNAIDS and WHO 2006. The first-line regimen inMozambiqueis madeup of Lamivudine, Stavudine, and Neviraprine. 339Treatment was scaledup from 100,000 in2003 to 810,000 by the end of 2005. 165 Table 6.6. Antiretroviral Coverage in Mozambique and NeighboringCountries, 2005 EstimatedNumber of ReportedNumber of Sites People, Aged 0-49, ART Coverage,December Providing ART, September Country and region Needing ART, 2005 2005 to December 2005 Mozambique 216,000 9 32 Tanzania 315,000 7 96 Malawi 169,000 20 60 Zambia 183,000 27 >110 Zimbabwe 321,000 8 48 SouthAfrica 983,000 21 183 Sub-SaharanAfrica 4,700,000 17 Source: UNAIDS andWHO 2006. 6.32 Antiretroviral coverage in Mozambique varies by region. Coverage declines the more northern the region, matching the prevalence pattern.340A March 2007 supervision mission found a considerable expansion o f the government program to rural areas over the previous six months, with 37 percent o f all patients receiving antiretroviral treatment now outside the provincial capitals.34'Yet 39 percent o f antiretroviral patients received treatment in Maputo city. At the end o f December 2006,59 percent o f those who had received treatment were women. 6.33 Access to antiretroviral treatment can allow AIDS victims longer and healthier lives and improve the economic situation o f the household and the probability of school enrollment for household children. No study on the impact for the labor force of antiretroviral treatment exists for Mozambique. Evidence from Kenya finds, however, that after six months o f treatment, the probability o f the patient participating in the labor force increased by 20 percent, and the number o f working hours per week increased by 35 percent.342Moreover, young boys and women living with the patient worked significantly less after the patient started the treatment while young girls and men worked the same. The findings suggest that treatment could be good for reducing young boys' labor. SUMMARY AND CONCLUSIONS 6.34 Mozambique's rapidly rising prevalence rate calls for Significantly increased efforts to prevent infection by changing risk-taking sexual behavior, particularly older men's. A main driver behind Mozambique's high incidence appears to be fairly older men having multiple sexual relations, while not always usingcondoms, with women both from younger age categories (particularly at ages 15 to 24) and with women o f their own age. Experience from Uganda suggests that 340 Table 6.4. 34' World BankAide-Memoire, MozambiqueHIV/AidsResponseMission, March 12-23,2007. 342 Thirumurthy and others 2005. InKenya273,000 people between0 and 49 years are estimatedto need treatment, 24 percent are covered, andthere are 250 reportedtreatment sites (UNAIDSand WHO 2006). Authors use longitudinal householddatathat were collectedin collaborationwith atreatment program. 166 efforts to persuade men to have fewer sexual relations and use condoms can slow incidence. 6.35 HIV/AIDS in Mozambique is dramaticallyfeminized, and the gender gap continues to grow. Women make up an estimated 58 percent o f the HIV/AIDS- infected population but outnumber men by four to one between ages 20 and 24. L o w female bargaining power and the HIV/AIDS stigma make women more vulnerable to infection-and hamper prevention work. This is exacerbated by intergenerational sex, with older men having sexual relations with younger women. Women's education and wealth determine their awareness o f HIV/AIDS and the perceived possibility o f discussing it openly with a sexual partner. The richer wealth quintile a woman belongs to in Mozambique, the more knowledge she has o f prevention methods and the more she is able to broach the topic o f prevention with her partner. 6.36 It is difficult to target HIV/AIDS-affected households for poverty reduction because only certain subgroups experiencing a prime-age adult death are worse off than unaffected households. In rural Mozambique widow-headed and high dependency ratio households are worse off than unaffected households. The labor constraint for affected households does not appear generally significant for Mozambique. To offset the potential poverty effects for HIV/AIDS-afflicted households, it may be more efficient to invest generally in labor-saving technologies, both for domestic tasks, such as fetching water and firewood, and for agricultural production. 6.37 The number of AIDS orphans is large and growing, and they are less likely to attend school. Evidence from neighborihg countries shows that the decline in enrollment probability depends on the sex of the deceased parent and the sex of the child who lost a parent: the negative effects are larger for a maternal death and a female child. This has not been investigated for Mozambique, but it may be a threat to the efforts o f closing the gender gap in education. 6.38 The HIV/AIDS epidemic risks lowering economic growth rates by as much as 1 percent annually. Two modeling exercises estimate that HIV/AIDS reduces per capita growth between 0.3 and 1percentage points a year. 167 7. THE POVERTY REDUCTIONSTRATEGY (PARPA): EVOLUTIONAND PROSPECTS 7.1 Mozambique has been a regional leader in developing a locally owned poverty reduction strategy and implementationprogram and in coordinating and harmonizing international support. Mozambique's first poverty reduction strategy was published in 1995, even before comprehensive national household survey data became available. The 1997 IAF survey and subsequent analysis led to the publication in April 1999 o f a new strategy entitled Action Guidelines for Eradication of Absolute Poverty. The first PARPA (a detailed poverty reduction strategy) was prepared in December 1999 as a fuller statement o f the Action Guidelines and implementation strategy. After public discussion, it was slightly revised and presented to donors as the government's poverty reduction strategy paper (PRSP) for combined donor support. Following the successful implementation o f PARPA I,PARPA I1was completed in 2006 and i s now under imp~ementation.~~~ 7.2 The detailedpoverty reduction strategy (PARPA-Plan0 de Accao para a Reducao da Pobreza Absolouta) articulates the Mozambican Government's strategic visionfor poverty reduction by promoting increasedproductivity and by improving the capacitiesand opportunitiesfor all Mozambicans. The first PARPA was prepared in December 1999, and after public discussion, it was revised and presentedto donors as the basis for donor support in2001-05. 7.3 PARPA I elaborated priority policies and programs that would foster human development and broad-based economic growth. Emphasizing the link between sustained growth and poverty reduction, the strategy incorporated policies and reforms to stimulate growth, identifyingphysical and human capital and rising productivity as fundamental macroeconomic determinants o f growth. The design o f PARPA I1was completed in 2006 and i s now being implemented in 2006-10. It maintained this strategic focus, organized around three main pillars: governance, human capital, and economic development. The vision common to both PARPAs includes consolidating national unity, developing each citizen's human potential, creating a working institutional system, and increasing the ability to create national wealth. 7.4 For PARPA I,citizen and civil societyparticipation were an afterthought, involving national chief security officers and individuals mainly in Maputo at a late stage. Nevertheless, PARPA Iitself opened space for a more participatory process in the development o f planning instruments (including the PARPA) and 343The PARPA documentsandthe World Bankand InternationalMoney Fundcomments (Joint Staff Assessment Note) are available inEnglishon the World Bank's POVNET website at httv://www,worldbank.ordvoverty. 168 committed to a permanent process involving representatives o f Government, the private sector, trade union confederations, civil society, and donors. To achieve this goal, the Observatorio da Pobreza (PO) was formally set up by the Mozambican Government with donor support on April 28, 2003.344 7.5 Opportunities exist to build on the successes of the PO but also of decentralization, and specifically to link the POs at national andprovincial levels to the bottom-up planning and monitoring of the consultative councils and forums at district levels and below. Given the many complexities, including the low capacity o f civil society and citizens particularly at the lower levels o f the state, further resources and investments will be neededto develop a fully functioning and participatory monitoring system in Mozambique. The effectiveness o f the POs and their influence on shaping and informing policy depend on an array o f issues, ranging from the measures to strengthen the broader enabling environment for participation and civic engagement (financial resources, information transparency and availability, enhanced government and civil society capacity), the actual spaces and opportunities to participate (scheduling, sequencing, and openness o f participation forums), and to the inclusiveness and representativeness o f the mechanisms for participation by stakeholders. Their effectiveness also depends on aligning monitoring and evaluation systems and cycles. PARPAI 7.6 The government set out in 1999 its key objective of reducing absolute poverty. PARPA I was based on the diagnosis o f poverty determinants in Mozambique from the analysis o f the first IAF, international experience with poverty reduction studies, and consultations with civil society and the private sector. It listed the main causes o f poverty: 0 Slow growth o f the economy untilthe beginning o f the 1990s. 0 L o w levels o f education, particularly o f women. 