PROTECTING POOR AND VULNERABLE HOUSEHOLDS IN INDONESIA THE WORLD BANK, JAKARTA OFFICE Indonesia Stock Exchange Building Tower 11/12th Floor J1. Jend. Sudirman Kay. 52-53 Jakarta 12910 Tel: (6221) 5299-3000 Fax: (6221) 5299-3111 Website: www.worldbank.org/id THE WORLD BANK 1818 H Street N.W. Washington, D.C. 20433, U.S.A. Tel: (202) 458-1876 Fax: (202) 522-1557/1560 Website: www.worldbank.org Printed in February 2012. Designed by Hasbi Akhir (hasbi@aisukenet.com) Cover photograph and photographs on pages 13, 27, 53, 65 and 69 provided by Anne Cecile Esteve/ Matahati Productions/World Bank. Photographs on pages 39 and 91 provided by Ryca C. Rawung. Photographs on pages 59 and 83 provided by Rythia Afkar. Copyright protection and all other rights reserved. Protecting Poor and Vulnerable Households in Indonesia is a product of the staff of the World Bank. The findings, interpretations, and conclusions expressed herein do not necessarily reflect the views of the Board of Executive Directors of the World Bank or the Governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of the World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. For any questions regarding this report, please contact Vivi Alatas (valatas@worldbank.org) or Jon Jellema (jellema@worldbank.org). Australian THE WORLD BANK | BANK DUNIA AID za Kingdom of the Netherlands PROTECTING POOR AND VULNERABLE HOUSEHOLDS IN INDONESIA * Protecting Poor and Vulnerable Households in Indonesia Foreword Over the past decade Indonesia has made impressive strides in poverty reduction, cutting the overall poverty rate by over two-fifths since the turn of the decade (1999/2000). Even today, however, nearly 30 million people live below the official poverty line while an additional 65 million remain vulnerable to falling into poverty. The Government of Indonesia is committed to tackling these challenges while further accelerating the pace of poverty reduction. Social assistance initiatives and social safety nets play a central role in Indonesia's poverty reduction strategy as complements to continued sustainable macroeconomic growth and the generation of more and better job opportunities. Well-designed and effectively-implemented social assistance programs provide two key functions. First, they protect the poor and vulnerable from chronic destitution and the risk of impoverishment stemming from negative economic shocks. Second, they promote independence and productivity by encouraging households to make wise investments and by providing more effective strategies for households to improve their own livelihoods. Indonesia delivers a range of social assistance programs prioritized for poor and vulnerable households. Until now, however, little was known about how well these programs protect and promote families and individuals. To provide answers to these and related questions, Protecting Poor and Vulnerable Households in Indonesia quantifies and analyzes patterns of public spending on social assistance and comprehensively reviews the effectiveness of each of Indonesia's main social assistance programs. The findings herein will help guide reforms for social assistance programs that work smarter and more efficiently to help those most in need. Emerging as a middle-income country with a strong recent record of growth and sound macroeconomic and financial management, Indonesia is well-placed and ready to take several steps forward in protecting and promoting the poor and vulnerable. This will require developing a new generation of social assistance programs, which expand upon and extend beyond the reach of today's programs, as well as knitting both new and old initiatives together into a coherent system that functions as a reliable social safety net for all households in all occasions. This report would not have been possible without close collaboration with partners in the Government of Indonesia, the research community and development partners. We look forward to further shared exploration and to understanding and applying what we have learned to find the right policy solutions for Indonesia. It is our sincere hope that this report will contribute to evidence-based policy making for Indonesia's social assistance programs. Together we can support Indonesian households who are paving their own way out of poverty and building a better future for themselves. Stefan Koeberle Country Director, Indonesia The World Bank 2 Acknowledgments Protecting Poor and Vulnerable Households in Indonesia is a product of the World Bank's Poverty cluster, a unit in the Poverty Reduction and Economic Management (PREM) group of the World Bank Office Jakarta. The team, led by Vivi Alatas (Senior Economist, EASPR), provides technical and policy advice based on sound empirical research and analysis to the Government of Indonesia in support of national poverty reduction goals. Support for this report has been generously provided by the Australian Agency for International Development, the Embassy of the Kingdom of the Netherlands in Indonesia. This report was prepared by a core team led by Jon Jellema (Economist, EASPR) and Hassan Noura (Economist, EASPR), overseen by Vivi Alatas. The program review was led by Jon Jellema with support from Rythia Afkar (Research Analyst, EASPR) and Edgar Janz (EASPR); the public expenditure review was led by Hassan Noura, supported by Dhanie Nugroho (Research Analyst, EASPR) and Retno Sri Handini (Consultant, EASPR), and overseen by Enrique Blanco Armas (Senior Economist, EASPR) and Anna Gueorguieva (Economist, EASPR). Additional input, analysis, and helpful comments by individuals and groups from the World Bank Office in Jakarta were made by Sjaiful Bahri, Victoria Beard, Budi Hidayat, Sjamsu Rahardja, Kiyoshi Taniguchi, and Hendratno Tuhiman (PREM); Claudia Rokx, Pandu Harimurti, Eko Pambudi, and Puti Marzoeki (Human Development group, Health cluster); Dandan Chen, Pedro Cerdan-Infantes, and Samer AI-Samarrai (Human Development group, Education cluster); and Robert Wrobel (Sustainable Development group). John Giles (DECHD) and co-author Elian Satriawan (the National Team for Accelerating Poverty Reduction, or TNP2K) kindly provided an advance summary of their impact analysis. Asmeen Khan (EAPVP) and Sara Giannozzi (EASHD) provided close collaboration and advance copies of analysis during their research on social safety net governance in East Asia. Research partners from other institutions provided very useful background papers and additional analysis. These included Irwanto and Santi Kusumaningrum from the Centre on Child Protection at the University of Indonesia on PKSA, Evi Martha and Bob Alfiandi from the Centre for Health Research at the University of Indonesia on the Jamkesmas program, and Dayl Donaldson on the history and evolution of social assistance in Indonesia. This report has benefitted greatly from excellent comments from Peer Reviewers Margaret Grosh (Lead Economist, LCSHD), Kathy Lindert (Sector Manager, ECSH3) and Andrew Mason (Lead Economist, EASPR), as well as from Ivailo Izvorski (Lead Economist, EASPR), Sudarno Sumarto and Julia Tobias (TNP2K), and Scott Guggenheim, Lisa Hannigan, Stephen Kidd, and Nicholas Freedland (AusAID). Invaluable editing support was graciously provided by Mia Hyun and Edgar Janz. Indispendsable logistical and production support was provided by Arsianti, Elisabeth Yunita Ekasari, and Deviana Djalil (PREM). We would like to thank Indra Irnawan (EXT) and Hasbi Akhir and his team at Aisuke Graphic House for designing the layout of the final report. Much of this report is based on work done in conjunction with partners from the Government of Indonesia. In particular, colleagues from Badan Perencanaan dan Pembangunan Nasional (Bappenas), the National Development Planning Agency (Ceppie Kurniadi Sumadilaga, Pungky Sumady, Rudy Prawiradinata, Endah Murniningtyas, Prasetiono Widjojo, Vivi Yulaswati, Utin Kiswanti, and