Report No. 25662-TP Timor-Leste Poverty Assessment Poverty in a New Nation: Analysis for Action (In Two Volumes) Volume I: Main Report May 2003 Poverty Reduction and Economic Management Sector Unit East Asia and Pacific Region Joint Report of the Government of the Democratic Republic of Timor-Leste, ADB, JICA, UNDP, UNICEF, UNMISET and the World Bank CURRENCYEQUIVALENTS CurrencyName = US Dollar FISCAL YEAR July 1-June 30 ABBREVIATIONS AND ACRONYMS ADB Asian DevelopmentBank CFET ConsolidatedFund for East Timor DPT Diphtheria, Pertussis, Tetanus ETTA East Timor TransitionalAuthority GDP GrossDomesticProduct JICA Japan International CooperationAgency MDG MillenniumDevelopmentGoal MICS MultipleIndicatorsCluster Survey MoPF Ministry of Planning and Finance NDP National DevelopmentPlan NGO Non-Government Organization PNG PapuaNew Guinea PPA ParticipatoryPotentialAssessment PPP PurchasingPowerParity PTA Parent TeacherAssociation SUSENAS Indonesian Socio-economicHousehold Survey TLSS Timor-LesteLiving StandardMeasurement Survey UN UnitedNations UNDP United Nations DevelopmentProgram UNICEF United Nations Children'sFund UNMISET United Nations Mission of Support in East Timor UNTAET United Nations Transitional Administration in East Timor Vice President Jemal-ud-dinKassum, EAP Country Director Xian Zhu, EACNF SectorDirector and Chief Economist Homi Kharas, EASPR SectorManager Tamar Manuelyan Atinc, EASPR Task Team Leaders Benu Bidani and Kaspar Richter, EASPR TABLE OFCONTENTS PREFACE ........................................................................................................................... ACKNOWLEDGEMENTS ............................................................................................ .*. i 111 EXECUTIVESUMMARY i 1. ...................................................................................................... ............................................................................................... Introduction..................................................................................................................... A NEW NATION 1 1 Poverty Assessment Project............................................................................................ 1 The National Development Plan..................................................................................... 3 4 2.Objectives and Organization ofthe Report..................................................................... TRANSITION TO INDEPENDENCE ................................................................... 7 Developments Since the Violence: Economic And Social Trends................................. 7 Developments since the Violence: The People's Perspective in2001........................... 9 3.Methodology Summary ....................................................................................................................... 15 WELFARE PROFILE ............................................................................................ ................................................................................................................. 16 16 Poverty Profile .............................................................................................................. Inequality ...................................................................................................................... 18 29 Summary andPolicy Issues ......................................................................................... 30 31 4.Research Issues............................................................................................................. ..................................................................................................... Employment and Poverty.............................................................................................. OPPORTUNITY 32 35 Rural Living Standards ................................................................................................. 37 Urban living standards .................................................................................................. 45 Summary and Policy Issues.......................................................................................... 50 51 5.Research Issues ............................................................................................................. BASICSOCIAL SERVICES ................................................................................. 52 Public Spending for Basic Services .............................................................................. 54 Education...................................................................................................................... 56 Health............................................................................................................................ 70 Summary and Policy Issues .......................................................................................... 80 82 6.Research Issues............................................................................................................. HOUSEHOLDSECURITY ................................................................................... 83 Disadvantaged Groups.................................................................................................. 84 Food Security................................................................................................................ 89 SummaryandPolicy Issues.......................................................................................... 94 94 7.Research Issues ............................................................................................................. ......................................................................... Millennium Development Goals., ................................................................................. DEVELOPMENT CHALLENGE 95 95 Poverty. Growth. and Inequality: Projections............................................................. 101 Determinants of Poverty ............................................................................................. 103 110 8.Summary and Policy Issues ........................................................................................ POVERTY MONITORING ................................................................................ 112 9. References .............................................................................................................. 116 ANNEX .......................................................................................................................... 121 TABLES Table 4.1:Per Capita Annual Production of DifferentCrops (kg/capita) ........................ 41 Table 4.2: Poverty Rates by Labor Status......................................................................... 47 Table 4.3: UnemploymentRates and Characteristicsofthe Unemployed....................... 48 Table 5.1:Structure of Expenditure by Sourceof FundsandSector (%)......................... Table 5.2: International Comparators: Sectoral Expenditure as Percentageof GDP .......54 55 Table 5.4: Monthly ExpenditureonPublic Primary Schools, 2001 (US Dollars)............61 Table 5.3: Education Spendingby Source FundsandProgram, FY2002 (%) .................56 63 Table 5.6: Repetition, Promotion andDropout Ratesby Grade (%)................................ Table 5.5: Gross andNet Enrollment Ratios .................................................................... 64 Table 5.7: Health Indicators for Timor-Leste ................................................................... 70 Table 5.8: Health Spending by Sourceof Funds, Program and Level of Service, FY2002, 72 Table 5.9: Reasonfor Outpatient Visit amongNon-poor andPoor ................................. (Yo)............................................................................................................................. 78 Table 6.1:Gender-AgeGroupsand Welfare (%) ............................................................. 85 Table 6.2: Female Headship andWelfare (%).................................................................. 87 88 Table 7.1: MillenniumDevelopment Goals inEast Asia ................................................. Table 6.3: Child Welfare andParental Living Status (YO)................................................ 98 102 Table 7.3: Simulations on ChangesinConsumption and Poverty (%)........................... Table 7.2: Poverty, Growth, andInequality - Scenarios, 2002-2007 ............................. 109 Table 7.4: FiveDimensions o f Public Action................................................................. 110 FIGURES Figure 2.1: Social Indicators 1999 and2001 .................................................................... 10 Figure2.2: ChangeinLiving StandardsBetween 1999and 2001 ................................... 11 Figure2.3: Economic andPower Status. 1999 and2001 ................................................. 12 Figure2.4: Change inCorruptionSince Violence in 1999............................................... 13 Figure2.5: Adequacy of Living Standards....................................................................... 14 Figure2.6: People's Priorities for the Future.................................................................... 15 18 Figure3.2: Poverty andHousehold Size........................................................................... Figure3.1:NationalPoverty Rates................................................................................... 19 Figure3.3: Poverty and Geography.................................................................................. 20 Figure 3.4: Poverty and Age of Household Head............................................................. 22 Figure3.5:Poverty andEducationLevel ofthe HouseholdHead................................... 22 23 Figure 3.7: Poverty and Land Size: Ruralversus Urban.................................................. Figure 3.6: Poverty and Employment Status of the Household Head .............................. 25 Figure3.8: Poverty andLivestock: Ruralversus Urban................................................... 26 28 Figure3.10: InequalityandGeography: The Gini Coefficient ......................................... Figure 3.9: Poverty and Infrastructure .............................................................................. 29 Figure3.11:Decomposition of Inequality........................................................................ 30 Figure4.1:Labor Force Participation Rates among the EconomicallyActive Population (15-64)....................................................................................................................... 35 36 Figure4.3: Output Per Worker. 2001 ............................................................................... Figure4.2: Employment Sectorby Gender ...................................................................... 36 Figure 4.4: Household Sources of Income inRuralAreas ............................................... 37 Figure 4.5: Daily Earnings inRuralAreas........................................................................ 38 Figure 4.6: Landholding per capita (ha) ........................................................................... Figure 4.7: Average Livestock Value Per Capita inRural Areas (US Dollars) ...............40 40 Figure 4.8: IrrigationinRural Areas................................................................................. . . 42 Figure4.9: Value of Per Capita Crop Sales andPoverty Rate ......................................... 43 Figure 4.10: Unemployment Rates................................................................................... 47 Figure 4.11: Urban Wage Rates........................................................................................ 50 Figure 5.1:Incidence of Public Spending inEducation ................................................... 58 Figure 5.2: School Participation Ratesby Age, 1998/99-2000/2001............................. 59 60 Figure 5.4: EverAttended School by Quintile and Age ................................................... Figure 5.3 : School Participation by Quintile andGender, 1999 and2001....................... 62 Figure 5.5: EnrollmentStatus of Children (School Year 2001/2002) .............................. 63 Figure 5.6: Reasons for Never Attending School, Ages 7-12........................................... 66 Figure 5.7: Language of Instruction inSchool ................................................................. 67 68 Figure 5.9: Children Under One Year Without Immunization......................................... Figure 5.8: Availability of Textbooks by Quintile............................................................ 71 73 Figure 5.11: Population Reporting Health Complaints inthe Last Month....................... Figure 5.10: Utilization o f Health Facilities..................................................................... Figure 5.12: Reasons for Not SeekingHealthCare Despite Having a Problem...............74 75 Figure 5.13: UtilizationRatesinthe Past 30 days............................................................ 76 Figure 5.14: Utilizationof Outpatient Health Services inLast Month............................. 77 Figure 5.15: Distributionof Type of Facility for Outpatient Care ................................... 77 Figure 5.16: One Way Travel Time to HealthFacility..................................................... 80 Figure 6.1:Household Food Security by Month.............................................................. Figure 5.17: Percentage Paying for Medical Services by Type of Outpatient Facility ....79 90 Figure 6.2: Poverty and InterviewDate............................................................................ 91 Figure 6.3 : Coping StrategieswhenNot Enough Food.................................................... 92 Figure 7.1: Languages....................................................................................................... 99 Figure 7.2: Poverty inEast Asia ..................................................................................... 100 BOXES Box 1.1: Timor-Leste's Vision for 2020............................................................................. 3 Box 1.2: Poverty Reduction Strategy: Empowerment........................................................ 5 Box 4.1: Poverty Reduction Strategy: Opportunities for Economic Participation...........33 Box 3.1:Constructing the Welfare Indicator.................................................................... 17 Box 4.2: Electricity........................................................................................................... 34 Box 4.3: Characteristics of FarmandNon-Farm Workers ............................................... 38 Box 4.4: How Large and Stable are RegionalDifferences?............................................. 44 Box 4.5: Jobs inDili/Baucau............................................................................................ 45 Box 4.6: Two Concepts o f Unemployment ...................................................................... 46 Box 5.1: Poverty Reduction Strategy: Improving Basic Social Service Delivery............53 Box 5.2: Who are the Out-of-School Children? ............................................................... 65 Box 6.1:Poverty Reductio11Strategy: Security ................................................................ 83 Box 7.1:MDGs-List of Goals, Targets andIndicators.................................................. 97 Box 7.2: Who are the Poor?............................................................................................ 104 Box 8.1: Poverty Data Sources....................................................................................... 113 PREFACE This report lays out the challenge of poverty reduction inTimor-Leste. It is based on the first nationally-representative household survey collected during August to December 2001. This work was conducted by the Poverty Assessment Project, a partnership between the Government o f Timor-Leste (with the Ministry o f Planning and Finance providing overall guidance), the World Bank, the Asian Development Bank (ADB), the Japanese International Cooperation Agency (JICA), the United Nations Development Program (UNDP), United Nations Children's Fund (UNICEF) and United Nations Mission o f Support in East Timor (UNMISET). The Poverty Assessment Project was launchedto provide up-to-date information on living conditions after the violence in 1999 as input into the National Development Plan. The Poverty Assessment Project comprised three data collection activities on different aspects o f living standards, which taken together, provide a comprehensive picture o f well-being in Timor-Leste on the eve o f independence: 0 Suco Survey This is a census o f all the 498 sucos inthe country and provides - an inventory o f existing social and physical infrastructure, and o f economic characteristics o f each suco, in addition to aldeia level population figures. It was completed between February and April 2001, and the report, written by the ADB, was publishedinOctober 2001. 0 Participatory Potential Assessment: This qualitative community survey assisted 48 aldeias to take stock o f their assets, skills and strengths, identify the main challenges and priorities and formulate strategies for tackling these withintheir communities. The field work took placebetweenNovember 2001 and January 2002. This activity was managed by UNDP and the report was finalized inMay 2002. 0 Household Survey: The Timor-Leste Living Standards Measurement Survey i s a nationally representative survey o f 1800 households from 100 sucos covering one percent o f the population. This comprehensive survey was designed to diagnose the extent, nature and causes o f poverty and analyze policy options for the country. Data collection was undertaken between end- August andNovember 2001. This report, written in two volumes, was a collaborative effort o f the members o f the Poverty Assessment Project, with the World Bank taking the lead in the analysis. The objectives o f this report are modest - to set a baseline for the new country on the extent, nature and dimensions o f poverty; to assist the decision making o f the newly elected government and its efforts in formulating, implementing and monitoring its Poverty Reduction Strategy. The objective was not to lay out the elements o f the poverty strategy but rather to present evidence on the basis of which the Timorese can define and refine their ownpoverty reduction strategy. We hope this isjust the start of a series of analysis to consider the effects o f government policies on different groups o f people, especially the poor. The preliminary analysis from the household survey was presented at a workshop inDili inFebruary 2002. The early results fed intothe NationalDevelopment Planpresentedby the Government at independence. Sector analysis for health, education and agriculture were also presented at the workshop and in more detailed discussions with the relevant Ministries. The full report was discussed with the Government inJanuary 2003. A series o f seminars was organized by the Ministry o f Planning and Finance during January 13- 24, 2003. The dissemination took place before the Ministries embarked on the prioritizing and sequencing of the National Development Plan for the FY2004 budget. Seminars were held at the Council of Ministers and several Ministries (Education, Health, Agriculture, Labor and Solidarity and Finance and Planning). A large workshop in Dili and three regional workshops in Baucau, Ainaro and Maliana were organized for Government officials from the center and districts, civil society representatives, including the Church, women's, students and youth groups, NGOs, Chefe de Sucos, and development partners. The results from the UNICEF sponsored Multiple Indicators Survey (MICS) were also presented by their staff andconsultants at these workshops, and at the Council o f Ministers and the Ministry o f Health seminars. The report was revised in light of the comments received and the health section was updated using the MICS results. .. 11 ACKNOWLEDGEMENTS This report is a result of a highly collaborative process betweenthe Government and our donor partners, ADB, JICA, UNDP, UNICEF and UNMISET. The Poverty Assessment Steering Committee chaired by Ms. Emilia Pires, Advisor, Ministry o f Planning and Finance (MoPF), provided overall guidance. We are very grateful to the Steering Committee members for their strategic guidance. The Steering Committee members included Emilia Pires, RobinBoumphrey (ADB ResidentRepresentative), Gwi Yeop Son (Deputy Resident Representative, UNDP), Sarah F. Cliffe (Chief of Mission, World Bank) and Mr. Takehara Masayoshi (JICA). Following the change of donor representativesinDili after Independence, the Steering Committee memberswere Meeja H a " (ADB Resident Representative), ShojiKatsuo, (Resident Representative, JICA), Haoliang Xu (Deputy Resident Representative, UNDP), Mr. Yoshi Uramoto (Special Representative, UNICEF) and Elisabeth Huybens (Country Manager, World Bank). We are very grateful to Emilia Pires for her leadership, constant support, and enthusiasm throughout this project. We would also like to express our thanla to Ms. Aicha Bassarewan, Vice Minister, MoPF, for her leadership during the dissemination of the Poverty Report. We are very thankful to the Statistics team (MoPF) for the great collaboration and partnership. The Statistics Office team did an outstandingjob in implementing the Suco Survey and the household survey under difficult conditions. The core team was led by Manuel Mendonca, Director of the National Statistics Office, and included Lourenco Soares (Data Manager), Elias dos Santos Ferreira (Field Manager) and Afonso Paixes (Field Manager). It was responsible for implementing the surveys, quality control and supervision, all o f which they managed with great skill. We are also grateful to the survey teams in charge of fielding the questionnaires. Their names are attached to this acknowledgement. Sonia Alexandrino from the Planning Office provided excellent logistical support in Dili, and David Brackfield, Advisor in the Statistics Office was always ready to lend a competent helping hand. The assistance from Gastao de Sousa and other staff of the Planning and External Management Assistance Division of the MoPF is gratefully acknowledged. The World Bank office inDili consistently provided outstanding support to us. Annette LeithandDianaIsaac always found a way to solve our problems, and the rest of the Dili team helped in innumerable ways, for which we are very thankful. The Timor-Leste Country team contributed greatly to the entire program of activities and we gratefully acknowledge contributions from Sofia Bettencourt, Gillian Brown, Lisa Campeau, Alfonso de Guzman, Adrian Fozzard, Dely Gapasin, Francis Ghesquiere, Ronald Isaacson, Natacha Meden, IanMorris, Janet Nassim and KinBing Wu. ... 111 The World Bank team comprised Benu Bidani, Kaspar Richter, Martin Cumpa, Juan Mufioz and Rodrigo Mufioz from Sistemas Integrales, Valerie Evans, David Madden, Kathleen Beegle, Paolo Nicolai and Wawan Setiawan. The Asian Development Bank team included Craig Sugden, Zacharias da Costa and Jessie B. Arnucu with Etienne van de Walle from the Manila office. The UNDP team included Antonio Assuncao, Jonathan Gilman, Janne Niemi, Sam Rao, Antonio Serra and Ian White. The JICA team included Charles Greenwald. The UNICEF MICS team included Yoshi Uramoto, Vathinee Jitjaturunt, Stemberg Vasconcelos, Raslied Mustafa, Peter Gardiner and Mayling Oey- Gardiner. This report was written by Benu Bidani and Kaspar Richter with superb overall assistance from Martin Cumpa. Background papers were written by Kin Bing Wu with inputs from Deon Filmer, Kathleen Beegle and Martin Cumpa on Education, Jean Foerster with analysis by Martin Cumpa on Agriculture, Janet Nassim with analysis by Martin Cumpa on Health, Kathleen Beegle and Martin Cumpa on Labor Markets, and by Kaspar Richter on the Welfare Profile, Disadvantaged Groups, and Food Security. Taranaki Mailei provided assistance with the task and the production o f the report. Walter Meza-Cuadra also helped in formatting the report. The peer reviewers were Pierella Paci and Lant Pritchett. This Report was prepared under the overall guidance o f Homi Kharas (Chief Economist and Sector Director, EASPR), Klaus Rohland (former Country Director), Xian Zhu (Country Director) and Tamar Manuelyan Atinc (Sector Manager, Poverty). The team greatly benefited from advice and guidance from Tamar Manuelyan Atinc. We are also very grateful to Sarah Cliffe (Chief o f Mission) and Elisabeth Huybens (Country Manager) for their consistent guidance and great support in the field and to Sanjay Dhar (Lead Economist) for his advice inheadquarters. We benefited greatly from the extensive comments received from the participants at the dissemination seminars, and the detailed written comments from the Ministry o f Health, Pierella Paci and Lant Pritchett (peer reviewers), Sofia Bettencourt, Elisabeth Huybens, ADB (Meeja Hamm and Craig Sugden), UNDP reviewers, Sam Rao, Caritas and Oxfam. The overall program of activities under the Poverty Assessment Project was funded jointly by the donor partners. The World Bank i s grateful to the Bank-Netherlands Partnership Program and the Norwegian Trust Fund for Environmentally and Socially Sustainable Development for financial support o f this project. Last but not least, our sincere gratitude goes to the people o f Timor-Leste who gave generously o ftheir time to help us collect the information on which this report i s based. iv Timor-LesteLivingStandardsSurvey TeamMembers 1 Altilis Moniz do Rosario Supervisor 2 Antoninho dos Santos Supervisor 3 Manuel da Silva Supervisor 4 Antonio Soares Supervisor 5 Felix Celestino da C. Silva Supervisor 6 Batista Leos Supervisor 7 De Francisco Barreto Supervisor 8 Tornas Gusmao Supervisor 9 Juliao da Cruz Enumerator 10 Samuel Fatima Enumerator 11 Armando Da Costa Enumerator 12 Gertudes de Amaral Enumerator 13 Henriqueta da Costa Braz Enumerator 14 Armando Martins Enumerator 15 Anibal Cardoso Enumerator 16 Miguel Pereira Enumerator 17 ManuelRibeiro Enumerator 18 Antonio C. Alves Enumerator 19 Raul Pinto Enumerator 20 Manuel Soares Pereira Enumerator 21 Rodolfo Soares Enumerator 22 Dilva do Rosario de F. da C. Enumerator 23 Januario Ximenes Enumerator 24 Amaro da C Tilman Enumerator 25 EvaFemandes Enumerator 26 DeliaNunes Enumerator 27 Doiniiiggos Moniz Enumerator 28 FiloinenaM.Guterres Enumerator 29 Julieta F. Silva Enumerator 30 Jaimito do Rego Enumerator 31 Gil Vicente Madeira Enumerator 32 Rogerio Castro Enumerator 33 Sebastiao Dias Saldanha Enumerator 34 Saozina da Costa Enumerator 35 Maria L.De Jesus Enumerator V Timor-Leste Living Standards Survey Team Members (contd.) 36 Antonio B.S. Dasilva Enumerator 37 Rogerio Babo Data Entry 37 InaciaVilena Data Entry 38 Silvina Suares Data Entry 39 Maria Odete Data Entry 40 Suzana Leong da Costa Data Entry 41 Eduardo Martinho Ximenes Data Entry 42 Maria Odete Baros Administration 43 Vicente Lean de Jesus Field Administration 44 Saul do Carmo Ximcncs FieldAdministration 45 Kintao de Deus FieldAdministration 46 Tomas Pereirra FieldAdministration 47 Rafael C. Lobato FieldAdministration 48 NicolauPereira FieldAdministration 49 Manuel da Costa Silva FieldAdministration 50 Alvaro Maia FieldAdministration vi EXECUTIVE SUMMARY 1. Timor-Leste became the first new nation o f this Timor-Leste emerges millennium on May 20, 2002 following a quarter century o f from a legacy of occupation and conflict. The country experienced a violence as theJirst new fundamental social and economic upheaval after its people nation of this voted for independence from Indonesia in a referendum in millennium. August 1999. The bulk o f the population was displaced inthe weeks following the ballot results and most o f the physical infrastructure was destroyed or rendered inoperable. Soon after the violence ceased, it began rebuilding itself with the support from UN agencies, the international donor community and NGOs. 2. Timor-Leste has achieved enormous progress in rehabilitating its economy, reconstructing its infrastructure, Itfaces the challenges of reintegrating its refugees and building the key elements o f a nation-building and sustainable political process in an environment o f internal poverty reduction. peace. Itnow faces many challenges o f nation-building and o f overcoming the deprivations affecting the lives o f the poor. On the eve o f independence, the government presented its vision for the year 2020 and its strategy for achieving this vision inits National Development Plan. The National Development Plan lays out a strategy for the next five years (2002-2007) with two overriding objectives: To reduce poverty in all sectors and regions o f the nation; and To promote economic growth that is equitable and sustainable, improving the health, education and well- beingo f everyone. Report sets a baselinefor 3. The Government's Poverty Reduction Strategy has four the implementation and main elements: (i) promoting opportunities for the poor; (ii) monitoring of the improving their access to basic social services; (iii) enhancing Government's Poverty security, including reducing vulnerability to shocks, and Reduction Strategy. improving food security (iv) and empowering the poor. The main objective o f this report is to support the Government's efforts in implementing and monitoring its NDP. It sets a baseline o f the pattern, extent, andnature o f poverty, which can be used to translate the broad elements o f the poverty reduction strategy into prioritized action plans consistent with the mediumterm expenditure framework. Welfare Profile Poverty, affecting two in 4. Poverty i s a complex phenomenon involving multiple fivepersons, is deprivations. We use an economic definition o f poverty, in predominantly rural, and which an individual is deemed poor if she i s unable to attain a minimal standard o f living. Based on this definition, two in higher in the than the East. five people in Timor-Leste are poor. Economic well-being varies across the country. Urban areas, especially DWBaucau, are better off than rural areas. While one in seven are poor in DiWBaucau, over four inten are inrural areas. And, poverty is concentrated in rural areas. Three quarters o f the population lives invillages but six inseven o f the poor, or 280,000 people, reside there. Poverty also increases from East to West. The three western districts (Oecussi, Bobonaro and Covalima) are home to one fifth o f the population but account for a quarter o f the poor. In contrast, the three eastern districts (Baucau, Lautem and Viqueque) account for a quarter o f the population butlessthan a fiftho fthe poor. 5. More human capital through better education leads to Education lowers lower poverty. For example, close to one in two persons are poverty. poor in households where the household head has not completed primary education. This compares to less than one in seven where the head has at least senior secondary education. Demographics matters too - larger households and families with a higher share o f children and elderly are poorer. 6. Assets are an insurance against economic uncertainty Poverty decreases with and a way for preparing for future expenses. They are held larger land size and largerlivestock primarily in the form o f housing, land and livestock. The incidence o f poverty decreases with larger land size, both in urban and rural areas. For families in villages and cities more livestock i s associated with less poverty. Urban-rural divide 7. Secure access to infrastructure services, ranging from exists in access to safe water, sanitation and electricity, i s essential for escaping infrastructure. poverty. Nationwide three in four persons live without electricity, three in five persons without safe sanitation and every other person without safe drinkingwater. There is a vast 11 * . urban-rural divide. In urban areas, 70 percent have access to each o f these services. In rural areas, the shortfall is 25 percentage points for drinking water, 37 percent for sanitation and 61 percent for electrification. Persons without infrastructure are in general poorer than those with access to infrastructure. For example, while only one in seven urban dwellers with electricity are poor, almost one in two without electricity live below the poverty line. 8. Inequality as measured by the Gini coefficient is 37 and it is higher in cities than in villages. Accounting for geography, gender, age and education o f the household head explains at most one-third o f overall inequality. There i s a need to better understand the determinants o f inequality. 9. It is also remarkable that despite the tragic events of In 2001, the population 1999, people's overall subjective assessment o f how their lives felt vastly more had changed in late 2001 was positive. In2001, the population empowered compared to felt substantially more empowered compared to Indonesian Indonesian times, and the times and economic well-being had improved primarily for the majority of the least well- bottom-third o f the population. People believed themselves to off had higher economic be powerless in 1999, with six inten placing themselves on the status. lowest category o f a nine-step ladder, and almost no one ranking themselves on the top four steps. In2001, the situation was substantially different. Only one in twenty people believe they are on the lowest step, and close to three in ten ranked themselves on the top five steps. In terms o f economic status the vast majority considered themselves as disadvantaged in 1999, with two thirds locating themselves on the two lowest steps. In2001, more thanhalf o f the least well-off improved by one or two categories, boosting the shares at the second and third lowest steps. Opportunities 10. The National Development Plan focuses on creating an enabling environment to generate opportunities for the Enabling environment economic participation o f the poor. The main elements are for the economic improving productivity in agriculture and the informal sector; participation of thepoor providing an enabling environment for private sector is essential. development; provision o f infrastructure and pro-poor public expenditure policies. Ensuring a conducive environment for private enterprises, which includes business regulations, land and property legislation, trade policy, labor legislation, and infrastructure issues, especially related to power and transport, 111 ... i s essential for employment generation. Improved security and job creation are urgent priorities on the Government's agenda. 11. In rural areas, agriculture is the dominant employer, Diversijkation out of accounting for four-fifths o f all employment. But it only agriculture into non- contributes one quarter o f non-oil GDP due to low output per farm activities, and worker. A key driver for higher rural living standards is improved crop mix, are diversification out o f agriculture into non-farm activities. key driversfor higher Among agricultural households, incomes are linkedto the level rural living standards. o f assets. Poor rural households have half the land per capita, half the value o f livestock assets, and less education than the non-poor. Higher land per capita translates into greater production o f crops per capita for the non-poor than the poor. The non-poor are more likely to produce higher paying crops (such as coffee, fruits and vegetables) and produce significantly higher amounts of these crops and o f the main staples (rice, maize and cassava). 12. Low productivity, overall, and especially among the Low productivity in poor, i s linked to the limiteduse o f key complementary inputs, agriculture is linked to the quality o f land, and access to markets and other low use of inputs, infrastructure. For example, while irrigation is limited overall, quality of land, and the non-poor have larger irrigated land holdings per capita access to markets. (0.13 ha) than the poor (0.04 ha). Among agricultural households, only 3 percent use fertilizers, manure or pesticides, and almost all o f the usage i s among the non-poor. The lack o f availability i s reported as the primary reason inlate 2001. 