0 Highdependency rates. 0 L o w productivity inthe family agricultural sector. 0 Lack o f employment opportunities. Poor infrastructure. Inaddition, itpointedto the vulnerability to natural disasters and economic shocks. 7.7 To address these causes, PARPA I focused on implementing priority policies and programs that would foster human development and broad-based economicgrowth. The strategy emphasized the link between sustained growth and poverty reduction and incorporated policies and reforms to stimulate growth, 344The UNDP was instrumentalincreating the PO, basedon its experience setting up POs inother countries. UNDP provided seed funding of about $1.5 millionto help set up the MozambicanPO andtrain the technical secretariat andhas continuedto support capacity building activities. 169 identifying investment in physical and human capital and rising productivity as fundamental macroeconomic determinants o f growth. The priority areas for action were: Education-Provide universal education, improve quality, and increase efficiency o f schooling. Health-Improve the access and quality ofhealthcare for women. Basic inpastructure-Establish connection betweenprovinces through better and new roads, expand access to energy, promote sustained water use, and increase the provision o f sewage. 0 Agriculture andruraldevelopment-Increase productivity and income. Goodgovernance-Introduce decentralization, increase institutional efficiency, and reduce corruption. 0 Macroeconomic andfinancing administration-Promote macroeconomic 7.8 To ensure that growth policies would favor and reach the poor, the strategy aimed at increasing allocationstopublic programs beneficial to thepoor, such as education, health, and infrastructure. The strategy emphasized the balance between regions and focused on service provision in areas with high concentrations o f poverty. Detailedpolicies and programs were elaborated for each subsector in the six priority areas. Institutional reform was identified as an overarching issue. 7.9 PARPA I also gave priority to structural policies that stimulate equitable growth. These policies included raising productivity and developing markets for small producers in rural family agriculture, acceleratingjob creation and increasing opportunities for nonagricultural activities, stimulating investment in labor- intensive industry and services, and improving conditions for micro and small enterprises. Recognizing the weakness o f the consultative process for PARPA I,the government committed to the development o f "tools and procedures for consultations" to create a "systematic, agile and effective model for carrying out c o n s ~ l t a t i o n s . ~ ~ ~ ~ ~ 7.10 The strategy, prepared by a unit in the Ministry of Finance and Planning (MPF), was seenfrom the beginning as an important instrument of government planning and budgeting. This contrasts with strategies in some other countries, which were seen as a plan for the donors and only later mainstreamed into regular government processes and workflows. The initial monitoring and evaluation strategy proposed using existing mechanisms-mainly the annual report to parliament-and collecting additional new household survey data to supplement the indicators in this annual report. A concise set o f indicators was provided for monitoring progress. 345GoM 2005, p. 102, Englishversion. 170 7.11 Thereview by the WorldBank and theInternational Monetary Fund cited as strong points of thefirst PARPA thepoverty diagnostics; the strong sectoral programs, including the prioritization andfocus on macrostability and growth; and thefinancingplan. Weaknesses seen at the time were the limitedidentification o f risks, the weak link with the budget process, and the plans for implementation o f the monitoring and evaluation program set out in the PARPA, including consultation with civil society duringthe implementation. 7.12 Implementation of PARPA between 2001 and 2005 went reasonably well. As expected, given the fragile public management capacit at the time, implementation was more effective in some sectors than ~thers?~`Monitoring and evaluation continued to be a problem during the whole period. The government addressed the consultation issue by creating the Poverty Observatory (discussed below). 7.13 One of the outstandingsuccesses of PARPA I was its usefor coordinating international donors and harmonizing donor programs with government objectives. The donor community accepted the PARPA wholeheartedly as the focal point for their own strategies. As a part o f an effort streamline the monitoring process, government and donor sector working groups (SWGs) were formed to support an annual review o f PARPA progress in each sector. Through the SWGs, donors partnered with the government to ensure alignment o f the budget with the PARPA such that international financing could be provided through pooled funds at the sectoral level (SWAPS) and eventually through general budgetsupport. PARPAI1 7.14 Afler evaluation of PARPA I by the government, civil society, and international partners, PARPA 11was drafted from inputs by international and local consultants, sectoral working groups, and the national and provincial poverty observatories created to follow PARPA'S progress and performance. Based on the conclusions from the evaluation consultation, it was decided that PARPA I1should continue the priorities o f PARPA I,while sharpening the focus on conditions for sustained economic growth, through district-based development, the productive sector growth, financial system improvement, small- and medium-sized enterprise support in the formal sector, and internal revenue collection systems and methods for allocating budgeted funds. 7.15 PARPA 11is organized around three main pillars: governance, human capital, and economic development. The common elements to the three pillars include consolidating national unity, developing each citizen's human potential, creating a functioning institutional system, and increasing the ability to create national wealth. Within each pillar the main objectives were defined as: 346See chapter 4. 171 Pillar 1: Governance --- Consolidate national unity,peace, justice, and democracy. Fightcorruption, excessive bureaucracy, and crime. - Strengthen international cooperation to benefit the country. Harmonious development o f the country. Pillar 2: Human capital - Expand and improve education, access to healthcare, access to - potable water, and adequate sanitation. Promote self-esteem o f population and increase awareness o f a - culture that values work, enthusiasm, honesty, and accountability. Help youth realize their potential and their creative and entrepreneurial abilities-and participate in society. Pillar 3: Economic development --- Foster rural development. Develop the national business community. Create an environment favorable to business investment. 7.16 Objectives were also defined for eight cross-cutting topics: gender, HIV/AIDS, the environment, food and nutrition security, sciences and technology, rural development, natural disasters, and demining. 7.17 PARPA 11aimed at maintaining the high average annual growth ratefor previous years along with continuing poverty reduction. It noted that economic growth might begin to depend more on productivity gains and on domestic saving, two significant challenges in the future. It also recognized the importance o f the consolidation o f peace and social stability. The integration and balanced development o f the nationwere a priority o f PARPA 11, as were regional and global integration. 7.18 Building on the successfrom PARPA I,PARPA 11specifically highlighted increasing donor alignment and harmonization, including specific activities and measures. Maintaining close coordination with the international community to continue the flow o f aid (particularly to the state budget) was emphasized as an important policy objective. The PARPA I1 process devoted significant time to reaching agreement through all sector working groups on a strategic matrix o f monitorable indicators, which all donors would use to monitor their own projects as well as government results. This agreement eliminated the need for ministries and project executing agencies to maintain multiple systems to satisfy each donor's requirements. 