Dinar Kharisma); Tim Nasional Percepatan Penanggulangan Kemiskinan (TNP2K), the National Team for Accelerating Poverty Reduction (Bambang Widianto, Suahasil Nazara, Sudarno Sumarto, Julia Tobias and Fiona Howell); Kementerian Sosial (Kemensos), the Ministry of Social Affairs (Makmur Sanusi, Harry Hikmat and the Directorate of Rehabilitation and Services for Children; Justina Dwi Noviantari and the Directorate of Rehabilitation and Services for the Disabled; Yulia Suhartini and the Directorate of Rehabilitation and Services for Elderly; the Center for Data and Information in the the Directorate General of Social Rehabilitation and Services; and Dwi Heru Sukoco, Anto Roy, and Harapan Lumban Gaol in the Program Keluarga Harapan Secretariat); Kementerian Pendidikan dan Kebudayaan (Kemdikbud), the Ministry of Education and Culture (Profesor Suyanto, Taufik Hanafi, Bambang Indriyanto, Nono Adya, and Keri Darwindo); and Kementerian Keuangan (Kemenkeu), Ministry of Finance (Lukas Lantip Ciptadi and Yani Farida Aryani in the Directorate of Treasury System). Protecting Poor and Vulnerable Households in Indonesia was produced under the overall guidance of Vikram Nehru (Sector Director, EASPR) and Shubham Chaudhuri (Lead Economist, EASPR). Strategic guidance and key comments were also provided by Stefan Koeberle (Country Director, Indonesia). This report also includes work from a sister report, Targeting Poor and Vulnerable Households in Indonesia. 3 * Protecting Poor and Vulnerable Households in Indonesia Table of Contents Foreword 2 Acknowledgments 3 Table of Contents 4 List of Abbreviations, Acronyms and Indonesian Terms 7 Key Messages 10 Executive Summary 12 Introduction 26 Indonesia Vulnerability Profile 26 Role of Social Safety Nets 29 Brief History of Social Assistance in Indonesia 32 Current Portfolio of Household-based Social Assistance Programs 33 Purpose of Report 36 1. Does Indonesia allocate the right level of resources to household social assistance? 38 Trends in Social Assistance Expenditures 38 Composition of Social Assistance Expenditures 45 Overall Adequacy 48 2. Do programs provide the right benefits? 52 3. Are benefits reaching the right people? 58 4. Do people receive the benefits at the right time? 64 5. Are programs implemented in the right way? 68 Spending on Support Operations 68 Socialization and Information Dissemination 71 Facilitation and Outreach 73 Monitoring and Evaluation 74 Budget Execution 76 Local Implementation Arrangements 78 Governance 79 6. Does Indonesia have the right programs and system in place? 82 Positive Impacts in Current Programs 82 Efficiency and Cost-effectiveness 86 Coordination Within and Across Programs 88 7. Recommendations for achieving a social safety net in Indonesia 90 Re-engineering Implementation 90 Covering More Risks and More Households 93 Protecting Households from Health Risks 97 Promoting Continuous Education to Break the Inheritance of Poverty 98 Protecting Incomes and Preventing Negative Coping 99 Transitioning to an Integrated Safety Net 102 References 104 4 List of Figures Figure 1. Indonesia Per Capita Consumption Distribution, 2011 13 Figure 2. Poor, Near-Poor and Newly Poor Individuals, 2010 14 Figure 3. Exposure to Poverty, 2008-2010 14 Figure 4. Public Expenditures on Household Social Assistance, 2001-2010 17 Figure 5. Household Social Assistance Expenditure Composition, 2010 18 Figure 6. Indonesia Per Capita Consumption Distribution, 2011 27 Figure 7. Poor, Near-Poor and Newly Poor Individuals, 2010 28 Figure 8. Exposure to Poverty, 2008-2010 28 Figure 9. Education Attainment by consumption quintile, 2000 and 2010 28 Figure 10. Public Expenditures on Household Social Assistance, 2001-2010 39 Figure 11. Expenditure Shares by level of government, 2010 39 Figure 12. Poverty Alleviation Expenditure by cluster, 2007-2010 41 Figure 13. Social Protection Expenditure by type, 2007-2010 41 Figure 14. Expenditures on Household SA (report definition) versus GOI-defined Social Assistance and Social Protection, 2005-2010 41 Figure 15. Sectoral Composition of National Public Expenditures, 2001-2010 43 Figure 16. Indonesia Debt to GDP Ratio, 2000-2010 44 Figure 17. Regional Debt to GDP Ratios, 1991-2010 44 Figure 18. Public Expenditures on Energy Subsidies, 2004-2010 44 Figure 19. Public Expenditures on Energy Subsidies and Household Social Assistance, 2008 and 2010 44 Figure 20. Household Social Assistance Expenditure Level and Composition, 2004-2010 45 Figure 21. Household Social Assistance Expenditure Shares, 2010 45 Figure 22. Social Assistance Expenditure Composition, ECA Averages 47 Figure 23. Income Gaps and Total Social Assistance Benefits Transferred, 2010 49 Figure 24. Projected Social Assistance Expenditures, 2011-2014 50 Figure 25. Social Assistance Expenditure Levels in International Comparison 50 Figure 26. Household Education Expenditures and Benefit Amounts by school level, 2009 54 Figure 27. Raskin Rice Procured and Total Raskin Purchases, 2006-2009 55 Figure 28. Raskin Promised Benefits and Actual Raskin Purchases 55 Figure 29. Quarterly HH Education Expenditures per Student versus BSM and PKH Benefit Amounts, by school level 66 Figure 30. Social Assistance Administrative Costs 70 Figure 31. Beneficiary Knowledge: Jamkesmas and PKSA 72 Figure 32. Indonesia's Monitoring and Evaluation Landscape 76 Figure 33. Budget Execution Rates in Social Assistance Programs, 2006-2010 77 Figure 34. Budget Execution Rate in Jamkesmas, 2006-2010 77 Figure 35. Budget Disbursement Profiles: PKH, Jamkesmas, and PKSA, 2009 77 Figure 36. Budget Disbursement Profiles: BSM, Raskin, and JSLU, 2009 77 Figure 37. Social Assistance Non-benefit Costs 87 Figure 38. Share of Total Social Assistance Benefits by consumption decile 87 Figure 39. Requirements for a Social Safety Net: Comprehensive Health Coverage 94 Figure 40. Requirements for a Social Safety Net: Continuous Education 95 Figure 41. Requirements for a Social Safety Net: Basic Income Protection 96 Figure 42. Proposed Social Assistance Expenditure Composition with BLT 101 Figure 43. Proposed Social Assistance Expenditure Composition without BLT 101 5 * Protecting Poor and Vulnerable Households in Indonesia List of Tables Table 1. Poverty and Vulnerability Headcount Rates, 2008-2011 13 Table 2. Indonesia Household-centered Social Assistance Program Summary 16 Table 3. Poverty and Vulnerability Headcount Rates, 2008-2011 27 Table 4. Who should safety nets cover? 30 Table 5. What can safety nets do? 30 Table 6. Indonesia Social Assistance at a Glance: the Major Household Programs 35 Table 7. National Public Expenditures on Social Assistance, 2004-2010 40 Table 8. Central Government Expenditures on Social Assistance by program, 2004-2010 46 Table 9. Social Assistance Benefit Evaluation 53 Table 10. Targeting Indicators in Social Assistance Programs 60 Table 11. Social Assistance Programs Received by poverty category, 2009 61 Table 12. Share of Social Assistance Complaints by identity 62 Table 13. Common Reasons for Social Assistance Complaints 62 Table 14. Social Assistance Benefit Timing 65 Table 15. Administrative Costs in Social Assistance Programs 70 Table 16. Decentralization and Social Assistance Integrity: the Brazilian example 78 Table 17. Summary of Positive Impacts in Social Assistance Programs 83 Table 18. Social Assistance Program Evaluation Matrix 85 Table 19. Proposed Social Safety Net: Key Assumptions 101 List of Boxes Box 1. The Role of Safety Nets in Middle Income Countries 31 Box 2. Analyzing Expenditure in Household Social Assistance in Indonesia 41 Box 3. Community-based development and PNPM 48 Box 4: Monitoring and Evaluation Activities in Five Indonesian Ministries 75 Box 5. Governance in Indonesia's Social Assistance Programs 80 Box 6. Integrated Social Assistance Architecture: Brazil 103 6 List of Abbreviations, Acronyms and Indonesian Terms AFC Asian Financial Crisis APBN Anggaran