13. Urban labor markets are characterized by both high Major urban areas are wages and high unemployment. A fifth o f the working age marked by coexistence population inDiWBaucau i s unemployed. Females have higher of high unemployment unemployment rates than males. Unemployment declines and high wages. sharplywith age: the unemployment rate among the youth (15- 24) i s 43 percent, and it drops to 17 percent for the 25-34 year olds and nine percent for the over 35 year olds. Thejobless are poorer than those in employment. The bulk o f the unemployed inDiliand Baucauare young educated males. Inspite ofhigh unemployment, wages are high. The wages for workers in Timor-Leste are two to three times higher than wages in Indonesia. The influx o f expatriates during the transition to independence fuelled a service sector boom inthe major urban centers, with attendant high real wage levels, urban concentration, and misallocated investment in the service industry. There is a difficult adjustment ahead as declining demand and high unemployment rates are expected to put downward pressure on urban wages. As painful as this transition may be, this correction needs to take place for the iv benefit o f long term growth prospects o f the economy. Evidence froin 2003 shows that private sector wages are falling. As the service sector starts downsizing, it will be crucial to provide a conducive private sector environment for sustainable job creation. Improved Basic Services 14. The National Development Plan gives priority to the Government budget delivery o f basic social services, especially primary and reflects emphasis on secondary education and primary health care, including social sectors. preventive programs such as immunization and public health. An important vehicle for achieving national development goals i s government spending. Public spending in the post- independence period broadly supports service delivery functions, with education accounting for one quarter of government spending, and health for another 10 percent in 2002. Government spending (CFET) in 2002 reached US$19 per capita in education, and US$7 per capita in health. These levels are substantially higher than most low income countries, but lower than middle income countries. Total spending, which includes donor and bilateral support, was US$58 per capita in education and US$32 per capita in health. These high overall costs reflect capital expenditures to set up education and health systems. These levels are higher than middle income countries but are closer to the levels of spending in some post-conflict countries immediately following the conflict. 15. The main issues lie in the allocation o f these funds for Intra-sectoral resource priority services that reach the poor and in ensuring allocation and sustainability inthe future. Both healthand education allocate a sustainability are main large share to tertiary services. In health, half o f CFET concerns. spending in FY2002 goes towards tertiary care. This has declined to 41 percent, more in line with the health policy objective o f 35-40 percent. Since poor households are more likely to visit primary health facilities than public hospitals, public hospital spending is regressive, benefiting the richer households. In education, only primary education i s progressive and receives about half o f the CFET education spending, whereas overall education spending i s regressive. As donor support phases out and operating costs shift towards the budget, ensuring the sustainability o f social expenditures will be important. This will likely involve cost recovery for tertiary education and health services for the non-poor. V 16. School reconstruction In education, impressive accomplishments to date are the rebuilding o f the school system and the sharp increase in and increased enrollment rates, especially for the poor, girls and rural enrollments are main children, partly due to the reduction in the cost o f schooling. achievements. Net enrollment rates increased from 65 percent in 1998/99 to 75 percent in2000/2001. The increase for girls was larger than for boys. 17. Nevertheless, education still faces several challenges. Developing apolicy The education sector i s only now developing a policy framework, large framework for the sector to guide its decisions in the school agepopulation, implementation o f the NDP. This policy needs to deal with high adult illiteracy, three issues. First, the school-age population is large and low internal efJiciency growing, and illiteracy high, with over seven in ten persons and low primary school over the age o f 30 never having attended school. Second, the enrollment are main internal efficiency o f the education system i s low, with a vast challenges. number of overage children in the school system, which is manifested in the large divergence o f the net (75 percent) and gross (113 percent) primary enrollment ratios. The repetition (20 to 30 percent) and drop out rates (10 percent) in primary school are high. At the current level o f internal efficiency, only two thirds o f those enrolled into first grade would reach Grade 4, and only one half would complete grade six. The cost per student for six years ofprimary schooling is about US$300, but the cost per graduate is about twice as highbecause o fthe high repetition and drop out rates. Finally, despite the increase, school enrollment remains low, with a quarter o f 6-18 year olds having never attended school. Improving education outcomes i s linked to both demand and supply side factors (availability o f qualified teachers and other inputs, language of instruction). The education sector faces a challenge in formulating and implementing its strategy, in prioritizing actions and costing them to achieve objectives in education within the medium term expenditure framework. In preparing education action plans, tradeoffs between expanding access, increasing quality and providing free schooling will need to be taken into account. Low health service 18. Health outcomes in Timor-Leste are among the lowest utilizationprevails with in East Asia. Immunization is one o f the most cost-effective, large rural-urban gaps equitable health interventions available. Once immunized, in access and choice of every child, rich or poor, is equally protected for life. Yet, only provider. one in ten children under the age o f 12 months received full DPT vaccination in 2001, and a year later this share was still no more than one in five. Health utilization rates are also low, but not because o f lack of need. Fewer than one inten people seek health care when sick. Many stay away from health Vi facilities because they are located only at far distances, especially in rural areas. There i s a large urban-rural divide in the choice o f health facility. Community health centers are the main provider o f health services inrural areas, reaching half o f the population. Public hospitals and private facilities play an important role in cities, with over two-fifths using private or church facilities for outpatient care. Outpatient care does not come for free. A third o f the population pays for transportation and medical services. On average an individual visiting a health facility spendsjust under US$2 per month for outpatient services and medications. The poor pay half the amount paid by the non-poor, but this represents a higher share o f expenditures for them. 19. Providing affordable, accessible health services Equitable provision of especially for the rural poor will be important. Another basic health services is challenge i s to constrain the use o f hospitals for services more main priority. appropriately offered in health centers, and to resist the demands o f urban populations for more hospital resources at the expense o f primary health care inrural areas. The Ministry o f Health developed a sector-wide policy framework from the outset and embedded the reconstruction effort within it. As a result, despite a slow start, it is well placed to achieve improved service delivery. Household Security Vulnerability is an 20. The National Development Plan outlines the main important dimension of elements o f a social safety net for the vulnerable. Key areas o f poverty. concern include: disadvantaged groups, such as widows and orphans o f the resistance; improving food security for households; and improving security o f livelihood caused by the lack o f recognition o f ownership and tenancy o f agricultural land, or lack o f access to forests or other community lands. This part focuses on two o f these issues: disadvantaged groups and food insecurity. 21. The analysis on disadvantaged groups focuses on Female-headed gender and parentless children. Male-headed households are households and consistently better off than female-headed households interms children without a of education, health and subjective well-being, but not so based father or mother are on consumption poverty - but we lack information on intra- more deprived. household distribution. For example, while one intwo children under 6 are immunized against measles in male-headed households, less than two in five are in female-headed vii households. Fatherless children experience higher poverty, with poverty being 6 percent higher than for children with living fathers. And parentless children (without a father or mother) have worse education and health indicators than those livingwithbothparents. Widespreadfood 22. Subjective assessments o f food adequacy suggest that shortages during food insecurity i s widespread. Close to nine in ten persons experience inadequate food provision at some point during the November to February' between rice and maize year. Food security is closely tied to not having enough rice harvests, are linked to and maize. Food availability i s aligned with the harvest cycle at higher poverty. the national and regional level. Food shortages are highest duringNovember to February,at the end ofthe rice harvestand before the maize harvest. Food insecurity during lean seasons i s also associated with higher poverty. The major urban centers typically have access to enough food all throughout the year, while other parts o f the country face greater fluctuation infood availability. Households have multiple ways o f dealing with food insecurity, which may lower vulnerability in the short term at the expense o f higher vulnerability over the longer term. Almost all households either change their diet or skip meals when faced with insufficient food. The striking result i s that children appear to take the brunt o f this adjustment. Ifthe situation required further adjustment, then households would undertake distress sales o f livestock and other farm assets. 23. These results point to the need to develop a policy response to deal with group-specific and seasonal vulnerability. It should be aimed at helpingpoor people manage risk better by reducing and mitigating risk and lessening the impact of shocks. Possible interventions range from support to traditional community structures, to targeted support for schooling and health care, and to improving access to productive resources and remunerative employment. The Development Challenge 24. Despite the progress since 1999, Timor-Leste faces Timor-Leste is among the daunting social challenges and remains one o f the least poorest countries in East developed countries in East Asia. Poverty is high and human Asia. and physical capital are depleted. One in five persons live on less than the international poverty line o f US$1-a-day. At US$1-a-day, Timor-Leste is the fourth poorest in East Asia with only Lao PDR, Cambodia and PNG showing greater deprivation. Life expectancy i s only 57 years. Timor-Leste is V l l l ... not just a new nation but one o f young people, with one intwo Timorese below the age o f 15. This nation o f over 800,000 people will grow rapidly as large young cohorts move through the reproductive ages, creating pressures on basic social services to ensure a healthy and productive life o f an expanding population and to generatejobs for the economically active. 25. While the social agenda is challenging, Timor-Leste has Despite a lack of overall the solid prospects o f future flows from the country's natural growthpoverty is resource wealth and can benefit from the support o f the projected to decline international donor community. Economic growth i s a modestly by 2007 due to necessary condition for poverty reduction. In Timor-Leste, expansion in agriculture. about one seventh o f the population lives within 10 percent o f the poverty line, suggesting that poverty is responsive to growth. Illustrative projections show that the impact o f economic growth on poverty depends crucially on agriculture, broad-based participation o f the population inthe opportunities o f an expanding economy, and modest population growth. The first scenario i s based on the growth rates o f the NDP with 2 percent average annual GDP growth over the plan period (2002-2007) and agriculture growing at close to 6 percent. Given current population growth rates, this translates into an average per capita growth rate just below zero over the plan period. In spite o f this contraction in per capita GDP, due to strong agricultural growth, the poverty rate at the national poverty line drops from 39.7 percent in2001 to 29.5 percent in 2007, about five percent below the Millennium Development Goal (MDG) target o f halving poverty within 25 years. Given past agricultural growth rates, and cross-country experience, it may be unlikely for agriculture to grow so strongly over the five year period. Simulating slower economic growth, through lower agricultural growth o f 3 percent over the plan period, would endanger progress towards achieving this goal, as would a widening in income inequality in agriculture. In these scenarios, poverty would stay above or just reach the MDG target. Finally, strong population growth can hamper progress inthe reduction of the absolute number of poor. Assuming the population expands by 3.2 percent annually rather than 2.4 percent as in the baseline scenario, an additional 46,000 persons will live below the poverty line by 2007. Enhancing human capital,promoting non- 26. Using a statistical model we highlight some o f the key farm activities, determinants of pro-poor growth. These results are merely encouragingproduction illustrative and should be interpreted with caution as this of high value crops and approach suffers from many limitations. Notwithstanding the expanding basic caveats, the results confirm some important messages o f the infrastructure lower report. Boosting male and female human capital, promoting poverty. i x lion-farm activities, encouraging the production o f high-value crops, developing services like irrigation, expanding sanitation and electricity infrastructure, creating a favorable business environment for private employers and improving market networks raise per capita incomes and lower poverty. Allocating aid and of- 27. The National Development Plan emphasizes that shore wealth towards allocating aid and off-shore wealth towards high priority highpriority development objectives will be critical. Public expenditure developmentobjectives is decisions should be driven by policy priorities on poverty, and critical. policy choices in turn have to be disciplined by resource and implementation realities over the medium term. A medium term expenditure framework can help linkpolicy priorities with the resource envelope. Poverty Monitoring A poverty monitoring 28. Timor-Leste has many varied data sources that present plan is critical to assess a coherent picture o f poverty and provide a baseline for progress on monitoring progress in poverty reduction according to the implementing the NDP. objectives o f the National Development Plan. The key challenge lies now in formulating a poverty monitoring plan that includes both quantitative and paGicipatory elemenisyeand lays out the institutional arrangements for data analysis and reporting to ensure that the collected data inform policy malting and program design. X Timor-Leste at a Glance: Social Statistics 1999 and 2001 11X? I! DPT(xof childrcn Measles ('K or c h i d m Colitraceptlve pre\alence under 12 months) wdcr 12liloiitlis) ('M. inarriednomen apes 15. 49) source 21101TLSS Feb I999 Fall 20111 Source I999 SUSENAS atid2001 TLSS 1 INFRASTRUCTURE HOUSING: DAMAGEI N 1999VIOLENCE AND REHABILITATION 10 Change insubjectivewell-being two years after the violence in 1999 2.11 H o w has life changed since 1999 inthe people's own assessment? Inthe survey, the people were asked to assess the changes since before the violence in 1999 along different dimensions: living standards, economic status, and power status. Living standards are closely linked to economic conditions, and, as evidenced in the previous section, they have remained difficult since the violence. In Figure 2.2, we show the responses o f all individuals aged 15 years or older when asked about the change inliving standards at the end o f 2001 compared to 1999. About three inten persons believe living standards have deteriorated, compared to only one in ten persons saying they have improved. This points to substantial material hardshipduringthe transition process. Figure 2.2: Change in Living Standards Between 1999 and 2001 I I Improveine~it Same Deterioratioii Source: 2001 TLSS 2.12 Living standards are important for both economic status and empowerment. Figure 2.3 displays the responses to "ladder questions", where persons are asked to rank themselves with regard to economic and power status, in 2001 and before the violence. Let us consider the economic dimension first. Looking back to before the violence in 1999, the vast majority view themselves as poor: one third o f the respondents believe they were on the lowest step, another third on the second lowest step, and another 30 percent between the third to fifth lowest steps. Less than two percent ranked themselves on the top four steps. By comparison, in 2001, the situation improved, especially for the lowest third. The share at the lowest step has significantly decreased, boosting the shares o f the second and third lowest steps, with the rest remaining unchanged. Overall, the lowest two thirds o f the respondents believe that their economic situation has improved or remained unchanged, while the situation for the highest thirdhas remained unchanged. 11 Figure2.3: Economicand Power Status, 1999 and 2001 ECONOMIC STATUS POWERSTATUS I n 1 Lawest 2nd 3rd 4th 5th 4th 7th 8th Highest source 20111 TLSS 1 2.13 The questions regarding power status reveal a clear picture. Today's population viewed themselves as powerless in 1999, with six inten placing themselves on the lowest step, and another two inten on the second lowest step. Essentially nobody ranked herself on the top four steps. The situation in2001 i s substantially different. Only one intwenty people believe they are completely powerless, and close to three inten believe they rank on the top five steps. These numbers suggest that, while the economic situation has improved primarily at the bottom tail, the advances inpower status have affected almost the entire population. 2.14 Corruption is a core poverty issue. For example, the World Bank's Voices o f the Poor recorded reports by poor people o f hundreds o f incidents o f corruption as they attempt to seek health care, educate their children, claim social assistance, get paid, access justice or police protection, and seek to enter the marketplace. In their dealings with officials, poor men and women are subject to insults, rudeness, harassment, and sometimes assault by officials. Harassment o f vendors in urban areas i s widespread. Politicians, state officials, and public servants are rarely viewed as effective, trustworthy, or participatory. Corruption also matters for the broader performance o f a country. It is an obstacle to economic and social development. It distorts the rule o f law and weakens the institutional foundation on which economic growth depends. These harmful effects are especially severe on the poor, who suffer most from economic decline, are most reliant on the provision o f public services, and are least capable o f paying the extra costs associated with bribery, fraud, andthe misappropriationo f economic privileges 2.15 People's perception on the change in corruption since 1999 i s shown in Figure 2.4. Overall, people feel corruption is less o f an issue in 2001 than in 1999. Only one fifth of the population aged 15 years or older believes corruption has worsened since the violence, compared to two fifths who feel corruption has declined. Across the board o f geographic and age-gender categories, more people believe corruption i s less prevalent in 2001. However, there are important differences. Most strikingly, in major urban centers, three in ten people feel corruption has become worse. Inrural areas, the issue appears to be larger in the west and center than in the east, and inrural mid- and highlands than in 12 rural lowlands. With regard to gender, men are more pessimistic than women about the progress made in corruption prevention, as are persons younger than 50 years o f age compared to those older than 50 years o f age. One possible explanation o f this pattern could be involvement in commercial and administrative tasks. Inhabitants of Dili and Baucau, and prime-age men are likely to be more exposed to such activities. Interestingly, the more optimistic view on change incorruption inthe rural east and rural lowland coincides with lower poverty than inthe other rural domains. Figure 2.4: Change in Corruption Since Violence in 1999 100 80 .-0c Y d d 60 0 a b 0 % a 8E 40 20 0 Major Other RuralWest Rural Center RuralEast UrbanCenters UrbanCenters I 1 I Source: 2001 TLSS More Same Less Subjective well-being: dimensions and priorities 2.16 What are the most pressing concerns o f the population in late 2001? A standard tool to assess subjective well-being are "adequacy" questions covering the different categories of family needs. InFigure 2.5, we display the answers of heads of households to questions regarding their family requirements. It shows the percentage shares o f each o f the three possible answers (less than adequate, just adequate, and more than adequate) along the dimensions o f food, shelter, clothing, health care, education, and income. 2.17 The striking feature i s one of widespread inadequacy and severe hardship o f everyday life. Whatever specific aspect o f living standards we consider, 99 in 100 people feel at best just adequately endowed, and between over one thirdto three quarters believe to be less than adequately covered. The concern is largest for clothing, followed by food, children's education, and housing, and least for the provision of health care. Inaddition, 13 more than three in four persons live in households where total income i s deemed inadequate'. Figure2.5: Adequacy of Living Standards 100 0 Food Housing Clothhg Health Education Income Care 0Less than adequate Just adequate Source: 2001TLSS. 2.18 In view of these substantial inadequacies in living standards, what are the personal and national priorities looking forward? TLSS asked individuals aged 15 or older to give first and second priorities from both the personal and national perspectives. The results are shown inFigure 2.6. 2.19 Top o f the list o f personal concerns are economic and social factors. Number one is employment, quoted by three fifths o f the interviewees. This is followed by improvements in social services (education, health care, and housing), and demand for products. In contrast, the main achievements o f the past two years (safety, political participation, and status incommunity) rank lowest interms o f importance for individual living standards for the future. The priorities for Timor-Leste's living standards are broadly in line with individual preferences. The bottom three categories are exactly the same, and the same three categories appear inthe top three, even iftheir internal ranking i s reversed. The most striking difference is the emphasis on education as key to national prosperity, listed by seven in ten individuals, compared to only three inten for personal preferences. Employment, housing, and demand for products are listed by fewer people as national priorities than as individual priorities. Overall, this suggests that the immediate individual economic concerns are viewed as less important for the national agenda. In both personal and national rankings, economic and social concerns dominate aspects linked to empowerment, perhaps a reflection o f the achievement inthis area over the past few years. It is not clear whether respondents viewed total income as a summary measure capturing other dimensions, or a separate dimension o f living standards itself. 14 Figure 2.6: People's Priorities for the Future NATIONAL PERSONAL 80 , 1 ~ 611 50 411 311 211 ill !I SUMMARY 2.20 Overall, at the time o f Timor-Leste's independence, the population feels more empowered compared to Indonesiantimes, but less secure about its economic well-being. When asked about their economic situation in end 2001 compared to before the violence in 1999, slightly more people believe their economic situation has improved than deteriorated, but the bulk feels little has changed. By contrast, seven in eight persons believed they had more power in 2001 than before the violence in 1999. The people's assessments confirm that progress has been achieved in safety, political participation, education and status in community, whereas economic factors like housing, demand for products, employment and infrastructure have worsened and remain priorities for the future. 15 3. WELFARE PROFILE 3.1 H o w do the poor differ from the non-poor? In this chapter, we investigate the characteristics o f the poor. The poverty profile includes information on where the poor live, what they do, how they earn a living, and what their living standards are interms o f health, education and housing. We also look at the distribution o f economic resources overall. This analysis i s important for two reasons. It provides insights on the characteristics o f the poor for the design o f poverty-reduction programs, and highlights the link o f poverty to other dimensions o f well-being. lo METHODOLOGY 3.2 Poverty i s a complex phenomenon involving multiple dimensions o f deprivation. It can mean lack o f access to resources and opportunities, poor health, malnutrition, illiteracy, lack o f safe drinking water and sanitation, deprivation o f basic rights and security and powerlessness. While these deprivations often go hand-in-hand, the correlations between these different dimensions o f poverty are far from perfect. The most commonly defined concept o f poverty i s an economic one, inwhich an individual is deemed poor ifhe is unable to attain a minimal standard o f living. Multiple decisions are involved inderiving a summary measure o f living standards (see Box 3.1 for a summary). There is a broad consensus that consumption i s a preferable indicator o f living standards than income. Following common practice in poverty analysis, the nominal consumption measure i s converted to real consumption to adjust for cost-of-living differences across regions, and to account for differences ininterview date. We follow standard practice and use per capita total household expenditure as basic welfare indicator and assume that households allocate resources equally among their members' '. 3.3 The poverty line i s the minimal standard o f living that an individual should attain so as not to be considered poor. Setting poverty lines i s often the hardest and most contentious step in constructing a poverty profile. Following common practice in East Asia, we have defined a poverty line as the minimum expenditure needed to purchase a food basket that provides 2100 calories per person per day and includes an allowance for non-food consumption needs (such as clothing and housing). The poverty line estimated for Timor-Leste i s US$15.44 per capita per month, or just over fifty cents per day12, or ''This lo chapter draws on Chapter 1 and Chapter 2, Volume 11. Throughout the report, we analyze to what extent the conclusions depend on adjustments made for differences inhousehold size and composition. 12 Most o fthe monetary values in the survey were reported inRupiah, since it was the dominant currency in use at the time of the survey. All Rupiah values inthe survey have been converted to U S Dollars usingan exchange rate o f 10,000 RupiahiUS Dollar, the approximate average exchange rate prevailing during the survey period. 16 Box 3.1: Constructingthe Welfare Indicator Income or Consumption? Income, together with assets, measures the potential claims o f a person or household, while consumption captures the level o f living interms o f what living standard individuals actually acquire. The main reason for preferring current consumption to income as an indicator of living standards is variability (Ravallion, 1994). In a mostly agricultural economy people receive income only infiequently and the amounts differ across seasons. Empirical evidence suggests that households in low income agricultural societies manage to smooth consumption in spite o f highly volatile income receipts (Deaton, 1997). Thus, consumption will most likely be a better indicator of current consumption than current income; and current consumption may also be a better indicator of longer term welfare, since it reveals information about incomes at other points intime. Per capita or equivalence scales? Households differ insize and composition. Inparticular, the needs of household members differ, particularly between adults and children. One option is to use a system of weights, whereby for example, children count as a fraction o f an adult in terms o f needs, and convert all households into the number o f equivalent adults. But there also exist economies o f scale in consumption. Some non-food items (for example, housing) have public goods characteristics, as their usage by one member o f the household does not reduce their value to other household members. Thus, because people can share goods and services without reducing their welfare, the cost o f attaining a given level o f welfare may be lower in larger households than in smaller households. Simply deflating household consumption by household size ignores these economies o f scale in consumption. The number o f equivalent adults can be adjusted for economies o f scale to get the number o f "effective" equivalent adults. We follow standard practice and use per capita total household expenditure as the basic welfare indicator and assume that households allocate resources equally among their members. For the poverty profile, it is important to conduct a sensitivity analysis to see to what extent the broad conclusions depend on assuinptions regarding equivalent scales. Cost of living differences: Prices o f goods and services vary considerably across different regions and this spatial variation in prices should be taken into account when comparing welfare levels across different parts of the country. In Timor-Leste, transportation is difficult and expensive, and local markets are not well connected, giving rise to possibly large variations in the cost o f living. Inorder to construct a price index to convert nominal consumption into real consumption, we follow standard practice in East Asia and use a Laspeyres Index based on a fixed consumption bundle. The price index i s constructed for five different parts o f the country: DiliiBaucau, other urban areas, the rural east, rural west and rural center. The Laspeyres price index for each region is computed by comparing the cost of buying a reference bundle in that region compared to a reference region. To represent the consumption pattern o f the poor, we take as the reference bundle the consumption basket of the 2'ldto 5" deciles according to per capita consumption. We test the sensitivity o f our poverty estimates to the choice of this index and findthat the results are remarkably robust. 1 2 0 100 0 80 -a -8 $ 0 6 0 3P 2 0 4 0 0 20 0 00 DiIilBaucau Other urban Rural center Rural east Rural west 17 US$1.5 in international dollars using the Purchasing Power Parityl3 adjusted exchange rates. POVERTYPROFILE Poverty Incidence 3.4 The incidence of poverty (or the headcount index) in the country as a whole is 39.7 percent, amounting to 329,000 individuals (see Figure 3.1). In other words, two in five individuals inTimor-Leste are not able to cover the food and noli-food consumption requirements. The poverty gap does not just count the poor, but measures their average consumption shortfall relative to the poverty line. It equals to 11.9 percent. This measure can be usedto calculate the minimumincome transfer neededto bringall of the poor just up to the poverty line assuming that the transfer is both perfectly targeted and fully consumed. This number totals to US$1.84 per person per month, or US$18.28 million overall per year. The severity measure of poverty, which incorporates inequality among the poor, by giving higher weights to the poverty gaps o f the poorest, equals 4.9 percent. Due to its sensitivity to the distribution among the poor, the severity measure reveals differences across population groups that are veiled by the other two poverty measures. Figure 3.1: National PovertyRates 45' Headcount Poverty Gap Severity Source: 2001 TLSS. l3 The Purchasing Power Parity (PPP) rates allow a standard comparison o f real price levels between countries, just as conventional price indexes allow comparison o f real values over time. The PPPs are generally derived form price surveys done by the International Comparison Program, ajoint program of the World Bank and UNagencies. Tiinor-Leste does not have a PPP yet, but the PPP has beencalculated using an altemate methodology which calibrates the price per calorie in Timor-Leste to Indonesia. A similar method has beenusedfor other East Asian countries with no PPPs (such as Vietnam). 18 3.5 What is the sensitivity o f poverty to assumptions regarding household size and composition? Using the per capita measure, poverty increases with household size (see Figure 3.2). In most cases, an increase in household size implies more children and elderly. Larger households with more non-earning dependents, such as children and the elderly, are less able to feed and clothe all householdmembers. Figure 3.2: Poverty and Household Size I 50 40 h 5 4- 2 30 -2 u 20 10 0 1 2 3 4 5 6 I 8 9 1Oplus HouseholdMembers (#) Source: 2001 TLSS. Geography 3.6 National poverty rates hide a remarkable variation across the country. Poverty in Timor-Leste increases from East to West. The three districts (Oecussi, Bobonaro and Covalima) that comprise the west are home to one fiftho f the population, but account for a quarter o f the poor. In contrast, the three districts o f the East (Baucau, Lautem and Viqueque) account for a quarter o f the population, but less than one fifth o f the poor. Poverty also rises with altitude above sea level, whereas coastal and landlocked sucos experience similar poverty. 3.7 These geographical patterns are partly, but not entirely, a reflection o f the degree o f urbanization. In line with experience in other developing countries, poverty in rural areas is higher than inurban areas (see Figure 3.3). Since three quarters of the population reside in rural areas, poverty i s overwhelmingly a rural phenomenon: six in seven of the poor live inrural areas, amounting to 280,000 persons. 19 Figure3.3: Poverty and Geography URBAK AND RURAL POVERTY 40 10l--+ Urban Rural Souice 2001 TLSS POVERTY AND REGIONS Major Urban Other Uiban Rural West Rural Center Rural East Soiirce 2001 TLSS POVERTY, ALTITUDE AND SEA ACCESS Flat Middle High Inland Seaside Source 2001 TLSS. 20 3.8 The investigation o f the sensitivity o f poverty rankings with regard to equivalence scales and poverty lines leads to these conclusions: 0 Ruralareasare substantially poorer thanurbanareas; 0 Other Urban Centers are substantially poorer than Dili and Baucau, the two major urban cities; 0 East i s the least poor, while the ranking between Center and West is ambiguous - both nationwide and inrural areas only; 0 Inrural areas, the ranking for the headcount indexbased on altitude is ambiguous. But for the poverty ap and severity measures, Highland is the poorest, and Flatland the least poorF4. Characteristics of the Household Head 3.9 A standard approach is to categorize households by the characteristics of the household head. The head i s in most cases the main provider, and his or her characteristics are o f special importance to the well-being o f the entire household. The head's features are also indicative o f characteristics o f the household in general, including size and composition. 3.10 Poverty i s linked to the age o f the household head. In Figure 3.4, we separate households into three groups depending on the age of the household head. We focus on male-headed households, as nine inten persons are members of such households. Almost two in three individuals live in households whose household head i s between 30 to 50 years old. The incidence of poverty is highest among old prime-age (30-49 years) adult- headed households and lowest among the young prime-age (15-29 years) adult headed households. l4Flatland refers to sucos below 500 m altitude and highland to those above 500 in in altitude. In Figure 3.3, flat refers to sucos below 100 inaltitude and middle to sucos between 100 and 500 m altitude. 