7.19 An action plan for collecting andprocessing the required monitoring and evaluation data was not specified. Some data are available through administrative data collection systems, such as the number o f grade repetitions or the percentage o f children immunized. But many require the collection o f household survey data. 172 PARPA I1proposed the collection o f this data every five years (the next DHS and IAFwould be in2009), not adequate for monitoringpurposes. 7.20 The World Bank-IMF review of PARPA II cited as strong points the strong commitment to intensify and expand the work on governance and capacity building, the recognition of the challengesof HIV/AIDS and the crediblestrategy for accelerating the national response, and the more active role of the private sector. It noted that the agriculture and rural development sections were vague, suggesting greater clarity as to how agricultural growth can be ensured through increased farm-level productivity and greater access to markets. It also called for a credible and specific data-collection plan and for the use o f SISTAFE to finally identify priority expendituresdefined inthe PARPA within the national budget and to be able to monitor progress inreal time. CIVIC ENGAGEMENTTHE PARPAPROCESS347 IN 7.21 The main objective of the PO is to monitor and evaluateperformance in the implementation of the PARPA. The PO is expected to make suggestions to the Government in order to maximize the impact o f PARPA i m ~ l e m e n t a t i o n . ~ ~ ~ Secondary objectives are to provide guidance to the government and to ensure transparent interaction between the government and its partners in the fight against poverty-and to ensure that information and data pertaining to the poverty reduction process are effectively disseminated. 7.22 ThePO is a consultativeforum comprising an ad hoc advisorygroup-the Opinion Council-and a technical secretariat of civil servants to coordinate its work located in the Ministry of Planning and Finance (MPF)).The Opinion Council has 60 members, nominally split evenly among 20 representatives from central bodies, 20 from international development partners, and 20 from civil society organizations. 7.23 Following the launch of the PO in 2003, six national consultations have also been held. ProvincialPOs (PPOs) have been organized inall but one province, and in some there have been up to three PPOs ( N a m ~ u l a ) .Focused working ~ ~ ~ groups based on specific topics have been establishedwithin this structure. 7.24 After the first consultation in 2004, the original 20 participating civil- society representatives established the "G20" group of civil society organizations to coordinate their input into the PO more effectively. The G20 now comprises 347This sectiondraws on two recent studies reviewing the progressofthe PO mechanisminMozambique:an evaluation(Rasappan, Mariappan, andRodolfo 2006), commissionedby the UNDP, andareview ofthe PO as a social accountabilitymechanism(Francisco andMatter 2007), commissionedjointly by the Swiss Development Cooperation andthe World Bank. 348Cited inFrancisco andMatter (2007, 26). 349The PO inNampulais called a"development observatory" rather than apoverty observatory. 173 more than 20 organizations, representing a wide variety o f groups, including religious groups, trade unions, private sector groups, and civil society groups more generally. The main result has been the development o f annual participatory poverty assessments, the annual poverty report (Relatdrio Annual da Pobreza), which as the G20's main inputto the PO process provides an alternative perspective on poverty in the country, aiming usually to present the voices and perspectives o f the poor. The report, the first and most extensive o f its kind, "incorporated the voice o f the common people, integrated into a participatory dynamic at provincial level. Therefore, it i s not meant to be seen as a confrontation with official figures, but rather as a contribution that i s rooted inthe opinions o f thousands o f people."350 7.25 Recent reviews concluded that the PO experiment has provided an importantJrst step toward ensuring greater civic engagement in monitoring and evaluating the PARPA. It may serve inthe longer term to ensure greater downward accountability and responsiveness o f government actions and performance. Mozambique i s one o f a handful o f countries-including Armenia, Tanzania, and Uganda-that have developed a well functioning institutional framework for participatory monitoring with an established permanent forum between the government and other stakeholders dealing with poverty reduction strategy monitoring. The extension o f the PO framework to the provincial level, as well as the experience with consultative councils at the district level, suggests a growing willingness and interest by the government to engage with civil society and citizens more broadly in the planning and monitoring o f public policy. With only a decade o f multiparty democracy experience, the PO system is an achievement. 7.26 Thepoverty observatorieshave reshaped thepoverty debateandpositioned civil society and citizens more generally as legitimate actors and stakeholders in public policymaking. With the annual poverty reports, and particularly the efforts that went into producing the 2004 report, the G20 helped bring attention to the multidimensional nature o f poverty in Mozambique and to participatory and qualitative data collection and analysis in shedding light on poverty's many dimensions. 7.27 The overall impact of the PO is less than itspromise because of structural constraintsin the PO and the broader monitoring and evaluation architecture. A recent self-assessment by civil society o f their role in poverty monitoring raised concerns about the definition o f the PO'Sroles and responsibilities, its structure and articulation with other monitoring and evaluation systems, the broader development infrastructure in Mozambique, and the quality and inclusiveness o f civic engagement and participation in the PO. The quality o f participation o f PO stakeholders in poverty monitoring, and particularly the G20, depends on the broader enabling environment for participation and civic engagement, the actual spaces and opportunities to participate, and on the inclusiveness and representativeness o f the mechanisms for participation. 350G20 2004. 174 Structure and Function of thePoverty Observatories 7.28 There is a lack of clarity about the internal make-up and structure of the PO. Guidelines have been developed for the PO, butthese are vague about the PO'S overall purpose, structure, and regular functioning. The division o f roles and responsibilities between the different actors involved in the POs-the government, donors, and civil society-is also unclear, as i s the relationship o f the PO with other more established actors on the development and poverty monitoring scene. There i s actually no legal basis for the PO or for participatory PRS monitoring more generally inMozambique, despite the fact that the objectives o f the PO suggest that itbe apermanentfeature. 7.29 Particularly worrisomefrom apublic accountabilityperspectiveis the lack of clarity of the results and recommendations expected of the POs and of how such recommendationsjit into the poverty planning and programming cycle of the government. N o systematic feedback mechanisms exist between the government and the PO. Some senior technical officials at the MPF are considering including participatory approaches in proposed annual impact monitoring reports for the PARPA. But these would be prepared independently o f the PO, thus duplicating the annual poverty report. 7.30 While the PO may be open to all citizens, membership of the G20 is not necessarily automatic and open to all citizens. Indeed, a clear weakness o f the G20 i s the lack o f representatives o f the informal private sector, which in Mozambique-like most poor countries-absorbs the majority of the working population.351Indeed, the active participants inthe POs are generally members o f a relatively small and mostly urban middle-class elite.352The poor, especially poor women, do not really participate inthe POs, although most claim to represent them. The G20, however, has tried to integrate the voices o fthe poor in its annual poverty reports, especially in the 2004 report, which included interviews with more than 10,000 individuals and 100 organizations. 