Pendapatan dan Belanja Negara (Central Government Budget) Askes Asuransi Kesehatan (Health insurance for government employees including military and pensioners) Askeskin Asuransi Kesehatan Masyarakat Miskin (Health insurance for the poor, now Jamkesmas) Bappenas Badan Perencanaan dan Pembangunan Nasional (National Development Planning Agency) BFP Bosa Familia Program BKKBN Badan Koordinasi Keluarga Berencana Nasional (National Family Planning Coordinating Agency) BLT Bantuan Langsung Tunai (Unconditional cash transfer) BN/bn Billion BOS Bantuan Operational Sekolah (School Operational Grants) BPS Badan Pusat Statistik (Centra Statistics Agency - Statistics Indonesia) BSM Bantuan Siswa Miskin (Cash transfer for poor students) Bulog Badan Urusan Logistik (National Logistics Agency) CCT Conditional Cash Transfer DG Directorate General Dinsos Dinas Sosial (Regiona level Kemensos) EAP East Asia and the Pacific ECA Eastern Europe and Central Asia ECD Early Childhood Development eKTP Kartu Tanda Penduduk Elektronik (Electronic identity card) GDP Gross Domestic Product GFC Global Financial Crisis (circa 2008) GOI Government of Indonesia HH/hh Household ICCU Intensive Coronary Care Unit ICU Intensive Care Unit IFLS Indonesian Family Life Survey Jamkesda Jaminan Kesehatan Daerah (Local leve health insurance scheme for the poor) Jamkesmas Jaminan Kesehatan Masyarakat (Health insurance scheme for the poor, formerly Askeskin) JPS Jaring Pengaman Sosial (Social safety net) JSLU Jaminan Sosial Lanjut Usia (Social cash transfer for the elderly) JSPACA Jaminan Social Penyandang Cacat Berat (Social cash transfer for the severely disabled) Kemdagri Kementerian dalam Negeri (Ministry of Home Affairs, MOHA) Kemdikbud Kementrian Pendidikan dan Kebudayaan (Ministry of Education and Culture, MOEC) Kemenag Kementerian Agama (Ministry of Religious Affairs, MORA) Kemenkes Kementerian Kesehatan (Ministry of Health, MOH) Kemenkeu Kementerian Keuangan (Ministry of Finance, MOF) Kemenkokesra Kementrian Koordinator Kesejahteraan Rakyat (Coordinating Ministry for Social Welfare) 7 E Protecting Poor and Vulnerable Households in Indonesia Kemenkomlnfo Kementrian Komunikasi dan Informatika (Ministry of Communications and Information Technology) Kemensos Kementerian Sosial (Ministry of Social Affairs, MOSA) KPP & PA Kementerian Pemberdayaan Perempuan dan Perlindungan Anak (Ministry of Women's Empowerment and Child Protection) LAC Latin America and Caribbean LHS Left hand side (of graph) LKSA Lembaga Kesejahteraan SosialAnak (Social Welfare Institution for Children) M&E Monitoring and evaluation Menpan Kementerian Pendayagunaan Aparatur Negara dan Reformasi Birokrasi (The State Ministry for Administrative and Bureaucracy Reform) MIS Management Information System MN/mn Million MTDP Medium-Term Development Plan NTS National Targeting System OECD Organization for Economic Co-operation and Development OPK Operasi Pasar Khusus (Special market operation for rice) PAUD Pendidikan Anak Usia Dini (Early Childhood Education) PKH Program Keluarga Harapan (Conditional Cash Transfer) PKSA Program Kesejahteraan SosialAnak (Social cash transfer for disadvantaged children) PL Poverty Line PNPM Program Nasional Pemberdayaan Masyarakat (Umbrella organization for all PNPM and community-driven development initiatives) PNPM-Generasi PNPM Generasi Sehat dan Cerdas (PNPM Healthy and Smart Generation Program) PNPM-Mandiri Program Nasional Pemberdayaan Masyarakat Mandiri (National Community Empowerment Program) PPLS Pendataan Program Perlindungan Sosial (Data collection for targeting social protection programs) PSE Pendataan Sosial Ekonomi Penduduk (Socio-Economic Population Survey) PT Pos Perseroan Terbatas Pos Indonesia (National post office system) Puska PA Pusat Kajian Perlindungan Anak (Centre on Child Protection) Puskesmas Pusat Kesehatan Masyarakat (Community health center) RAS Red de Apoyo Social (Social Support Network, Colombia) Raskin Beras Miskin (Program for sale of subsidized rice to the poor) RHS Right hand side (of graph) Rp Indonesian Rupiah RPJM Rencana Pembangunan Jangka Menengah (Medium-Term Development Plan, MTDP) SA Social Assistance SD Sekolah Dasar (Elementary school) SJSN Sistem Jaminan Sosial Nasional (National Social Security System) SMA Sekolah Menengah Atas (Senior secondary school, now SMU) SMERU SMERU Research Institute SMP Sekolah Menengah Pertama (Junior secondary school) SMU Sekolah Menengah Umum (Senior secondary school, formerly SMA) SP Social Protection 8 SSN Social Safety Net Susenas Survei Sosio-Ekonomi Nasional (National Socio-Economic Survey) Taspen Tabungan dan Asuransi Pegawai Negeri (Civil servant pension savings and insurance) TKPK Tim Koordinasi Penanggulangan Kemiskinan (Coordination team for poverty reduction) TNP2K Tim Nasional Percepatan Penanggulangan Kemiskinan (National Team for Accelerating Poverty Reduction) UCT Unconditional Cash Transfer UKP4 Unit Kerja Presiden Bidang Pengawasan dan Pengendalian Pembangunan (The Presidential Working Unit for Supervision and Management of Development) UPPKH Unit Pelaksana Program Keluarga Harapan (PKH implementation unit) US$ United States Dollars Yanrehsos Pelayanan dan Rehabilitasi Sosial (Social Rehabilitation & Services) 9 * Protecting Poor and Vulnerable Households in Indonesia Key Messages Though absolute poverty is declining, 40 percent of the Indonesian population remains highly vulnerable to shocks that threaten to push them into poverty. In 2011, 12.5 percent of Indonesians lived below the national poverty line, but a large portion of the population is clustered just above the poverty line and is prone to entering poverty. Estimates show that half of all poor households in recent years were not poor the year before, and over four-fifths of next-year's poor will originate from the 40 percent of households with the lowest expenditure levels. Social assistance programs play an important role in helping poor households escape destitution while reducing the likelihood that vulnerable households will be pushed into poverty. The Government of Indonesia has developed several household-based social assistance (SA) programs targeting the poor and near-poor; these households make up roughly the bottom 25 percent of the population. Household-based program development has been rapid and these initiatives have, with varying degrees of success, provided some protection for the poor and vulnerable. Indonesia also has a range of complementary programs and policies that extend beyond the household to "protect and promote" the poor and vulnerable, including community- driven development programs, job creation and employment strategies, and plans for social security. Despite demonstrated promise, much work remains to be done in the loose collection of household-based programs. The current range of SA programs does not go far enough in protecting the 40 percent of the population with the highest risk of falling into poverty. In addition to significant gaps in both risk and population coverage, all of the household-based programs have been limited in their effectiveness due to (a) an insufficient ability to find and prioritize poor or vulnerable households; (b) a total benefit package that is sometimes underfunded, sometimes inadequate for addressing the particular household need or risk, and sometimes delivered with less-than-optimal timing; (c) a passive and implicit reliance on poorly-equipped local implementation partners combined with little explicit financial or technical support; (d) weakly-monitored and insufficiently-detailed implementation procedures; or in many cases a combination of all four of these. The first step on the way to a dynamic and responsive social safety net should be reform within these currently available programs. Meanwhile, Indonesia will need to go beyond program reform to create a social safety net that is capable of providing consistent, high-quality, and