21 Figure 3.4: Poverty and Age of Household Head 50 , 15-29 30-49 50 plus Source: 2001 TLSS Education 3.11 All over the world, education i s an important predictor of poverty. Taking grade level completed as education indicator, we find that education levels o f household heads are low. Close to three in five individuals live in families where the head has not completed primary education. No more than one in five has a household head that has finished at leastjunior secondary education. Figure 3.5: Poverty and Education Level of the Household Head 50 40 30 20 10 0 None or Primary Junior Senior Tertiary Pre-school Secondary Secondary Source: 2001 TLSS. 22 3.12 Furthermore, as expected, poverty declines with the education level o f the household head (see Figure 3.5). For example, close to one in two persons are poor in households where the household head has not completed primary education. This compares to less than one in seven where the head has at least senior secondary education. Finally, the average age of the household head drops off as we move from none or pre-school to secondary or tertiary education. This reflects the general increase in school enrollment and attainment over the last decades. Employment 3.13 Jobs and income generation are at the core o f the livelihood o f families around the world. In Timor-Leste, the challenge o f sustainable employment creation i s especially urgent inview o fthe recent legacy. Many workers had formal employment inthe bloated Indonesian public sector before 1999. The vast majority o f these jobs disappeared with the move towards independence. Today, most families depend fully on farming (cultivation, husbandry, forestry, and aquaculture), and only few can supplement this income with receipts from householdbusinesses. Household farm Wage Household Other employnient business Source: 2001 TLSS 3.14 What i s the link between poverty and the occupation o f the household head? We limit our attention to heads inthe age bracket from 15 to 64, typically considered as the economically active phase in life. Economic activity can be o f different types. We distinguish four broad categories (see Figure 3.6): self-employment in agriculture (household farm), remunerated work as an employee for somebody else, self-employment innon-agriculture (household business), and others. Almost seven inten individuals live with heads that, over the course o f the last 12 months, have only worked on their farm. Almost half of them are poor. For one in ten persons, household resources were at least 23 partly gained from a household business (but not from remunerated employment), and for one in seven from wage employment. These two groups experience substantially less poverty, with less than two in ten falling below the poverty line. Finally, the residual category contains heads that are not pursuing any o f these three activities, including living off wealth. They are worse off than those engaged inremunerated employment or household businesses, but still far better off than heads who dependentirely on farming. Assets 3.15 While farming income and wage earnings move families toward self-sufficiency, opening the door to acquiring assets i s the key to their achieving economic security. Assets are an insurance against economic uncertainty and a way o f preparing for future expenses. InTimor-Leste, material assets come inthe form o f land and livestock. We will now look inturn at each o f them to explore the link to poverty. Land 3.16 Land is the most important factor of productioninagriculture, the primary source o f income for three quarters o fthe population. Land access is determined by a traditional system o f land tenure. Households claim to own 95 percent of the land under their control. Four fifths o f this land was inherited and two-thirds o f it i s held on the basis of customary right. Only 4 percent o fthe land plots are disputed. 3.17 Land holdings are widespread, with 86 percent o f the population living in households with access to land. Among those with land access, land holdings are typically limited to one or two plots. The size o f the land holdings i s small: the average area per person is 0.4 hectares, the median area per person i s only 0.22 hectares, and fewer than one intwenty persons with land access hold more than one hectare per capita. Families make full use o f their landholdings, with 95 percent cultivating their land over the last year. The quality o f the land varies - about one fifth is irrigated and less than two fifths are flat. There i s a wide urban-rural gap. As expected, rural households are more likely to own land, and on average have access to 70 percent more land than the urban citizens. Their land i s also o f higher quality, as it i s more likely to be irrigated and level. 3.18 Inorder to investigate the relationship betweenpoverty and land size among the land holders, we turnto Figure 3.7. It depicts the link between the poverty headcount and per capita land size up to one hectare, which covers 95 percent o f the land holding population. For per capita land size o f less than 0.4 ha, poverty i s higher in rural than in urban areas, while the ranking reverses for larger land holdings. More importantly, poverty decreases with larger land size, both in urban and rural areas: as expected, more land is linked to lower poverty15. l5 We find a similar pattern for the relationship betweenpoverty and the estimatedsales value of the land. 24 Figure 3.7: Poverty and Land Size: Ruralversus Urban 60-'i ' + Rural '* Urban 0 I I 0 2 4 6 8 1 Per capitaLand (Ha) Source 2001 TLSS Livestock 3.19 Apart from land and housing, livestock is the most valuable household asset. Cattle, pigs, chicken and other animals are life-enhancing and life-supporting, feeding both people and soils. For many, livestock is one o f the few means o f creating assets and escaping poverty. Nine in ten rural dwellers live in households that own animals. The value o f livestock assets i s about US$lOO per capita, which amounts to four times the monthly per capita expenditures. About one inten persons in rural areas have per capita livestock holdings in excess o f US$200. In cities, 70 percent o f the population live in households which own animals, but the value o f the livestock i s only about half that in rural areas. 3.20 Are livestock holdings related to poverty in farming households? In Figure 3.8, we display the poverty headcount relative to livestock holding per capita, separating rural from urban areas. We find that for families in both villages and cities, more livestock is associated with less poverty. However, this relationship does not hold for all values of livestock assets. For example, for animal assets between US$IOO to US$300, poverty appears broadly unchanged inrural areas, even though it i s declining inurban areas. This reminds us that, while livestock is a key factor in the livelihood o f families and communities, it i s only one o f many determinants. 25 Figure3.8: Povertyand Livestock Ruralversus Urban 60 - 1 Rural + f 3 2 20 Urban I I 0 100 200 300 400 Per capita Livestock (US Dollars) Source: 2001 TLSS. Infrastructure 3.21 The importance of infrastructure for development can hardly be overstated. The experience o f many low income countries shows that living standards improve dramatically as access to services, such as safe water, sanitation, electric power and transport, expand. However, quantity i s no substitute for quality. L o w operating efficiency, inadequate maintenance, and lack o f attention to the needs o f users can result in gains made from initial infrastructure investments evaporating quickly. Timor-Leste received a boost in infrastructure during the Indonesian occupation.. Yet inadequate institutional incentives for maintenance, together with the destruction accompanying the violence, took a severe toll on households' access to services. The considerable effort to rehabilitating infrastructure could only begin to redress this situation, especially inrural areas. 3.22 Lack o f infrastructure services, from safe water, sanitation to electricity, i s clearly an important dimension of poverty. The numbers tell a stark picture. Nationwide, three in four persons live without electricity, three infive persons without safe sanitation and every other person without safe drinking water (see Figure 3.9). There i s a vast divide between urbanand rural areas. Inurban areas 70 percent o f the population has access to these services. The shortfall in rural areas relative to the urban share is 25 percentage points for drinking water, 37 for sanitation and 61 for electrification. In urban areas, almost half the population has access to all three services in contrast to only 4 percent in rural areas. 3.23 The evidence confirms that lack of infrastructure is a key constraint o f the poor. Persons without access to infrastructure are in general also poorer than those with access 26 to infrastructure. This holds especially in urban areas. These differences are sharpest with respect to access to electricity in both urban and rural areas. For example, while only one in seven urban dwellers with electricity are poor, almost one in two without electricity live below the poverty line. The corresponding gap for rural areas i s only half as large (17 percent compared to 34 percent). For drinking water, just under one in five urban dwellers with safe drinkingwater are poor, compared to two in five urban citizens without drinking water. In rural areas, the differences are much less sharper with no differences in the headcouiit index across groups with and without drinking water, but differences persist for the poverty gap and poverty severity measures. 27 Figure 3.9: Poverty and Infrastructure 80 70 60 I 3 50 E L o 40 $0 30 p. 20 I0 0 DrinkingWater Sanitation Electrification All Three Source 2001 TLSS CIRBANAREAS 60 Diinking Water Sanitation Electrification Source 2001 TLSS RCIRALAREAS 6ou- DiifikingWater Sanitation Electrification Source 2001 TLSS 28 INEQUALITY 3.24 So far, our atteiitioii had been primarily on the lower half o f the distribution. N o w we ask how the rich fare relative to the poor. We find evidence o f considerable inequality. For example, the poorest two fifths o f the population, raiilted on the basis o f per capita expenditure, have an expenditure share o f no more than 18 percent, and have monthly per capita expenditures below US$15.49, which isjust above the poverty line o f US$15.44. By contrast, the riches two fifths o f the population have an expenditure share o f about two thirds, and have monthly per capita expenditures o f no less than US$18.22. The most popular inequality indicator, the Gini coefficient, i s displayed in Figure 3.10.l6 Comparing inequality by the different geographical categories, we found a significant difference along the East-West dimension, and smaller variations for the other groupings. Inequality is higher inurban than inrural areas. Figure 3.10: Inequality and Geography: The Gini Coefficient 42 ~~ ~ 39 a 36 2 B2 Y 33 30 27 National Rural Urban Major Other West Center East Flat Mid High Seaside Inland Urban Urban Y 2 - Rural Rural Rural Source:2001 TLSS. 3.25 What can account for the variation in inequality? In Figure 3.11, we isolate five principal characteristics o f the household that may be seen as potential explanations of the structure o f inequality, using a decomposable inequality measure." The first two are geographical features, namely degree of urbanization and the classification o f regions lGThe Gini index increases with inequality. A Gini index of zero indicatesperfect equality, and an index of 100 perfect inequality. l7The General Entropy (GE(a)) class of inequality measureswhere the parameter a determines the weight given to distances o f expenditures at the tails o f the distribution. GE(0) is identical to the mean log deviation and gives more weight to the lower tail. GE(1) is the Theil index and applies equal weights across the distribution. The decompositions are presentedfor GE(0). 29 INEQUALITY 3.24 So far, our attention had been primarily on the lower half o f the distribution. Now we ask how the rich fare relative to the poor. We find evidence of considerable inequality. For example, the poorest two fifths of the population, ranked on the basis of per capita expenditure, have an expenditure share of no more than 18 percent, and have monthly per capita expenditures below US$15.49, which isjust above the poverty line o f US$15.44. By contrast, the riches two fifths of the populationhave an expenditure share of about two thirds, and have monthly per capita expenditures of no less than US$18.22. The most popular inequality indicator, the Gini coefficient, is displayed in Figure 3.10.16 Comparing inequality by the different geographical categories, we found a significant difference along the East-West dimension, and smaller variations for the other groupings. Inequality is higher inurban than inrural areas. Figure3.10: Inequality and Geography: The Gini Coefficient 42 ~ National Rural Urban Major Other West Center East Flat Mid High Seaside Inland Urban Urban Y Rural Rural Rural Source: 2001TLSS. 3.25 What can account for the variation in inequality? In Figure 3.11, we isolate five principal characteristics of the household that may be seen as potential explanations of the structure o f inequality, using a decomposable inequality measure." The first two are geographical features, namely degree of urbanization and the classification of regions '` The Gini index increaseswith inequality. A Gini index of zero indicatesperfect equality, and an index of 100perfect inequality. l7 The GeneralEntropy (GE(a)) class of inequality measures where the parameter a determines the weight given to distances of expenditures at the tails of the distribution. GE(0) is identical to the mean log deviation and gives more weight to the lower tail. GE(1) is the Theil index and applies equal weights across the distribution. The decompositionsare presentedfor GE(0). 29 into Major Urban Cities, Other Urban Centers, Rural West, Rural Center and Rural East. The last three dimensions are linked to the household head: gender, age (five groups: under 25,25-34, 35-44, 45-54, 55 plus), and education (five groups: no primary, primary, junior secondary, senior secondary, and tertiary). For example, separatingurban and rural differences accounts for 13 percent of overall inequality. The largest contributions to explaining inequality come from urbanization and education. If we control for all five features we explain no more than one third o f the observed inequality. The implication i s that the real story of inequality is to be found within geographic, gender, age, and education groups. Figure3.11: Decompositionof Inequality 40 30 Urban-Rural Regions Regions Regions, Regions, and Gender Gender Gender, Age andAge andEducation Note: Regionsconsiders Major Urban Centers, Other UrbanCenters, Rural West, Rural Center, andRural East. Source: 2001 TLSS. SUMMARY AND POLICY ISSUES 3.26 Poverty i s widespread in Timor-Leste with two fifths of the population unable to cover basic food and non-food needs. Living standards vary across the country. Urban areas, especially the two major cities Dili and Baucau, are better off than rural areas. While three quarters of the population live inrural areas, six in seven poor reside there. Poverty also increases froin East to Center and West, and, but less distinctly, from Lowland to Highland. More human capital through better education leads to lower poverty. Demographic characteristics matter too - larger households and families with a higher share of children and elderly are poorer. Inrural areas, valuable landand livestock holdings imply low poverty. Secure access to infrastructure services, ranging from safe water and sanitation, to electricity, is essential for escaping poverty. Inequality i s 30 considerable and mostly within-group. Accounting for geography, gender, age, and education o f the householdhead explains at most one third o f overall inequality. RESEARCH ISSUES 3.27 Timor-Leste's poor share a number o f characteristics, including rural residence, low education and farming, which are inline with features o f the poor inmost developing countries. Country-specific are the findings on the geographical distribution o f poverty, and more research i s neededto better understand the differences between East and West, and Low- and Highland. Furthermore, since the fielding o f TLSS in late 2001, Timor- Leste has undergone important changes, including the reduction ininternational presence and the reflux o f emigrants. It will be important to assess the repercussions o f these economic and social changes on poverty, including the urban-rural divide. Finally, in view of the importance o f rural livelihoods for poverty, future work should establish a more detailed poverty profile of farming communities and explore the importance and origin o f intra-regional differences inliving standards. 31 4. OPPORTUNITY 4.1 Economic growth i s a precondition for sustained poverty reduction. The NDP presents economic growth, together with poverty reduction, as its paramount objective. Creating opportunities for the poor i s the first o f four pillars o f the Government's Poverty Reduction Strategy (see Box 4.1). It involves foremost making markets work for the poor. Providing an enabling environment for the private sector i s essential for employment generation and the prosperity o f small and micro-enterprises. The NDP stresses priority policies and legislationto improve the policy environment. This includes business regulations, trade policy and regulations, land and property legislation, labor legislation, the efficiency o f law and order services, and infrastructure issues, especially related to power and transport". Power sector issues present an important constraint to private sector development (see Box 4.2). Clarifying property rights, in particular for assets such as land and other natural resources, i s crucial to provide incentives to invest and to enable the poor to benefit from the returns to these assets. The Stability Program of January 2003, which outlines the key priorities for the next year in implementing the National Development Plan, highlights govemance andjob creation as key areas. Under the general theme o f service delivery for poverty reduction, priority is also given to interventions in the agricultural sector to improve food security, market access and distribution. Inthis chapter, we discuss the role that employment, assets, andproductivity play in shaping rural and urban livelihoods. The first part presents the employment structure, and the second section highlights key aspects o f economic activity in villages and cities.Ig Economic infrastructure, while o f crucial importance to poverty reduction, is not discussed since the survey had limitedinformation on this aspect. See World Bank (2002) for a discussionof the key elements of improvingthe business environment. This chapter draws on Chapter 4, Volume 11, and Foerster, J. (2002). 32 Box 4.1: Poverty Reduction Strategy: Opportunitiesfor Economic Participation The Government's Poverty ReductionStrategy has five mainelements: -. Agriculture: The strategy aims to improve productivity inagriculture, which provides the livelihoodof the majority ofthe poor. Rehabilitation and construction of irrigation systems, introduction of water harvestingtechniques, wider distribution of improved seeds, fruits, protection of livestock and sustainable management of forest and other natural resources through community participation. Improvementsinmarketing and infrastructure are also planned. Informal sector: Increasingopportunities and improvingproductivity inthe informal sector through training, introduction of appropriatetechnologies, and other support services including micro-credit are planned. Private sector development:Providing an enabling environment for private sector development, where priority policies andlegislation are beingdraftedto improve the policy environment andto encourageboth domestic and foreign private investment. Infrastructure:Provision of infrastructure, including roads andbridges, ports and airports, electricity, communication, andpostal services. Pro-poor macroeconomic policies and public expenditure policies. Source: Planning Commission (2002). Main messages Agriculture is the dominant sector of employment, accounting for four-fifths of all jobs nationwide, and for nine in ten jobs in rural areas. Overcoming the deprivations faced by low-productivity, volatile subsistence farming householdsrequires atwo-track approach. Reduce dependency on agriculture through promotion of non-farm employment opportunities and diversification out of agriculture. This includes activities carried out during slack periods, sectors with close ties to agriculture (farm equipment, agro-processing, transportation, marketing, and food processing), and ensuring that there are no barriers to rural-urban migration. 0 Improve agricultural productivity through building up of the poor's human and physical assets (education, land, livestock) and improving the returns to these assets with the use of better seeds, fertilizers, diversifying into higher valued crops, improved farming technologies, and upgradedinfrastructure (access to markets and credits). Investmentsin human capital will also help shift labor out of agriculture into non-farm work. Urban areas are characterized by the coexistence of high wages and high unemployment. Some 15- 20,000 young persons enter the job market each year. A lean public sector will not be able to provide them with employment. The discontent of unemployed youth, fuelledby material hardship, can pose a threat to the fragile social stability. A policy response should focus on these areas: 0 Generatejobs inthe private sector. This requires the promotion of a favorable businessclimate (providing a transparentlegal andregulatory environment, includingclarifying property rights, and developing basic business services such as accounting, finance, insurance and infrastructure), especially for small- and micro enterprises and labor-intensive sectors. e Ensure that regulations do not distort the labor market and discourage employment. Wages should be allowed to adjust to levels o f demand and supply as the economy adjusts to the withdrawal ofthe large international presence. 33 Box 4.2: Electricity A reliable source of electricity is an important condition for a vibrant private sector. Electricity supply in Timor-Leste is entirely diesel based and costs have risen sharply in line with international oil prices. The electricity tariff effective August 2001 was 24.9 cents per kwihour for businesses and US$l/month for the first 25kwihours and 24.9 cents for each additional kw/hour for residential users. This rate i s high by international standards but this reflects the high unit costs. Since billing began, the rate of payment of invoices has been low. Inspite ofthe high cost, supply remainserratic. The government is spending a significant share of its budget for electricity subsidies. In FY2002, of the 20 percent allocated to economic services, over half went to power (11 percent). From a poverty perspective, the high share of government expenditures allocated to power utility operating subsidies is an issue of concern. The figure below shows that the distribution of the electricity subsidy is strongly regressive, that is, it benefits the rich more than the poor. Two fifths of the population that has electricity belongs to the richest quintile. The householdsurvey only asks ifthe householdwas electrified but not how muchelectricity was consumed. The exact distribution of the subsidy across income groups is therefore difficult to quantify. However, to the extentthat the richare likely to consume more power, the subsidy would be even more regressivetlian shown. Inprinciple, a higher and better enforced revenue collection among rich households could be a counterbalancing factor, but at the present time, fee collection is low, even in Dili. Hence there i s little doubt that electricity subsidies are benefiting the rich far more than the poor. Furthermore, the opportunity cost of electricity subsidies - 16% of actual government expenditures in FY2001 and at least 11% in FY 2002 - is huge, diverting resources that could be applied inpoverty reduction programs. 100 0 0 20 40 60 80 100 Cum. percentage of population (rank by per capita expenditure) Source: WorldBank (2002), WorldBank (2002b) and 2001 TLSS. 34 EMPLOYMENT POVERTYAND 4.2 Secure employment i s the key to escaping poverty. The population views it as the main priority for improving living standards and among the top priorities for the nation as a whole (Figure 2.6). The overall labor force participation rate is 60 percent (Figure 4.1). This figure is comparable to numbers under the Indonesian occupation.20 Labor force participation rates are higher for menthan women, lowest inDili, and highest inthe rural center. They are lowest among the youngest (15-24 years). Nine inten prime aged males (25-54 years) participate inthe labor force, while female participation rates peak after the child-bearing years. Figure 4.1: Labor Force ParticipationRates among the Economically Active Population (15-64) 80 70 -9 60 50 5 z 40 30 20 10 0 National Men Women DilU Other Rural Rural Rural Rural Baucau urban center east west Source: 2001 TLSS. 4.3 Agriculture i s the main sector o f employment, accounting for four-fifths o f all jobs. Only 4 percent o f the labor force i s employed in industry. By comparison, in 1998, 70 percent o f all workers were employed in agriculture and 10 percent inindustry. This indicates a reduction in formal employment opportunities since 1999, and a shift of employment back towards self-employment in agriculture. The sectoral composition o f employment differs between urban and rural areas. Nine inten meno f prime working age between 15 and 64 are employed inagriculture inrural areas, while less than a quarter in Dili are employed inagriculture (Figure 4.2). Other urban areas are closer to rural areas interms oftheir employment structure. Women are more likely to work inservices inall areas than men. 2oThe participation rates from the Indonesianlabor force survey were 62.5 percent in 1995, 61.5 percent in 1996, 61.1 percent in 1997, and 71.9 percent in 1998. 35 EMPLOYMENT POVERTYAND 4.2 Secure employment is the key to escaping poverty. The population views it as the main priority for improving living standards and among the top priorities for the nation as a whole (Figure 2.6). The overall labor force participation rate i s 60 percent (Figure 4.1). This figure is comparable to numbers under the Indonesian occupation.20 Labor force participation rates are higher for menthan women, lowest in Dili, and highest inthe rural center. They are lowest among the youngest (15-24 years). Nine inten prime aged males (25-54 years) participate inthe labor force, while female participation rates peak after the child-bearing years. Figure 4.1: Labor Force Participation Rates among the Economically Active Population (15-64) 80 70 60 30 20 10 0 National Men Woineii Dili/ Other Rural Rural Rural Rural Baucau urban center east west Source: 2001 TLSS. 4.3 Agriculture is the main sector o f employment, accounting for four-fifths o f all jobs. Only 4 percent o f the labor force i s employed inindustry. By comparison, in 1998, 70 percent o f all workers were employed in agriculture and 10 percent in industry. This indicates a reduction in formal employment opportunities since 1999, and a shift o f employment back towards self-employment in agriculture. The sectoral composition o f employment differs betweenurban and rural areas. Nine inten meno f prime working age between 15 and 64 are employed inagriculture inrural areas, while less than a quarter in Dili are employed inagriculture (Figure 4.2). Other urban areas are closer to rural areas interms oftheir employment structure. Women are more likely to work inservices inall areas than men. 2oThe participation rates ffom the Indonesian labor force surveywere 62.5 percent in 1995, 61.5 percent in 1996, 61.1 percent in 1997, and 71.9 percent in 1998. 35 Figure4.2: EmploymentSector by Gender DISTRIBUTIONOF MEN 15-54 ACROSS SECTORS DISTRIBUTIONOF WOMEN 15-54 ACROSSSECTORS I 0 0 , ,j 1n 'ill 'ill 811 811 711 711 % 60 p BO 50 50 e 411 2 411 311 311 211 211 111 ill 0 11 4.4 While agriculture provides the bulk o f employment, its contribution to non-oil GDP is less than one quarter. Productivity, defined as output per worker, inindustry and services i s more than ten times as highas inagriculture (see Figure 4.3). Figure4.3: OutputPerWorker, 2001 7,000 I 6,000 5,000 4,000 3,000 2,000 1,000 0 Agriculture Industry Services Source: 2001 TLSS and PlanningCommission (2002). 4.5 The move away from formal employment has also reduced wage earnings. Only one in ten workers receive wages or salaries. Male workers are slightly more likely (13 percent) to be wage employees than females (9 percent). By comparison, in 1998, 21 percent o f male workers and 8 percent o f females were wage employees. Wage employment is strongly associated with higher living standards. Only 3 percent o f the male wage-workers belong to the poorest quintile, and over one fifthto the top quintile. 36 RURAL LIVINGSTANDARDS 4.6 Timor-Leste i s a rural country. Three quarters o f the population live in villages, and three infour rural households rely exclusively on income from agriculture. Figure 4.4 shows the patterns o f labor income by region and poverty status. The overall diversification in income sources i s limited. However, in all three regions, the non-poor depend less on farming income than the poor. Diversifying income sources outside o f agriculture is an important mechanismto cope with farm income fluctuations. Inaddition, non-farm income sources yield higher earnings. Earnings from wages are eight times higher than from farming, and three times higher than from non-farm self-employment (see Figure 4.5). 4.7 These numbers illustrate that non-farm employment i s vital for raising living standards in rural areas. Nevertheless, agriculture will remain the main sector o f employment and source o f income for the majority of the population for decades to come. Improving productivity and incomes in agriculture are an important part o f expanding opportunities for the poor, and have to be pursued in parallel with promoting the rural non-farm economy. Figure 4.4: Household Sources of Income in Rural Areas 100 90 10 0 Non-poor L- Poor. '/ \Non-poor ' Poor Non-Poor Poor c Y -/ 2 I Center East West Source: 2001 TLSS. , Non-farm Self-employment 37 Figure 4.5: Daily Earningsin Rural Areas 4.0 1 I 3.5 3.0 2.5 2.0 1.5 1.o 0.5 0 Self-employed Self-employed Wage employment agriculture non-agriculture Source: 2001 TLSS. Box 4.3: Characteristicsof Farm and Non-Farm Workers Non-farm workers are only a small fraction o f the workforce in Timor-Leste. Among the working age group (15-64 years) who are employed, just under one fifth engage in non-farm work as their primary occupation. Wage work predominates, with 12percent o f the employed work-force, and self- employed business employment accounts for the remainder. Four inten non-farm workers are from Dili, another quarter from the Rural Center, and the other regions account for about one tenth each. The most important distinguishing feature between farm and non-farm worker is education: persons engaged in non-farm wage work have over eight years o f education, those engaged innon-farm self- employment half that, and farmers have 2.4 years o f education (see Figure). This pattern is consistent across regions. 1 0 9- n S o u r c e 211111 T L S S In addition, non-farm wage workers are also younger than farm workers (33 years versus 37 years), especially in urban areas and the Rural West. Finally, two-thirds o f all non-farm workers are male, but the pattern differs across wage work and self-employed workers. Three out of every four wage workers is male, but only 1 in 2 non-farm self-employed workers is male. The fraction o f males in non-farm self-employment i s highest inDili (two-thirds) but only between a third and forty percent in the other regions. 38 Agriculture 4.8 Agriculture, comprising crop and livestock activities, fisheries and forestry, accounted for about 32 percent o f non-oil GDP between 1993-1998 (World Bank, 2002a). Crop cultivation dominated the sector, with livestock, fisheries and forestry contributing a much smaller role. In 2000, the share o f agriculture declined to about 26 percent o f non-oil GDP. The destruction o f infrastructure and assets (livestock, food and seed stocks), displacement o f large numbers o f the population, and the elimination o f production subsidies for crops such as rice, inputs such as fertilizers, fuel, and cooking oil, are some o f the factors that contributed to this decline. The decline in agricultural output, in conjunction with a shift of employment towards agriculture noted above, implies that agricultural productivity declined between 1999 and 2002. The recovery in production o f most major crops (except rice) by 2001 has been threatened by the reports o f drought inthe south o f the country this year. 4.9 Agriculture in Timor-Leste i s dominated by subsistence farmers. They produce for self-consumption, produce with basic inputs including unpaid family labor, small plots of land, basic tools, and rely mostly on rainwater. Inthe following, we characterize agriculture with regard to assets (land and livestock), crop production, factors o f production, and crop sales. Assets 4.10 Land i s the main asset o f rural The average rural household owns 1.2 ha o f land, or 0.4 ha per capita. Land ownership i s fairly widespread, with only 6 percent of the rural population being landless. But, land ownership i s distributed unequally. Among rural landholders, the poor own half as much land per capita than the non-poor (Figure 4.6). The Gini coefficient for per capita land holdings i s 0.55, which i s significantly higher than the distribution o f consumption (0.37). As shown in Chapter 3, poverty declines with higher landholding. Most households have no formal titles to land, yet only 4 percent o f land plots are disputed. The majority owns their landon the basis o f customary rights and report having inherited it. A large proportion of households in other urban areas also relies on agriculture for their livelihood and for inany purposes will be combined with the rural sample. Average land ownership in other urban areas (0.25 ha per capita) is lower than inrural areas (0.38 ha per capita). 39 Figure4.6: Landholdingper capita(ha) ^^...,,,.... . , 0.5 .z!?b,"0.4 0.3 4- 2 g0 0.2 0.1 0.0 Rural Center East West Non-poor Poor Source: 2001 TLSS. 4.11 Besides land, most rural households also own livestock. Chickens and pigs are the most common livestock, with seven inten rural residents holding them. One infive rural residents own horses, while one in ten hold cows and buffaloes. The average value of livestock assets owned by the non-poor i s double that o f the poor on a per capita basis (Figure 4.7). The value of livestock assets held by households inthe East exceeds those inthe Center and West by 75-85 percent. Thus, the non-poor own more assets - twice the land per capita, and twice the livestock holdings than the poor and also have greater human capital, another important asset. As noted in Chapter 3, more livestock is associated with lower poverty. Figure4.7: Average LivestockValue Per CapitainRuralAreas (US Dollars) Rural Center East West Non-poor Poor Source: 2001 TLSS. 40 CropProduction 4.12 Agriculture data are notoriously difficult to collect, and estimates o f production and yields have to be treated with caution. Higher cultivable land per capita translates into greater production o f crops per capita for the non-poor than the poor. The non-poor produce more per capita of the main staples (rice, maize, cassava)22.They also produce significantly higher amounts o f the higher-value crops, such as coffee, fruits, vegetables (Table 4.1). Compared to the poor, the non-poor produce in per capita terms 50 percent more rice, one quarter more maize, and one fifth more cassava. They produce four and a half times the amount o f coffee cherries, and seven times the amount o f fruit. Poverty rates by different crops produced shows that coffee producing households are the least poor. Insummary, a larger fraction o f non-poor households grow the higher value crops (coffee, vegetables, and fruit) and they produce more o f all crops, especially the higher value crops. Table 4.1: Per CapitaAnnualProductionofDifferent Crops (kgkapita) Total Per capita production (kg) (tons I year) National N o n Poor poor Gogo rice 3,622 4.4 5.5 2.6 Rice 53,845 65.0 75.2 49.5 Maize 64,93 1 78.4 85.5 67.6 Cassava 48,056 58.0 62.5 51.2 Coffee cherries 19,285 23.3 33.7 7.5 Coffee dry bean 14,134 17.1 20.2 12.3 Kidney bean 3,722 4.5 4.7 4.2 Sweet potato 24,705 29.8 31.5 27.3 Potato 968 1.2 1.1 1.3 Taro 13,111 15.8 17.1 13.9 Squash 8,932 10.8 14.0 5.8 Mungbean 1,786 2.2 2.7 1.3 Soy bean 819 1.o 1.o 1.o Coconuts 2,115 2.6 3.3 1.4 Peanuts 1,468 1.8 2.1 1.3 Vegetables 1,860 2.2 2.7 1.6 Bananas 19,138 23.1 23.9 21.9 Other fruit 3,052 3.7 5.6 0.7 Source: 2001 TLSS. 22 The differences are even larger when you compare the poorest and the richest quintiles. The richest quintile has larger plots o f land for all crops (except soy bean). The differences are marginal for peanuts, coconut, squash, cassava and maize, but larger for crops such as rice, upland rice, coffee, vegetables and fruit. These differences are especially large for higher paying crops such as coffee, vegetables and fruit. 