7.3 1 These concerns areperhaps even greater at theprovincial level,partly due to the lack of standardization of PO practices.353Coordinating and sequencing activities between the different levels have also been lacking, which means that the findings and recommendations o f PPOs are not necessarily reflected in the reports o f the national PO. PPOs in Sofala and Nampula had not developed clear links with the provincial development strategies, which are supposed to draw inputs and development plans from lower administrative levels. As Francisco and Matter noted intheir review of the Nampula PPO: "Surprisingly the., Observatories inNampula , '"Francisco and Matter 2007,36. "* Francisco and Matters 2007,37. 353Rasappan, Mariappan, Rodolfo, and others (2006,4). 175 have given significant attention to the national PARPA and so far totally ignored the very Strategic Development Planfor Nampula set for 2003-2007."354 7.32 The G20 has proposed linking the PPOs to the broader decentralization program in Mozambique, especially the efforts to increase and ensure broader public and social accountability of government actions and performance. The consultative councils at district and lower levels are a good foundation for more bottom-up planning and monitoring processes. The work o f these councils should be integrated with the monitoring efforts at the provincial levels (PPOs). The relationship between the PPOs and the provincial assemblies, once created, will also need to be better clarified. And sequencing i s needed to integrate these levels with the national level ina bottom-up fashion. ThePlethora of MonitoringMechanisms 7.33 Alongside the PO is the annual donor-focused Joint Review of the Performance Assessment Framework (PAF), organized through the joint government-donor sector working groups. Operating to a large extent in isolation o f the PO, this annual event has a major impact on government programs and budgets and on government performance. Civil society groups have been invited to these reviews since 2005, but only as observers. Joint working groups between the government and donors alone-without involving parliamentarians or other stakeholders on an equal basis-"[frustrate] nongovernmental stakeholders and [exclude] them from important dialogue forums,"355 with potentially severe implications for domestic accountability. The strong technical expertise o f the donor community dominates these reviews. As a result, the opinion council o f the PO-and specifically the G20-has thus far been largely sidelined from the assessments o f government progress. This raises the question o f whether the government's focus on donor alignment through the PAF and the joint review mechanism-and ultimately the role o f donor aid in the political economy- undermines domestic systems o f poverty monitoring. 7.34 The impact of the PO depends on the coverage and coherence of the government's PARPA monitoring and evaluation system. The monitoring and evaluation system i s weak on both accounts. The PARPA is not the guiding, planning, and budgeting instrument for the country because there is no system to link the budget and expenditures with PARPA objectives, actions, and results. It was only with PARPA I1 that the instrument was approved by the national assembly. Instead, annual plans and expenditures are set out in the Economic and Social Plan (Plano Econdmico e Social, PES) and the corresponding state budget (Orpmento Geral do Estado, OGE), drawing on the Five-Year Plan, all approved by the national assembly. Because the PO'S role is defined only in terms o f the PARPA, its role in monitoring these plans and budgets is unclear. Although the ~~ 354Francisco andMatter (2007, 33). 355Matter 2007, 3. 176 PARPA i s being integrated into state planning and budgeting, and from 2009 i s supposed to coincide entirely with the country's Five-Year Plan, a concrete and specific timetable for aligning the systems (and benchmarks to measure the alignment) and for organizing the new financial and administrative system-the SISTAFE- around PARPA is lacking. Alignment at this point depends on the quality o f strategic planning and administration at ministries and agencies. The open question i s whether and how the PO will have a more central role in poverty monitoring and evaluation as a result o f this integration. 7.35 This exclusion could undermine thefuture sustainability of the PO. Both recent reviews o f the PO process pointed to the need to "regularize and formalize the PO with the objective to transform it into a more formal mechanisdprocess with a specific purpose, processes'' and clearer definitions of roles and responsibilities o f the different actors.356To provide value-added recommendations o f strategic use, the PO process needs to be better integrated into the overall monitoring and evaluation framework. Specifically, additional mechanisms are also required "to ensure that feedback comin from the PO members i s channeled in the right direction and becomes actionable." F57 Participant Frustration 7.36 Civil society stakeholders interviewed both at national and provincial levels complained of the lack of ftnancial support for the PO process, undermining their ability and willingness to participate. Limited funds also underminedthe PO'Sability to collect and conduct analysis o f monitoring data.358 Some civil society organizations representatives also complained that government information and data on impacts, which the POs need to provide an independent assessment, are not freely available. As noted by one CSO representative: "How can we give usefulfeedback when we don't even know what is going 7.37 Low capacity-both of the government and civil society-is also a strong impediment to meaningful participatory monitoring and presents a major obstacle to effective civil society participation in the PO. For the government, working directly with citizens and civil society usually requires a different mindset and approach, and often requires different methodological skills in processes o f monitoring and evaluation data collection and analysis (such as participatory appraisal and qualitative techniques). For civil society, in addition to the financial and information constraints mentioned above i s a lack o f qualified personnel and 356Francisco andMatter 2007,47. 357Rasappan, Mariappan, andRodolfo 2006, 8. DFID and Irish Aid are preparingamultiyear Civil Society SupportMechanismfor monitoringand advocacy on governance, which shouldbe available for strengtheningcivil society participation inthese poverty monitoring initiatives. 359Rasappan, Mariappan, andRodolfo 2006, 5. 177 technical expertise to ensure their effective participation in what are essentially highlytechnical and information-driven processes.360 Recommended Actions 7.38 Increasing the effectiveness of the poverty observatories. The PO experiment has provided an important first step toward ensuring greater civic engagement in poverty monitoring and evaluation and thereby greater public participation innational discussions on poverty policy. 7.39 Ensuring that transparency, accountability, and responsiveness remain key parts of the PARPA process. Institutionalization o f the PO process would help guarantee that the principles o f accountability and transparency are adhered to. Seeking a legal basis for the PO would contribute to this. From a public accountability perspective, much more clarity i s required on expected results-how recommendations from the POs serve the poverty planning and programming cycle o f the government. There is as yet no systematic feedback mechanism between the government and the PO. A clear weakness in the structure o f the POs is the lack o f broad representation beyond the small and mostly urban middle-class elite. At the provincial level, concerns about the role o f the PO and its articulation with local strategies, plans, and budgets are even more pronounced. The work o f local consultative councils should be integrated with the monitoring efforts o f the provincial POs. 7.40 Building opportunities for hitherto excluded constituencies to participate and strengthening the role of civil society. Weak civil society capacity dampens civic engagement with the PO. Upgrading civil society organization skills i s requiredto build up a cadre o f qualifiedpersonnel and technical expertise and thus ensure credibility in their partnerships with the government. The PO has considerable potential as a mechanism to bring together key stakeholders in the poverty planning, implementation, and monitoring process. But the poor, especially poor women, have scant participation currently inthe POs, and ways must be found to incorporate civil society organizations that represent their views. 360Matter 2007, 4. 178 s L g a m a . a . L a a 9L a . a Y c % a, 2 a, 0 wu Y scd n m w a 3 .-cd0 m0 -0 5 0 6 $ 0 0 0 REFERENCES AfriMAP. 