comprehensive coverage. The current range of SA programs provides partial and non-guaranteed protection to the poor and vulnerable from some, but not all, of the risks faced. There are risks that are not yet covered by any program - for example, risks due to sudden job loss or underinvestment in early childhood education. However, even among the important risks that are addressed by current programming, the likelihood that an eligible household will consistently receive all benefits is small, while the facilitation, outreach, and information dissemination that are necessary to ensure households with any type of background use programs effectively are not consistently provided. A true social safety net will involve system-wide planning and coordination between programs and agencies in order to ensure that all types of eligible households are reliably protected for all important risks. Indonesia confronts these challenges from a position of strength and can create gains for all through better protection of vulnerable households. Indonesia benefits from a strong macroeconomic and fiscal position and an administration committed to poverty reduction and social protection, allowing it to undertake comprehensive reforms from an enviable position of strength. In addition to ensuring that poor households are more effectively protected from shocks, such reforms will contribute to Indonesia's continued economic strength by promoting pro-poor investments in human capital and a healthy, educated, and productive workforce. An effective and efficient social safety net will also enable further government policy reform by alleviating the burdens that reform can create for the least well-off. 10 The following recommendations outline some of the steps necessary for the creation of a social safety net system in Indonesia: 1. First, spend better by improving programs and achieving a more optimal mix of initiatives. Increase the benefit level and delivery schedule of the cost-effective conditional cash transfer program (Program Keluarga Harapan, PKH); institute a package of radical reforms for stopping leakage and improving targeting in the subsidized rice program (Beras untuk Keluarga Miskin, Raskin), which delivers too little at high cost; upgrade capacity for the pilot cash transfers targeting highly vulnerable groups; re-engineer the scholarship program (Bantuan Siswa Miskin, BSM); and redefine an appropriate benefit package for the health fee waiver program (Jaminan Kesehatan Masyarakat, Jamkesmas) in order to provide financially sustainable and reliable health care utilized by all poor households. 2. Then, scale up to protect more households from health risks, promote continuous education and protect from shocks threatening welfare. Expand Jamkesmas and BSM to reach all vulnerable households, and introduce a pilot early childhood education program. Scale up PKH to reach all chronically poor households and the collection of programs that target marginalized populations. Right-size Raskin to cover only poor households. Fill existing gaps in the social safety net by adding a coordinated emergency response system, featuring a revised version of BLT (Bantuan Lansung Tunai, Unconditional Cash Transfer) that includes conditions for community service. Such expansion to all vulnerable households is estimated to require an increase in social safety net spending levels from 0.5 percent to just less than 1 percent of GDP. 3. Integrate the social safety net by consolidating program support operations under a single roof and encouraging single window household access to all services. Consolidate support operations (e.g. socialization, complaints handling and M&E) under one roof and develop a single National Targeting System (NTS). Create a reliable public face for the social safety net under a single agency with employees that perform outreach and socialization activities and can encourage and facilitate access to all initiatives available in the social safety net and beyond. 11 Executive Summary Despite strong economic growth and falling poverty in the last decade, there are many households on the edge of poverty. The last decade in Indonesia has seen a return to strong economic growth, and the poverty rate has fallen from 23.4 percent (1999) to 12.5 percent (2011). Declining poverty, however, partially masks a high degree of vulnerability: much of Indonesia's population is clustered just above the 2011 poverty line of Rp 233,000 per month (about US$ 27 at 2011 nominal exchange rates). Around 24 percent of Indonesians live below the official near-poor line of 1.2 times the poverty line while 38 percent of the population lives below 1.5 times the poverty line and is almost equally vulnerable (Figure 1 and Table 1). Even relatively small shocks to these vulnerable households can be enough to push them into poverty. Vulnerable households experience income insecurity and frequently fall in and out of poverty. In Indonesia recently, approximately half of all poor households are chronically poor, or consistently measured as poor in all of three consecutive years. The remaining poor households (in any given year) are households that are highly likely to be moving into and out of poverty. For example, of those who were not measured as poor in 2009, 12.6 million had fallen into poverty status by 2010; these 12.6 million individuals made up half of all poor individuals in 2010. Over four-fifths of these poor households originated from the group of vulnerable households below 1.5 times the poverty line (the bottom 40 percent). This high level of income churning among vulnerable households, and the large population movements into and out of poverty, are a stubborn feature of poverty: in the last three years, over a quarter of all Indonesians have been in poverty at least once while 43 percent fell below the official near-poor line at least once (Figure 3). 1 Statistics Indonesia (Badan Pusat Statistik, BPS) defines the poverty line as the amount required to obtain 2,100 calories per day from local food commodities and a small amount for other basic necessities, such as clothing, housing, and transportation. In 2010, the poverty line was around Rp 211,000 per month or Rp 7,033 per day. Near-poor is defined as 1.2 times the poverty line. In 2010, the near poor line was around Rp 250,000 per month, or Rp 8,400 per day. 12 ―鰍 F igurel·賽口I Indonesia浴一叫 Per capita一11-h_ consumpt!on黑一目―-―口― &&”『’&u&‘。n&&0&！訛區 0 200000 400000 600000 800000 Monthly Household Per Cap-ta Consumptlon(Rp) Tablel·Poverty Poverty Rate(o,o) 叩d Yulner中ility Povert, L.ne ou.tiP．。，005,009之。，o,o,, 湍寫’&a&e&‘織滬口農蠶 jou几e馴Susenas,varlous ye己rs 陶弼：The natlonal povertyl-ne 19 around Rp 2BB700 per person per month-n 2011 -13 E Protecting Poor and Vulnerable Households in Indonesia Figure 2. Poor, Near-Poor and Newly Poor Individuals, 2010 100 - 80 72 _0 S 60 6 0 40 0 P 2 20 Poor in 2010 Near Poor in 2010 Existing Poor (Already Poor in 2009) Newly Poor (Not Poor in 2009) Figure 3. Exposure to Poverty, 2008-2010 100 - m 80 - 60 - 40 - 20 - 0 Number of Times Poor Number of Times Near 2008-2010 Poor 2008-2010 Never Poor Poor Once Poor Twice Poor Three Times Sources: Susenas and World Bank calculations. Indonesia's challenge is double: helping poor households escape impoverishment while protecting the 40 percent of Indonesians who are highly vulnerable. Policies and programs must be tailored to fit the Indonesian context, which is characterized by a high level of vulnerability and churning near the poverty line and marginal but frustratingly slow improvements in social indicators among poor households. Social safety nets, which consist of nol contributory cash or in-kind transfer programs targeting the poor and vulnerable, are designed to directly respond to such challenges. They are one component in a social protection suite, which typically also includes social insurance, active labor market programs, and provision of high-quality, low-cost education and health services accessible to all. Safety nets serve three main functions: 1. Protect households from destitution and catastrophic human capital loss: Social safety nets can provide direct income support and reduce inequality. They can also reduce the likelihood of poor and vulnerable households resorting to negative coping strategies, such as pulling children prematurely from school to enter the workforce. 