41 Factors of production 4.13 What explains the higher production of the non-poor than the poor? One key factor is the availability of more, and better quality land. One important dimension of land quality is irrigation.23Just under a fifth of all plots are irrigated and a quarter of all irrigated plots have year-round irrigation in 2001 (Figure 4.8). The substantial investments in reconstruction in 2002 would have expanded the plots under irrigation. The limited availability of irrigation throws farmers at the mercy of the vagaries of weather, and their fortunes are linked to rainfall availability. Irrigationi s used primarily to grow rice, but not exclusively. The East has the largest fraction of irrigated plots, and 44 percent of the irrigation is year round.24 And, while irrigation is limited, the non-poor have larger irrigated landholding per capita (0.13 ha) than the poor (0.04). Figure 4.8: Irrigation in Rural Areas 50 45 40 -," 3 35 -98 30 0 25 2 20 15 10 5 0 Rural Center East West I I With year round irrigation Source: 2001 TLSS. 4.14 Achieving higher yields is also linlted to the use of labor and other inputs and better access to markets. Besides land, the main factor of production is labor - within agriculture, 98.5 percent are self-employed working solo or assisted by family or unpaid household workers. The use of other inputs, such as fertilizer, pesticides, manure, and improved seeds i s very limited. Among all agricultural households, only 3 percent of households used fertilizer, manure, or pesticides, and almost all of them are non-poor. Over three quarters of the farmers inlate 2001 report that they do not use such inputsdue to lack of availability. Households inthe rural East and rural West are more likely to use 23 Data are also available on the slope of the land. Steep slopes are associated with higher poverty in the East, but with lower poverty rates inthe Center, where householdsgrow coffee onthese slopes. Inthe West there is little correlation betweenslope andpoverty rates. 24 Land holding in other urban areas is smaller, but the poorlnon-poor difference persists among landowners. 42 these inputs. Land preparation i s undertaken primarily with basic tools, with very limited use o f equipment such as tractors. 4.15 Access to credit from formal sources i s still very limited. In late 2001, for the country as a whole, 12percent o f the population borrowed money inthe past year. While 14 percent o f the poor borrowed, only 11 percent o f the non-poor borrowed. Just under seven out o f ten people borrowed once, a quarter borrowed twice. Nine out o f ten people who borrowed did so from friends and relatives and the loans were primarily for consumption. Only 2 percent borrowed for agricultural inputs and four percent for a non- agricultural business. Crop Sales 4.16 Most farmers are subsistence farmers who produce for self-consumption. Overall, about two-thirds o f crops are self-consumed. Integration into markets allows farmers to engage in the production o f higher earning cash crops. At the national level, we find that households with higher crop sales per capita are less poor (Figure 4.9). Poorer farmers sell more low-paying crops such as rice, maize, taro, squash and differenttypes o f beans. Non-poor farmers vend high-valued crops, such as coffee, vegetables and fruit. This relationship does not, however, hold up for the East, where households tend to sell less, on average, yet are less poor than households inthe Center and West (Box 4.4). Figure 4.9: Value of Per Capita Crop Sales andPoverty Rate I I I I I 0 Per capita Sales (US Dollars) 450 Source: 2001 TLSS. 43 Box 4.4: How Large and Stable are RegionalDifferences? The design o f efficient policies that are tailored to the specific conditions o f a country presents a challenge for government and development agencies. Inthis context, the extent o fregional differences is one essential input into policy priorities. If regions are marked by sharp and stable differences in living standards, then a geographically differentiated strategy might be appropriate (for example, fiscal transfer rules for regional transfers that transfer more per capita to poorer regions, incentives for investment etc). While other factors have to be taken into account in the design of such policies, establishing the extent and nature o f regional gaps is an important first step. InTimor-Leste, the percentage ofthe population living inpoverty is the lowest inDililBaucau, followedby the Rural East and Other Urban Centers. Poverty i s highest in Rural Center and Rural West. Are these differences significant? A first question i s the extent to which the rankings depend on the precise level of the poverty line. Analysis shows that in rural areas, the ranking between Rural West and Rural Center is ambiguous, and Rural East is unambiguously the least poor. Nationwide, DiliiBaucau is the best offregion. However, the differences between DiliiBaucau and Rural East narrow under alternate assumptions. For example, excluding housing from the consumption measure and poverty line leads to only slightly higher poverty in the Rural East relative to DiliBaucau. But other welfare indicators demonstrate that DiWBaucau is better off than the other parts o f the country. It has substantially better access to infrastructure services (safe drinking water, sanitation, electricity, and access to markets), higher education outcomes (lower illiteracy rates, higher primary and secondary net enrolment rates), and superior immunization coverage for children under the age o f one. Foodsecurity indicators also confirm that the major cities are the bestoff: no more than one third report that their food consumption was less than adequate in contrast to two-thirds of the population in Rural East. The average distance to an everyday market inthe Rural East is over 25 kmin contrast to just 1.6 kin in DiliiBaucau. In summary, while the consumption poverty rankings between the major cities and the Rural East narrow under altemate assumptions, other noli expenditure indicators of welfare confirm that DililBaucau i s the most well-off region. Turning to rural areas, why does Rural East have the lowest poverty in spite of its more limited access to markets? First, being furthest away from the border to Indonesia, the Rural East was relatively protected during the violence in September 1999, as confirmed by survey indicators o f destruction o f housing and of livestock values compared to 1999. The worst hit areas were Rural West and Other Urban Centers. Almost 6 in 10 houses in Rural West were destroyed during the violence in 1999 compared to less than one tenth in Rural East. Average per capita livestock holdings in2001 were only 17 percent o f their 1999 value in Rural West and 40 percent inOther Urban areas, but almost three quarters in Rural East. Subjective well-being data also corroborate that the population inthe Rural East reported the lowest downward mobility in terms o f both economic well-being and power between 1999 and 2001. Second, the population in Rural East currently has substantially higher values o f some type of assets - average livestock values per capita in Rural East are the highest, almost 80 percent higher than in Rural Center, and almost double that in Rural West. Per capita land holdings in Rural East are 0.33 per capita, lower than in Rural Center (0.5 l), but the amount o f irrigated land per capita is significantly higher in Rural East (0.15 ha per capita) than in Rural Center or Rural West (around 0.05 ha per capita). Third, evidence from the suco survey indicates that some areas in Rural East have two harvests of rice, with the second harvest in October. Thus, the timing o f the survey with respect to the second harvest might partly explain the lower poverty. 0 Finally, employment data from the household survey indicate that while labor force participation rates in Rural East are lower than in Rural Center or Rural West, largely due to lower female participation rates, the average hours worked by those inthe labor force are significantly higher (44 hours per week compared to 36-37 hours). This may reflect additional labor needs during harvesting time. The data point to a number o f factors that explain the relative prosperity inthe Rural East compared to other rural areas. Yet, it i s difficult to assess whether this reflects a transitional position or a truly structural phenomenon. While consumption poverty indicators rank Rural East lowest, other welfare indicators show that Rural East i s not consistently better off than other rural areas. Insummary, while DiliiBaucau i s clearly the most prosperous region, other geographic differences are less robust and may be affected by transitional factors. URBAN LIVING STANDARDS 4.17 Labor market issues differ for rural and urban segments. While the main concerns in rural areas are low productivity and lack of off-farm employment. urban workers are faced with the fundamental problem o f job security. With the withdrawal o f the large international presence in major urban centers, many service sector jobs will be lost. As noted above, the service sector i s the main employer in DiWBaucau (Box 4.5). This section focuses on two aspects o f the urban labor market: unemployment and wages. Box 4.5: Jobs in Dili/Baucau Most jobs in Dili/Baucau in late 2001 were in the service sector, which employed seven in ten workers. Wholesale trade, retail, restaurants and hotels is the largest sub-sector within services employing one in five workers. While one in three women work inthis sector, less than one in five men do. Almost three in ten jobs are in community, personal, social services, including in education and health, and these jobs account for almost two in five jobs that women hold. One in ten people work intransportation and communications, andjust under one inten inconstruction. Thesejobs are dominated by men. Agriculture, the predominant employer in Timor-Leste as a whole, accounts for only one in five jobs in the major urban cities. This pattem is broadly consistent across age cohorts, although the 25-44 year olds are less likely to work inagriculture. While two out of five uneducated workers work inagriculture, only one intwenty amongthose with senior secondary educationdo. 100Y 90% Other 80% Other community, social and personal services 70% Public adiniiiistration, health 60% and educatioii storage and 50% 0Transportation, coininuiiications 40% Wholesale, retail, restaurants and hotels 30% Construction 20% Agriculture 10% no, "I" Total Men Women Four in ten people are self-employed. Older workers (45-64 years) are more likely to be self- employed, with more than half engaged infamily businesses. Women are also more likely to be self- employed. One in four workers are employees in the private sector, and another one in five in the public sector. A third of all 15-24 year olds work as employees in the private sector. Higher education allows workers to access more formal jobs. Four out of five workers with at least some senior secondaryeducation are employees comparedto only one fifth amongthose with no education. 4.18 Unemployment i s largely an urban phenomenon. According to international standards, defined by the International Labor Organization, the unemployed are persons who are part o f the labor force and, inthe last 7 days, did not work but were looking for work. This international definition may not provide an adequate picture ofjoblessness in 45 developing countries due to the importance o f seasonality and the discouraged worker effect, i.e. take account o f individuals who are willing to work but have ceased to look actively for jobs. This definition i s more appropriate for wage and salaried workers. In rural areas, where most individuals work as self-employed in the household farm, this concept o f unemployment i s particularly difficult to apply - underemployment and low productivity jobs are the main issues. Our data allow for an alternate classification o f workers as "jobless" based on the self-reported main occupation o f every working-age individual. Overall, there i s a considerable but imperfect overlap between the two definitions (see Box 4.6). Box 4.6: Two Conceptsof Unemployment ~ We present two alternate definitions o f the unemployed. First, we use the standard international definition. Second, we use the information on the self-reported main occupation of every working- age individual. The latter definition implicitly has a longer time frame over which the individuals response is based, in contrast to the international definition that refers to the past seven days. The unemployed are identified as those responding "jobless", and the out of the labor force as those reporting "pensioner", "housewife", and "student". The table shows the working age population (15- 64 years) divided into three groups - working, unemployed, and out-of-the labor force. There is considerablebut imperfect overlap between the two definitions. Overall, about four in five working- age individuals are classified into the three groups (working, unemployed, or being out of the labor force) identically by both definitions. However, only about one third of those classified as unemployed in DilUBaucau according to the international definition label themselves as jobless, and halfconsider themselves to be out of the labor force. Composition of the Labor Force, (%) I nte r n at io na I class ific a tioti W o r k i n g U n e m p l o y e d O u t o f L F T o t a l N B t i o s a 1 S e If-re p o rt e d act iY ity W o r k i n g 4 7 1 4 5 2 J o b l e s s 0 1 2 3 O u t o f L F 10 I 3 4 4 5 T o t a l 5 1 3 4 0 1 0 0 D iIiiB a 18 E a 11 Self-reported a c t i v i t y W o r k i n g 3 4 1 2 3 8 .Io b l e s s 0 3 6 1 0 O u t o f L F 5 5 4 3 5 2 T o t a l 3 9 I O 5 2 1 0 0 S o , , r c c 211111 T L S S 4.19 Are those without jobs poorer than those with jobs? Table 4.2 compares poverty rates in Dili and Baucau o f the working, the jobless, and those out o f the labor force for both definitions o f unemployment. The working are further divided into those employed inside and outside o f agriculture. For population aged 15-64, poverty i s highest amongst those working in agriculture. It i s between two to three times as high as for the unemployed or jobless, confirming the close connection o f poverty and dependence on agriculture. Poverty among those without jobs is higher than those employed in non- agriculture jobs, but the difference i s relatively small. However, among the 15-34 year olds, the unemployed are substantially poorer than those working in non-agricultural 46 jobs, especially according to the self-reported classification. The unemployed have a poverty rate almost twice as high according to the international definition, andthree times as highaccording to the occupational definition. Table 4.2: PovertyRates by Labor Status DiliiBaucau 15-64 15-34 Iiiteriiatioiial classificatioii Workiiig 13.8 9.6 Agriculture 30.0 25.7 Non-agriculture 10.1 6.8 Uiieinployed 11.5 12.7 Out o fLF 12.0 11.5 Self-reported activity Workiiig 13.0 9.7 Agriculture 30.5 26.4 Noli-agriculture 9.1 5.7 Jobless 14.8 17.1 Out ofLF 12.0 10.6 Total 12.6 11.1 Source: 2001 7zSS. 4.20 While the poverty incidence differs somewhat between the two concepts o f unemployment, the definitions closely coincide with regard to unemployment rates by different characteristics, In Table 4.3, we focus on unemployment in DiWBaucau according to the international definition. Workers in those major urban centers face the highest unemployment rates, with one fifth o f the workforce being unemployed (Figure 4.10). Women have higher unemployment rates - one in four womeii are unemployed, compared to one in seven men. Unemployment rates decline sharply with age: the unemployment rate among the youth (15-24) is a staggering 43 percent, it declines to 17 percent for the 25-34 year olds, and 9 percent o f the over 35 year olds. Figure4.10: Unemployment Rates 20 18 16 14 8 12 S g 10 2 $ 8 6 4 2 0 Dili Other urban Rural National Source: 2001TLSS. 47 4.21 To look at the joint impact o f worker characteristics on unemployment, we turn to a multivariate analy~is.~'The model includes as explanatory variables personal characteristics (age, gender, education, marital status), and household variables (household composition and assets). Separate estimations are runfor the 15-64 year olds and for the 15-34 year olds in DiWBaucau. The results confirm the previous findings. Among the working age population, women are more likely to be unemployed. Controlling for other characteristics, they have a six percent higher probability o f being unemployed. Unemployment declines also strongly with age -those older than 24 years of age are 13 to 16 percent more likely to be unemployed than the 15-24 year olds. Controlling for other characteristics, education does not have a significant effect on employment.26 These results are confirmed for the 15-34 year olds sub-group. In this case, 25-34 year olds are 10 percent less likely to be unemployed than the 15-24 year olds. Table 4.3: UnemploymentRatesand Characteristics of the Unemployed Unemployment rates Compositionof the Unemployed DililBaucau National DililBaucau National Total 19.7 5.3 100 100 Gender Men 17.6 4.6 63 57 Women 25.0 6.8 37 43 Age groups 15124 43.0 14.9 50 56 25134 16.5 5.0 31 27 35144 10.8 2.3 13 10 45/54 8.3 1.o 5 3 55164 5.6 1.6 1 3 Education No school 11.0 2.0 12 20 Primary 17.2 6.6 20 27 Junior high school 27.1 8.4 17 15 Senior high school or more 23.7 14.9 50 38 Source: 2001 TLSS. 4.22 Next, we focus on the pool of the unemployed (Table 4.3). The bulk o f the unemployed are young and educated males. Half the unemployed in DiWBaucau are young (15-24 years) and another third are between 25 and 34 years. While men face a lower likelihood o f unemployment, they also represent the bulk of the labor force and 25A probit model is estimated for individuals in the labor force. The dependent variable takes a value 1 if the person is unemployed and zero ifthe person i s working. 26 The coefficients are positive and in some cases near significant, suggesting that in DiliIBaucau, unemployment is a problem of the educated. 48 therefore o f the unemployed. Two inthree unemployed are male. Half o f the unemployed have higher secondary education or more. 4.23 Inspite of highunemployment, wages are high.The large influx of international agencies, NGOs and other foreign employers inflated earnings. Figure 4.11 shows hourly wages for urban employees, in addition to those for production manufacturing workers for Indonesia.27As a cross-check, we have also included the rates posted for the Timor- Leste civil service, which are inline with our estimates on TLSS hourly wages for public sector employment. The wages for workers in Timor-Leste are on the order o f two to three times higher than the Indonesian wages. These findings confirm evidence from other sources. For example, unskilled farm labor wages in the coffee industry are estimated to be three times higher now compared to rates in Indonesia (World Bank 2002) 28. According to a widely held view within Timor-Leste, this differential is justified inview of the significantly higher cost-of-living compared to Indonesia. These high wages, however, erode Timor-Leste's competitiveness. The coexistence o f high wages and unemployment i s a puzzle. Civil service wages were set initially at rates starting at US$S5 per month, which were about three times the average in Indonesia. These relatively high wages at the lower grade levels may have set the levels for other private sector wages leading overall to an uncompetitive real wage.29 More recent information indicates that private sector wages have started to fall. In early 2003, one of the largest employers in Dili reduced the pay for non-skilled workers by one quarter to $90/month. With public sector wages remaining unchanged, one implication o f this adjustment i s that the private-public wage gap i s increasing. 4.24 Highurban wages, fueled by a service sector boom triggered by international presence, and reduced public sector employment are likely to have contributed to the decline in wage employment in overall employment from 1998 to 2001. These high wages have direct implications on private sector employment opportunities. Private employers, unable to meet the high wage costs, may exportjobs and shift towards labor- reducing technologies, inturnreducing the growth o f private enterprises inTimor-Leste. 27The Indonesianfigure is a national average. Wages inBali and other easternprovinces are even lower. 28See World Bank (2002) for a discussion of this issue. 29ETTA salaries correspond reasonably closely to the mid-points of the salaries agreed by the NGOs and it seem that the NGOs usedETTA scales as a reference point (World Bank, 2002). 49 Figure4.11: UrbanWage Rates 3 1.o c"ba 0.8 -$ 0.6 m 0.4 3 0.2 0 m v1 n n m m .-C [r, --5 h m e, P3 v 0 e, Y .*@ z" & Source: TLSS (2001), ETTA, UNTAET andNGO rates from World Bank (2002) SUMMARY AND POLICY ISSUES 4.25 Employment andjobs are at the core o f improving living standards. Before 1999, formal employment inthe bloated Indonesianpublic sector was common, while now only few people have regular incomes outside o f agriculture. The public sector employed 28,000 people under the Indonesian occupation, whereas the current payroll is only half that number. The NDP einphasizes the need to keep a lean, disciplined and transparent public sector. Some 15-20,000 young persons enter the job market each year, far more than the anticipated vacancies in the public sector. Creating an adequate number o f both formal and informal employment opportunities to meet the needs of the country's youth i s one o fthe key challenges for Timor-Leste. 4.26 The Government's Poverty Reduction Strategy rightly emphasizes improvements in agricultural productivity. Agriculture is the dominant employer, accounting for four- fifths o f all employment. But it only accounts for a quarter o f non-oil GDP, pointing to low output per worker inthe sector. A key driver for higher living standards inrural areas i s access to non-farm employment opportunities and diversification out o f agriculture. Non-poor households in rural areas are much more likely to receive income from non- farm enterprises and/or wage earnings. While improving employment opportunities outside o f agriculture i s important, it remains essential to raise agricultural productivity. Among agricultural households, living standards are primarily determined by the distribution o f assets. Non-poor households have higher human capital and twice as much 50 land and livestock per capita. A larger fraction produce higher value crops, and they produce more per capita o f the main staples and o f higher value crops. Households that are better integratedinto markets have lower poverty rates. 4.27 Urban labor markets experience the twin existence o f high wages and high unemployment. The influx o f expatriates during the transition to independence fuelled a service sector boom in the major urban centers, which yielded high real wage levels, urban concentration, and misallocatedinvestment inthe service industry.The coexistence o f high wages and highunemployment i s a puzzle. Civil sector wages, set at three times the average in Indonesia, may have set the levels for other private sector wages, leading to an overall uncompetitive real wage. There i s a difficult transition ahead as declining demand and high unemployment rates are expected to put downward pressure on urban wages. Recent evidence indicates that private sector wages are falling. As difficult as the transition may be, it needs to take place for the longer term growth prospects o f the economy. Looking forward, as the service sector starts downsizing, it will be crucial to provide a conducive private sector environment for sustainable job creation. As new sources for jobs become available, wage levels should be allowed to adjust to the new conditions o f demand and supply. While labor costs and labor productivity may be an important restraint on private investment, other issues related to private sector development are also important (clarifying property rights, the legal and regulatory environment, availability o f business services, such as accounting, finance, insurance, and availability o f infrastructure). The December 2002 riots in Dili have brought security and law andorder to the forefront. Job creation is one component for stability. Ensuring that the skills supplied by the education system match the skills demanded in the labor market i s a key challenge for Timor-Leste. Labor regulations need to be able to maintain adequate worker protection while keeping sufficient labor market flexibility to ensure competitiveness and creation o f employment. RESEARCH ISSUES 4.28 This poverty assessment has only provided the first step in understanding the sources o f employment and productivity growth in rural and urban sectors. Agriculture will remain the main source o f livelihood for the bulk of the population for the foreseeable future. More research is needed to identify the determinants, and constraints, o f agricultural productivity across regions. A flourishing non-farm sector i s critical to improving rural living standards, yet it employs only a small share o f the labor force. Investigating the essential drivers for expanding non-farm employment i s an area for future research. Generating jobs in urban areas is a high priority. Reviewing the regulation and legislation governing the labor market, analyzing business conditions for small- and micro enterprises, identifying the mismatch o f skills demanded and supplied and the role o f private firms inon-the-job training are areas for further investigation. 51 5. BASIC SOCIALSERVICES 5.1 Access to basic social services, like education, health, water and sanitation i s essential for a decent life. It also improves the ability o f persons to contribute to a prosperous country. The poor often lack access to these basic services. The benefits of improved human development outcomes are well known. Education raises the productivity o f labor, the most important asset o f the poor. There i s ample evidence o f the effects o f education in improving productivity and output in farming and wage employment. Educated farmers are more likely to adopt new technologies and to get higher returns on land.30 Raising the human capital o f poor children greatly improves their chances o f escaping poverty later in life. There are other positive effects of education. For example, educated mothers tend to have healthier children, as they are more likely to be better nourished and immunized. A healthier life in turn reduces the time lost in school, or at work, due to illness. As international experience demonstrates, the case for improving the access and quality o f the basic services for the poor is compelling. 5.2 Timor-Leste's need for raising human development standards cannot be overstated. Literacy rates are low and health i s poor. Only one in twenty persons speak Portuguese, and no more than four in five Tetun. The population ranks education and health as top priorities for the future, on par only with employment. In its poverty reduction strategy, the NDP places high importance on providing social services to the poor, particularly o f quality primary education and health care, including preventive programs such as immunization and public health (Box 5.1). The Stability Program of January 2003 reiterates the Government's commitment to service delivery for poverty reduction by focusing on primary and secondary education, vocational training and expanding basic health services by increasing the number o f mobile clinics and health posts in inaccessible areas. This chapter first reviews spending on social programs, and then provides a more detailed analysis of the two largest social sectors, education and health. The experience o f the two sectors has been different. A key lesson o f the reconstruction program in Timor-Leste i s that there i s a tradeoff between developing a coherent policy framework and rebuilding the infrastructure3 * . While health paid early attention to developing a medium-term strategy, creating sustainable institutions with strong management capacity, it was slower at the initial stages in achieving physical reconstruction targets. By contrast, education achieved rapid progress in reconstructing schools and enrolling children, but the emergency response was not embedded ina policy framework, which i s now hampering implementation. 30At the macroeconomic level, education is one ofthe maindeterminants o f a country's aggregate output. 31 See Rohland and Cliffe (2002) for a discussion o f the lessons learned from the Timor-Leste reconstruction program. 52 Box 5.1: Poverty Reduction Strategy: Improving Basic Social Service Delivery The delivery o f basic social services, particularly quality primary and secondary education and primary health care, including preventive programs such as immunization and public health, is given priority inthe National Development Plan. 1 In education, the Plan presents programs aimed at increasing enrolment rates, particularly for children from poor families, improving the quality o f learning and teaching, through increasedprovision o fteaching materials andteacher training, and for adult literacy. 2 Inthe health sector the delivery of basic health services will focus on the needs of women and children by expanding preventive, curative and educational programs at the community level. 3 Provision o f drinking water in urban areas will, eventually, be made on a cost-recovery basis, whilst community ownership and operation i s already the norm inrural areas, with the State supporting initial investments. 4 The Government recognizes that it cannot be the sole provider o f basic social services if it is to achieve the desired levels o f coverage. The Government intends to strengthen partnerships with the Church, NGOs and the private sector, as well as communities, in education, health and rural water supply. Public-private partnerships in the education and health sector will be explored with a view to the expansion of cost-effective, quality services for all. Source: Planning Coinmission (2002). Main Messages Public Spending: Public spending in the post-independence period broadly supports service delivery functions, with education accounting for one quarter o f government spending (CFET) and health for 10 percent. The challenge will lie in maintaining adequate allocations to these highpriority sectors as external financing declines. Within the sectors o f health and education, there is concern about the share o f resources allocated to tertiary services, which largely benefit the non-poor. For example, hospital spending is regressive, but half of CFET health spending was allocated to tertiary care in FY2002. The NDP specifies capping it at 40 percent, a policy that the Ministry is implementing. Education: Education enrolments increased dramatically over the period 1998199-2OOOlO1 with a narrowing o f the gaps between rich and poor, and boys and girls. But the education sector faces several challenges: +++ Developing a sector strategy to guide decisions; Providingquality education to the large school-going population; Improving the efficiency o f the education system by reducing the number of ++ over-age children, drop-out and repetition rates; and Enrolling the quarter o f school-aged children who have never attended school. Establishing Portuguese and Tetun as languages o f instruction, with a sufficient number o f adequately trained teachers and appropriate pedagogic materials. Health: Health indicators are among the lowest inEast Asia and immunization rates, already low to begin with, declined between 1999 and 2001. Utilization rates remain low, and distance to the health facility is cited as an important reason for not seeking health care especially in rural areas. Providing affordable, accessible health services, especially for the rural poor, will be a key challenge. Limiting the expansion o f hospital services will release resources for primary health care in less well served rural areas. The Ministry of Health is well placed to focus on service delivery since it has developed a policy framework since the outset. 53 PUBLICSPENDINGFORBASIC 5.3 Government spending is a powerful vehicle for achieving national development goals. Budgetary spending in 2002 reflects the changing policy priorities with a strong focus on social sectors.33In spite o f large start-up costs for the establishment o f core public sector institutions, the social sector's share o f Consolidated Fund for East Timor (CFET) spending, the Government's recurrent budget, increased from 29 percent of total CFET expenditures in 2001 to nearly 40 percent o f expenditure in FY2002 (Table 5.1). Education accounts for one quarter o f CFET spending, and health for almost one-tenth. The aggregate pattern o f expenditure broadly follows the CFET structure. Table 5.1: Structure of Expenditure by Source of Funds and Sector (%) 2001 2002 CFET Total (all sources) CFET Total (all sources) U S $ Share US $ Share US $ Share U S $ Share (millions) (millions) (millions) (millions) Health 3.1 6 8.9 3 6.0 9 26.8 9 Education 10.1 20 42.3 16 16.0 25 47.7 17 Other Social 1.6 3 69.7 27 3.2 5 1os.2 30.6 11 Total Social 14.8 29 120.9 47 25.1 40 3 1 Total 51.3 100 258.2 100 63.4 100 286.6 100 5.4 The pattern o f sectoral expenditure is broadly supportive o f the immediate objectives o f the pre-independence period: establishment and support to the core institutions o f government and restoration of basic social services. Spending on education as a share of GDP and in per capita terms i s significantly higher than the average for low-income countries (Table 5.2) but comparable to countries o f similar income levels. FY2002 CFET spending on education amounted to US$19 per capita.34 When all external financing is taken into account, per capita spending increases to US$58. The situation i s similar inthe health sector, where CFET expenditures are higher than l o w income countries, both as a share o f GDP and on a per ca ita basis, with US$7.3 for CFET spending but lower than in middle income countriesY5. However, when all 32 This section draws on World Bank (2002b). Updated analysis on public spending is available in the forthcoming Public Expenditure Review. 33The fiscal year 2002 runs from July 2001 to June 2002. Public spending in Tiinor-Leste is disbursed through four channels - the Consolidated Fund for East Timor, which accounted for 22 percent of programmed expenditures in 2002; the Trust Fund for East Timor, which is the Government's capital program and accountedfor 19 percentof expenditures; bilateral projects, which accountedfor 39 percentof programmed expenditures; and the assessed contribution to the UN covering some international staff and some government operating costs, which accountedfor 19 percentof programmedexpenditures. 34 Education spending was $11 per capita in Vietnam and $8 per capita in Uganda, $13 per capita in Pakistan and $14 per capita in India. These are countries with similar levels of income. Expenditures in countries with higher incomes, show higher levels of spending $36 per capita inthe Philippines and $28 - per capita inSri Lanka, and $1I per capitain China. 35Health spendingwas $1 per capita in Vietnain,$5 per capita inUganda, $4 per capita inPakistan, $2 per capita in India, $12 per capita in Sri Lanka and $15 per capita in the Philippines, and $14 per capita in China. 54 sources o f financing are considered, health expenditures are significantly higher than all other low and middle income countries, at US$32 per capita and 7 percent o f GDP. In FY2003, CFET spending on health is budgeted at $9 per capita and i s $26 per capita for all sources o f funding. Table 5.2: InternationalComparators:Sectoral Expenditureas Percentageof GDP CFET Total Low Middle East Asia Per capita Share Per capita Share Income Income Pacific spending o f GDP spending of GDP Education 19.3 4.1 57.6 12.4 3.4 3.8 2.5 Health 7.3 1.6 32.4 7.0 1.3 3.1 1.7 Source: Ministry of Finance (Note: CFET datapom FY02 Revised Budget); WorldDevelopment Indicators 2000/01 (Note: Education and Defense data refers to 1997; Health to 1990-98). 5.5 The high levels o f external financing in both these sectors reflects inflows to support reconstruction and rehabilitation programs which will draw to a close inthe next two to three years36. As operating costs shift to the budget, the challenge will lie in ensuring adequate allocations to the priority sectors o f education and health. For example, for health spending to reach the international US$12 per capita benchmark, CFET spending would have to increase by 50 percent, with health's share o f the budget rising from 10 percent to over 15 percent. Consequently, choices will have to be made regarding the prioritization of programs in case the overall budgetary envelop does not increase in line with the reduction in external financing. As public spending is cut back, services with the greatest social returns and poverty reduction impact should be protected. Clearly, considerable attention should be paid to the forward planning o f expenditures inorder to assess the future cost implications o f policy decisions and ensure that sufficient resources are allocatedto the Government's stated policy priorities. 