2006. "Mozambique: Justice Sector and the Rule of Law." Africa Governance Monitoringand Advocacy Project, Open Society Instituteof Southern Africa, September 19,2006. Alderman, H.2001. "The implications o f Private Safety Nets for Public Policy: Case Studies ofMozambique andZimbabwe." Africa Region HumanDevelopment Working Paper Series. World Bank, Washington, D.C. Alderman, H.,H.Hoogeveen, andM.Rossi. 2005. "Reducing Child Malnutrition in Tanzania: Combined Effects o f Income Growth and Program Intervention." Background paper for Catching the Window to Address Malnutrition in Tanzania. World Bank, Washington, D.C. Anglewicz, P., S. Bignami-van Assche, P. Fleming,A. van Assche, and C. van de Ruit. 2005. "The Impact of HIV/AIDSon Intra-household Time Allocation in Rural Malawi." Paper presentedat the IUSSP Seminar on Interactions betweenPoverty and HIV/AIDS, December 12-14, Cape Town, South Africa. Arndt, C. 2003. "HIV/AIDS, HumanCapital and Economic Growth Prospectsfor Mozambique." Africa RegionWorking Paper Series No.48. World Bank, Washington, D.C. Arndt, C., and F.Tarp. 2000. "Agricultural Technology, Risk,and Gender:A CGE Analysis o f Mozambique." WorldDevelopment287: 1307-1 326. Arndt, C., M.Barslund, and J. Sulemane. 2006. Seasonality in Calorie Consumption: Evidenceporn Mozambique.Ministry of Planning and Development Discussion PaperNo. 13E. Maputo, Mozambique: MinistCrio da Planificaqiio e Desenvolvimento. Arndt, C., S. Robinson, and F. Tarp. 2005. "Trade Reform and Gender inMozambique." Nordic Journal of Political Economy 321:73-90. Austral Consultoria e Projectos. 2005. "Pesquisa Nacional Sobre Governaqiio e Corrupqiio." World Bank, Austral Consultoria e Projectos, Maputo, Mozambique. Barnes, J. 2005. "Local Government and Development Planning: Towards Decentralised Planning and Budgeting inMozambique." Paperpresentedat the UNCDF Comparative Workshop, Performance Budgeting: The Challengesof Implementation, September 13-1,5, Dar-es-Salaam, Tanzania. 181 Barreto, A., and others. 2002. Impact0 Demogrhjko do HIV/SIDA em Moqambique Actualizaqco, an0 2000.Maputo, Mogambique: InstitutoNacional de Estatistica, MinstryofHealth, Ministry ofPlanning andFinances, Centro de Estudosde PopulagBo-UEM, Conselho Nacional de Combate ao HIV/SIDA, Faculdade de Medicina-UEM, MinistCrio da EducaqBo. Barslund, M.2007. "Children's nutritional status and dietary diversity: Evidence from demographic and health data for Mozambique." Study commissioned by the World Bank Financed by the Belgian Poverty Reduction Program. World Bank, Washington, D.C. Benfica, Rui.2006. "An Analysis o f Income Poverty Effects in Cash Cropping Economies in Rural Mozambique: BlendingEconometric and Economy-Wide Models." Ph.D.Dissertation. Department o f Agricultural Economics, Michigan State University,East Lansing, Michigan. [Retrieved from http://www.aec.msu.edu/fs2/mozambique/dissertation~benfica.pdfl. Benfica, Rui, D. Tschirley, and D.Boughton. 2006. "Interlinked Transactions in Cash Cropping Economies: The Determinants o f Farmer Participation and Performance inthe Zambezi River Valley ofMozambique." Paper presentedat the 26th International Association o f Agricultural Economists Conference, August 12-1 8, Gold Coast, Australia. [Retrieved from http://www.aec.msu.edu/fs2/mozambique/cashcropping-benfica.pdfl Benito-Spinetto, M.,and P. Moll. 2005. "Mozambique: Macroeconomic Developments, Economic Growth, and Consequences for Poverty." Background paper for the 2005 Country Economic Memorandum. World Bank, Washington D.C. Boughton, Duncan, DavidMather, Christopher B.Barrett, RuiBenfica, Danilo Abdula, David Tschirley, and Benedito Cunguara. Forthcoming. "Market Participation by Rural Households in a Low-Income Country: An Asset-Based Approach Applied to Mozambique." Boughton, Duncan, David Mather, David Tschirley, Thomas Walker, Benedito Cunguara, and EllenPayongayong. 2006. Changes in Rural HouseholdIncome Patterns in Mozambique, 1996-2002, andImplicationsfor Agriculture 's Contributionfor Poverty Reduction.MINAG/DE/DAP ResearchReport No. 61E. Department o f Agricultural Economics, Michigan State University. [Retrieved from http://www.aec.msu.edu/fs2/mozambique/wps61e.pdfl. Bourguignon, F. 2003. "The Growth Elasticity o f Poverty Reduction: Explaining Heterogeneity across Countries and Time-periods." InT. Eichler and S. Turnovsky, eds., GrowthandInequality. Cambridge, Mass.: MIT Press. 182 Braathen, E. 2003. "Democratic Decentralization inMozambique?" InM.Newitt, with P. Chabal andN.Macqueen, eds., Community and the State in Lusophone Afiica. London: King's College. Case, A., and C. Ardington. 2004. "The Impact o f Parental Death on School Enrollment and Achievement: Longitudinal Evidence from South Africa." Paper presented at the IUSSP Seminar on Interactions between Poverty and HIV/AIDS, December 12-1 4, Cape Town, South Africa. CCM, GDM, FDC and G20.2007. "Seminario sobre a AvaliaqBo das Organizaqoesda Sociedade Civil nos Observatorios da Pobreza." Unpublishedpresentation from Civil Society Workshop inBeira, Sofala, 29-30 March 2007. Centro de FormaqBoJuridica e Judiciaria. 2005. "Relat6rio do Estudo sobre a Biografia da ProfissCioJudicial." Chambers, R., and G. Conway. 1991. Sustainable rural livelihoods: practical concepts for the 22st century. IDSDiscussion PaperNo. 296. Brighton, UK:Instituteo f Development Studies. Chapoto, A., and T. S. Jayne. 2005. "Impact o f HIV/AIDS-related Mortality on Rural Farm Households inZambia: Implicationsfor Poverty Reduction Strategies." Paper presentedat the IUSSP Seminar on Interactions between Poverty and HIV/AIDS, December 12-14, Cape Town, South Africa. Chemonics. 2006. Mozambique General Services Contract-Land Tenure Services Final Report. Report prepared for the Land Component of the MillenniumChallenge Corporation (MCC) Compact. Washington, D.C.: Chemonics. Chilundo, A,, B. Cau, M.Mubai, D.Malauene, and V. Muchanga. 2005. "Land Registration inNampula and Zambezia Provinces, Mozambique." Securing Land Rights inAfrica Initiative Report No. 6. London, UnitedKingdom: International Institutefor Environment and Development. Collier, E. 2006. Towards Gender Equality in Mozambique: A Profile on Gender Relations-Update 2006. Swedish International Development Cooperation Agency (SIDA) and the Swedish Embassy, Maputo, Mozambique. CTC Consulting. 2003. "Appraisal ofthe Potential for a Community Land Registration, Negotiation and Planning Support Programme inMozambique." Department for International Development, Maputo, Mozambique. Coughlin, P. 2006. "Agricultural IntensificationinMozambique: Infrastructure, Policy and InstitutionalFramework-When do Problems Signal Opportunities?" Maputo, Mozambique. 183 Datt, G.,D. Jolliffe, and M. Sharma. 2001. "A profile of poverty inEgypt." Afiican Development Review 13 (2): 202-237. De Sousa Santos, B., andJ. C. Trindade, eds. 2003. Conzito e Transformaqao Social: Uma Paisagem das Justiqas em Mocambique.Porto, Portugal: Ediqaes Afrontamento. De Walque, D. 2006. Who GetsAIDS and How? Thedeterminants of HIV infection and sexual behaviors in Burkina Faso, Cameroon, Ghana, Kenya and Tanzania. Development Research Group WorkingPaper Series 3844. World Bank, Washington, D.C. Deaton, A. 1997. TheAnalysis of HouseholdSurveysA microeconometric approach to developmentpolicy. Baltimore, Md.: Johns Hopkins University Press for the World Bank. Deininger, K., and L.Squire. 1996. "A New Data Set Measuring Income Inequality." WorldBank Economic Review 103: 565-91. Demery,L.2003. "Analyzing the Incidence ofPublic Spending." InF.Bourguignon and L.A. Pereirada Silva, eds., TheImpact of Economic Policies onPoverty and Income Distribution: Evaluation Techniquesand Tools.New York: Oxford University Press. Donovan, C., and D.Mather. 2007. "The Impact of Prime-age Adult Mortality on Rural Household Income, Assets and Poverty in Mozambique." Background paper for the Mozambique Poverty, Gender and Social ImpactAssessment.World Bank, Washington, D.C. Eberlei, W. 2006. "Unfolding Communicative Power inPoverty Reduction-Stakeholder Participation in PRS Processes." Social Development Department,World Bank, Washington, D.C. Unpublishedmimeograph. ECON Analysis. 2005. "Impacts of Extension Services in Rural Mozambique." Consultant Report for the 2005 Rural Development Strategy. World Bank, Washington D.C. Evans, D., and E.Miguel. 2005. "Orphans and Schooling inAfrica: A Longitudinal Analysis." Paper presentedat the IUSSP Seminar on Interactions between Poverty and HIV/AIDS, December 12-14, Cape Town, SouthAfrica. Finney, C. E., and E.Kleemeier. 2003. "Background Paper on the Water Sector." Background paper to the Mozambique Public Expenditure Review. Government o f Mozambique and World Bank, Washington, D.C. 184 Fox, L.,and M.S. Gaal. 2007. Working Out of Poverty: Job Creation and the Quality of Growth in Afiica. Directions inDevelopment. Washington, D.C. Fox, L.,C. Arndt, F. Mafambissa, S. Samuel, J. Arbache, R.Benfica, and M.Sekkel. 