2. Promote opportunities, livelihoods, and better jobs: Social safety nets can also be used to ensure that poor and vulnerable families increase investments in productive assets, including in human capital like education and health. These investments not only sever the transmission of poverty to future generations but leave households and families better prepared in terms of ex ante risk reduction strategies like saving and other financial management tools. 3. Preparing for progressive reforms: Safety nets may help government replace inefficient redistributive policies in other sectors, or successfully reorient macroeconomic policy and structure to improve growth. For example, reorienting spending towards progressive transfers and providing consumption support during the acute inflationary environment that follows a subsidy reduction can help sustain pro-poor reforms. 14 Executive Summary U Indonesia has introduced a range of SA programs forming the potential foundation of a true social safety net. The first generation of programs was borne of the 1997-98 Asian Financial Crisis (AFC) when the government introduced a number of temporary initiatives to protect the poor from the large negative shocks buffeting the Indonesian economy. A second generation of more permanent programs was introduced in 2005 to help usher in fuel subsidy cuts; savings from reduced subsidy spending were channeled to programs to help poor and near-poor households cope with the inflationary shock caused by the increase in regulated fuel prices. More recently, the government has piloted and expanded programs that have a greater emphasis on the promotion of health and education services by poor and vulnerable families. Programs launched over the past decade, but especially those introduced during the 2005 reforms, could provide the foundation for a true social safety net targeting poor and near-poor households Today, social assistance is concentrated in eight household-based programs which are all primarily designed, funded, and executed by the central government. A temporary unconditional cash transfer program (BLT) was deployed in 2005-06 to mitigate the inflationary impact caused by fuel price adjustments and again in 2008-09 to protect vulnerable households from further fuel price adjustments and the effects of the global financial and food price crises. Raskin distributes subsidized rice to 17.5 million families across the country. Jamkesmas provides health service fee waivers for 18.2 million poor and vulnerable households. A scholarship program (BSM) provides cash assistance to approximately 4.6 million students across the country. PKH - a conditional cash transfer - provides income support and investment in health and education services for over 800,000 extremely poor households in pilot areas. Finally, there are cash transfers with facilitated services for highly vulnerable groups including at-risk children (Program Kesejahteraan Sosial Anak, PKSA), the disabled (Jaminan Social Penyandang Cacat Berat, JSPACA) and vulnerable elderly (Jaminan Sosial Lanjut Usia, JSLU). Each of these eight programs has a unique government authority and provider located primarily in one of five central government agencies (Figure 3). The Government of Indonesia has demonstrated a commitment to strengthening social assistance programs as part of its broader social protection and poverty reduction strategy. The current administration's Medium-Term Development Plan (MTDP) for 2010 to 2014 aims to accelerate poverty reduction and reduce income inequality; MTDP goals include a headcount poverty rate between 8 and 10 percent by 2014. The MTDP lays out strategies to achieve this goal, one of which is the development of a "family-centered" social assistance system and reforms to priority programs. Simultaneously, the government is expanding and improving other programs with social protection and poverty reduction elements including: social security reform, community-based programs, credit provision for micro- and small-enterprises to stimulate job creation, and other active labor market programs. This report, the first comprehensive assessment of its kind in Indonesia, assesses the extent to which current social assistance programs are providing an effective social safety net for poor and vulnerable households. The government and its development partners require an analytical base to inform their decisions about social assistance policy reform and program design and delivery.2 To support this, the report uses all available qualitative and quantitative data (including the most recent) to assess the extent to which the current collection of SA programs is providing effective safety net functions: protecting the poor and vulnerable; promoting good behaviors, and enabling reforms effectively and efficiently. In order to answer this overarching question, six intermediate questions are asked': 1 Does Indonesia allocate the right level of resources to household social assistance? 2 Do programs provide the right benefits? 3 Are benefits reaching the right people? 4 Do people receive the benefits at the right time? 5 Are programs implemented in the right way? 6 Does Indonesia have the right programs and system in place? 2 Though this report focuses solely on household-centered social assistance programs, it recognizes that improvement in other areas mentioned will be critical for continued reduction of poverty and vulnerability. 3 Throughout this report "right" is used as shorthand to indicate effectiveness or efficiency and is not meant to be taken as a normative indicator of "correct", "proper", or even "meeting a pre-defined standard". For example, the "right" time to deliver benefits is when they are needed and when they can and will be used as intended; similarly the "right" benefits are not a certain percentage of median incomes, but rather benefits that allow households to achieve what the program intends for them to achieve. The report will clarify this usage in the course of elaborating on each of the six different "rights" mentioned here. 15 * Protecting Poor and Vulnerable Households in Indonesia 0 -0 C> C) C) o00 C cn c .)( N CIc E E E o E C- C> e D c c 0 c 0 *- 00E-o m 0 >0 t 0 Qj 0 0 0 .. 0. o À|? E' 0 9 0 0 ** > o o N * Ifl u U uu C) ID ID > ' 0)C) V * 0 u- ud « uu U o 16 Executive Summary U 1. Does Indonesia allocate the right level of resources to household social assistance? Spending on social assistance has significantly increased over the past decade, supported by fiscal consolidation. From a low base in the early 2000s, Indonesia's aggregate national public expenditures on SA programs permanently increased from 2005, in line with the proliferation of individual initiatives beginning then. At the same time, the Government of Indonesia (GOI) has also been increasing its expenditures on social insurance, but these mainly cover civil servant pension and health premiums. Overall, of the 1.2 percent of GDP spent on social protection (social assistance plus social insurance) in 2010, about one-third went to household-based social assistance and two-thirds to social insurance. Increased fiscal space - a result of starkly declining debt payments4 - has left room for further increases in SA spending. However, regressive energy subsidies, which in some years cost as much as 4.5 percent of GDP, continue to dwarf spending on SA programs. Indonesia spends 0.5 percent of GDP on SA, which