36 I t may be more valid to compare spending to other post-conflict countries in the years following the conflict. While it is complicated to make cross-country comparisons because o f data comparability, the data for a limited sub-set o f post-conflict countries shows that the share o f education and health inGDP and per capita spending levels are more comparable, though there is a range. For example, in Lebanon, per capita health and education spending were US$50, and the shares o f GDP were between 2-3 percent. Health spending in Nicaragua was US27 per capita, with a GDP share o f 6.8 percent, and education spendingper capita was US$13 per capita and the GDP share was 3 percent. Rwanda health spending was US$4.2 per capita andthe share in GDP was 2 percent. 55 EDUCATION Public SpendinginEducation 5.6 International experience suggests that one o f the key determinants o f the poverty- orientation o f spendinginthe social sectors i s the distribution o f spending between levels o f service delivery. Expenditures on lower level services, which are more accessible to the poor, tend to be progressive and spending on higher levels o f service tend to regressive. A similar pattern i s found inTimor-Leste. Table 5.3: EducationSDending bv Source Funds and Program. FY2002(YO) SectorA'rogram CFET TFET Bilateral Total Early Childhood Education 1 0 Primary and Secondary, o f which 77 90 29 57 Primary Education 54 90 3 39 Technical and Vocational 3 17 9 Non-formal and Language 1 6 3 University 10 47 25 Administration and Management 7 10 2 5 Total 100 100 100 100 Source: Ministry of Finance. 5.7 InFY2002, 54 percent of CFET expenditures are allocatedto primary education, 23 percent to secondary education and 10 percent to tertiary services (Table 5.3). When external financing37 i s taken into account, the share o f primary education in total spending drops significantly, to 37 percent o f total, while the share of tertiary education increases to 25 percent, both as support to the development of national institutions and scholarships abroad. Figure 5.1 plots the cumulative percentage o f beneficiaries against the cumulative percentage o f the population for primary and secondary schooling. Unit costs for public schooling by level have been calculated for the analysis in the public expenditure note. Since the unit costs o f primary schooling are constant, the distribution o f beneficiaries (students inprimary school) i s identical to the distribution o f the subsidy. Public spendingon primary education i s progressive, with the lower primary grades (1-3) more progressive than the upper primary grades (4-6) because the poor tend to drop out before completing primary education. As fewer poor children attendjunior secondary and senior secondary school, the better off students capture the benefits o f public spending on these levels. As a consequence, junior secondary school i s regressive and senior secondary school is more regressive thanjunior secondary education. The top quintile o f the population has 48 percent o f the senior secondary students. Tertiary education i s highly regressive, and the top quintile of the population has 65 percent of all tertiary students. Owing to the relatively high share o f education spending on secondary and ''It should be noted that the information on external financing is about commitments, which may differ quite significantly from disbursements. Informationon disbursements is not available. 56 tertiary services, the overall pattern o f education spending i s regressive, with the richest quintile benefiting from 35% o f education subsidies (Figure 5.1). Obviously, the regressive nature o f education spending would be even more marked if external financing i s taken into account, since this significantly increases the subsidy at university level. Bilateral spending, in particular, i s very skewed towards university education, with half o f all spending allocated to the sector. 5.8 The public sector i s the main provider o f schooling with 87 percent o f all students attending public school. Close to one in ten pupils attend religious schools and the remainder private secular schools. The poor are much more likely to attend public schools. Analyzing the incidence o f education by the type o f school (public, private or religious) shows that public spending on primary school i s progressive, but that on religious and private primary schooling i s not3*. The breakdown for public spending on junior and senior secondary schooling indicates that public schooling is distributed more equally than religious schooling for both levels o f schooling. 5.9 This analysis presents only the average incidence of spending. It is likely that increases in spending (marginal incidence) favor more the poor than the rich. However, this depends on a greater understanding o f the determinants o f enrollment. There is clearly a need for some reconsideration o f the allocation o f public spending between levels o f education, with a larger share o f spending allocated to primary and junior secondary education accompanied by measures intended to increase poorer children's enrollment in secondary and university education. The policy response favored by most countries in these circumstances i s to increase cost recovery from higher level services, while providing targeted subsidiesto support students from poor families. 38In 2001, there was a substantial degree of public subsidy for religious schools. Principals and filltime teachers tend to receive the same public salaries as their counterparts in public schools, while the schools continue to charge fees to finance part time or supplemental teachers to improve the learning environment. 57 Figure5.1: IncidenceofPublicSpendinginEducation a. Primary Schooling 1. Secondary Schooling 1 20 I 40 I 60 I BO 100 1 Cum. percentageof population (rank by per capita expenditme) Cum percentageof population (rank byper capita expenditure) c. Tertiary Education d. All Education 100 m :5 p 80 2Ee, 60 5 P L0 5 40 5 .Em - 20 V3 / --.I-- 0 ! I I I I I 0 20 40 60 80 I00 0 20 40 GO 80 10 Cum. percentage o f population (rank by per capita expenditure) Ciun percentage o fpopulat~oii(Iank by per capita expenditure) 58 AccomplishmentsinBuildinganEducationSystem39 5.10 Within about 18 months after the destruction, the school system, by and large, was rebuilt. By early 2001, about 86 percent o f classrooms were rehabilitated and useable. 922 schools were in operation, o f which 82 percent offered primary education, 11 percent junior secondary education, 3 percent, senior secondary education and the rest other types of education4'. 5.11 School Participation: Inaddition to rebuilding schools, school participation rates increased dramatically between 1999 and 2001. The largest increase in enrollment between 1998/99 and 2001/02 was among children between the ages of 5 and 14 (Figure 5.2). These increases in enrollment, especially by the poor, girls and rural children has resulted in narrowing the gaps in school participation rates between the richest and the poorest quintiles, boys and girls (Figure 5.3), and urban and rural areas. This is remarkable given the destruction in the violence of 1999 that affected 95 percent o f the schools and led to an exodus o f teachers. About 20 percent o f primary school teachers and 80 percent o f secondary school teachers, who originally come from other parts of Indonesia, left the country. Many migrants who had higher levels o f education and skills also left. This has created a shortage of teachers especially at the secondary school level. Figure 5.2: School ParticipationRates by Age, 1998/99-2000/2001 100 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Age Source: 2001 TLSS. 1 3-1998199 -I-2001/02 1 39This section draws on Chapter 5, Volume 11. 40Data are from the School Mapping Survey (2001). 59 Figure 5.3: School Participation by Quintile and Gender, 1999 and 2001 AMONG 7-12 YEARS OLD n i1 100 1 I 1999 2001 1 I Male Male Female Female Male Male Female Female Poorest Richest Poorest Richest Poorest Richest Poorest Richest Source: 2001 TLSS AMONG 13-15 YEARS OLD loo - i 1999 2001 I 90 -- n Male Male Female Female Male Male Female Female Poorest Richest Poorest Richest Poorest Richest Poorest Richest Source: 2001 TLSS Cost of Schooling 5.12 The reduction of the cost o f schooling by means o f the abolition o f school fees, PTA contributions, and requirements for uniforms i s likely to have contributed to increasing enrollments. The average monthly expenditure for attending public primary school in2001 was US$0.56 in contrast to US$1.55 in 1995 (in 2001 exchange rate and prices). Table 5.4 shows the distribution and levels o f school expenses for public primary schools in2001. In2001,the poorest quintile spent US$0.3 1 per month per student while the richest quintile spent US$0.91 per month. Tuition fees, Parent Teacher Association (PTA) fees and textbook costs were very low for the bottom four quintiles. The main 60 expenditure was on educational materials besides textb~olts.~'In contrast in 1995, monthly expenditure fees ranged from US$0.82 for the lowest quintile to US$2.67 for the richest q ~ i n t i l e ~ ~ . Table 5.4: Monthlv Exnenditure on Public Primarv Schools. 2001 (US Dollars) Tuition PTA Uniforms Textbooks Other Meals Extra Other Total education and classes materials transport Poorest 0.003 0.002 0.178 0.001 0.098 0.000 0.000 0.028 0.310 Q2 0.019 0.004 0.234 0.008 0.184 0.006 0.000 0.034 0.488 Q3 0.018 0.004 0.338 0.008 0.162 0.010 0.009 0.041 0.588 Q4 0.030 0.006 0.348 0.004 0.256 0.027 0.002 0.040 0.712 Richest 0.140 0.017 0.393 0.018 0.234 0.048 0.000 0.060 0.911 Note: All Rupiuhrdue.v/mnr rhc .SUIWJ~ were converted to USDiillum u.singon exchungerule of1 0,OOORupiuWUSDiillur. Source: 20111 TLXY. 5.13 The effect o f reducing the cost o f schooling i s reflected inthe regression analysis that shows that household resources (represented by nominal household expenditure) had a much weaker relationship with school enrollment in 2001 than in 1999 or 1995, after controlling for age, gender, and urbadrural residence.43For every 10 percent increase in household resources, enrollment rose by about 2 percentage points in 1995; 1.6 percentage point in 1999,0.28 percentage point in2001. ChallengesinEducation 5.14 While progress on reconstruction and school enrollment has been very impressive, the sector faces a number o f urgent concerns. The education sector is still developing a sector-wide policy framework to guide sector decisions for the implementation o f the National Development Plan. Such a medium-term strategy has to ensure that issues related to the quality of education, including curricula development, teacher training and management, are addressed, as they received too little attention duringthe initialemergency response. It also has to spell out specific policies to deal with three challenges:(i) demographic context; (ii) internal efficiency o f the education system; and (iii) challenge o f bringing the out-of-school children into the system. 5.15 Large SclzoolAge Population and High Adult Illiteracy: Timor-Leste i s a young nation and they are a young people with about 45 percent o f the population under the age o f 15. This large cohort o f school going children will put pressure on the education system. Inaddition, the adult population has very low educational attainment. Overall, 41The questionnaire in 2001 asked for expenditures on uniforms and other clothing, while data for 1999 report only uniforms. To the extent this might have caused any ambiguity, we do not discuss this category. 42 Fees accounted for 13 percent o f household spending on public primary education per capita o f the poorest quintilein 1995, PTA charges for 9 percent, uniforms for 52 percent, textbooks for 16 percent, and other instructional materials 10 percent. 43Expenditures are innominal terms because appropriate deflators for pre-2001 values were not available. For all other statistics using per capita expenditure, value are in real t e r m adjust for temporal and spatial price differences. 61 57 percent had no or little schooling, 23 percent only primary education, and 18 percent secondary education and 1.4 percent higher education. A larger share o f the younger generation has attended school compared to the older generation. About 72 percent o f over 30 years o f age have never attended school, whereas 31 percent o f the 19-29 year olds have not attended school (Figure 5.4). Within each age group, the rich are more likely to have attended school. The older the generation and the poorer they were, the least opportunity they had for education. As a consequence, adult illiteracy rates are high.44The implications are that the pool of well-educated persons who could be recruited to teach in the schools i s very small, posing a constraint to efforts to improve education quality. Figure 5.4: Ever Attended School by Quintile and Age 90 I Age 19-29 Age 30+ I 43 ~ Q 1LI]Richest 4 Source. 2001 TLSS I 5.16 Overage children: One measure of the internal efficiency of the system i s the match between grade and age o f the child. Table 5.5 shows the gross and net enrollment ratios for different levels o f schooling. There i s a big divergence between the gross and net enrolment rates. Although many students who were not enrolled in 1998/99 enrolled in school in 2000 and 2001, most of them attended lower grades in primary education. For example, in 2000/1, over 70,000 students enrolled in Grade 1, more than double the estimated number o f 6 year olds. More recent data from August 2002 MICS survey covering the school year 2001/2002 confirm the overage children phenomenon and show the primary net enrolment rate constant at 75 percent. 44Illiteracy rates are also highest among the poor and the older generation. Among children 13-15, these disparities across income groups have been removed. 62 Table 5.5: Gross and Net Enrollment Ratios 1998199 1999100 2000101 Net enrollment ratio Primary (7-12 yrs) 65 57 75 Jr. secondary (13-15 yrs) 24 21 22 Sr. secondary (15-16 yrs) 11 11 16 Secondary(13-17 years) 27 25 30 Gross enrollment ratio Primary (7-12 yrs) 90 85 I13 Jr. secondary (13- 15 yrs) 44 42 47 Sr. secondary (15-16 yrs) 22 22 29 Secondary (13- 17 years) 34 33 38 Source: 2001 TLSS. 5.17 This can be seen even more clearly in Figure 5.5, which plots the education profile for 6-18 year olds. As the figure shows, despite the increases in enrolment, a third o f all 8 year olds and a quarter o f the 9 year olds have never attended school; overall a quarter of all 6-18 year olds have never attended school. There are clearly a vast number o f over-aged children in the education system, as is manifested in the divergence inthe gross and net enrolment ratios. Figure 5.5: Enrollment Status of Children (School Year 2001/2002) 100 80 20 0 6 7 8 9 10 11 12 13 14 15 16 17 18 Age I Behind0Not inschool now 1 Source: 2001 TLSS. 63 5.18 The poverty dimension i s partly manifested through the grade-age misalignment. Poorer students are more likely to be older at any given grade. For example, only 10 percent o f poor students started Grade 1 at age 7 and 26 percent started at age 9. In contrast, 29 percent o f the children in the richest quintile started Grade 1 at 7 years. Through combination o f late enrollment and repetition, this pattern i s maintained across grades. Although more boys started Grade 1 at age 7 (22 percent versus girls' 14 percent), girls over-took boys by Grade 3 due to lower repetition rates. Rural children were by far worse off than urban children. Only 16 percent o f rural children started Grade 1at age 7, compared with 28 percent o f urban children. By Grade 4, only 6 percent were o f the right age, compared with 12 percent o f urban children. 5.19 High repetition and dropout rates. Table 5.6 shows the distribution o f repetition, promotion and dropout for each grade at the primary and secondary levels. Between 20- 25 percent o f children repeated and around 10 percent dropped out each grade inprimary education and junior secondary education. Compared to primary and junior secondary education, senior secondary education has lower dropout and repetition rates. This i s likely due to the fact that students who move up to secondary education level are more persistent and also tend to come from wealthier families who do not need their labor to support the family. The data also shows that girls tend to have lower repetition, lower dropout rates and higher promotion rates. A cohort flow analysis found that at this level o f internal efficiency, only 67 percent would reach Grade 4, 47 percent would complete Grade 6, 53 percent would drop out. On average, the dropouts would complete 4 grades. The cost per student for 6 years of primary education is about US$300. The cost per graduate, however, is twice as much because o f the repetition and dropout rates. Table 5.6: Repetition, Promotion and Dropout Rates by Grade (%) Primary Grades G-1 G-2 G-3 G-4 G-5 G-6 Males Repetition 20 24 25 25 25 23 Promotion 70 68 66 67 66 68 Dropout 11 9 9 9 10 9 Females Repetition 20 23 24 24 23 20 Promotion 70 69 68 68 69 72 Dropout 10 8 8 8 9 8 Secondary Grades G-I G-8 G-9 G-10 G-11 G-12 Males Repetition 23 25 24 9 10 11 Promotion 71 68 69 87 86 87 Dropout 6 6 7 3 4 2 Females Repetition 21 23 24 9 8 8 Promotion 75 70 70 89 90 90 Dropout 5 7 6 2 3 2 Source: School Mapping 2001. 64 5.20 This highlevel o f wastage has serious implications. From the educational point of view, the levels of skills acquired by those who have enrolled are likely to be low because about half o f them are not in school long enough to learn. From the fiscal perspective, this entails high levels o f spending without educating as many children as it should. The cost per graduate i s the key measure of efficiency o f resource use. The large number o f children who are still out o f school and a larger younger cohort that need to be educated inthe future are bearing the real cost o f inefficient use o f resources. 5.21 Out-of-sclzool Cizildren As indicated above, there are also a large fraction of children who do not attend school. Over a third (36%) o f all 6-14 year old children do not attend school, and 61 percent o f all 6-9 years olds do not attend school (Box 5.2). The reasons for never attending school are important to consider in developing successful strategies to reach education objectives. Among children 7-12, about 27 percent or so considered that they were not of the right school age (Figure 5.6). Demand side issues seem to be more o f the determining factor. About 32 percent o f the poorest and 26 percent of the richest had "no interest" iiischooling. O n the supply side, "school too far" i s a key factor cited for non-attendance. Among children aged 13-15, the lack o f interest i s cited as the major reason for never attended school for this age group. Box 5.2: Who are the Out-of-SchoolChildren? ~~ One third of all 6-18year olds did not enroll in the 2000/2001 school year, but a largepart of them do not attend due to young age. Over twofifths of those children are 6 and 7year olds, and over half oj them say they are not attending school because they are below the school age. Another twentypercent are 8 and 9year oldsfor whoni lack of interest is the main reason. For older children, lack of interest and the need to work at home or in agricultural work are important reasons. Distribution of Out-of-School Children by Age A s 6 7 8 9 10 11 12 13 14 15 16 17 18 6-18 YO 2 6 1 5 1 2 8 5 3 3 2 3 5 4 6 8 100 Just over halfthe children are males, and they are more likely to comefYom the lowestper-capita consumption quintile. Distribution of Out-of-School clzildren by Quintile Ouintile 1 2 3 4 5 National YO 27 21 21 21 10 100 Just under half of the out-ofschool children live in the Rural Center and another fifth live in the Rural East. In both these regions, then share among the out-of-school children exceeds their share of the school age population. Urban areas only contribute I 5 percent of out-ofschool children, lower than their share among thepopulation of school age children. Distribution of Out-of-School Children by Area Area Dili/Baucau Other urban Rural center Rural east Rural west National %ofschool age population 13 10 40 19 19 100 Yoof out-of-school children 8 7 46 21 18 100 I 65 5.22 Absenteeism. The number o f days that students were absent in 2000/01, among those who attended, i s a good indicator o f whether they find it worthwhile to go to school. Primary school students from poorest two quintiles have the lowest absenteeism within the last 3 months, while the top quintile has the highest. Twenty-two percent o f students of the poorest quintile and 46 percent in the students in the richest quintiles reported absenteeism in the three months o f the school year. The vast majority has no more than six days absence within a three month period. Similar patterns also hold for junior secondary and senior secondary education level. 5.23 The principal reason for absence from school at all levels was illness. Inprimary school, two-thirds o f students across all quintiles reported illness as the reason for absence, in secondary school the percentage increases to 78 percent, and for senior secondary students it i s 82 percent. At the primary level, distance from the primary school weighs more heavily for the lower four quintiles but does not affect the richest quiiitile at all. At the secondary school, distance to school and the need to work at home affect the poorest quintile disproportionately more. Figure 5.6: Reasonsfor Never Attending School, Ages 7-12 Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Total Below school age No interest Work 0Schooltoo far No teacherlsupplieskoo expensive Other ~ I Source: 2001 TLSS. Supply Side Factors 5.24 Access and quality o f schooling are important determinants in the decision to go to school. While quality i s difficult to quantify, the household survey asked about several dimensions o f it. 66 5.25 Language of Instruction: This issue poses a major challenge. Portuguese and Tetun are the official languages in Timor-Leste. Inthe education sector, the policy has been to introduce Portuguese progressively as the language o f instruction. Starting inthe 2000/2001 school year, instruction would be inPortuguese for students ingrades 1and 2, and it would be introduced as a second language for higher grades. This has created a number o f problems since only 5 percent o f the population speaks Portuguese, and consequently few teachers speak Portuguese. Only 158 teachers were approved for teaching Portuguese, 44 percent o f whom live in DiWBaucau. Figure 5.7 shows the language o f instruction in school in 2001 by quintile. For practical purposes in the transition, instruction i s divided almost evenlybetweenTetun and Baliasa Indonesia, with about 8 percent in Portuguese. Tetun i s more commonly used inthe schools attended by the poorest quintile, whereas a higher proportion of schools attended by the rich use Portuguese. Figure 5.7: Language of Instruction in School 100 I -- I-- -- 80 *3s 60 a 40 20 0 Poorest 42 43 44 Richest Tetun Bahasa 0Portuguese 1 Source: 2001 TLSS. 5.26 Sclzool Access. As indicated above, distance froin school and means o f transportation affects the decision to go to school. The data shows that the majority o f children who attend school walk there (94 percent or more inthe bottom four quintiles), and even three quarters of the richest quintile walk to school. On average, the one-way time to primary school i s 24 minutes, and it increases to 49 for junior secondary school. 5.27 School Infrastructure and Quality. Figure 5.8 shows the distribution o f textbook availability (inthree categories) by quintiles. It turns out that across all quintiles, about half o f the students do not have a complete set o f textbooks. The vast majority o f them obtained the books first from the school. The second most common way to obtain the books i s by purchasing second-hand books. About 81 percent o f students across all 67 quintiles have a desk to work on and chair to sit on but 20 percent do not. The majority found their teachers present either all the time (63 percent) or almost all the time (31 percent), but still about 7 percent experience teacher absenteeism. The average student- teacher ratio i s 62 for public primary schools in2000/200 1, with a wide range from 17 (in a school in Dili) to 243 (for a school in Turiscai sub-district o f M a n ~ f a h i ) ~ ~ . The variation across districts ranges from 40 in Covalima to 89 inAileu. Data for 2001/2002 show that with a significant increase in the number o f teachers particularly in underserved districts, this ratio has declined to 47, with a narrowing in the range across districts. The average ranges from 44 inErmera to 52 inManufahi. 46 The average hours o f homework per week can also be used an indicator o f school quality, and the rich tend to spend more time on homework. The quality o f teaching i s low and upgrading teacher skills i s a priority. Figure 5.8: Availability of Textbooks by Quintile Poorest 42 43 44 Richest Yes, complete Onlysome None 1 ~~ Source: 2001 TLSS. Determinantsof Enrollment 5.28 Enrollment decisions are determined by demand and supply side factors. We undertake a multivariate analysis o f the determinants of school enrolments to allow us to disentangle the effects o f different variables. A model o f school p a r t i ~ i p a t i o nis~ ~ estimated separately for primary school children (6-12 years) and secondary school age 45See East Timor HumanDevelopment Report (2002), UNDP. 46The district level numbers are based on a background paper (Timor Leste Education Sector Expenditure Review (2002)) for the World Bank Public Expenditure Review (2003). 47 A probit model is estimated, which takes the value 1 if the child i s in school and zero otherwise. Marginal impacts, which calculate the effect of the variable on the probability of participation, are estimated. 68 children (13-18 years). The regressions are run independently for urban and rural areas. The model includes as explanatory variables characteristics of the child (age, gender, mother tongue), education and age of parents, characteristics o f the household head (gender and occupation), demographic composition o f the household, per capita expenditure, access to school, and school quality variables as proxied by the percentage of children with textboolts, with chairs and desks, the language o f instruction, teacher absenteeism, and the cost o f schooling, which is proxied by the median cost o f public primary schooling inthe suco. 5.29 What determinesparticipation in for 6-12 year olds? As expected, age is a strong predictor o f participation and i s positively associated with being in school in both urban and rural areas. A twelve year old i s 45 percent more likely to attend school than a six year old. Girls are more likely to attend school in rural areas. Inline with evidence from other studies, children o f more educated parents are more likely to go to school - this i s especially true if either father or mother have some secondary education. Having a mother alive i s very important, especially for children inrural areas. Access to schooling matters in rural areas - having a primary school in the community increases the probability o f school participation by 11 percent, and a greater distance to school deters school participation. This effect i s stronger for girls. The cost o f schooling matters only for girls in rural areas. School quality variables generally have the right sign but are usually not significant. Household wealth i s a significant determination o f school participation for 6-12 year olds girls inrural areas only. 5.30 For secondary school participation, age remains an important predictor, but now older children are less likely to go to school. Girls are less likely to continue to go to school. Parents education i s important, especially the father's education. Whether the mother i s alive is particularly important in rural areas. Household size is negatively associated with secondary school enrollments in urban areas. Again, the presence o f a secondary school in the community enhances the probability o f participation by 23 percent. Household wealth enhances school participation particularly inrural areas. 69 HEALTH 5.31 Timor-Leste faces a host o f health challenges. The burden o f disease is largely due to communicable diseases, such as malaria, tuberculosis, respiratory tract infections and childhood infections. Maternal mortality is high- it is estimated to be inthe range o f 800-840 per 100,000 births, which in turn is linked to poor reproductive health. According to the MICS survey in 2002, only one in four women who gave birth had skilled personnel assisting them. Infant and child mortality rates are also estimated to be high (Table 5.7). Lack of safe drinking water and sanitation also contribute to poor health outcomes. One in two people live without safe drinking water and three in five people without sanitation facilities. Life expectancy i s low in the range o f 57 years. As pointed out in Chapter 3, immunization rates for children had declined in 2001 from already low levels since 1999. Against this backdrop, Timor-Leste seeks to restore access to basic services for all its people. This section first examines the impact o f public spending in health on the poor. It then turns to the pattern and costs o f health care utilization. Table 5.7: Health Indicators for Timor-Leste Timor Low Income Leste Countries 2002 2000 Infant mortality rate (per 1000 live births) 88 76 Male 99 n.a. Female 77 n.a. Under 5 mortality rate (per 1000 live births) 125 115 Male 142 n.a. Female 108 n.a. Maternal mortality rate* 420 n.a. Life expectancy ** 57 59 Male 56 58 Female 59 60 Anthropometrics *** Malnutrition (weight for age) 43 n.a. Stunting (height for age) 47 n.a. Wasting (weight for height) 12 n.a. **** Percentageof Data arefor I999 **Life expectancyjgures arefor 2001 moderately malnourishedchildren under 5years old. Sotirce. UNDP (2002), UNICEF (2002) and WorldBank S I M database 5.32 Figure 5.9 shows the percentage o f children aged under 1 year with no immunizations in 20014*. The MICS survey shows improvement during 2002, with increased coverage rates for all immunizations, but they remain significantly lower than the Ministry o f Health statistics. In general, immunization coverage i s lowest in the Rural East, and highest in DilUBaucau. Generally, children o f less than one year old in 48 The figures are much lower than the service statistics collected by MOH, which indicate 38 percent coverage for measles immunization under age 1 by December 2001. The latest MOH numbers show 53 percent DPT3 coverage for children under 1 year o f age. 70 urban areas are more likely to be immunized. Urban children are much more likely to have a DPT immunization (38 percent) than children in rural areas (22 percent). Complete DPT immunization i s even lower, with less than one inten children receiving the full set o f DPT injections. Complete DPT immunization rates doubled by August 2002. However, once again, there i s a significant discrepancy between these survey- based numbers and the Ministry o f Health statistics, which show 53 percent immunization for complete DPT immunization among the children under a year o f age. This pattern is consistent with experience from several countries, as administrative data are often more optimistic than household survey data as a result o f institutional incentives for over-reporting. In addition, administrative data may underestimate the number o f children to be immunized, adding to the bias o f over-stating immunization coverage. However, household survey data that rely on recall o f mothers for immunization data as inTimor-Leste may suffer fromrecallproblems leadingto under-reporting. The Ministry o f Health i s undertaking a review o f its Health Management Information System in2003 to look at the consistency o f data sources. The Demographic and Health Survey and the Health Seeking Behavior Surveys planned for 2003 will provide updated information on progress inimmunization. Figure5.9: ChildrenUnder OneYear Without Immunization 90 I 80 60 BCG Measles Polio DPT , Total Urban Rural I Source: 2001 TLSS. 5.33 Malnutrition i s a serious problem in Timor-Leste, with over four in ten children under the age o f five being moderately or severely malnourished in 200249. Wasting, as measured by weight for height, i s used as an indicator o f short-term access to adequate food and i s therefore affected by seasonal food availability. Over one inten children are moderately or severely wasted. Stunting, which is measured by height for age, is an 49Malnutrition is measuredby the proportionof underweight children, which is basedon weight for age. 71 indicator o f longer-term nutritional deficiency over multiple seasons. One intwo children are moderately or severely stunted. This evidence points to a widespread prevalence o f chronic malnutrition. PublicSpendinginHealth 5.34 The program structure used in the health sector allows a rough breakdown o f spending by level o f service.50 Tertiary - hospital - services are allocated just under half o f CFET health sector spending (Table 5.8). This pattern contrasts with the health policy objective, which i s to restrain hospital spending to the range o f 3540% o f CFET expenditures. The pattern o f spending i s even more skewed in favor o f tertiary services when TFET spending i s taken into account, rising around two-thirds o f combined expenditure. However, the bulk o f TFET spending on tertiary services are one-off payments for the rehabilitation and re-equipment o f hospitals, rather than on-going operational costs. More recent data for FY2003 from the Ministry o f Health shows that the share o f spending allocated to hospital services has declined to 41 percent5'. Health spending on hospitals i s also highdue to the cost o f expatriate doctors. Table 5.8: Health Spending by Source of Funds, Program and Levelof Service, FY2002, ("/) Program/Level of care CFET TFET Bilateral Total By Program Support Ongoing Service Delivery 68 17 25 32 Range and Quality of Service 25 66 35 48 Policy andManagement 8 16 40 20 Total 100 100 100 100 ByLevelofCare Primary and Secondary Care 51 34 39 Tertiary Care 49 66 61 Total 100 100 100 Source: Ministry of Finance and Ministry of Health. 5.35 The incidence analysis from the household survey shows that poor households are more likely to visit primary health facilities than public hospitals. Given the difficulty in deriving unit costs by level o f provision, the graph shows just the distribution o f beneficiaries across different levels o f facilities (Figure 5.10). Mobile clinics are the most pro-poor, while community health centers are neutral. Public hospitals are strongly regressive. Mobile clinics, used by 10 percent o f the population, are progressive. Larger This section is taken from World Bank (2002b). Unfortunately it is not possible to analyze bilateral projects by service level. "SeebackgroundpaperontheHealthSectorfortheTiinor-Leste PublicExpenditureReview. 72 unit costs for public hospitals would make the distribution of spending even more unequal. These data supports the reallocation o f public spending away from hospitals, as put forward inthe NDP,to lower level services inorder to benefit more the poor. Figure 5.10: Utilization of Health Facilities 100 ~ I 0 - / I_--__ I I 0 20 40 60 80 100 Cum percentage of population (rank by per capita expenditure) Health Access and Utilizations2 5.36 The Clznllenge of Building n Health System. As indicated above, Timor-Leste's health indicators are low. Inthe past, the population had limited interactions with health services. Low levels of health provider utilization were not a sign o f good health but o f the general political situation. A sharp decline inmonthly contact rates from 14.3 percent in1997 to 6.8 percentin 1998 is indicative ofthe instability anddistrust ofgovernmentin those years.j3 The challenge for Timor-Leste i s to build a health system that creates demand for health care, particularly preventive services, across all sections o f society, and to assure the supply o f quality services to meet that demand. 5.37 The household survey confirms the sober view o f health service utilization and access. It sets an early baseline against which to measure progress as the health system develops. TLSS data were collected one year after the first health project came into effect, and at a time when responsibility for service provision at the district level was being transferred from the international NGOs to the newly appointed Ministry o f Health district management teams. Recruitment o f health workers had only recently been completed, and vehicles including the motorbikes for the mobile clinics had just been delivered to the districts. The process o f engaging doctors from overseas was inits early stages, with no more than two or three already in position. The district management advisors had not yet been appointed. With implementation o f the directions already set out by the Ministry o f Health inits policy framework, inparticular the emphasis on basic 52The health analysis from the TLSS is basedon Nassim (2002). 