2006. Job Creation in Mozambique: Is Labor Law Reform theAnswer? Discussion Paper 35E. DirecqBoNacional de Estudos e Anhlise de Politicas, MinistryofPlanning andDevelopment, Mozambique andPoverty andEconomic Management Department Africa Region, World Bank. Fox, L., E. Bardasi, and K.van den Broeck. 2005. "Poverty inMozambique: Unraveling Changes and Determinants." Africa RegionWorking Paper SeriesNo. 87. World Bank, Washington, D.C. Francisco, A. A. S., and K.Matter. 2007. Poverty Observatory in Mozambique-Final Report. Report commissioned by the Swiss Agency for Development and Cooperation (SDC) and by the Social Development Department o f the World Bank. Richterswil, Switzerland: Gerster Consulting. G20.2004. Annual Poverty Report 2004: Fighting the Causesof Poverty. Maputo, Mozambique. ,2005. Annual Poverty Report 2005. Maputo, Mozambique. Gaventa, J., A. Shankland, and J. Howard, eds. 2002. "Making rights real: exploring citizenship, participation and accountability." IDSBulletinVol. 33 No. 2. InstituteofDevelopment Studies, Brighton, UK. Gibson, J., and S. Rozelle. 2003. "Poverty and access to roads inPapuaNew Guinea." Economic Development and Cultural Change 52(1): 159-1 86. Golan, A,, G. Judge, and D.Miller. 1996.Maximum Entropy Econometrics: Robust Estimation with Limited Data. Series inThe Financial Economics and Quantitative Analysis. West Sussex, England: John Wiley & Sons Ltd. GoM (Governo de Moqambique). 2000. "Plano de AcqBo Para a PobrezaAbsoluta, 2001-2005 PARPA I." Governo de Moqambique, Maputo, Moqambique. Available at www.worldbank.org/poverty. , 2005. "Plano de AcqBo Para a PobrezaAbsoluta, 2006-2009 PARPA TI." Governo de Moqambique, Maputo, Moqambique. Available at www,worldbank.ordDoverty. .2006. "UNGASS Declaration o f Commitment on HIV/AIDS-Progress Report Reporting Period 2003-05." Draft. 185 .2007. "Extension Master Plan 2007-20 16-Final Draft". Ministryof Agriculture, National Directorate of Extension. Maputo, Mozambique. GoM (Governo de Moqambique) and PAP (os Parceiros para o Apoio Programatico). 2007. Revis60 Conjunta-Anexo IV Relatorios dos Gruposde Trabalho.Maputo, Mozambique: Governo de Moqambique and Parceiros para o Apoio Programatico. GoM (Governo de Moqambique), M I S A U (Ministry o f Health), and National Directorate o f Health. (2005). Report on the Update of the HIV Epidemiological Surveillance Data-2004 Round.Maputo, Mozambique. Greer, J., and E. Thorbecke. 1986. "A methodology for measuring food poverty applied to Kenya." Journal of DevelopmentEconomics 24 (1): 59-74. Haddad L.,H.Alderman, S. Appleton, L. Song, and Y. Yohanes. 2003. "Reducing Child Malnutrition: How Far Does Income Growth Take Us?" The WorldBank Economic Review 17 (1): 107-131. Handa, S., and K.R. Simler.2005. "Quality or Quantity? The Supply Side Determinants o f Primary Schooling ina Poor Rural Economy." University o fNorth Carolina and International Food Policy Research Institute. Handa, S., K.R. Simler, and S. Harrower. 2004. Human Capital, Household Welfare, and Children's Schooling in Mozambique. .International FoodPolicy Research Institute ResearchReport 134. Washington, D.C. Hardle, W. 1990.Applied nonparametricregression.Cambridge, UK: Cambridge UniversityPress. Hawkins, K.,F. Mussa, and S. Abuxahama. 2005. "Milking the Cow: Young Women's Constructions o f Identity, Gender, Power and Risk in Transactional and Cross- generational Sexual Relationships." A PEER research study, Options Consultancy Services and Population Services International. Maputo, Mozambique. Heltberg, R., K.Simler, and F. Tarp. 2001. Public Spendingand Poverty in Mozambique. WIDER DiscussionPaper No. 2001/63. Helsiniki,Finland. Hodges, A,, and R. Tibana. 2005. A Economia Politica do Orgamento em Moqambique, Lisbon, Portugal: Principia. Howard, J., J. Jeje, D.Tschirley, and others. 1998. WhatMakesAgricultural Intensijkation Profitablefor Mozambican Smallholders?An Appraisal of the Inputs Subsector and the 1996/97DNEWSG 2000 Program. VolumeI: Summary. Research report 32E. Maputo, Mozambique. [Retrieved from 186 http://www.aec.msu.edu/fs2/mozambique/wps32.pdfl. Ibraimo, M.2005. Evoluqclo das Disparidades no Bem-Estar dosAgregados Familiares e o Acesso aos bens e Sewiqos 1997e 2003. Discussion Paper 14P. Maputo, Mozambique: Direcqilo Nacional de Estudos e Anhlise de Politicas, MinistCrio de Planificaqilo e Desenvolvimento, Moqambique. .2006. "Participation and Decentralization." Background paper for the Mozambique Poverty, Gender and Social Assessment (unpublisheddraft). World Bank, Washington, D.C. IFAD (International Fundfor Agricultural Development). 2005. "Agriculture Development inRepublic o f Mozambique." Agriculture Support Programme FormulationReport, Working Paper 2. International Fundfor Agricultural Development, Rome. Ilal, A., and M.,Chisvo. 2005. "Avaliaqilo Final da Unidade de Desenvolvimento Comunitario." Maputo, Mozambique: Organizaqilo Holandesa de Desenvolvimento, Programa MAMM. IMF (International Monetary Fund). 2004. Review of the Economic and Social Planfor 2003. IMF Country Report No. 041135. Washington, D.C. ,2005. Republic of Mozambique: Poverty Reduction Strategy Paper Annual Progress Report. IMF Country Report No. 051310. Washington, D.C. .2006. Republic of Mozambique: ThirdReview Under the Three-Year Arrangement Under the Poverty Reduction and Growth Facility and Requestfor Modijkation of Performance Criteria. IMF Country Report No. 06/46. Washington, D.C. INE(Instituto Nacional de Estatistica). 2000. Impact0 Demografico do HIV/SIDA em MocambiqueActualizacao. Maputo, Mozambique. .2004. Inque`ritoDemogrbfico e de Sazide 2003. Maputo, Mozambique: Instituto Nacional de Estatistica e MinistCrio de Safide, Moqambique. .2006a. Inque`rito Integrado a Forqa de Trabalho IFTRAB-2004/2005. Maputo, Mozambique: InstitutoNacional de Estatistica, Moqambique. .2006b. TheNon-profit Institutions in Mozambique: Outputs of thefirst national census 2004/05. CINSFLU 2004. Maputo, Mozambique: InstitutoNacional de Estatistica. 187 INE(Instituto Nacional de Estatistica), MISAU(MinistryofHealth), and Measure DHS+/ORC Macro. 2004. MozambiqueDemographic and Health Survey 2003. Maputo, Mozambique: Instituto Nacional de Estatistica and Ministryo f Health. INE(Instituto Nacional de Estatistica), MISAU(MinistryofHealth), MPF (Ministryof Planning and Finances), CEP (Centro de Estudos de Populacao da Universidade Eduardo Mondlane), CNCS (Conselho Nacional de Combate ao HIV/SIDA), Faculdade de Medicina-UEM. 2004. Impacto Demografico do HIV/SIDA em Mocambique-Acualizacao Rondade Vigilancia Epidemiologica 2002. Maputo, Mozambique: Instituto Nacional de Estatistica. INE(Instituto Nacionalde Estatistica), MISAU(MinistCrio da Saude), MinistCrio do Plano e Finanqas, Centro de Estudos de PopulaqBo-UEM, Conselho Nacional de Combate ao HIV/SIDA, Faculdade de Medicina-UEM, MinistCrio da EducaqBo. (2002). Impacto Demogrbfico do HIV/SIDA em Moqambique (Actualizaqco,ano 2000). Maputo, Moqambique: Instituto Nacional de Estatistica. INEF(Institute for Development andPeace). 2006. "Civil Society and Stakeholder Involvementin PRS Monitoring." Social Development Department, World Bank, Washington, D.C. Mimeograph. James, R. C., C. Arndt, and K.R. Simler. 2005. Has Economic Growth in Mozambique Been Pro-Poor? Discussion Paper No. 8E. Maputo, Mozambique: National Directorate o f Studies and Policy Analysis (DNEAP), MinistCrio de Planificaq5o e Desenvolvimento. Jolliffe, D., G. Datt, and M.Sharma. Forthcoming. "Robust poverty and inequality measurement in Egypt: Correcting for spatial-price variation and sample design effects.'' Review of DevelopmentEconomics. Jones, S. 2006. GrowthAccountingfor Mozambique 1980-2004. Discussion Paper No. 22E. Maputo, Mozambique: National Directorate o f Studies and Policy Analysis, MinistryofPlanning andDevelopment, Republic ofMozambique. .2007. "Sustaining Growth inthe Long-Term.'' International Monetary Fund, Washington, D.C. Kanbur, R., ed. 2001. Q-Squared: Combining Qualitative and Quantitative Methods in PovertyAppraisal. Delhi: PermanentBlack. Kaufmann, D., A. Kraay, and M.Mastruzzi. 2006. GovernanceMatters V: Governance Indicatorsfor 1996-2005. Washington, D.C. [Retrieved from http://www.worldbank.org/wbi/governance/govmatters5]. 188 Kennedy E., and P. Peters. 1992. "Household food security and child nutrition: the interaction o f income and gender o f household head." WorldDevelopment, 20 (8): 1077-1 085. Khan, S., L.Hidalgo, C. Ismael, S. Meershoek, H.Ibraimo, A. Tiago, A. Miguel, andN. Junusso. 2004. "Mozambique: Investir naNutrigBo C Reducir a Pobreza." MinistCrio da Sahde, RepartigBo de NutriqBo, Maputo, Mozambique. Khandker, S. R. 1998. "Fighting Poverty with Microcredit: Experience inBangladesh." World Bank, Washington D.C. Klasen, S. 2002. "Low Schooling for Girls, Slower Growth for All?" WorldBank Economic Review 6 (3) 345-373. Klasen, S., and S. Woltennann. 