is low in comparison to regional peers and middle-income developing countries. National expenditures on SA programs are estimated at almost Rp 30 trillion (US$ 3.3 billion) in 2010, equivalent to 2.6 percent of total national expenditure or 0.5 percent of GDP (Figure 4). According to the MTDP for 2010 to 2014, modest expansion plans for most of the household-based SA programs results in national expenditures flatlining at their current relative level (0.5 percent of GDP). The average developing country, on the other hand, spends around 1.5 percent of GDP on social assistance. The average for East Asian countries is 1 percent. Latin America countries - where safety nets are relatively comprehensive - spend, on average, 1.3 percent of GDP. Central government spending consistently accounts for almost 90 percent of total Indonesia-wide public SA expenditures. Sub-national governments account for just over 10 percent of total national SA expenditures, the majority of which appears to be absorbed by staff salaries and general administration in support of the major GOI programs. Figure 4. Public 50,000 - Expenditures on Household Socialr 40,000 - Assistance, 2001-2010 30,000 --3 20,000 --2 10,000 -- 1 0 =' 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Nominal expenditure (LHS) Share of total expenditure (RHS) Share of GDP (RHS) Sources: Kemenkeu, BPS and World Bank staff. 4 Indonesia's debt-to-GDP ratio declined from almost 90 percent in 2000 to just 25 percent by the end of 2010 and is expected to decline further in the years ahead. 17 E Protecting Poor and Vulnerable Households in Indonesia Figure 5. Household Disarged* Elderpy* Other SP Other Social Assistance 08 0.. Kemensos Expenditure Chi dren*tres- po Composition, 2010 PK 4% BSM 14% Raskin 53% Jamnkesmas 18% Sources and Notes: Kemenkeu and World Bank staff* Refers to all spending in the directorate implementing these programs. The majority of SA spending goes to income relief for poor and vulnerable households; smaller amounts are spent promoting productive behavior and human capital investment. Raskin, the single largest program, accounting for 53 percent of total SA expenditures, aims to protect households from food insecurity by delivering regular in-kind transfers (Figure 5). Both Jamkesmas and BSM scholarships - the next two largest programs, together accounting for about 32 percent of total SA expenditures - protect by providing income or no-cost healthcare services. Each could promote regular and effective healthcare or education service utilization, but as this report will show the promotive elements in both BSM and Jamkesmas are underdeveloped. The cash transfers designed to promote livelihoods and investments in human capital are allocated much smaller resource shares: PKH is allocated 4 percent and programs for marginal groups 2 percent of total national SA expenditures. These pilot programs are not yet allocated sufficient resources to reach all eligible beneficiaries, although it is an open question whether implementing agencies could effectively absorb the increased spending necessary for full coverage. In contrast, regions like Eastern Europe and Latin America, where safety nets are more mature, tend to allocate a significant majority of SA expenditures to targeted cash transfers for vulnerable families and marginal groups. Current SA expenditures appear low given the Indonesian risk and vulnerability profile described above. First, many social assistance programs do not yet havere to nondo osos o reach all eligible beneficiaries. Second, programs officially target poor and near-poor households, not the additional vulnerable households that are at risk of falling into poverty. In addition, each program prioritizes beneficiaries idiosyncratically, meaning many beneficiaries of one program will not receive other programs and few households are transferred benefits from all available programs and interventions. Third, total benefits transferred by major government SA programs represent just 60 percent of the cumulative income gap of tepoor households and just 10 percent of what would be needed to close the cumulative income gap of all vulnerable households living below 1.5 times the poverty line. Taking into account the actual allocation rules and targeting outcomes reduces these ratios further: as not all benefit spending reaches only intended poor and vulnerable households, actual benefits received by these households are a smaller proportion of their cumulative income gap. Finally, the obvious majority of spending is absorbed by an in-kind transfer (Raskin) with relatively small benefit levels and high levels of redistribution to non-poor households. If all programs were consistently reaching the same eligible households (as well as at least some vulnerable households), significantly more resources would need to be devoted to existing programs. Meanwhile, larger SA resource shares would need to be shifted from Raskin to programs that consistently deliver more significant benefits. 2. Do programs provide the right benefits? The main SA programs protecting the poor and vulnerable deliver only a fraction of the benefits promised or needed. In 2010, Raskin - the largest program by expenditure - promised beneficiaries 14 kilograms per month but only delivered an average of 3.8 kilograms per month. These amounts, when purchased at actual Raskin prices, represent between 2 and 3 percent of poor household expenditure, the lowest benefit level provided by any Indonesian SA program. Jamkesmas is generous by design, protecting households from health shocks by offering a fee waiver for nearly all medical services available at public hospitals and primary care centers. The program, however, does not provide enough of the facilitation and outreach that could make the benefit packages effective for poor households. For example, Jamnkesmas 18 Executive Summary U can not address costs that households identify as serious impediments in accessing health services (transport, lost wages, childcare, food and lodging for companion or chaperone). While cash transfer programs offer the right type of benefit, the amount is often not enough for households to invest in education services. Neither scholarship programs nor conditional cash transfers provide sufficient benefits for the needs of target households. For example, secondary education expenditures (including placement fees, transportation, and uniforms among others) can be as high as 20 percent of a poor household's annual income, which puts it well beyond the reach of beneficiary households even after transfers. A household receiving both PKH and BSM might find the total transfer adequate, but implementing agencies have in the past targeted different households and individuals. In addition, benefit amounts for most programs have never been adjusted for a rising cost of living and have remained unchanged at their initially set levels (going back as far as 2005 in some cases), meaning their real value to beneficiaries has declined by as much as 30 percent over time. The unconditional cash transfer (BILT) was successful in easing policy reforms and providing beneficiaries with the right benefits to help them cope with shocks. It provided beneficiary households with cash amounts equal to approximately 15 percent of regular expenditures. These transfers were more than enough to cover increased expenditure on fuels. Benefits continued for one year as shocks from government policy reverberated through the rest of the macroeconomy, allowing beneficiaries time to readjust spending patterns to new relative prices. Although BLT served as a good example of how SA can provide benefits that ease policy reforms, the government has used the BLT program only twice in over 6 years (since 2005). 