53Saadah, Pradhanand Surbakti (2000). 73 services, and on a 60/40 division o f resources for primary and hospital care, tangible improvements should sooii be coming through. The health sector i s now manned; those with long-term responsibility for central and district health management are inplace, and doctors are being hired froin overseas to fill the gap before the return from training of doctors from Timor-Leste. Thus, despite slower progress in the early stages in meeting its reconstruction targets, early attention to developing a health policy framework and embedding the reconstruction efforts in that are paying big dividends now. The Ministry o f Health has been well placed in planning and in prioritizing its policy actions in implementing the NDP, and has been focusing its attention on service delivery. For a time, as the government re-establishes the health system, equity and efficiency objectives will both be served by an emphasis on providing preventive and simple curative services at the community level. 5.38 Morbidity and Treatment Seeking. The poor rate their health as slightly better than the n ~ n - p o o r .The poor are also less likely to report suffering a health complaint in ~ ~ the last 30 days, and, conditional on reporting a complaint, are also slightly less likely to seek treatment (Figure 5.1 1). Rural residents report slightly higher levels o f health problems thanpeople iiiurban areas, but are much less likely to seek treatment. The main reasons given for not seelting treatment in spite o f having a health complaint are that the complaint was not serious enough, or that the distance to the facility was too far (Figure 5.12). The poor are more likely than the non-poor to say the facility was too far away. Not surprisingly, access to facilities i s not a major issue for urban residents, though it is a deterrent to seelting health care in rural areas. There is no difference among men and women. Among the population reporting health problems, 40 percent are children under the age o f 14 years. Figure5.11: PopulationReportingHealthComplaintsinthe LastMonth -- Total Noii poor Poor Urban Rural Males Females With health complaiiits0With health coinplaiiits seeking treatment l Source: 2001 TLSS 54 This, at first, seems unexpected, but may reflect that fact that the wealthy have more interactions with healthproviders that couldraise reports of illness or morbidity. 74 Figure 5.12: Reasons for Not SeekingHealth Care DespiteHavinga Problem 100 80 8l 60 c 252 40 20 0 Total Non-poor Poor Urban Rural B N o t serious enough Health facility too far U N Otransport O H e a l t h workers not present m o t h e r reasons I Source: 2001 TLSS 5.39 Utilization of Health Facilities The survey asks individuals about their utilization o f different types o f providers (health facilities, private practitioners) for outpatient care inthe past month, the extent of self-medication and inpatient hospitalization rates inthe past year. As Figure 5.13 shows, the populationrelies on healthfacilities, with very little use of private practitioners for outpatient care. Self-medication i s used by 7 percent o f the population. Inpatient utilization rates are low at 1%. Only about 12 percent o f the population reported utilizing a health facility for outpatient care - treatment or preventive -inthemonthprecedingthesurvey,only4percentmorethanthoseseekingtreatment for illness (Figure 5.14). This low overall utilization, inparticular for preventive services, i s the main challenge for the health sector.j5 Large differences are found between the poor and non-poor.j6 There i s also a big difference between non-poor and non-poor females (14.9 percent and 9.1 percent respectively).j7 Contact rates are highest in Dili (14 percent). 55 The outpatient contact rates are similar to that prevailing in Indonesia before East Timor voted for independence-in 1997 the rate was 14.3 and in 1998, 6.8 percent (Saadah et al, 2000). 56 The differences are greater when considering the bottom quintile (8.5 percent) and the richest quintile (15.1 percent). See Nassim (2002). s7 See Nassim (2002). 75 Figure5.13: UtilizationRatesinthe Past30 days 12 10 8 m % Y 6 al B2 4 2 0 Facilities Private Traditional Self-medication Outpatient care practitioners practitioners Source: 2001 TLSS. 5.40 Figure 5.15 shows the distribution o f facilities/providers used for those who sought outpatient care inthe past 30 days.58 At present, three quarters o f the population utilizes public facilities for outpatient care. There i s much greater use o f private facilities among the non-poor, with over 29 percent using private or church facilities incontrast to 14 percent among the poor. Given the distribution o f the poor, and facilities, there i s a big urban-rural divide in the choice o f health facility. Urban residents rely on hospitals and private facilities. Half o f the rural inhabitants, in contrast, rely primarily on community health centers, while public hospitals and private facilities are usedby less than a fifth o f the population each. Mobile clinics are most important to the lowest quintile- 20 percent use this facility for outpatient treatment. As the health system i s being developed, mobile clinics are being used in place o f health posts to give time to judge whether improved transport, and/or increased demand for better staffed and equipped community centers will make healthposts increasingly unnece~sary.~~ 58The pattern o f utilization for those seeking treatment when illis similar to the outpatient utilization rates, hence, inthis volume we only report the former. 59However, the new mobile clinics are delivered by nurses on motorbikes, rather than the vehicles often with doctors operated by the NGOs. Whether the mobile clinics provided on motorbikes will be as important inproviding health care to the poorest groups, and inrural areas generally is an open question. 76 Natioiial Noli-poor Poor Urban Rural Male Feinale Source: 2001 TLSS. Figure 5.15: Distribution of Type of Facility for Outpatient Care National Non-poor Poor Urban Rural Community health center Private hospital/clinic Source 2001 TLSS 5.41 This reliance on government services, and the existing socioeconomic differences in using government services, underscore the importance of the Government's commitment to equity objectives - to reaching the poor, and those in rural areas - ifthese differences are not to become magnified as the health system develops. An important signal i s provided by the urban use o f public hospitals for outpatient treatment - though the hospitals tended to be the base o f NGO operations, and this pattern may be changing. The challenge for policy is to constrain the use o f hospitals for services more appropriately offered in health centers, and to resist the demands o f urban populations, 77 typically politically more strongly organized, for resources for hospitals at the expense o f primary health care inless well served rural areas. 5.42 Table 5.9 shows the reasons people report for outpatient visits. Just over half o f the visits are for medications, with medical check-ups being the second most common reason given (28 percent). The poor are more likely to say they attended for medications than the non-poor. There was little demand for preventive services, except among the better-off. Only 2.5 percent o f outpatient visits were for pre- or post natal care or delivery. Table 5.9: Reasonfor OutpatientVisit amongNon-poor and Poor Non-poor Poor Total Iinmunization 0.4 0.5 0.5 Medical check-up 30.5 21.1 27.8 Consultation 10.8 6.5 9.6 Medications 47.5 63.8 52.2 Injections 5.1 5.7 5.3 Treatment for itijury/illness 2.9 1.5 1.2 Prenatal care 1.1 0.5 0.9 Delivery o f baby 0.4 0.0 0.3 Postnatal care 0.2 0.0 0.1 Other 1.1 0.5 0.9 Source: 2001 TLSS. 5.43 Costs: Transportation and Services While public health care i s nominally free, there are costs associated with getting outpatient care. The costs include travel costs (both time and money) and costs for medicines or services, in some cases even while visiting public facilities. On average, individuals pay a little less than US$2 per person per month for the monetary costs o f health care. Monetary costs are highest in Dili and the Rural Center and lowest inthe Rural West. Among those visiting health facilities, the average one way travel time to the health facility i s 62 minutes (Figure 5.16). The most common mode o f transport was walking, for two-thirds o f the population. This could be a constraint to seeking care when individuals are sick and unable to walk or intaking small children for preventive or curative care. Travel times vary by location and type o f facility, being particularly highinthe rural Center. 78 Figure 5.16: One Way Travel Time to HealthFacility 90 1 80 70 60 + 50 $ '2240 30 20 10 0 Source: 2001 TLSS. us z 5.44 Average transport costs are less than a dollar, though two thirds o f the population does not pay for transportation cost to get outpatient care. Among those that pay, the cost is twice as high". There are also costs incurred for services and medications - on average US$l, with the non-poor paying US$1.16, and the poor paying 65 cents. They indicate both a willingness to pay, probably at least for medicines. About two fifths pay for medical care and this varies by the type o f provider. Eveninpublic health facilities, a quarter pay for health care, and over halfpay at mobile clinics. 5.45 With regard to medicines, there has been some concern about the extent o f self- medication. TLSS indicates that nearly 10 percent o f the non-poor and 2.5 percent o f the poor had bought medicines without prescription in the past month, chiefly from kiosks and street vendors, not from pharmacies. 6oExcludestwo observations with very hightransport costs. 79 Figure5.17: PercentagePayingfor MedicalSenices by Type of OutpatientFacility Public Coininunity Public Mobile Private Church National hospital health health clinic hospital/ clinic centre post clinic Source: 2001 TLSS. 5.46 Maternal Health. Only 8 percent o f currently married women are using contraception. These statistics are similar to the rates from the SUSENAS data, which show low contraceptive use inTimor-Leste with a rate o f 11 to 13 percent among married women 1997-1999. There are only slight urbadrural differences with 9 percent currently using in urban areas and 7.7 percent in rural. The most common reasons for not using contraception were that the woman "wants children" (3 1 percent), religious beliefs (28 percent), and fear o f side effects (13 percent). Similarly low rates are found in the MICS survey. The MICS data also point to the high fertility levels - women have more than seven children, among the highest inthe world. Furthermore, the MICS survey highlights low levels o f antenatal care and deliveries attended by skilled personnel. As previously noted, only one in four women are assisted by a skilled medical practitioner during delivery, one half i s assisted by family members or relatives, and one in five had no assistance at all. SUMMARY AND POLICY ISSUES 5.47 Public spending in the post-independence period broadly supports service delivery functions, with education accounting for one quarter o f CFET spending and health for another 10 percent. CFET per capita spending is US$7.3 for health and US$19 for education, while overall per capita spending for 2002 on education equals US$58, and US$32 for health, substantially more than most low and middle income countries. While these highcosts reflect capital expenditures to set up education and health systems, at this stage, Timor-Leste i s in the remarkable situation o f allocating large sums on social sectors. One issue lies in ensuring sustainability in the future. As external financing 80 declines, operating costs will shift to the budget, which will pose a challenge in maintaining adequate allocations to priority sectors o f education and health. Whether these levels o f public spending can be sustained in the future needs to be assessed. Cost- recovery measures should be considered for those who can afford to pay, while maintaining affordable services for the poor. 5.48 A second aspect o f this is inallocating public resources for services that reachthe poor. Within education and health, there i s some concern about the proportion o f spendingallocated to tertiary level services, which largely benefits the rich. Community health centers, the facility most used by the population are by and large neutral, while the incidence o f hospital spending i s regressive. Implementing the policy directions laid out inthe National Development Plan which caps this share at 40 percent is a priority and recent data show that the share o f CFET spending on hospitals i s in this range. Public primary schooling is progressive in that the share o f poor beneficiaries exceeds their share inthe population, while secondary and tertiary education are regressive. 5.49 In education, by 2001, the scliool system was largely rebuilt. Enrollments increased dramatically between 1998/99 and 2000/200 1,especially for the poor, girls and rural children, which led to a narrowing o f the gaps between the poorest and richest quintiles, girls and boys and urban and rural areas. The reduction in the private cost of schooling was significant and likely contributed to increasing enrollments. Despite the accomplishments, implementation i s hampered by the lack o f a policy framework for the sector. While access expanded dramatically, the quality o f education, including curricula development and teacher training, received far less attention. Inaddition, education faces several challenges - (i) the sector has a large school going population and high adult illiteracy; (ii) the internal efficiency of the education system is low, with a large fraction of overage children in the system; and high repetition and drop out rates. The cost per student for 6 years o f primary education i s about US$300. The cost per graduate, however, i s twice as much because o f the repetition and dropout rates; and (iii) a quarter o f 6-18 year olds have never attended school. 5.50 The National Development Plan's focus on improving the access, especially for the poor and improving the quality o f learning and teaching in primary and secondary education i s appropriate. Despite the narrowing inthe enrolment gaps between the poor and the rich, the out-of-school children are more likely to be poor. The challenge lies in developing a sector-wide policy to guide implementation o f the strategy, and in prioritizing actions and costing them to achieve the objectives in education within the medium term expenditure program. For example, there are trade-offs between expanding access and increasing quality, between expanding access and providing free schooling, and between expanding primary schooling and having an adequate supply o f teachers in the future. Supply side issues, which affect demand for education, loom large. There is an insufficient supply o f adequately trained teachers and lack o f appropriate learning materials. These trade-offs will need to be assessedinthe context o f the prioritized action plans that the Ministry i s preparing currently. Given the large cohort o f school age children, financing the education needs in a sustainable manner in the future is a key priority to consider now and these should include measures to improve the internal efficiency o f the education system. 81 5.51 Health outcomes in Timor-Leste are among the lowest in East Asia. Immunization, which i s one o f the most cost-effective health measures, had abysmally low coverage in 2001. Health utilization rates, however, are also low. Only 8 percent sought health care despite having a health complaint. While the lack o f seriousness was one major reason, distance to the health facility was cited by two-fifths o f the population as the major reason for not seeking care. This is particularly a concern in rural areas. Public health facilities are the main provider o f health services, especially in rural areas. In urban areas, the private sector plays an important role, with over two-fifths using private facilities or church clinics. There are some costs associated with getting outpatient care. O n average, the total cost to households o f visiting a health facility isjust under US$2 per person per month. Only one third o f individuals pay for transportation and for medical services. The poor, on average, pay half the amount paid by the non- poor. The amount still represents a larger share o f the expenditures o f the poor. Payments for services are higher in private and church facilities, and even public facilities charge fees. 5.52 The National Development Plan stresses the delivery o f basic health services, particularly for women and children focusing on expanding preventive programs. These are priorities. The Ministry o f Health i s also benefiting from the attention paid in the early days to develop a coherent health policy framework and embedding the reconstruction actions within it. The challenge lies in prioritizing actions within the expenditure constraint and in assessing tradeoffs in different policy priorities in health. But, given the strong foundation, the Ministry of Health was well-positioned in the prioritizing and sequencing exercise undertaken as part o f this years planning process. RESEARCH ISSUES 5.53 Ineducation, analyzing the determinants o fenrollments, repetitionrates anddrop- out rates will help in devising strategies to improve education outcomes. Understanding the factors contributing to improving learning outcomes will be critical in focusing resources on the most cost-effective inputs. Education sector work that is currently ongoing will address many o f the issues identified here: teacher training, language o f instruction and the strategies for expanding secondary school graduates. In addition, it will provide guidance on the appropriate role of public finance in education, with a medium term perspective taking into account budgetary and capacity constraints. 5.54 Inhealth, utilization rates for health services are very low. More work is required to understandwhether this is linked to lack o f information, distrust, fees, limited access, or other factors. Such analysis will have to take into account the effects of the major administrative reorganization inthe health sector undertaken since the time of the survey. A health care financing study can examine options for providing affordable health services for all ina sustainable manner. 82 6. HOUSEHOLDSECURITY 6.1 Indeveloping countries, households are exposed to many unforeseen changes of events threatening their livelihoods. Such vulnerability to poverty i s an important dimension of deprivation, and can itself become the cause o f poverty. Inthis chapter, we look at two aspects o f vulnerability that are emphasized in the Government's Poverty Reduction Strategy (Box 6.1). The first part looks at groups that are especially disadvantaged in dealing with adverse circumstances. The second section turns towards the temporal dimension o f vulnerability andpresents evidence on food security. Box 6.1: Poverty ReductionStrategy: Security ~~ Security has been a major concern of the people of Timor-Leste over the past twenty five years, particularly interms of security of personand property, but also interms of food security and security of livelihood, and protection against natural disasters. The National Development Plan outlines the broad framework of a social safety net for the vulnerable. The emphasis will be on partnership, with the Government supporting community, NGO and Church initiatives. The NDP highlights a number of key areas o f concern. Disadvantaged groups, including widows and orphans of the resistance, veterans, child- soldiers andthe traumatized, deserve particular attention. Overall food availability inthe country i s to increase and food security at the householdlevel to be improved. The distribution of food to the vulnerable during times of food stress should be continued, together with employment inpublic works, particularly road maintenance,as a self-targeted form of assistance. Since women and children are particularly at risk, this includes school feeding and targeted provision of milk and food supplements for pregnant women and young children. Insecurity of livelihood or employment, caused by the lack of recognition of ownership and tenancy of agricultural land, or lack of access to resources such as forests or other community lands, is to be addressed. Programs, aimed at improving economic participation, will target those affected by economic shocks, including those laid off as UN mission and supporting services wind down. Source: Planning Commission (2002). Main Messages The analysis shows that female-headed households, widows, and parentless children experience severe hardship. Designing appropriate policy responses to provide support to these disadvantagedgroups will be an important element of any poverty alleviation strategy. Food insecurity is widespread in Timor-Leste and i s aligned to the harvest cycle, with highest levels of insecurity experienced between November and February at the end of the maize harvest and before the rice harvest. Interventions to improve the availability of food duringthe course of the year are critical to improvinghouseholdwelfare. 83 DISADVANTAGED GROUPS 6.2 In many Asian countries, some groups are excluded from the benefits of economic developments. Parentless children, elderly, widows, and women are often found to be vulnerable, as economic, social, cultural, and institutional barriers combine to result in low living standards. These groups depend particularly on cooperation from others. Identifying disadvantaged groups i s a first step towards developing support strategies that prevent poverty, marginalization, and social disintegration. Inthis section, we take a closer look at the TLSS evidence on social and economic inequities experienced by specific groups". We use demographic and family characteristics to categorize the population, and investigate whether particular household groups, or segments within a household, are especially disadvantaged. Gender 6.3 Gender i s an important aspect in the debate on development. While quantitative data may be mixed on the link between poverty and gender, there i s growing evidence that societies that discriminate on the basis o f gender tend to experience more poverty, slower economic growth, and a lower quality o f life than societies in which gender inequality is less pronounced. In all countries, but particularly in the poorest, giving women and men the same rights - allowing them equal access to education, jobs, property and credit, and fostering their participation in public life - produces positive outcomes, such as decreased child mortality, improved public health, and a strengthening o f overall economic growth. 6.4 Attempts to estimate the number o f women living in poverty has generated a considerable amount o f debate around the world. The main stumbling block i s the lack o f an acceptable indicator for gender comparisons. The basic poverty measures inthis report are based on household resources, and incorporate the essentially arbitrary assumption o f equal distribution within the household. They do not capture any female poverty deriving from intra-household inequality. With this caveat in mind, it i s nevertheless useful to ask whether, under this "conservative" assumption, there i s evidence for gender bias in poverty. This part draws on Chapter 6, Volume 11. 84 Table 6.1: Gender-Age Groups and Welfare ("A) 0 to 6 ?to 14 15 to 49 50 or older Female Male Female Male Feinale Male Female Male Poverty Headcount 42.6 44.7 49.1 45.7 36.0 35.3 31.1 33.1 Poverty Gap 13.0 13.5 15.1 14.1 10.8 10.2 9.3 8.8 Severity 5.5 5.6 6.4 6.0 4.5 4.1 3.8 3.3 Immunization BCG 52.2 55.8 Polio 57.9 61.1 DPT 53.3 57.0 DPT3 8.3 9.1 Measles 51.7 49.0 Vitainin A 6.5 7.6 Health No health complaints last month 73.2 72.7 86.8 87.4 79.0 83.0 61.3 58.2 Subjective health status (1 to 5) 4.0 3.9 3.9 3.9 3.6 3.6 Education Net Primary EnrollmentRate 63.4 60.5 Net Primary ClassEnrolhnent Rate 19.2 17.0 Schooling 82.1 77.7 47.9 66.2 2.9 12.8 Grade completed (1 to 6) 1.9 2.3 1.0 1.2 Literacy 49.8 67.3 6.1 14.3 Subjective Welfare Happiness(1 to 5) 3.16 3.18 3.07 3.15 Changeinliving standard since violence (1 to 3) 1.80 1.81 1.88 1.83 Economic status(1 to 9) 2.40 2.41 2.16 2.36 Changein economic status since violence (-8 to 8) 0.12 0.15 0.08 0.07 Power status(1 to 9) 3.78 3.90 3.45 3.69 Change inpower status since violence (-8 to 8) 2.14 2.24 1.86 2.00 Note: This table is lakcn,frr,m Tables6.2 atid 6.3 in VolirnreII. Source: 2001 TLSS. 6.5 The evidence on gender bias inTimor-Leste i s mixed. First, women do not live in poorer households than men (see Table 6.1). Household demographics differ little within each age category, implying that this result i s robust to changes in equivalence scales. This finding comes however with a strong caveat. TLSS provides no information on the gender allocation o f consumption within the household. More research into intra- household distribution i s required to conclude that this household-level finding translates into an absence o f gender bias at the individual level. 6.6 In addition, we also find little systematic differences across gender-age groups. Immunization rates are higher for boys, and education indicators better for girls, but the gaps are statistically insignificant. For adults, male educational standards are generally higher, which says more about gender inequalities in the past than today. Finally, subjective indicators tend to rank men higher than women, especially for those 50 years or older, but the differences are small, and the evidence on changes since the violence inconclusive. 85 Female Headship 6.7 The analysis so far focuses on characteristics o f gender-age groups cutting across households. It does not capture deprivations linkedto particular households features. One salient household characteristic i s the gender o f the household head. In this section, we focus on differences inwelfare between male and female-headed households. We want to explore, whether, as a result o f economic and perhaps cultural constraints, female-headed households experience lower welfare than male-headed households. 6.8 In Timor-Leste, cultural values in general, and traditions of family life specifically, are primarily based on catholic beliefs. In this context, female headship arises for two main reasons. First, some families have lost their male breadwinner as a result of the years of violence during the Indonesian period and the time o f the referendum. Second, women have a higher life expectancy than men. In consequence, almost all female heads are widows.62 6.9 Both factors suggest that female-headed households have fewer household members than male-headed households, while the second aspect implies that female heads are on average older than male heads, and in turn are likely to have lower child shares. Overall, more than one in seven household heads are women. Female headed households are indeed smaller than male headed households (3.1 members relative to 5.3 members), so in terms o f population, about one in ten individuals live in households whose head i s a woman. For male headed households, seven in ten individuals have a head who i s younger than 50. The corresponding number for female headed households is only 5 in 10. The child share inmale headed households i s on average 20 percent higher than infemale headed households. 62 However, one third of all widows are not heads of household. In Chapter 6, Volume 11, we analyze widows separately. 86 Table 6.2: Female Headship and Welfare (%) 0 to 6 7 to 14 15 to49 50 or older Female Male Female Male Female .Male Female Male head head head head liead liead head head Poverty Headcount 31.9 44.5 43.8 47.7 26.2 36.6 19.4 34.9 Poveity Gap 8 9 13.6 13.7 14.6 6.4 10.9 5.6 9.8 Severity 3 8 5.7 5.8 6 2 2 4 4.5 2.2 3.9 Immunization BCG 392 54.9 Polio 40.3 60.6 DPT 40 7 56.1 DPT3 5.6 8.9 Measles 37.5 51.1 Vitamin A 2 4 7.3 Health No health coinplaints last month 70.7 73.1 83.7 87.5 77.2 81.4 55.3 60.7 Subjective health status (Ito 5) 3.9 3.9 3.8 3.9 3.3 3.6 Education Net PrimaryEnrollment Rate 57.1 62.4 Net PrimaryClass EnrollinentRate 12.8 18.6 Schooling 76.2 80 2 53.5 57.3 2.3 9.2 Grade completed (Ito 6) 2.1 2.1 1.0 1.1 Literacy 54.1 58.9 5.5 11.3 Subjective Welfare Happiness(Ito 5) 3.03 3.19 2.97 3.14 Change inliving standard since violence (Ito 3) 1.85 1.80 1.94 1.83 Economic status (1 to 9) 2.17 2.43 1.87 2.35 Cliange in economic status since violence (-8 to 8) 0.22 0.13 .0.04 0.10 Power status (Ito 9) 3.68 3.86 3.26 3.64 Change inpower stahis since violence (-8 to 8) 2.12 2.20 1.85 1.95 6.10 The welfare comparison i s shown in (Table 6.2) As before, the findings on poverty are subject to the caveat o f lack o f information on intra-household distribution. Poverty i s between one third to one half higher for male headed households. However, as discussed in the previous paragraph, male and female headed households differ both in size and composition. Especially, allowing for economies o f scale can reverse the ranking as male headed households are one third larger than female headed households. We conclude that the poverty rankings o f male and female headed households are not robust to changes inequivalence scales across a plausible range. 6.11 With regard to other dimensions o f well-being, including education, health, and subjective well-being, male-head households are consistently better o f f than female- headed households. In male-headed households, children under 6 have significantly higher immunization rates, and children o f school age report less health complaints and better educational indicators. The same holds for both prime age adults and the elderly. Finally, the subjective welfare indicators suggest that adults in male-headed households feel to have a higher economic and power status. Better welfare in male headed households may not be linked to gender bias. It could simply reflect that female-headed households are deprived o f one important breadwinner. 87 ChildrenWithout Fatheror Mother 6.12 The counterpart o f female headship, from the point o f view o f the children generation, is boys and girls without living fathers. In any country, one o f the most disadvantaged groups is children without parents. In Timor-Leste, as a legacy o f a long history o f violent conflict, over one inten children have only one or no living parent. The largest group is the children without fathers, accounting for four in five o f the children without at least on parent.63This part discusses the welfare o fparentless children. 6.13 A simple way to identify the impact o f having lost a parent is to compare the welfare o f children with and without fathers and mothers. We separate three groups: those with both parents alive, those whose father has died and whose mother is still alive, and those whose mother has died and whose father i s still alive.64 The categories represent 89 percent, 6.5 percent, and 3.5 percent o f all children under the age o f 15, re~pectively.~~ us first consider the two largest groups, children with both parents Let alive versus those with a living mother and a deceasedfather. Table 6.3: Child Welfare and Parental Livinc Status (YO) Father and Father dead, Father alive, mother alive mother alive mother dead Poverty Headcount 45.3 51.2 42.5 Poverty Gap 13.8 15.7 13.6 Severity 5.8 6.8 5.7 Education Schooling 66.6 63.6 57.9 Enrolled in age-specific school 63.6 52.8 51.8 Enrolled in age-specific grade 19.4 10.2 11.9 Immunization BCG 55.0 27.8 50.6 Polio 60.7 36.6 50.6 DPT 56.0 47.2 39.9 DPT3 9.0 4.2 0.0 Measles 50.7 47.1 44.6 Vitamin A 7.4 0.0 1.8 Health No health complaints last month 40.5 31.6 34.3 h'ole: This ruble i.s rukenjroi11 7iihle,s 6.8 und 6.9 in ViiIiiuie 11, (`hildren ure d~$izedu.su p / 13y e w s oryounger, l ~ i ~ ~ i i i i i ~ i zrure.r~ i n u ~ irq2r lo children under 5 yeur.~of oge. Source: 21101 TLSS. 6.14 Fatherless children live in households without the typical main breadwinner, so we expect high poverty. This i s indeed the case (Table 6.3). Child poverty rates are Out of the childrenwith both natural parents alive, more than nine in ten o f these children live together with both o fthem, and almost all of them with at least one ofthem. 64Among the children below the age of 15, 19 in 20 childrenhave a livingmother. 65We do not have a sufficient number of observations on orphanedchildren (1.O percent o f all children) to present reliable statistics. 88 around 15 percent higher for those without a living father than for those where the father has deceased. This ranking i s robust to changes in the equivalence scale. In terms of education, we find that children without fathers are worse off than children with both parents alive: they are less likely to have received any schooling; have a lower net enrollment rate, both for primary school as a whole and for each primary school grade. With regard to child health and immunization, children with both parents report fewer health complications during the last month, and children less than 5 years o f age with living fathers are betterimmunizedthan other children. 6.15 Turning to children with a living father and a deceased mother, there is little difference in terms o f poverty compared to children with both parents alive. However, education, immunization, and health indicators show consistently that children without mother are worse off than those whose both parents are alive. 6.16 This examination i s preliminary only and calls for more research to uncover the impact o f child care arrangements on the welfare o f parentless and orphaned children. Nevertheless, these numbers suggest that the presence o f fathers increases welfare for the children involved. In addition, we find that with regard to education and immunization, parentless children, being with a deceased mother or father, are consistently worse o f than children with both parents. FOOD SECURITY 6.17 Poverty means more than inadequate consumption, education, and health. It also means dreading the future. Living with the risk that a crisis may descend at any time, not knowing whether one will cope, i s part o f life for poor people. Poor people are often among the most vulnerable in society because they are the most exposed to a wide array o f risks. L o w income implies poor people are less able to save and accumulate assets, which inturn restricts their ability to deal with a crisis when it strikes. Poor people have developed elaborate mechanisms o f dealing with risk, some o f which offer short-term protection at long-term cost, preventing any escape from poverty. 6.18 Risk is a pervasive characteristic o f life in developing countries. While it is beyond the scope o f this report to discuss the multiple sources o f vulnerability comprehensively, TLSS allows us to explore one issue o f vulnerability in more detail: food security." Prevalence 6.19 Food security refers to assured access to enough food at all times for an active and healthy life. Ideally, we would want to base empirical evidence on data collected over the entire course o f the year, covering the different stages o f the agricultural season. Yet, TLSS surveyed households only between late August to early December, and did not measure dietary intake or malnutrition. However, the survey included a range o f 66The chapter draws on Chapter 7, Volume 11. 89 questions on the perception o f food security. While these subjective indicators raise questions with regard to the comparability o f responses, they nevertheless give instructive pointers bothto the extent and pattern o f food insecurity. Figure 6.1: Household Food Security by Month Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec ,Sucosreportingharvestofrice Sucos reporting harvest o f maizeI ~ ~%/ ' Populatioii reporting low food security Source: 2001 TLSS and Suco Survey. 6.20 Subjective assessments o f food adequacy suggest that food insecurity i s widespread. Close to nine in ten persons experience inadequate food provision at some point duringthe year, while fewer than one intwo have too much food during any month inthe year. Foodsecurity is closely tied to having enoughrice andmaize. Foodshortages are aligned with the harvest cycle, as shown in Figure 6.1. They are greatest during November and February, at the end o f the rice harvest and before the maize harvest. 6.21 Major urban centers typically have access to just enough food all throughout the year, while other parts o f the country face greater fluctuation in food availability, and experience food shortage about twice as often as food excess. One interpretation o f this evidence is the difference in capacity to keep consumption constant over the year. The greater reliance on non-agricultural income sources allows households inDiliand Baucau to keep consumption constant at an adequate level across the year. Two factors could explain this ability o f consumption smoothing. First, urban incomes are likely to be less variable as they depends less on the agricultural seasons. Second, as they receive a higher share o f income in cash, city dwellers may be able to engage more in saving and dis- saving o f income. 