2004. "The Impact o f Demographic Dynamics on Economic Development, Poverty and Inequality inMozambique." Background paper for the Mozambique Country Economic Memorandum. World Bank, Washington, D.C. Knight,R.2002. Camponeses'Realities: Their Experiences and Perceptions of the 1997 Land Law. Oxford, UK: Oxfam. Kyed, H.M.,and L.Burr.2006. "New Sites o f Citizenship: Recognition o f Traditional Authority and Group-based Citizenship inMozambique." Journal of Southern Ajkican Studies (32) 3. Lanjouw, P. 1994. Regionalpoverty in Pakistan: How robust are conclusions? Washington, D.C.: World Bank. Lindelow, M.2003. "The Utilization of Curative Healthcare in Mozambique: Does IncomeMatter?" Centre for Studies o fAfrican Economies Working Paper Series. World Bank, Washington, D.C. Macamo, J. L.2006. "Mozambique Governance Indicators: Trends and Comparisons, 1996-2005." World Bank, Maputo, Mozambique. Unpublishedmimeograph. Malena, C., R. Forster, and J. Singh.2004. Social Accountability: An Introduction to the Concept and Emerging Practice. Social Development PaperNo. 76. Washington, D.C. Mather, D.,H.Marrule, C. Donovan, M.Weber, and A. Alage. 2004. Analysis ofAdult Mortality within Rural Households in Mozambique and Implicationsfor Policy. MinistryofAgriculture andRuralDevelopment, Directorate ofEconomics ResearchPaper Series, Report No. 58E.Maputo, Mozambique. 189 Matter, K.2007. "Literary Review on International Experiences o f Participatory PRS Monitoring." Background report for Francisco and Matter, Poverty Observatory in Mozambique. Social Development Department, World Bank, Washington, D.C. Meneses, M.P., J. Fumo, G. Mbilana, and C. Gomes. 2003. "As Autoridades Tradicionais no Context0 do Pluralism0 Juridico." InB. Sousa de Santos and J. C. Trindade, Confictoe Transformaggo Social: UmaPaisagem de Justiga. Porto, Portugal: Afrontamento. Menon, R., M.J. Wawer, J. K.Konde-Lule, N.K. Sewanlambo, and C. Li ,1998. "The Economic Impact o f Adult Mortality on Households in Rakai District, Uganda." InM.Ainsworth, L.Fransen, and M.Over, eds., ConJFontingAIDS; Evidence JFomDeveloping Countries. Washington, D.C.: The World Bank, the European Commission, and United States Agency for International Development. MISAU(Ministryo fHealth). 2000. Revisgodas Mortes Maternas em Mogambique 1998/99.Maputo, Mozambique: MinistCrio de Sa6de. ,2003. Atenggo Integrada as Doengas da Infdncia 1998-2002. Maputo, Mozambique: MinistCrio de Sahde. ,2004. Relatorio Anual de 2003.Maputo, Mozambique: Direcqiio de Recursos Humanos. ,2005. "Controlo de Malhria em Moqambique." MinistCrio de Sahde, Maputo, Mozambique. MISAU(Ministryof Health) andNational Directorate o fHealth. 2005. Report on the Updateof the HIV Epidemiological SurveillanceData-2004 Round. Maputo, Mozambique: Ministry o f Health. MPF (Ministry o f Planning and Finance). 1999. "Conta Geral do Estado." Ministry o f Planning and Finance. Maputo, Moqambique. .2000. "Conta Geral do Estado." MinistryofPlanning and Finance. Maputo, Moqambique. ,2001, "Relat6rio de Execusstio Fiscal." Ministry o f Planning and Finance. Maputo, Moqambique. .2002. "Relat6rio de Execussiio Fiscal." Ministry o f Planning and Finance. Maputo, Moqambique. ,2003. "Relat6rio de ExecussiioFiscal." Ministry o fPlanning and Finance. Maputo, Moqambique. 190 .2004. Poverty and Well-Beingin Mozambique: TheSecondNational Assessment. Maputo, Mozambique: Ministryo f Planning and Finance. .2004b. "Relat6rio de ExecussiloFiscal." MinistryofPlanningand Finance. Maputo, Moqambique. .2005. "Conta Geral do Estado." Ministry o f Planning and Finance. Maputo, Moqambique. .2006. "Orqamento Geral do Estado." Ministry ofPlanning and Finance. Maputo, Moqambique, MPF (MinistryofPlanning and Finance) and Direcqilo Nacional de Planificaq3o e Orqamento. 2002. Mapeamento da Pobrezaem Moqambique: Desagregaqgo das Estimativasda Pobreza e Desigualdade aos Niveis de Distrito e Post0 Administrativo. Maputo, Moqambique: Ministry of Planning and Finance and Direcqgo Nacional de Planificaqilo e Orqamento. MPF (Ministry o f Planning and Finance), IFPRI (International Food Policy Research Institute),and Purdue University.2004. Poverty and Well-Being in Mozambique: TheSecondNational Assessment 2003. Maputo, Mozambique: Ministryof Planning and Finance. MPF(MinistryofPlanning and Finance) and (INE) Instituto Nacionalde Estatistica. 1996197.Inquerito Nacional Aos Agregados Familiares Sobre as Condicoesde Vida.Maputo, Mozambique: Ministryof Planning and Finance (MPF). .2002103. Inquerito Nacional Aos Agregados Familiares Sobre Orcamento Familiar. Maputo, Mozambique: Ministry of Planning and Finance. MPF (Ministry o f Planning and Finance) and M E C (Ministry o f Education). 2003. "A Despesa Publica com a Educaqilo em Moqambique." Ministryo f Planning and Finance and Ministryo f Education, Maputo, Mozambique. MPF (Ministry o f Planning and Finance) and MISAU(Ministryo f Health). 2004. "A Despesa Publica com a Saude em Mozambique." Ministry o f Planningand Finance and Ministry o f Health, Maputo, Mozambique. MPF (Ministry of Planning and Finance), UEM(Eduardo Mondlane University), IFPRI (International Food Policy Research Institute). 1998. Understandingpoverty and well-being in Mozambique: Thefirst national assessment (1996197).Washington, D.C: International Food Policy Research Institute. Moser, C., A. Norton, T. Conway, C. Ferguson, and P. Vizard. 2001. To Claim our Rights: Livelihood Security,Human Rights and SustainableDevelopment. London: Overseas Development Institute. 191 Mukherjee, S., and T. Benson. 2003. "The determinants o f poverty inMalawi, 1998." WorldDevelopment 31 (2): 339-358. Muzima, J., and C. Arndt. 2004. "TendCncias Sobre a EficiCncia no Subsistema de EducaqEioPrimiria em Mogambique: EstimagEio de Matrizes de Transiq3o 1992- 2000." Ministryo f Planning and Finance, DNPO, Maputo, Mozambique. Mimeograph. Naaera and Nicholajsen. 1998."Agricultural household behavior and labor market components. Macroeconomic Reforms and Regional Integration in Southern Africa project." Development Economics Research group, University o f Copenhagen. Mimeograph. Nhate, V. 2005. "Orphans in Mozambique: Vulnerability, Trends, Determinants, and Programme Responses." National Directorate o f Planning and Budget, Ministry o f Planning and Development, Maputo, Mozambique. Nhate, V., and K. Simler. 2006. Small grouppoverty estimates: The case of disabled people in Mozambique. National Directorate o f Studies and Policy Analysis (DNEAP)Discussion PaperNo. 25E.Maputo, Mozambique: MinistCrio de PlanificaqEio e Desenvolvimento. Norfolk, S., 2004. "Examining access to natural resources and linkages to sustainable livelihoods: a case study o f Mozambique." FA0 LSP Working Paper 17.FAO, Rome. Norfolk, S. and C. Tanner. 2006. "Mozambique Country Case Study on Improving Tenure Security for the Poor." Paper presented at Nairobi Conference on Poverty and Land Tenure Security, November, Nairobi, Kenya. Price Waterhouse Coopers. 2006. "Final Evaluation o f the FirstPhase o f the National Agriculture Development Programme - PROAGRI (1999-2005). Volume I:Main Report." Maputo, Mozambique. Rasappan, A,, M.Mariappan, and B. J. Rodolfo. 2006. Poverty Observatories: Lessons Learned. UNDP Moz/03/002 Project Evaluation. Maputo: United Nations Development Programme and Government o f Mozambique. Ravallion, M. 1994.Poverty comparisons. Chur, Switzerland: Hanvood Academic Publishers. . 1998."Poverty lines theory andpractice" LivingStandards Measurement in Study Working Paper No. 133. World Bank, Washington, D.C. 192 .2005. "Why Should Poor People Care about Inequality?" Background paper for the WorldDevelopmentReport2006. World Bank, Washington, D.C. Ravallion, M., and B. Bidani. 1994. "How robust is a poverty profile?" WorldBank Economic Review 8 (1): 75-102. Ravallion, M.,and M.Lokshin. 2003. "On the utility consistency of poverty lines." World Bank Policy ResearchWorking Paper No. 3157. World Bank, Washington, D.C. Ravallion, M.,and S. Chen. 1997. "What CanNew Survey Data Tell us about Recent Changes inDistribution and Poverty?" WorldBank Economic Review 112: 357- 82. Ruffer T., and J. Knight. 2007. "Informal Sector Labour Markets inDeveloping Countries." Oxford Policy Management. SADC (Southern African Development Community). 2006. National Report of Mozambique.Report on National Meeting of the SADCInitiative on Land and Agrarian Reform. Maputo, Mozambique: Southern African Development Community. Sen, A. 1999. Developmentas Freedom.New York: Knopf. Serrano, R. 2002. Participation, Transparency,and DownwardAccountability in District Planning in Mozambique-Final Report,August 2002. Report prepared as an input for the design o fthe Decentralized Planning and Finance Project. Water and Urban Group 1, Easternand Southern Africa, World Bank, Washington, D.C. Seuane, S., and M.Rivers-Moore. 