3. Are benefits reaching the right people? A significant number of poor households are excluded from beneficiary lists. Overall, the poorest households are more likely to receive SA benefits. However, less than half of the poorest and most vulnerable 40 percent of households receive BLT and Jamkesmas (for example), while 20 to 25 percent of total benefits from both programs go to the richest 40 percent. Over 70 percent of the vulnerable receive Raskin, but Raskin also has high coverage of the non-vulnerable, a result of local-level Raskin sharing among all households. In a comparison of targeting outcomes, and with 100 percent representing perfect targeting according to program design, BLT performs the best at 24 percent better than random, with Jamkesmas and Raskin at 16 and 13 percent respectively. BSM performs quite poorly: the poorest 30 percent of students receive less than double the amount of BSM benefits received by the richest 30 percent. Indonesian program targeting, as measured by coverage of the poor, is in line with international benchmarks, but leakage to the richest households is much higher in Indonesia than elsewhere. Each program has developed its own beneficiary eligibility rules and targeting in practice has often strayed from these official guidelines.5 For example, BLT was meant to use a mix of data collection methods, but each step in the data collection procedure was carried out with significant revisions: statistical assessment of poverty status was not in-line with international best practice while community-based assessment was in most cases neither consultative nor transparent. Raskin is meant to use official lists of the poor to select beneficiaries, but in practice communities distribute the rice as they see fit, often sharing it out amongst many or all households. Jamkesmas is also meant to use official lists of the poor but there is considerable variation in beneficiary identities at the local level, with local health officials sometimes choosing beneficiaries, or households selecting themselves based on previous healthcare use. Different targeting approaches mean different beneficiaries for each program even though all major SA programs target the same populations. Poor socialization and mistargeting have undermined support for SA programs. The percent of communities experiencing protests over the programs ranged from 25 percent for Askeskin (now Jamkesmas), to 56 percent for BLT, with those not receiving assistance being the most likely to complain. Mistargeting, nepotism and a lack of transparency in, and poor socialization of, beneficiary selection were the main sources of complaints. The nature of the community protests suggests that improved targeting of programs would improve satisfaction and buy-in. Indonesia represents a complex targeting environment and improved data collection can enhance outcomes in all the household-based programs. Nearly 240 million individuals are dispersed across at least 18,000 islands and over 500 districts (each of which has considerable ownership and operational control of public spending and social sector programs since decentralization) in Indonesia. Targeting should be able to identify the chronically poor, the near-poor, and 5 Refer to Targeting Poor and Vulnerable Households in Indonesia (World Bank, 2012a) for a detailed review and discussion of targeting practices in Indonesia. 19 * Protecting Poor and Vulnerable Households in Indonesia the especially vulnerable (but not currently poor) in all these localities and across a consumption distribution that is tightly compressed near the poverty line. In 2011, a large survey which collected data from nearly 45 percent of Indonesian households has allowed BPS to meaningfully update its list of poor, near-poor and vulnerable households and families; it is hoped the PPLS1 1 (Pendataan Program Perlindungan Sosial, Data collection for targeting social protection programs) survey will also serve as the foundation for an initial Cluster 1 eligibility database and a unified beneficiary registry. This massive improvement in data collection, which combined results from previous lists of poor households with 2010 population census results and community nomination, is expected to result in significant targeting improvements over previous methods. 4. Do people receive the benefits at the right time? The largest SA programs performed well in delivering benefits to households when needed. BLT was well- timed, reaching households during the month when the largest increases in fuel prices occurred and were quickly spent. Jamkesmas is always available to households if they can cover the supplementary costs of access. Raskin is also continuous, with subsidized rice delivered monthly. However, local-level implementation practices - with rotation and sharing of rice amongst households regardless of strict eligibility - negatively impact Raskin's dependability for poor and vulnerable households. Implementation issues often prevent benefits from reaching beneficiaries at the right time. PKH faced bottlenecks in early years because of partial and slow management information systems (MIS) systems resulting in delayed and ill-timed payments. These problems have since been addressed, but PKH's effectiveness would be enhanced further through better synchronization of benefit amounts with the chronological profile of a household's needs; this is especially true for education expenditures, which are predictably larger at the beginning of a school year. PKSA, JSLU and JSPACA payments only reach beneficiaries in the second half of the year, resulting in benefit-bunching that reduces any consumption-smoothing effects and encourages large one-time expenditures. Lessons from the PKH experience can be useful in improving the delivery of benefits from these other programs. Timeliness is sometimes weakened when design issues reinforce the negative effects of slower implementation. BSM is delivered in one lump-sum payment that arrives more than one year after enrollment and thus are not available to students in the final year at each level of schooling. The cash transfers provided to families through these programs, therefore, are absent at the beginning of the school year and during primary-to-secondary or within- secondary transition years, which is precisely when the greatest risk to, and sharpest increases in the costs of, continued education occur. Similar problems have been identified in the PKH program, which did not deliver payments just prior to the academic calendar when parents needed to pay school registration fees. This problem, which is a likely explanation for why the program did not have an impact on school enrollment rates among beneficiaries, will be fixed in upcoming payment cycles. 5. Are programs implemented in the right way? Most of the larger programs probably spend too little on administration and support operations. BSM, Jamkesmas and, to a lesser extent, BLT spend too little on administration to ensure good performance. The smaller cash transfer programs have higher administrative costs even when measured on a per beneficiary basis, and these costs are reasonable given the pilot status and small scale of the programs. Raskin - like most food programs around the world - spends much more on administration overall, although these expenditures are for physical transportation, distribution and packaging of rice rather than on support operations for beneficiaries. Weak socialization and lack of accountability controls result from underfunding of support processes. Too little effort is spent on the content, delivery, and oversight of safeguarding or supporting operations. All programs suffer from inadequate socialization guidelines, leading to reduced program transparency and legitimacy and heightened potential for corruption. Knowledge on eligibility rules, program objectives, and beneficiary rights and responsibilities is usually spread thinly among beneficiaries, eligible households, communities, and local-level