90 Food Security and Poverty 6.22 Agriculture i s o f overwhelming importance for living standards. About seven in ten persons live with heads o f households who work on a household farm, and over three quarters are with heads whose main occupation i s farming. Given this dependence on agricultural seasons, what i s the implication o f the intra-year cycle o f food security for poverty? 6.23 The subjective food security indicators showed that food availability was closely aligned to the harvest cycle. August was the last month o f the plentiful season, and lack o f food became more severe from September untilthe end o f the year, and had its peak in January. O n the basis o f this pattern, we would expect poverty to broadly show an increase from early in the survey to the end o f the survey. In Figure 6.2, we display the national pattern, linking the average poverty headcount to a count o f the days in the survey. We find indeed a strong dependence o f the poverty headcount to the timing o f the interview. Fewer than one in ten persons live below the poverty line at the beginning o f the survey. The share of the poor rises continuously until about three months after the start o f the survey, or about mid-November, peaking at about 45 percent. This share then remains fairly constant duringthe last month. Figure 6.2: Poverty and Interview Date 60 - ___ 15 40 65I 90 115 # Days InterviewPeriod Source: 2001 TLSS. 6.24 This strong evidence for seasonality of poverty raises an immediate question. In the analysis o f the poverty profile, we argue that about two fifth o f the population live below the poverty line. In view o f the intra-year fluctuations o f living standards, this estimate is specific to the survey period. H o w representative i s therefore this poverty rate o f 40 percent for the year as a whole? In the absence o f information o f consumption behaviour throughout the year, we have to rely on subjective food security for a rough assessment. We compare the average value o f food security for the survey period with the 91 annual average. Taking as weights the percentage shares o f interviews conducted in August, September, November, and December, we calculate that the share of not having enough food for the survey period i s 30 percent. The annual average for this variable is 34 percent. Overall, this comparison suggests that the "survey" poverty rate i s fairly close, and possibly slightly lower, to the "annual" poverty rate. Copingwith Food Shortage 6.25 What happens when a family i s faced with a risk o f food shortages? And how does a household respond to a food crisis? Farmers have always been exposed to weather risks, and for a long time have developed ways o f reducing, mitigating, and coping with these risks (Besley 1995, Dercon 2002). Traditional risk management covers actions taken both before ("ex-ante") and after ("ex-post") the risky event occurs (Siege1 and Alwang 1999). These strategies are ofteii costly, as they lower vulnerability inthe short term at the expense o f higher vulnerability over the longer term. For example, diversifying crops may help the farmer to reduce the exposure to complete crop failure, butmay also contribute to low productivity andhence to keeping his family inpoverty. 6.26 I s food security associated with more ex-ante coping? In DiWBaucau, the distinguishing feature of households in terms of food security i s being employed in the non-agricultural sectors. By contrast, outside Major Urban Centers, dependence on agriculture is almost universal, and food security is related not just to being more diversified, but also having more assets and outputs, interms of savings, livestock, crops, andjobs. Figure 6.3: Coping Strategies when Not Enough Food 100 .......................... ..... ../ ............................... . / /!!!,Iiiiii///:..!:.iiiiii!i::.i.i.lliii ,::..,..........:...-::. 80 20 0 First Second Third Ate less food Changeddiet Sold livestock or assets Borrowed money Others ~ Source:2001 TLSS 92 6.27 TLSS asked households what actions they undertook in response to a food shortage. Figure 6.3 shows the ex-post actions taken by families when faced with lack of food. Household heads were asked to give up to three responses, ranked by degree o f importance. Almost all families (99 percent) reportedtwo actions, and close to 90 percent three actions. The need to resort to multiple strategies i s in itself an indication o f vulnerability. The number o f coping strategies i s linked to poverty: o f those engaged inat most two actions, only one in four are poor, compared to almost one in two for those reporting three strategies. 6.28 Separating out the coping strategies suggests a sequencing o f responses. At first, the household head experiences anxiety about food insufficiency, leading to decisions to reduce the household's food budget by altering the quality or variety o f food consumed by the family. Overall, almost all households either change their diet or skip meals when faced with insufficient food. These two actions were not just most widespread, but also took priority over other responses. 6.29 Only if the situation requiredfurther adjustment, then households also undertook distress sales o f livestock and other farm assets. Every other household reported this response, most o f them as third action. Selling productive assets i s clearly a last resort. It makes ends meet today at the cost o f lowering the future income stream. Furthermore, it requires having marketable assets in the first place. For example, only one quarter o f those without livestock holdings reported asset sales, compared to over half for those owning animals. 6.30 Other strategies played a minor role. Private transfers are informal ways in which individuals exchange cash, food, and clothing, informal loans and assistance with work and child-care. Only about one infifth families obtained resources from friends, relatives, and neighbors. Over half o f the households receiving private transfers state this only as the third line o f response. Food aid, either from government, NGOs, or the international community, was irrelevant - only one ina hundredpersons benefited from such relief. 6.3 1 The overwhelming importance o f dietary adjustments compared to reliance on asset sales and support from others or is also related to the nature o f the risk. Food insecurity is related both to the agricultural cycle and weather-related production risks, and is a "covariate" risk. It concerns many households in a community or region at the same time. Under great stress, informal arrangements tend to break down, as the members o f the community, or "risk pool", are commonly affected. The income o f the village as a whole i s reduced, triggering a collapse o f community-based informal insurance arrangements (Morduch 1998). For example, as farmers attempt to sell livestock to make ends meet after a drought, livestock prices will fall as supply outstrips demands. Similarly, the family's neighbors and friends are faced with the same negative income shock, and are likely to be reluctant or incapable to provide loans or grants to them. 6.32 When households cut back on meals or change nutrition, who suffers the most? The survey asked families to identify up to three household members, who are affect 93 most in case o f a food shortage. The striking result i s that children appear to take the brunt o fthe adjustment. They account for between three fifthto three quarters ofthe three most affected individual, even though they represent just over half o f all household members. Since malnutrition at young age can lead to long-term health problems, this points to a potentially permanent detrimental consequence o f even occasional food shortages. SUMMARY AND POLICY ISSUES 6.3 3 Temporal and group-specific vulnerabilities are important dimensions o f poverty. The analysis on disadvantaged groups confirms evidence from other countries. Female- headed households, widows, and parentless children experience severe hardship. Possible interventions range from support to traditional community structures; transfers or income-generating activities to widows and targeted support for schooling and health care. 6.34 Subjective assessments o f food adequacy suggest that food insecurity is widespread. Food availability i s aligned with the harvest cycle at the national and regional level. Major urban centers typically have access to just enough food throughout the year, while other parts o f the country face greater fluctuation in food availability, and experience food shortage about twice as often as food excess. Food insecurity during the lean seasons i s also associated with higher poverty. Households have multiple ways o f dealing with food insecurity, which may lower vulnerability in the short term at the expense o f higher vulnerability over the longer term. Almost all households either change their diet or skip meals when faced with insufficient food -to the detriment o f especially children. 6.35 Overall, policies should be aimed at helping poor people manage risk better by reducing and mitigating risk and lessening the impact o f shocks. These comprise multiple measures, ranging from developing human resources, improving access to productive resources and remunerative employment, expanding markets, infrastructure, credit, and institutions, to sound governance and trade and macroeconomicpolicies. RESEARCH ISSUES 6.36 The analysis points to groups that face severe hardships - female headed households, widows and parentless children. More research i s required to fully explore the complicated dynamics between family structure, community support, and welfare. This would help indesigning appropriate policy responses that complement, not displace, family and community support structures. 6.37 Food insecurity i s widespread. The findings on food security call for more survey work explicitly designedto capture the temporal dimension o f food security and poverty, and to investigate household coping strategies. Understanding the underlying causes o f food security (lack o f cash incomes which allow households to purchase food during periods o f shortfall, lack o f availability o f food inmarkets, or lack o f storage) would help design appropriate policies. 94 7. DEVELOPMENTCHALLENGE 7.1 The world's newest country is faced with a daunting challenge o f economic and human development. In spite o f the impressive progress made during transition since 1999, the legacy o f four centuries o f colonial rule, a quarter century o f occupation and conflict, and the destruction following the referendum on independence, are still visible. With independence, the people of Timor-Leste have gained the opportunity, and taken on the responsibility, to meet the development challenge o f overcoming the multiple deprivations burdeningtheir lives. 7.2 This chapter lays out Timor-Leste's development challenge. First, we take stock o f where Timor-Leste stands today with regard to human development, drawing on the latest available indicators for the Millenium Development Goals (MDGs). The second section i s forward looking and investigates the overaching MDG on poverty. The third part presents scenarios linking progress in poverty reduction to aggregate growth and inequality. The final part summarizes many key messages o f this report, asking what kind o f policy and economic changes leadto lower poverty. MILLENNIUM DEVELOPMENTGOALS 7.3 O n September 27, 2002, Timor-Leste became the 191Stmember o f the United Nations, two years after member states o f the UnitedNations unanimously adopted the Millennium Declaration. The Millennium Development Goals (MDGs) are part o f the road map for implementing this declaration. They commit the international community to an expanded vision o f development, where human development i s at the center for sustaining social and economic progress. The l e y development indicators contained in the NDP explicitly draw on the global MDGs. 7.4 The MDGs comprise seven goals, each o f which addresses a main dimension o f poverty.67The goals are set in transparent and quantifiable terms. The MDGs provide only a global blueprint that has to be tailored to national circumstances. On the basis o f such localized numbers, countries, together with their development partners, can chart a course o f action to achieve the goals andtrack progress. Overview 7.5 Box 7.1 presents the MDG goals, targets and indicators. Table 7.1 shows the latest available MDG indicators for Timor-Leste and other East Asian countries. We 67An eighth goal concerns the global developmentpartnership. 95 present information on 6 o f the 7 goals, 8 out o f the 11 targets, and 13 out o f the 31 indicators. Many o f the indicators represent work inprogress. Insome cases observations are sparse and still being compiled, or are not yet adequately collected. This information allows us to rank Timor-Leste relative to other countries in East Asia." The numbers confirm that Timor-Leste i s among the poorest countries in East Asia. The ranking i s very low for child mortality, contraceptive prevalence rate, and education; below average for poverty and environmental sustainability; and average for gender equality. 7.6 These MDGs provide only a global blueprint that has to be tailored to national circumstances. For example, one goal concerns literacy. The population i s ethno- linguistically diverse, with more than 30 languages or dialects in use. Timor-Leste has adopted Portuguese and Tetun as official languages, with English and Indonesian accorded the status o f working languages. As shown inFigure 7.1, no more than one in twenty are fluent in Portuguese, and only one in ten speak Tetun as a mother tongue, though it i s more widely spoken by four in five people. This poses unique challenges o f communication between and within the Government andthe people. ~~ 681tis important to bear inmindthat MDG targets of countries are formulated interms of either meeting a certain improvement relative to 1990, or reaching a specific level by 2015. 96 Box 7.1: MDGs-List of Goals, Targets and Indicators Goal 1. Eradicate extreme poverty and hunger Target 1. Halve, between 1990 and 2015, theproportion ofpeople whose income is less than $1-a-day I.Proportionofpopulationbelow$1perday 2. Poverty gap ratio (incidence x depthof poverty) 3. Share of poorestquintile in national consumption Target 2. Halve, between 1990 and 2015, theproportion ofpeople who suffer from hunger 4. Prevalenceof underweight children(under five years of age) 5. Proportion of population below minimum level of dietary energy consumption Goal 2. Achieve universal primary education Target 3. Ensure that, by 2015, children everywhere, boys andgirls alike, will be able to completeafull course of primary sclzooling 6. Net enrolment ratio in primary education 7. Proportion of pupils starting grade 1who reachgrade 5 8. Illiteracy rate of 15 to 24 years old Goal 3. Promote gender equality and empower women Target 4. Eliminate gender disparity inprimary and secondary education,preferably by 2005, and to all levels of education no later than 2015 9. Ratio of girls to boys in primary, secondary andtertiary education I O . Ratio of literate females to males of 15 -to 24 years old 11.Ratio of women to men in wage employment inthe non agricultural sector 12. Proportion of seats held by women in nationalparliament Goal 4. Reduce child mortality Target 5. Reduce by two thirds, between 1990 and 2015, the under-five mortality rate 13. Under-five mortality rate 14. Infant mortality rate 15. Proportion of 1 year-old children immunized against measles Goal 5. Improve maternal health Target 6. Reduce by three quarters, between 1990and 2015, the maternal nzortali@ratio 16. Maternal mortality ratio 17.Proportion of births attended by skilled health personnel Goal 6. Combat HIV/AIDS, malaria and other diseases Target 7. Have halted by 2015 and begun to reversethe spread of HIV/AlDS 18. HIV prevalenceamong 15-to 24 years old pregnantwomen 19. Contraceptive prevalencerate 20. Number ofchildren orphanedby HIViAIDS Target 8. Have halted by 2015 and begun to reversethe incidence of malaria and other major diseases 21, Prevalenceand death rates associatedwith malaria 22. Proportion of populatioii in malariarisk areas using effective malariaprevention and treatment measures 23. Incidence oftuberculosis (per 100,000 people) 24. Proportion of tuberculosis cases detected and curedunder directly observedtreatment short course I Goal 7. Ensure environmental sustainability Target 9. Integrate theprinciyles of sustainable development into countrypolicies andprogrammes and reverse tlze losses of environmental resources 25. Proportion of land area coveredby forest 26. Land areaprotected to maintainbiological diversity 27. GDP per unit of energy use (as proxy for energy efficiency) 28. Carbon dioxide emissions(per capita) Target 10. Halve by 2015 theproportion ofpeople without sustainable access to scfe drinking water 29. Proportion of population with sustainable access to an improved water source Target 11.By 2020 to have acltieved a significant improvement in tlze lives of at least 100 million slum dwellers 30. Proportion of people with access to improved sanitation 3 1. Proportion of people with access to secure tenure (urban/rural) 97 Figure 7.1: Languages hlOTHERTONGUE LAKGUAGESSPOKEN 35 90 so 711 611 511 40 ill 7.0 I O 0 Telim liidolleri ,I, PonllgiiCsC EIlgilSIl Souice 211111 TLSS MDGon Poverty 7.7 Inthe nexttwo sections, we take a closer look at the MDG on poverty. This first, overarching MDG i s to halve the proportion o f people living in poverty by 2015 compared to 1990. Differences in poverty across countries can reflect differences in economic development, the distribution o f assets, the quality and responsiveness o f state institutions, the degree o f inclusiveness in societies, and risk management. Highlighting the diversity in outcomes i s important. It allows the identification o f successes and failures in poverty reduction, and thereby enhances our understanding o f what causes poverty and how best to reduce it. Awareness o f these differences will help policymakers set priorities, concentrating actions where they are most needed. 99 Figure 7.2: Poverty in East Asia I 35 ~ Timoi- Cembodia Indonesia Malaysia Philippines Thailand EAP EAP Leste China Lao PDR PNG SouthKorea Vietilam less China Source 2001 TLSS and World Bank (2002a) POVERTY HEADCOUNT$YDAY 90 r- -- Tiinoi Cambodia Indonesia Malaysia Philippines Thailand EAP EAP Leste China Lao PDR PNG South Korea Vietnam less Source 2001 TLSS and World Bank (2002a) GIN1COEFFICIENT 50 40 ,..50530 er 20 I O 0 T ~ n o iCambodiaIndonesia Lao Malaysia PNGPhilippines South Thailand Vietnam Leste PDR Korea Source 2001 TLSS andWorld Bank (2002a) 100 7.8 H o w does Timor-Leste compare to other countries in the region? For international comparisons o f poverty, we use the international poverty lines fixed at roughly US$1-a-day and U S $ 2 - a - d a ~ .The country and regional estimates o f the ~ ~ poverty headcounts for these two different poverty lines, based on the latest available household surveys, are shown in Figure 7.2. The US$1-a-day estimates indicate substantially higher poverty in Timor-Leste than in East Asia as a whole (20 percent versus 12 percent). Out o f the eleven countries listed, Timor-Leste i s the fourth poorest country, with only Lao PDR, Cambodia, and PNG showing an even starker deprivation. At US$2-a-day, the difference to East Asia is equally pronounced (63 percent versus 42 percent). Lao PDR and Cambodia are still poorer, but PNG now has lower poverty than Timor-Leste. Overall, the figures confirm Timor-Leste's status as one o f the poorest countries inEast Asia. While the discussion inthis section was just based on one poverty indicator, this conclusion would still hold if we considered other standard poverty measures. 7.9 Poverty reduction takes place within a broader process o f the distribution o f returns to economic activity. Obviously, poverty and inequality are very closely linked - for given economic resources, the more unequal-its distribution, the larger the percentage o f the population living in poverty. Figure 7.2 also shows the Gini inequality coefficient. The value o f 38 places Timor-Leste in the East Asian context in the middle rank, with consumption inequality substantially higher than in South Korea and Indonesia, and substantially lower than inPNG and the Philippines7'. POVERTY,GROWTH,AND INEQUALITY: PROJECTIONS 7.10 Can Timor-Leste meet the MDG challenge? In Timor-Leste, there i s a large clustering o f the population inthe neighborhood o f the poverty line, which suggests that poverty would be very responsive to growth. About a seventh o f all individuals lie within 10 percent o f the poverty line. Economic growth, especially in agriculture, can have a strong effect on lifting those just below the poverty line out o f poverty. However, overall economic growth rarely translates into equal increases in income for all the people in a country. The overall impact o f aggregate growth on poverty depends also on population growth and on how the additional incoiiie i s distributed within a country. If an economic 69These poverty estimates differ from the national poverty rates, as they are based on other poverty lines. To be precise, the poverty lines are set at US$l.O8 and US$2.15 per person per day for all countries. They use Purchasing Power Parity (PPP) exchange rates for 1993 to convert local currencies to constant values. The Timor-Leste national poverty line, evaluated at PPP, is equal to about US$l.S-a-day. Therefore, the poverty headcount at the national poverty line equals roughly to the midpoint (40 percent) to the numbers at US$1-a-day and US$2-a-day. Furthermore, national poverty lines typically allow for spatial cost o f living differentials within countries, which are omitted in the calculations o f Figure 7.2 to maintain a consistent methodology across countries. 'OThe Gini index increases with inequality. A Gini index o f zero indicates perfect equality, and an index of 100 perfect inequality. The Gini coefficient i s 38 based on nominal per capita consumption expenditure and 37 based on per capita consumption expenditure adjusted for spatial cost-of-living differences. Cross- country comparisons o f inequality are however beset with a variety o f problems relating to differences in the definition of the underlying measure of welfare, recall periods, survey design and survey implementation. 101 expansion i s accompanied by lower inequality, then this pro-poor growth will led to fast advances inpoverty reduction. 7.11 This interdependence between poverty, inequality, economic and population growth i s borne out by four illustrative projections for Timor-Leste. The first MDG foresees the halving o f poverty over a 25 year period. This implies that by 2007, the last year covered by the NDP, no more than 35 percent o f the population lives below the national poverty line.71Will Timor-Leste meet this target? The simulations indicate that this depends crucially on three factors: high growth, low inequality, and moderate population growth. Table 7.2 presents four cases. The first scenario fdly incorporates the assumptions o f the NDP mediumterm economic framework. GDP contracts during 2002 and 2003, largely due to the phased withdrawal o f international personnel and the winding down o f reconstruction investment. The economy then recovers to reach a 5.6 percent growth rate by 2007, and less than 2 percent growth annually over the NDP period. However, assuming the population expands at a rate similar to the first half o f the 1990s, this impliesan average per capita growth rate o fjust below zero. Table 7.2: Poverty,Growth, and Inequality Scenarios, 2002-2007 - Actual Projections NDP-Baseline Sliiggish Growth Rising Iiieqiialitv ExpandingPopulation 2001 2002-03 2004-07 2007 2002-03 2004-07 2007 2002-03 2004-07 2007 2002-03 2004-07 2007 Real GDP growth 183 -44 4 2 G I -2 2 3.3 5 2 -14 4 2 6 1 -14 4 2 6 1 Real per capita GDP growth 159 -38 1 8 3 7 -4.6 0.9 2 8 -38 I 8 3 7 -46 I O 2 9 Gini coefficient 370 363 35.6 355 366 362 363 375 392 404 363 35.6 35.5 Headcount 39.7 3 9 2 336 295 418 415 399 402 376 3 5 1 400 36.2 32.8 Poverty gap 119 1 1 5 9.5 8.1 126 125 1 1 9 127 129 127 11 9 10.4 9 1 Severity 4 9 4 7 3 7 3 1 5.3 5.2 4 9 5 5 6 3 6 5 4 9 4.1 3.6 Number of poor ('000) 329 337 309 282 358 382 381 345 346 335 347 345 328 Nore: ADP-Baseline rq,revcnli llie a.s.~impIionsof rhe NW nrcdiimr lenri econourii /ro,mcork Sh&h Gwwh modifiesNDP-Baaelinchy mduci,f,qrhu onniml gmwh rale in a,cr.rrciillwchy 3pwcenl Rhinp lmqmlip u,od$e,s NDP-Boselinehy incrembip die (iini coqflcicnl in ogricidlwe by 1.SprcenI annrtally. ExpondinnPophriun rtiodifieaNDP-Boielinc hj, inct'cminp (heannita1grcJlnhliilu Ofthe puprrlalion froa 2.4Io 3 2 pcmcnl Sowee Siaff esriniarcs 7.12 Inspite of a slight contraction inper capita GDP over the entire period, poverty rates are simulated to fall thanks to strong agriculture growth, the prime source o f livelihood for four infive poor. The NDP assumes for this sector an annual growthrate o f close to 6 percent over the entire period. This leads overall to a reduction o f poverty to just below 30 percent by 2007, about five percent less than the MDG target. Taking into account population growth, the absolute number o f poor in 2007 would drop about 40,000 below the level of 2001. Inequality declines too, as agriculture catches up relative to industry and services. 7.13 Given past agricultural growth rates and international experience, it may be unlikely for agriculture to grow at 6 percent per annum over the Plan period. Therefore, we simulate an alternative scenario with sluggish economic growth, especially in 71 This translates into reducing the headcount to 20 percent by 2026. A more ambitious target would be to halve the share of the poor by 2015, the standard reference point for reaching the MDGs. The target headcount index for 2007 would then be 31 percent. 102 agriculture. In this second projection, we stipulate a slower overall recovery due to a lower growth path for agriculture, which expands three percent less than assumed by the NDP. With per capita GDP now contracting annually by around 1percent, the number o f poor increases by almost 60,000, and the headcount remains unchanged at 40 percent. Inequality still declines, as agriculture outgrows the other sectors, but the reduction is smaller than inthe first case. 7.14 The poverty-reducing impact o f growth can also be offset by a rise in inequality. InScenario 3, we assume the same growth path as for the NDP-scenario, but incorporate a widening in the income distribution in agriculture, assuming that agricultural growth leaves out subsistence farmers. The Gini coefficient now rises above the 2001 level. Poverty reaches 35 percent in 2007, about one percent above the MDG target, and the numbero fpoor remains unchangedcompared to 2001. 7.15 Finally, progress in poverty alleviation i s dependent on population growth: the more mouths there are to fill, the less i s available for each o f them. In the first three scenarios, we assumed that the population will expand at 2.4 percent, matching the experience during the first half o f the nineties. However, the 2002 MICS found that women in Timor-Leste have fertility rates that are among the highest in the world, with on average more than seven children born at the end o f a woman's child-bearing age. Scenario 4 takes the assumptions o f the NDP-scenario, modified by incorporating population growth o f 3.2 percent, one third higher than previously. While the share o f the poor inthe total population falls to 33 percent by 2007, inline with the MDGtarget, and inequality drops, the absolute number o f the poor remains unchanged at about one third o f a million. DETERMINANTSPOVERTY72 OF 7.16 The previous section emphasized the importance o f broad-based growth for poverty reduction. Yet, what are the economic, social, and policy changes required to boost economic activity? How can agricultural growth o f 6 percent be achieved? This part presents an analysis to disentangle the various determinants on poverty. Two-way tables, as shown in previous chapters o f this report, are only o f limited value to identify such sources o f pro-poor growth. Box 7.2 summarizes key characteristics o f the poor. Although these two-way relationships are informative about associations between factors, they cannot answer the key question whether these relationships hold up when other influences are held constant. For example, there i s a clear correlation between the education o f the household head and poverty. But this link could be due to third factors relatedto both education and poverty, like occupation or household assets. 7.17 The standard tool to address this issue i s to conduct a multivariate analysis o f the determinants o f living standards. Such examination can be helpful in identifying correlations between variables, such as those between consumption, characteristics o f the household head, household demographics and assets, and community features. In this section, we analyze the determinants o f one particular dimension o f living standards: ~~ ~ 72This section draws on Chapter 8, Volume 11. 103 Box 7.2: Who are the Poor? The poor live in large households and have higher Andreside predominantly inrural areas, particularly dependency ratios.. . the center... 1611 I7 ill 5 0 DiliiBaucau Oilier urban Ruralcciilcr Rural cast Run1west Have less assets - less human capital, land and livestock,. .. And more limited accessto infrastructure services.. . Experience higher food insecurity.. .. And are more vulnerable, for example, children living infemale headed households have lower enrolment and immunization rates. xi! 711 711 611 hO 50 B 511 5 5411 40 2 3(' B 3 0 2 1 211 2(' II! I1 1! 104 household consumption per capita and the implied probability o f being consumption- poor. SimulationModel 7.18 Inthis section, we describe the basic approach to modelin the determinants o f poverty and deriving simulations. We adopt a three-step procedure." First, we regress the real per capita consumption on a range o f determinants. Then, we derive from this regression the predicted poverty headcount. We allow for regional differences by estimating the regression separately for DiWBaucau, Other Urban Centers, Rural West, Rural Center, and Rural East. Finally, we use this estimated model to predict the impact o f changes o f these determinants on poverty. 7.19 In order to estimate the regression, we have to specify the determinants o f consumption. The selection o f variables i s driven by five considerations. First, the empirical analysis i s obviously limited to factors that are observed and measured in the TLSS andthe Suco Survey. As such, it cannot identify all o fthe various determinants and correlates o f poverty. In particular, the role o f exclusion and social capital inpromoting poverty cannot be adequately analyzed due to gaps inthe available data sets. Second, the bivariate analysis on the welfare profile suggested a number o f key drivers for consumption and poverty that we should take account o f in the analysis. Third, we also include a set o f community level determinants, both at the Aldeia (12 variables) and Suco level (10 variables). This not only ensures that the household level factors are purged from observed community-level determinants, but it also allows us later to simulate the impact o f community level variables on household consumption. 7.20 The determinants can be grouped into the following categories: a. Household demographics: household size (number o f persons) and number of persons in these age groups (under 6, 7 - 14, 15 - 49, and 50 plus). b. Head characteristics: gender, age, age squared, five education categories (no schooling, lower primary (year 1 - 3), upper primary (year 4 - 6), lower secondary, andpost-lower secondary (including university)), and six occupation categories (housework, farmer, non-farin worker, trader, teachedcivil servant, and other). c. Spouse characteristics: indicator variable for spouse present, age and age squared, and the five education categories. d. Agriculture and assets: value o f total crop production, livestock holdings, and savings, all in Rupiah per capita; land holding per capita (hectare); andthree indicators for crop mix (coffee, rice, and maize). e. Housing: indicator variable for house ownership, and number o f years livedinthis dwelling. 73This approachfollows Chaudhuri (2000), Datt and Jolliffe (2001), Hentschelet a1(ZOOO), IFPRI (1998) and Ravallion (1996). 105 J: Infrastructure: three indicator variables on household access to safe drinkingwater, sanitation, and electricity. g. Access: minutes from dwellingto vehicle passable road, indicator whether this road i s accessible during the rainy season, and distance in kilometer from aldeia to suco center (from Suco Survey). h. Aldeia: twelve indicator variables on community facilities (primary school, secondary school, health center, church, kiosk, shop, everyday market, periodic market, bank, mill, vehicle passable road, paved road). i.Suco:indicatorvariableonirrigation,alsointeractedwhetherhouseholdis rice producing; indicator variable on presence o f major private employer (more than five employees); ratios o f number o f teachers per student and number o f classrooms per teacher; ratios o f number o f midwives and traditional birth attendants per population and days in month o f operating health service per population. j. Community Leaders: average characteristics of respondents in Suco Survey in terms o f years o f age, years o f education, and years lived in suco. 7.21 Our simulations illustrate the impact on poverty o f changes o f both policy variables and other determinants. Looking at factors beyond those directly under the control o f decision makers i s also important for policy purposes, as it can give useful information for targeting public resources to population or regional subgroups. Yet, they are unlikely to provide us with the key counterfactual living standard, resulting from a particular policy or economic change, due to seven caveats. First, we only consider one dimension o f living standards. Other welfare outcomes are also important and have to be taking into account when assessingthe relative merits o f policy interventions. Second, the quality o f the simulations can only be as good as the underlying model. Our model accounts overall for three fifth o f consumption variability, implying that two fifth are due to factors we do not control for. Furthermore, our estimates do not as such uncover causal relationship, but only conditional correlations. In particular, our model draws only on data from one point intime and cannot reveal dynamic interactionbetween factors. Third, the simulations are conducted under the "ceteris paribus" assumption, implying that the considered change in the determinant does not affect the model parameters or other variables. This assumption may be defendable for marginal or incremental changes, but it becomes implausible for large policy reforms. For example, changing the occupation o f one person from farmer to trader i s unlikely to affect market outcomes. By contrast, if many farmers are involved, the remuneration o fthese occupations and prices o fproducts will adjust, and households, even those originally unaffected, will modify their behavior inresponse. Such "general equilibrium effects" make it difficult to predict the impact of major policy and economic adjustments. 7.22 Fourth, the impact o f a change in one determinant i s likely to differ across households. However, our model accounts only for the differential impacts by regions through separate regional parameters, applying the same mean effect to all households affected by the change within regions. Fifth, the determinants differ with regard to both the extent to which they are amenable to policy decisions, and the time horizon in which they are likely to adjust. One the one hand, factors that are directly affected by policy 106 with a fast response time can contribute most to poverty alleviation over the short horizon. This group includes infrastructure and health variables. On the other hand, some determinants, like demographic variables, are more removed from policy intervention and slow to change, yet they still may be important for reducing poverty from one generation to another. Sixth, the simulations concentrate only on the potential benefits in terms o f poverty reduction, but ignore any cost differences across the various interventions. For example, we will find that expanding electricity to the entire population reduces poverty by more than providing all households with basic sanitation. Yet, the first intervention may well be more costly than the second one. This difference could be large enough so that in the end, for a given level o f resources, poverty will drop more if the government invests into basic sanitation rather than electricity. Finally, Timor-Leste has already changed substantially since the time o f the survey - when it still was called East Timor. Yet, the model reflects the economic environment duringlate 2001. Poverty Simulations 7.23 The findings froin the simulations are shown in Table 7.3. We consider simulations over five groups o f variables (demography, education, agriculture, infrastructure, and economy). The results are presented for six different populations: nationwide, and urban and rural separately, both for the total population and for the "affected" population only, i.e. those households for whom the value o f at least one right- hand side variable was changed. The table displays the percentage changes o f both per- capita consumption and poverty. In our discussion o f the results, we concentrate on the poverty impacts. 