2005. Aspecto do Ge'nero e o Impact0 do HIVISIDA sobre os direitos das mulheres e das crianqas no access0 a terra e recursos naturais. Maputo, Mozambique: Centro de FormaqBoJuridica e Judiciaria. Shah, M.K.2003. UsingCommunityScorecardsfor Improving Transparency and Accountability in the Delivery of Public Health Services: ExperienceJFomLocal Initiativesfor Health (LIFH)Project. Lilongwe, Malawi: Care International. Simler, K.,and H.Ibraimo. 2005. "Estimating Changes inChildren's Nutritional Status When Samples Are Not Comparable: An Example from Mozambique." Paper prepared for the 10thAnnual Conference on Econometric Modelling inAfrica, July 6-8, Nairobi, Kenya. Sitoe, E., and C. Hunguana. 2005. Democratic decentralization is needed to keep the MozambicanMiracle StillAlive. CEDE Policy Brief.Maputo, Mozambique: Centro de Estudos de Democracia e Desenvolvimento, Universidade Eduardo Mondlane. 193 Sutton, R. 1999. "The Policy Process: An Overview." OD1Working PaperNo. 118. The OverseasDevelopment Institute,London. Tanner, C. 2005. "Land Rights and Enclosures: Implementing the Mozambican Land Law inPractice." Paper presentedat the International Conference of the Changing Politics of Land inAfrica: domestic policies, crisis management, and regional norms, University of Pretoria, 28-29 November. Tanner, C., and S., Baleira. 2006. "Mozambique's legal framework for access to natural resources:the impact of new legal rights and community consultations on local livelihoods." LSP Working Paper No. 28. Foodand Agriculture Organization of the UnitedNations, Maputo, Mozambique. Tarp, F., K.R. Simler, C. Matusse, R.Heltberg, and G. Dava. 2002. "The robustnessof poverty profiles reconsidered." Economic Developmentand Cultural Change 51 (1): 77-108. Thirumurthy, H.,J. Graff Zivin, and M.Goldstein. 2005. TheEconomic Impact of AIDS Treatment:Labor Supply in WesternKenya.NBERWorking Paper Series, No. 11871. Washington, D.C.: National Bureauof Economic Research. Thomas, D. 1990. "Intra-Household ResourceAllocation: An InferentialApproach." The Journal of Human Resources, 25 (4): 635-664. Thomas, D., J. Strauss, and M.Henriques. 1991."How Does Mother's Education Affect Child Height?" TheJournal of Human Resources (26) 2: 183-21 1. Tschirley, D., Danilo Abdula, andMichael T. Weber. 2006. TowardImproved Maize Marketing and TradePolicies to Promote HouseholdFood Security in Central and SouthernMozambique. MINAG/DE/DAP ResearchReport No. 60E. DepartmentofAgricultural Economics, Michigan State University. Tvedten, I., M.Paulo, and C. Rosfirio. 2006. `Opitanha' SocialRelations ofRural " Poverty inNorthern Mozambique." CHR. MichelsenInstitute. UEM(Universidade Eduardo Mondlane). 2006. "Avaliaqgo Social Combinada sobre Pobrezae GCnero em MoGambique-Niassa, Nampula, ZambCzia e Gaza Componente Qualitativa." Background paper for the Mozambique Poverty, Gender and Social Assessment. Departamentode Arqueologia e Antropologia da Universidade Eduardo Mondlane, Maputo, Mozambique. UNAIDS(Joint UnitedNations Programme on HIV/AIDS). 2004. 2004 Report on the Global Epidemic. Geneva, Switzerland: Joint UnitedNations Programme on HIV/AIDS. 194 .2006.2006 Report on the Global Epidemic. Geneva, Switzerland: Joint United Nations Programme on HIV/AIDS. UN.AIDS(Joint UnitedNations Programme onHIV/AIDS)and WHO (World Health Organisation). 2004. "Epidemiological Fact Sheets on HIV/AIDS and Sexually Transmitted Infections: Mozambique." Joint UnitedNations Programme on HIV/AIDSandWorld HealthOrganisation. .2005. "AIDS Epidemic Update-December 2005." Joint UnitedNations Programme on HIV/AIDS and World Health Organisation, Geneva, Switzerland. .2006a "AIDS EpidemicUpdate-December 2006." Joint UnitedNations Programme on HIV/AIDS and World Health Organization, Geneva, Switzerland. .2006b. 2006 Report on the Global AIDS Epidemic. Geneva, Switzerland: Joint UnitedNations Programme on HIV/AIDS and World Health Organization. .2006c. "Progress on Global Access to HIV Antiretroviral Therapy. A Report on "3 by 5" and Beyond", Joint UnitedNations Programme on HIV/AIDS and World Health Organisation, Geneva, Switzerland. UNCDF (UnitedNations Capital Development Fund). 2004. AvaliaqEo Independente do Impact0 de Programa (PIA) do Prograrna de Desenvolvimento Local do UNCDF. Maputo, Mozambique: Ebony Consulting International (ECI). UNDP (UnitedNations Development Programme). 2000. MozambiqueNational Human Development Report 2000-Education and human development: trajectory, lessons and challengesfor the 21st century. Maputo, Mozambique: United Nations Development Programme. .2006a. Human Development Report 2006-Beyond scarcity: Power,poverty and the global water crisis. New York: UnitedNations Development Programme. ,2006b. Mozambique National Human Development Report 2005-Reaching for the Millennium Development Goals. Maputo Mozambique: UnitedNations Development Programme. UNICEF (UnitedNations Children's Fund).2004. The State of the World's Children 2006, Excluded and Invisible. New York: The UnitedNations Children's Fund. .2006. Childhood Poverty in Mozambique:A situation and trends analysis. Maputo, Mozambique: UnitedNations Children's Fund. UNPP (UnitedNations Polulation Division). 2006. WorldPopulation Prospects Database. New York: UnitedNations. [Retrieved from http://esa.un.org/unpp/]. 195 USAID(UnitedStates Agency for InternationalDevelopment). 2005. "Corruption Assessment: Mozambique." Washington, D.C.: Management Systems International and UnitedStates Agency for International Development. van den Broeck, K. 2005. "Healthcare Use inMozambique." Background paper for the Mozambique Integrated Poverty, Social and Gender Assessment. World Bank, Washington, D.C. .2006. "Determinants of Stunting and the Role of Child Care Practices in Mozambique." Background paper to the Mozambique Integrated Poverty, Social and Gender Assessment. World Bank, Washington, D.C. Walker, T., D.Tschirley, J. Low, M.Pequenino Tanque, D. Boughton, E. Payongayong, and M.Weber. 2004. "Determinants of Rural Income inMozambique in2001- 2002." International Development Collaborative Working PapersMZ-MINAG- RR-57E.DepartmentofAgricultural Economics, Michigan State University. [Retrieved from http://ideas.repec.org/s/msu/icpwrk.htm1]. Water and Sanitation Program-Africa Region. 2002. UpgradedFamily Wells in Zimbabwe: Household-Level Water Suppliesfor Multiple Uses.Blue Gold Series, FieldNote 6. Nairobi, Kenya: World Bank. .2004. "Self Supply: A FreshApproach to Water for Rural Populations." World Bank, Nairobi, Kenya. WHO (World Health Organization). 1985.Energy andprotein requirements. Technical Report Series No. 724. Geneva. .2005. The WorldHealth Report 2005: Making Every Mother and Child Count. Geneva. .2006. The WorldHealth Report 2006: Working Togetherfor Health. Geneva. World Bank. 1997. "Mozambique Agricultural Sector Memorandum." World Bank, Washington, D.C. .2000. WorldDevelopment Report 2000/1: Attacking Poverty. New York: World Bankand Oxford University Press. .2004. "Primary healthcare inMozambique: Service Delivery ina Complex Hierarchy." Africa Region HumanDevelopment Working Paper Series No.69. World Bank, Washington, D.C. .2005a. Poverty and Social Impact Analysis. Primary School Enrolment and Retention - TheImpact of School Fees. Report No. 29423-MZ. Washington D.C. 196 .2005b. Better Public Spending on Health to Reach the Millennium Development Goals. Washington, D.C. .2005c. "Mozambique RuralDevelopmentStrategy: Promoting Smallholder Agricultural Growth." Environment,Rural, and SocialDevelopmentDepartment, Washington, D.C. .2005d. Mozambique Country Economic Memorandum: SustainingGrowth and Reducing Poverty. ReportNo. 32615-MZ. Washington, D.C. .2005e. WorldDevelopmentReport 2006: Equity and Development. New York: The World Bank and Oxford University Press. .2006a."Mozambique: Agricultural DevelopmentStrategy-Stimulating Smallholder Growth." AFTS1.Agriculture, Environment and SocialDevelopment Unit. Country Department2. Africa Region.World Bank, Washington, D.C. .2006b. "Mozambique Country IntegratedPoverty, Gender and Social Assessment." Concept note. World Bank, Washington, D.C. ,2006~.The Human Resourcesfor Health Situation in Mozambique. Africa RegionHuman DevelopmentWorking Paper Series, No. 91. World Bank, Washington, D.C. .2007. "Mozambique Country PartnershipStrategy,2008-20 11-Empowering Institutions and Citizens for Shared Growth." World Bank, Washington, D.C. World Bank and African DevelopmentBank. 2004. Millennium Development Goalsfor Water and Sanitation, Country Assessment Mozambique. Washington, D.C. Yamano, T., and T. S. Jayne. 2004. "Measuring the Impactsof Working-Age Adult Mortality on Small-scale Farm HouseholdsinKenya." WorldDevelopment 32 (1): 91-1 19. .2005. "Working-Age Adult MortalityandPrimary SchoolAttendance inRural Kenya." Economic Developmentand Cultural Change 53 (3): 619-53. 197 198