7.24 Demography. Household size, composition, and, in urban areas, gender o f the head matter for poverty. Reducing household size by one for all households with more than one member lowers poverty by 7 percent nationwide, and more inurbanthan inrural areas. By contrast, changing householdcomposition by replacing one child up to age 6 by one prime-aged adult reduces poverty by about twice as much inhouseholds with at least one child, and the effect i s larger in rural than in urban areas. Finally, male headed households have lower poverty only in urban areas. These findings imply that, compared to urban households, rural families are less affected by size and gender o f the head, but more by age composition. While demographic characteristics evolve over generations, this informationcan still beusedfor targeting public assistance or investmentprograms. 7.25 Education. Building human capital o f heads and spouses leads to lower poverty. This is confirmed inSimulations 4 and 5, where we look at the impact of lifting all heads, and all spouses, to at least four years o f schooling. This i s a large experiment - it affects about seven in ten heads and spouses - with substantial payoff: poverty drops by about 12 to 15 percent nationwide. Inview o f the large number o f affected people, it i s clearly unrealistic that the returns to education remain unchanged, casting doubt over the point estimates. Nevertheless, even for small changes, three messages remain. First, education lowers poverty. Second, the overall gains are larger from the increase o f the education grade o f spouses than from those o f heads. Third, while the effect o f spouse education is the same for urban and rural areas, head's education matters about twice as much incities 107 than invillages. One possible explanation i s that heads are the main breadwinner, and the returns to education o f occupations are higher inurbanthan inrural areas. 7.26 Agriculture. Non-agricultural activity, high-valued crops, and irrigation are three main exits from rural poverty, as shown in Simulations 6 to 12. Again, we illustrate the impact o f large reforms, so our main focus i s on the direction rather than the point estimate of the changes. Switching heads from being farmers, accounting for about three quarters o f the population, to traders triggers a drop inpoverty o f one quarter (Simulation 6). The benefits are especially marked in urban areas, suggesting that trading i s more profitable incities than invillages. The right crop portfolio i s essential: maize farmers are poor, while coffee farmers are non-poor. For given inputs and crop production, improving the crop mix to high value crops, as simulated through cultivating coffee or ceasing maize productionreduces poverty by 15 percent (Simulations 7 and 8). However, keeping the crop mix constant, boosting productivity, cultivating more land, or increasing livestock holdings, reduces poverty by no more than 1to 4 percent (Simulations 9 to11). Finally, expanding around-the-year irrigation to all sucos lowers poverty in affected areas, representing about two fifth o f all households, by about 10percent (Simulation 12). 7.27 Infrastructure. Sanitation and electricity are important to improve living standards. Providing basic sanitation to all households lower poverty by 9 percent among the newly covered families, and by up to 20 percent among those households in cities (Simulation 13). Giving electricity to all households, a financially more expensive intervention than expanding sanitation, reduces poverty by-more than one quarter among the beneficiaries (Simulation 14). By contrast, improving access to vehicle passable roads has little payoff, partly because most households are already within less than 10 minutes walking distance to such roads (Simulation 15). 7.28 Economy. Infrastructure other than irrigation can also provide substantial benefits to communities. This i s illustrated in two simulations, even though a high covariation o f community factors makes it problematic to isolate a particular intervention. Presence o f major private employers reduces the poverty headcount by almost one tenth, and more than one fifth in urban areas (Simulation 16). Establishing full-fledged periodic markets inall sucos is associated with poverty reductions of more than 30 percent inrural areas (Simulation 17). 7.29 As the government has limited resources, not all o f these policies can be implemented at the same time. This raises the question o f prioritization. Should the measure with the largest estimated impact be taken first? No, these simulations are only illustrative, and the other aspects highlighted before need to be taken into careful consideration. In Table 7.4, the measures are classified in terms o f four additional dimensions apart from their likely impact on poverty. First, the impact o f a change inone determinant differs across households. Some households are directly affected from a measure, other only indirectly, and others not at all. Second, the determinants differ with regard to both the extent to which they are amenable to policy decisions, and the time horizon inwhich they are likely to adjust. Third, the simulations concentrate only on the potential benefits interms o f poverty reduction but ignore any cost differences across the various interventions. Fourth, most measures have an impact on more than one MDG, 108 and such synergies would have to be taken into account when assessing the relative merits of policy interventions Table 7.3: Simulations on Changesin Consumptionand Poverty (%) Description Indicator Eutire Population Affected Population National Rural Urban Natioiial Rural Urban Demography 1 Reduceby one the nuiiiber of PCC 5.0 4.0 7.2 5 1 4.0 7.2 household members POV -6.7 -5.9 ..10.6 -6 7 -5.9 -10.6 POP 100 100 IO0 I00 100 100 2 Replace o m child aged 0-6 by PCC 3.8 6.5 -1.4 8 6 11.6 2.8 one adult aged 15-49 POV -9.1 -10.5 -1.3 -13 8 -15.0 -7.1 POP 100 IO0 I00 72 72 71 3 Move headship from feiiiale to iiiale PCC 0.3 -0.2 1.3 2.5 -1.9 14.3 PO\' -0.3 0.1 -2.6 -3 8 2.0 -30 9 ................................................................................................................................................................................................................................................................sIO0 POP 100 100 I O 10 9 Education 4 Increasehead's education to 4-6 PCC 8.0 6.8 10.4 I3 9 9.9 30.5 years ofprimary school POV -1 1.5 -9.9 -20.4 -15.3 -12.8 -31.2 POP 100 100 100 66 72 46 5 Increasespouse'seducation to 4-6 PCC 8.6 10.4 5.2 13.5 14.1 11.4 years of priinaiy school POV -15.0 -15.3 -13.3 -18 8 -18.8 -18.8 ................................................................................................................................................................................................................................................................5 POP 100 100 100 7.!..................................................................55 77 Agriculture 6 Move head`soccupation koin fanner PCC 20.1 17.8 24.6 30 9 21.7 79.3 to trader POV -26.1 -23.3 -41.8 -30 3 -25.8 -64.0 POP 100 100 100 16 86 44 7 Introduce coffee for all crop-producing PCC 12.0 8.6 18.8 23.6 14.1 58.7 households PO\' -16.1 -14.7 -24.0 -23.6 -21.5 -35.5 POP 100 100 IO0 59 64 45 8 Abolish maize for all crop-producing PCC 8.4 12.3 0.6 12.4 14.8 1.7 households POV -14.3 -16.9 0.0 -16 3 -18.8 0.0 POP 100 100 100 78 86 52 9 Increasecrop production by 50% for PCC 0.9 1.5 -0.2 I 2 1.6 -0.6 all crop-producing households POV -0.7 -1.0 1.1 -0 7 -1.1 1.3 POP 100 100 IO0 85 95 55 10 Increaselaidholdings by 0.1 Iia per PCC 1.3 1.8 0.5 I8 1.9 1.2 capita for all land-holding liouseholds POV -3.4 -3.7 -1.4 -3 6 -3.8 -1.7 POP 100 100 100 86 95 55 11 Iiicrease animal holdings by 50% for PCC 1.5 2.3 -0.2 I8 2.5 -0.3 all animal holding households PO\' -1.6 -1.6 -1.4 -1 8 -1.8 -1.7 POP 100 100 100 85 90 71 12 Expand around-the-year irrigation PCC 5.8 5.1 7.4 14 8 11.8 22.7 to all sucos POV -4.1 -2.9 -10.6 -9 5 -6.8 -25.2 ................................................................................................................................................................................................................................................................r POP 100 100 100 41 43 36 Infrastructure 13 Expand basic sanitation to all PCC 4.4 4.7 3.7 8 4 7.0 16.6 households PO\' -5.5 -5.0 -8.3 -8 9 -7.6 -20.0 POP 100 100 100 58 67 30 14 Expand electricity to all households PCC 12.9 17.0 4.9 20 9 20.2 28.8 PO\' -22.3 -23.8 -14.1 -25 7 -25.5 -28.0 POP 100 100 IO0 74 89 28 15 Reducetime to nearest road by 10% PCC 0.0 0.0 0.1 0.1 0.1 0.1 for all households POV -0.1 -0.1 -0.1 -0 I -0.1 -0.1 ............................................................................................................ POP 100 100 100 80 81 77 Econoiny 16 Expandprivate employer (more than 5 PCC 11.1 14.9 3.6 14 8 17.3 6.7 employees) to all sucos POV -7.7 -6.7 -13.1 -9.2 -7.6 -22.1 POP 100 100 100 19 87 53 17 Expand periodic inarket to all sucos PCC 22.9 22.7 23.2 28.2 28.6 27.5 PO\' -25.0 -26.2 -18.0 -30.4 -32.3 -20.7 POP 100 100 IO0 81 80 85 iteudcoinif,und POP,for~nil,irlulbn. 109 Table 7.4: Five DimensionsofPublicAction Measure Description Poverty Population Policy costs MDGs Subgroup Influence Agriculture Farmerto trader -26 Farmers Low High Environment Economy Suco markets -2s -22 Rural areas High Low Infrastructure Electricity Families wio electricity High High Environment Agriculture Coffee -16 Farmers Low High Environment Education Spouse education -15 Spouses High Low Education, Gender, Health Agriculture No maize -14 Poor farmers Low High Environment Education Headeducation -12 Heads High Low Education Demography Child to adult -9 Families Low Low Environment Economy Private employer -8 Rural areas Low Low Gender, Health Demography Household size -1 Families Low Low Infrastructure Sanitation -5 Families wio sanitation High High Gender, Health Agriculture Irrigation -4 Rural areas High High Environment Sour.~c.20111 TLSX World Bank. SUMMARY AND POLICY ISSUES 7.30 Timor-Leste faces an enormous development challenge. Poverty is high and human and physical capital are depleted. One in five people in Timor-Leste live below US$1-a-day, and three in five below US$2-a-day. Timor-Leste i s among the least developed countries in East Asia on most social indicators. Timor-Leste is not just a young nation, it i s also a young people: one intwo persons are below the age o f 15. This nation o f about 830,000 people will grow rapidly as large young cohorts move through the reproductive ages. 7.31 Illustrative projections show that even a relatively modest target o f halving the poverty headcount over 25 years depends on strong per-capita growth, especially in agriculture, and a broad participation o f the population in the opportunities of an expanding economy. We highlight some o f the key determinants o f pro-poor growth, using a statistical model that pulls together some of the key findings o f this report. However, we have to interpret these results with great caution, as the approach suffers from a number o f limitations. With those qualifications in mind, the simulations confirm important messages. Lowering the dependency ratio and size o f households, boosting male and female human capital, promoting non-farm activities, encouraging the production o f high-value crops, developing extension services like irrigation, constructing sanitation and electricity infrastructure, creating a favorable business environment for private employers, and improving market networks all help to lower poverty. Future work planned under the public expenditure study will help cost different policy options. 7.32 While the social agenda i s daunting, Timor-Leste has the solid prospects o f future flows from the country's natural resource wealth, and the commitment o f donors including the World Bank to support its development. Achieving sustained social improvement will raise the quality and pace o f long-term growth and ensure that the economy will not develop wholly dependent on oil and aid. As emphasized inthe NDP, 110 allocating aid and off-shore resources towards highpriority development objectives inan effective manner will be critical. The MDGs can provide the framework for prioritizing and monitoring human development. A "localization" o f MDGs to Timor-Leste will be an important tool for malting the poverty reduction policies effective. This involves identifying indicators that are appropriate for assessing progress against the MDG goals inTimor-Leste and setting targets that are realistic (see Chapter 8). This can be done in the context o f action plans being prepared by the Ministries on the basis o f the NDP. Communities and civil society will have to play a central role in this process. Furthermore, the medium term expenditure framework can enhance the realism o f the anti-poverty policies. The key challenges will be to ensure adequate linkages from the poverty reduction strategies to the operational budget level. Budget decisions should be driven by policy priorities on poverty, and policy choices in turn have to be disciplined by resource and implementationrealities over the mediumterm. 111 8. POVERTYMONITORING 8.1 Monitoring and evaluation systems enable the government to assess whether a poverty reduction strategy i s effective inreducing poverty. Monitoring concerns the issue o f measuring the progress towards poverty reduction goals. The evaluation o f policies and programs determines the extent to which improvements in outcomes are due to specific public actions. The National Development Plan highlights the importance o f monitoring and evaluation as an essential element for assessing progress towards the goals articulated inthe Plan. 8.2 Monitoring progress on poverty reduction for Timor-Leste will entail institutionalizing a system o f data collection, analysis and reporting on a set o f well- defined indicators. This involves defining key indicators, tracking them over time, and seeing what changes have taken place. As part o f the implementation o f the National Development, work i s ongoing on developing such a set o f measurable indicators and targets. In compiling such a data base, a number o f issues will have to be considered. First, it is important to include various types o f information, ranging from quantitative and qualitative surveys to administrative and budget data. Box 8.1 provides an overview over the existing data sources that can be drawn on to provide a baseline against which progress can be monitored. 8.3 Second, the objective o f data collection should determine the type, frequency, and level o f disaggregation. The principal purpose o f a monitoring system i s to track the changes in poverty over time to assess the overall effect o f the development strategy. Given the multi-dimensional nature o f poverty and the commitment to the Millennium Development Goals, this requires monitoring both consumption-based poverty and other dimensions o f well-being, like access to basic services (such as education, health, safe drinking water). Disaggregated information by region and household characteristics is essential for understanding how overall trends are related to the welfare o f specific groups. The changes in poverty can then be compared to changes in public expenditures to assess the impact o f government policies. A more ambitious objective o f understanding the determinants o f poverty requires additional information. Aspects related to how the poor earn their living, gain access to assets and credit, such as land, education or transfers from family and friends, are all relevant in this context. This requires different types o f data, from detailed household surveys to community surveys o f social and economic infrastructure. The same holds for other dimensions o f poverty. For example, to investigate why some children go to school and others do not, it is important to know the characteristics o f the child, like age and gender, o f the household, like household size and composition, asset ownership, educational attainment and occupation o f household members, inaddition to information about the distance and cost o f going to the school and the quality o f the school, such as the qualification o f teachers 112 and the student-teacher ratios. This requires information from household and school surveys. Box 8.1: Poverty Data Sources Tintor-Leste Living Standards Survey: The nationally representative household survey o f 1800 households provides data on consumption, education, health, labor markets and agriculture. The survey was conducted end August-November, 2001. The survey is representative for the big cities (DiliiBaucau), other urban centers and rural areas. Within rural areas, it provides information for the East, Center and West. These data were collected by the Statistics Office in the Ministry of Planning and Finance. Survey of Sucos: The survey o f all 498 sucos in Timor-Leste provided an inventory of infrastructure and population characteristics in early 2001. These data were collected by the Statistics Office inthe Ministry o f Planning and Finance. Multiple Indicators Cluster Survey: The nationally representative survey of 4000 households provides data on maternal and child health, in particular on infant and child mortality, education, water and sanitation, child health and malnutrition and reproductive health. The data collection was completed during August-mid September 2002. These data were collected by the Statistics Office inthe MinistryofPlanning andFinance. School Mapping: Data were collected in 2001 on all schools in Timor-Leste. Information was collected on the enrollment by grade and by age, number o f students, the number o f teachers (for both public and private schools), and the percentage of classrooms inoperation. Adnziizistrative Data: Ministries collect performance indicators that relate to the provision of services. Data often relate to the outputs (such as the number of classrooms, number o f teachers) and sometimes to outcomes, such as enrolment rates or immunization rates. The relevance, quality and timeliness o f the data collected are uneven across Ministriesand in some cases are at odds with the data collected through household surveys. Budget Data: Budget allocations, and more importantly, actual spending by different Ministries on their programs provide important input indicators on whether the resources are being spend, and on the different components (e.g on teacher salaries, purchase o f textbooks, school construction) on which they are being spent. 8.4 Third, monitoring activities need to be carried out by institutions that are competent and that have strong links to key decision-makers, if they are to be useful in the design and implementation o f the NDP. Much monitoring and evaluation takes place without adequate development o f in-country capacity and without strong links to key decision-making processes. Precious opportunities to learnwhat works and what does not are then lost. It i s therefore important to build capacity and in particular strengthen the processes that provide policy makers and others with feedback on the impact o f policies and programs. 8.5 Furthermore, dissemination o f results is critical for use. Results that are not widely disseminated, through mechanisms tailored to different groups in civil society, will not be used, and the resources that were spent ingetting such results will be wasted. Non-governmental actors, be they research institutions, civil society organizations, 113 special-interest and advocacy groups, or others, have an important role to play in the design o f the monitoring and evaluation system, in actually carrying out monitoring and evaluationactivities, and inusing the results. 8.6 Looking forward, it will be important to build on the existing database for up-to- date assessments o f the progress in the implementation o f the NDP. Different data tools can contribute to a rich monitoring system: 0 Population Census: The Census planned for 2004 i s critical. It will present updated information on the characteristics o f the population, and provide the master sample frame from which the sampling for future surveys can be undertaken. Indesigning the Population Census, it would be useful to consider the option o f developing "poverty maps", which combine household surveys and the Population Census to provide expenditure-poverty estimates for small geographic units, like sucos. 0 Living Standards Survey: Inview o f the rapidly evolving economic situation in Timor-Leste, it would be desirable to conduct a second integrated household survey over the next two years to obtain updated estimates o f expenditure- poverty, other indicators, and their determinants. To ensure comparability with the poverty estimates for 2001, it will be important to maintain as much as possible the questionnaire o f the first household survey, and to conduct the survey during the same time period in the year (August - November) to avoid biases arising from seasonality. The household survey should be jointly fielded with a community (suco) and price survey to provide information on the economic environment o f households. These surveys should be implemented at regular intervals o f three to five years. 0 Special Purpose Surveys: In addition to the living standards survey, special purpose surveys may be needed (such as the Demographic and Health Survey or Multiple Indicators Cluster Surveys) that focus on child andmaternalhealth. * 0 Administrative Data: While expenditure-poverty estimates are typically collected on a multi-year cycle it i s generally desirable to collect some indicators (either through administrative sources, or community level data collection) on a more frequent basis. Such regular monitoring can provide an early indication o f emerging economic problems. For example, information o f prices o f key commodities and rural wages could be collected on a monthly basis, health and education data from administrative data on a semi-annual or annual basis. This will also ensure the provision of timely information as input into the annual budget planning process. 0 Participatory Suvveys: The formulation o f Timor-Leste's vision and the NDP was informed by extensive consultations with its people. Continued involved o f different stakeholders in monitoring and implementation o f the NDP, through a systematic plan that lays out the different elements and methodology for this consultation would be desirable. 114 8.7 In summary, Timor-Leste has many varied data sources that present a coherent picture o f poverty and provide a baseline for monitoring progress inpoverty reduction, as outlined in the National Development Plan. The key challenge lies in formulating a monitoring plan that includes both quantitative and participatory elements, and the institutional arrangements for data analysis and reporting to ensure that the data collected inform policymalting and program design. 115 9. REFERENCES Arrow, Kenneth J., 1950, "A difficulty in the concept of social welfare", Journal of Political Economy, 58,328-46. Atkinson, Anthony B., and Franqois Bourguignon, 1987, "Income distribution and differences in needs" in George R. Feiwel, ed., Arrow and the foundations of the theory ofeconomic policy, New York University Press, 350-70. AusAID, 2001, "Employment Patterns and Skill RequirementsinEast Tiinor, and Their Implicationsfor Technical and Vocational Education and Training". Bartholomew, D.J., 1982, Stochastic Models for Social Processes, 3`d edition, John Wiley. Beegle, Kathleen and Martin Cumpa, 2002, "Labor Markets, Employment and Poverty in East Timor", mimeo, World Bank, Washington, DC. Besley, Timothy, 1995, "Savings, Credit and Insurance", Handbook of Development Economics, 3A, 2123-2207, Handbooks in Economics, vol. 9. Amsterdam, New York and Oxford: Elsevier Science, North Holland. Bhalotra, Sonia and Christopher Heady, 2001, "Child Farm Labour: The Wealth Paradox", Bristol Discussion Paper 00/492, Department of Economics, University of Bristol, UK. Birdsall, Nancy, Allen Kelley, and Steven Sinding, 2001, Population Does Matter: Demography, Growth, and Poverty in the Developing World, Oxford University Press, NewYork. Blackorby, Charles, and David Donaldson, 1987, "Welfare ratios and distributionally sensitive cost-benefit analysis", Journal of Public Economics, 34,265-90. Canagarajah, Sudharshan and Helena Nielsen, 1999, "Child Labor and Schooling in Africa: A Comparative Study", SP Discussion Paper 9916, World Bank, Washington, DC. Case, Anne, and Angus Deaton, 2002, "Consumption, Health, Gender, and Poverty", Working Paper 212, ResearchProgram inDevelopment Studies, PrincetonUniversity. 116 Chaudhuri, Shubham, 2000, "Empirical methods for assessing household vulnerability to poverty", mimeo, Department of Economics and School of International and Public Affairs, Columbia University. Conlisk, John, 1990, ``Ranking mobility matrices", Journal of Mathematical Sociology, 15, 173-91. Datt, Gaurav and Dean Jolliffe, 2001, "Poverty in Egypt: Modeling and Policy Simulations", mimeo. Deaton, Angus, 1997, TheAnalysis of Household Surveys: A microeconometric approach to developmentpolicy, Published for the World Bank, The John Hopltins University Press, Baltimore and London. Deaton, Angus, and John Muellbauer, 1986, "On measuring child costs: with applications to poor countries", Journal of Political Economy, 94,720-44. Deaton, Angus, and Salinan Zaidi, 1998, "Guidelines for Constructing Consumption Aggregates for Welfare Analysis", LSMS Working Paper 133, World Bank, Washington, DC. Dercon, Stefan, 2001, "Assessing Vulnerability and Poverty", Jesus College and CSAE, Department of Economics, Oxford University. East Timor Transitional Administration, Asian Development Bank, World Bank and UnitedNations Development Programme, 2001, "The 2001 Survey of Sucos: Initial Analysis and Implications for Poverty Reduction", Timor-Leste. Foerster, Jean, 2002, "Agriculture and Poverty in East Timor", miineo, World Bank, Washington, DC. Foster, James, J. Greer, and Eric Thorbecke, 1984, "A class of decomposable poverty measures", Econometrica, 52, 761-65. Hentschel, Jesko, J.O. Lanjouw, P. Lanjouw, and Javier Poggi, 2000, "Combining Census and Survey Data to study Spatial Dimensions of Poverty: A case study of Ecuador", in David Bigman and Hippolyte Fofack, ed., Geographical Targeting for Poverty Evaluation: Methodology and Applications, World Bank, Washington, DC. Howes, Stephen, and Jean Olson Lanjouw, 1995, "Making poverty comparisons taking into account survey design: How and why", Policy Research Department, World Bank, Washington, DC, and Yale University, New Haven. International Food Policy Research Institute (IFPRI), Ministry of Planning and Finance, and Eduardo Mondlane University, 1998, Understanding Poverty and Well-Being in Mozambique: TheFirst National Assessment (I996-97). 117 Indonesian 1990 Population Census on Timor-Leste, Central Bureau of Statistics, Jakarta, Indonesia. Kremer, Michael, A. Onatslti, and James Stock, 2001, "Searching for prosperity", NBER Working PaperNo. 8250, NBER, Cambridge, Massachusetts. Lanjouw, Peter, Giovaniia Prennushi, and Salman Zaidi, 1996, "Building blocks for a consumption-based analysis of poverty in Nepal", mimeo, World Bank, Washington, DC. Morduch, Jonathan, 1999, "Between the Market and State: Can Informal Insurance Patch the Safety Net?", World Bank Research Observer, 14 (2), 187-207, World Bank, Washington, DC. Muiioz, Juan, 2001, "Timor Lor0 Sa'e Living Standards Survey Sampling Design and Implementation", mimeo, World Bank, Washington, DC. Narayan, Deepa, Raj Patel, Kai Schafft, Anne Rademacher and Sarah Koch-Schulte, 2000a, Voices ofthe Poor: Can Anyone Hear Us?,Published for the World Bank by Oxford UniversityPress, New York. Narayan, Deepa, Robert Chambers, Meera Kaul Shah, and Patti Petesch, 2000b, Voices of the Poor: Crying Outfor Change,Published for the World Bank by Oxford University Press, New York. Nassim, Janet, 2002, "Health and Equity in East Timor: The Starting Point", mimeo, World Bank, Washington, DC. Planning Commission, 2002, East Timor National Development Plan, Dili, Timor-Leste. Pradhan, M., and Robert Sparrow, 2000, "Basic education outcomes during crisis - An analysis usingthe 1995, 1997, 1998 and 1999 Susenas", mimeo. Pradhan, Menno, Asep Suryahadi, Sudarno Sumarto, and Laiit Pritcliett, 2000, "Measurements of Poverty in Indonesia: 1996, 1999, and Beyond", SMERUWorking Paper, Social Monitoringand Early Response Unit,Jakarta. Quah, Danny, 1993, ``Empirical Cross-section Dynamics in Economic Growth", European Economic Review, 37,426-34. Quah, Danny, 1994, "Convergence einpirics across economies with (some) capital mobility", mimeo, London School of Economics, London. Ravallion, Martin, 1994, "Poverty comparisons", Harwood Academic Press Fundamentals of Pure and Applied Economics, volume 56, Chur, Switzerland. 118 Ravallion, Martin, 1996, "Issues in Measuring and Modelling Poverty", The Economic Journal, 106, 1328-1343. Ravallion, Martin, 1998, "Poverty lines in theory and practice", LSMS Working Paper 133, World Bank, Washington, DC. Rohland, Klaus and Sarah Cliffe, 2002, "The East Timor Recoiistruction Program: Successes, Problems and Tradeoffs", CPR Working Paper 2, World Bank, Washington, DC. Saadah, F., M. Pradhan, and S. Surbalcti, 2000, "Health Care During Financial Crisis: What can we learn from the Indonesian National Socioeconomic Survey?", HNP Discussion Paper, World Bank, Washington, DC. Shorrocks, Anthony F., 1978, "The measurement of mobility", Econometrica, 46, 1013- 24. Siegel, Paul and Jeffrey Alwang, 1999, "An Asset-Based Approach to Social Risk Management: A Conceptual Framework", Social Protection Discussion Paper 9926, World Bank, Washington, D.C. UNICEF 2002, MultipleIndicators Cluster Survey (MICS), draft Report, UNICEF, Timor-Leste. UnitedNations Development Programme, 2002, East Timor HumanDevelopment Report 2002, UNDP, Timor-Leste. United States Department of Agriculture, USDA Nutrient Database for Standard Reference, http://www.iial.usda.aov/fnic/cai-bidnut search.pl, Agricultural Research Service. World Bank, 2001, WorldDevelopment Indicators, World Bank, Washington, DC. World Bank, 2001a, World Development Report 2000/2001: Attacking Poverty, Oxford University Press, New Yorlc. World Bank, 2002, East Tinzor: Policy Challengesfor a New Nation, Country Economic Memorandum, Poverty Reduction and Economic Management Unit, East Asia and Pacific Region, Washington, DC. World Bank, 2002a, East Asia Update, East Asia Rebounds, But How Far?, Regional Overview, East Asia and Pacific Region, Washington, DC. 119 World Bank, 2002b, East Tiinor Public Administration: Public Expenditure Management and Accountability Note, Poverty Reduction and Economic Management Unit, East Asia and Pacific Region, Washington, DC. 120 ANNEX 121 TABLE A.l: PROFILE BY REGION National Dili/ Other Rural Baucau Urban Center East West Poverty, inequalityand expenditure Headcount 39.7 13.9 38.4 49.3 32.0 47.5 Gap 11.9 3.8 10.0 15.8 9.4 13.2 Severity 4.9 1.6 3.7 6.9 3.8 5.2 Food povertyrates (food expenditurevs food line) Headcount 39.5 34.3 35.2 42.6 34.1 45.5 Gap 1I,6 10.4 9 6 13.0 10.2 12.2 Severity 4.7 4.1 3.4 5.4 4.0 4.8 Gini * 37.0 36.4 36.0 35.8 32.6 29.7 Per capita expenditure(US Dollarsper month) 24.2 40.1 25.9 20.5 24.6 18.9 Food (%of total expenditure) 63 41 57 75 65 69 Purchases 34 36 33 35 31 37 Home production 24 3 21 34 30 29 In-kind 4 1 3 6 4 4 Rent (%of total expenditure) 22 42 23 15 14 20 Others (YOoftotal expenditure) 15 18 20 I O 20 10 Householdcomposition ** Households i x 4.9 6.0 4.9 5.0 4.3 4.5 Dependencyratio ("h) 125 102 117 139 117 123 Children(% householdsize) 45 41 43 49 43 45 Infrastructure Drinkingwater (%population) ai 50 84 51 48 32 50 Sanitation(%population) bi 42 86 52 39 25 30 Electrification("hpopulation) ci 26 90 50 I O 16 9 Housingdamageand rehabilitation Damagedinviolence (% liouseholds) 29 28 48 17 I O 58 Totally damaged(as % o fdamaged) 85 52 82 90 90 88 Rehabilitated (as% ofdamaged) 62 49 60 65 75 63 Totally rehabilitated(as% ofdamaged) 21 11 24 22 48 18 Access Distanceto the aldeiacenter (Ian) 1.9 1.3 1.4 I.3 4.7 0.7 Distawe from the aldeiacenter (kin) Everydaymarket 20.6 1.6 6 2 28.9 25.7 14.5 Periodic market 8.5 3.8 4.3 9.5 11.6 4.8 Vehicle-passable road 0.7 0.1 0.4 0.8 1.4 0.1 Pavedroad 3.1 0.1 0.6 5.3 3.1 1.1 Education Iliteracy rate (% of 15 and older) 51 20 46 58 55 62 Educationofthehead(years) 3.1 6.7 3.4 2.4 2.6 2.2 Primary, Net enrollmentrate di 73 77 80 71 70 75 Junior Secondary,Net enrollment rate di 25 49 43 15 23 23 122 TABLE A.l: PROFILEBY REGION National Dili/ Other Rural Baucau Urban Center East West Health Immunization (%childrenless than 1 year-old) BCG 33 52 47 28 18 35 Measles 6 I 13 4 0 9 DPT (complete) 9 16 20 3 4 17 Polio (complete) 6 I O 13 3 2 9 Health complaints last mnth("A population) 22 22 17 22 26 20 Utilization rites ("hpopulation) Public outpatient care 12 14 I O 12 11 I O Privatecare 2 2 3 1 4 2 Traditional care practitioner 2 1 2 1 6 3 Self-medication 7 9 6 8 5 5 Employment e/ Participation rate 60 48 59 68 55 61 Femaleparticipation rate 40 29 41 50 30 38 Unemploymentrate 5.3 19.7 4.4 3.7 2.1 3.1 Weekly working hours fl Mean 40 47 41 37 44 36 Median 41 48 42 38 48 36 Weekly workinghours ("A) fl Up to 15 3 2 1 2 4 2 I 6 to 35 29 18 23 33 13 45 36 to 50 54 44 64 58 56 41 More than50 15 35 12 I 26 12 Agriculture gi Per capita land(ha) hi 0.38 0 26 0.48 0.33 0.29 Per capita irrigated(ha) Ill 0.08 0.06 0.06 0.15 0.05 Among land holders: Average per capitaland (ha) hi 0.41 0.29 0.51 0.36 0.31 Average per capitairrigated land (11%) Ill 0.09 0.06 0.06 0.16 0.06 %of irrigated land hi it 18 12 17 28 13 Averagevalue ofland perha (US Dollars) *** 7.2 1.7 6.6 10.2 4.5 Medianvalueof landper ha(US Dollars) *** it 1.5 1.3 1.6 1.2 1.1 Production uses(as %o ftotal value) Sales 29 30 45 3 24 Barter 2 2 2 0 1 Lost 4 5 3 9 3 Payments 1 1 1 1 0 Self-consumption 62 61 48 81 72 Subsistence (% population) .if 33 29 1 1 81 32 Agricultural inputs ("Ahouseholds) Use ofmanure, fertilizers, pesticides or herbicides 3 3 0 5 5 Purchasedor receivedmaize, rice or bean seeds 25 18 17 33 35 Use any ofthe above inputs 21 20 17 37 39 123 TABLE A.l: PROFILE BYREGION National Dili/ Other Rural Baucau Urban Center East West Livestock gi Averageper capitavalue of2001 livestock(US Dollars) 95 87 81 143 74 Medianper capitavalue of2001 livestock(US Dollars) 35 42 35 44 21 Averageper capitavalue of 1999 livestock(US Dollars) 221 214 137 196 434 Median per capitavalue of 1999 livestock(US Dollars) 62 65 51 57 116 Among2001 livestock holders: Average per capitavalue of2001 livestock(US Dollars) 106 96 88 151 93 Medianper capitavalue of2001 livestock(US Dollars) 40 48 39 48 28 Average per capitavalue of 1999 livestock(US Dollars) 242 237 147 206 524 Medianpercapitavalue of 1999 livestock (US Dollars) 72 78 56 60 173 Subjective well-being Life compared to 1999(% populatioii 15andinore) MLIC~ better 29 37 35 26 25 31 Same 60 52 55 66 62 56 Muchworse 11 11 10 8 13 14 Economic mobility(%population 15 and more) Downward 23 22 25 23 12 34 None 43 34 42 45 58 26 Upward 35 44 33 32 29 40 Power mobility (% population 15 andmore) Downward 6 3 5 6 3 10 None 10 10 9 11 6 12 Upward 85 86 87 82 91 78 Food security Foodcomisinnptionless than adequate, last month (% population) 59 35 52 68 66 54 Months with lowfood consumption, lastyear 3.6 1.8 3.7 3.7 4.2 3.9 Months with not enoughriceor maizeto eat duringlast year 3.6 1.8 3.7 3.8 4.2 3.9 Note: Thedirrricts of Oecussi, Rohonaro rind < '?JW Lima conslilute Ihr West; Baucau, Luutem and Viyueyuerepresenr the Imsl: andAilcu, Ainaro, Dili,Ermera, Liyuiccr,Akrinifiihi undManuluro hr1,iiig 10 rhe ('ewer. a/Borlled wufer, rap warer,pump, proreciecl iwll orpr(~1ected.spring. h/Flush toiler, rradirionul lurrine or .sepricrank d P u h l i c or priimte. d/ 211Oi/O2 ucudemicyeur. e/ (bi7sider.speopIe uged 15-64 undo Ius/ week ~ c c u lperiod.l .f/ Excludespeople who did not work la.sr w e b hiir do have a joh. g/ Excludes Dili/Buucau. h/ C'un.sider.s land c/a.s.sifi as Annual cr0p.s or,jrriiou~and Piontarion. i/ Wei,qhred,firs/ by ureu wirhin the household uiid then by household. j / A household i s considered a ~uh.si.sraii~e **** l