program implementers. Therefore, bottom-up monitoring of the targeting and benefit distribution process is limited while intra-community jealousy and misunderstanding are often high. SA programs - with the exception of the pilot Kemensos (Kementerian Sosial Ministry of Social Affairs) cash transfers - do not include an explicit facilitation or outreach process. This limits beneficiaries' effective access and leads to increased capture by those already familiar with the services offered, especially for Jamkesmas and BSM. 20 Executive Summary U Few programs have embedded monitoring, evaluation, or complaint resolution mechanisms that function efficiently. All programs have descriptions (in regulations and manuals) of program monitoring arrangements and some details regarding the content of monitoring procedures and reports. However, program monitoring and reporting is most often carried out by local-level implementers and delegated with very little financial support, technical support, or systems for quality control. Monitoring and reporting most often produces information that is not useful for evaluating service delivery performance or household outcomes. Likewise, complaints and grievances processes are usually described but remain only weakly functioning and they are mostly unfamiliar to households and front-line providers. Both shortcomings constrain implementing agencies' ability to quickly and effectively remedy unwanted or unintended program outcomes. Some programs have weak budget execution and most exhibit unsmooth yearly disbursement. Many SA programs exhibited low budget execution rates in their early years, but some now disburse close to 100 percent of allocated budgets. Jamkesmas is an exception: it has seen a steady decline in its budget execution ratio in recent years partly as a result of underutilization and confusion caused by the proliferation of competing local schemes and corresponding regulations. Most SA programs exhibit slow and therefore unsmooth budget disbursement: benefit payments are often "bunched" in the second-half of the fiscal year making them less useful for consumption smoothing. The main reason for the delay is long bottom-up beneficiary identification and verification procedures, meaning payment authorization letters are rarely sent to the Treasury before May. Disbursement of funds to intermediaries typically begins in May or June and to beneficiaries shortly thereafter. PKH has in recent years exhibited the smoothest budget disbursement profile, helped by a strong MIS and advanced disbursement of funds followed by reconciliation. Other public financial management issues include lack of performance-based budgeting and bottom-up funds monitoring. Budget audit documents focus on budget execution rather than outcomes, and there is a lack of capacity to support performance-based budgeting. Leakage of funds is not yet a major issue in most programs - Raskin may be an exception - but benefit deductions and other fees are common during implementation and there are no efforts at rights and awareness campaigns that could encourage bottom-up funds monitoring. Finally, implementation is also affected by local-level politics and revisions. Local governments, agencies, service providers, and broader communities are asked to support various stages of most programs. Targeting, beneficiary verification, socialization, funds channeling, facilitation, monitoring and evaluation, and the complaints and appeals process are all areas where these actors may be involved. However, weak socialization and inconsistent follow-up mean that local actors are free to revise implementation procedures to suit what they feel is needed or desired by the community. This often means minimum service standards in each of the above-mentioned processes cannot be guaranteed and both implementation and outcomes will vary widely from region to region. 6. Does Indonesia have the right programs and systems in place? Conditional and unconditional cash transfer programs have effectively protected households from shocks, promoted good health and education behaviors and facilitated reform. BLT effectively protected households from the shock of fuel price increases and helped facilitate much needed subsidy reforms by delivering cash transfers at the right time. These transfers were spent on basic necessities and also provided a cushion for other good behaviors related to nutrition, education, child labor and health. BLT will benefit from further institutionalization and codification as an automatic stabilizer that is triggered by pre-defined crisis events as well as better provision of monitoring, a system for complaints and grievances, and clearer divisions of authority and incentives between implementing agencies. Although confined to a small set of households, the PKH pilot program has also produced positive impacts. Monthly household consumption increased by 10 percent (over and above initial levels); the largest shares of this increase went to food, especially high-protein foods, and health care. PKH's presence even produced more pre-natal visits and child weighings in non-beneficiary households living in PKH areas. PKH did not have an effect on drawing more children into school (enrollment rates), encouraging them to stay (dropout rates), or encouraging them to continue (transition rates) due to poor timing, relatively small benefits, and lack of outreach to school-leavers. PKH will benefit from continued attention to the entire benefit delivery process and management of the MIS system monitoring all subprocesses; the design and intensity of its collaboration with service providers and local governments; and capacity and quality upgrading in its facilitator corps. Other SA programs, however, are struggling to meet their overarching objectives. Jamkesmas has increased utilization of health services, but the effects are much larger for non-poor households and households with previous experience with the healthcare system. For private or public facilities and for primary or secondary (hospital) care, households in the richest quintile with Jamkesmas saw their utilization rates increase at much higher rates than households in the poorer quintiles with Jamkesmas. Poor beneficiaries are not taking advantage of Jamkesmas' nearly 21 * Protecting Poor and Vulnerable Households in Indonesia unlimited benefits due to lack of awareness of services provided and inability to meet supplemental costs of access. If Jamkesmas (in collaboration with service providers and community groups) can do a better job recruiting beneficiaries into the healthcare system and providing enough information for effective use, service providers in the Jamkesmas network will need to develop plans for increasing both the quality and quantity of services provided; otherwise, Jamkesmas benefits are likely to continue to be in name only. BSM and Raskin are not likely to significantly protect households or promote good behaviors because of design and implementation weaknesses. BILT and PKH produce effective benefits from reasonable levels of public monies provided; Raskin and the BSM programs are not cost effective. BLT and PKH spend reasonable amounts on all the support processes necessary to distribute cash transfers relatively efficiently (5 and 16 percent, respectively, of the total amount of benefits provided) and they deliver proven outcomes. The smaller cash transfers also deliver benefits relatively efficiently, although their effectiveness is less well known. In contrast, while BSM delivery looks efficient - i.e., with minimal overheads - the program achieves very little and is less well-known and less used by target groups. On the other hand, Raskin spends the most (there is a built-in administrative cost of approximately 25 percent, but actual non-benefit expenditures may be higher or lower) to deliver rice, but beneficiaries end up with a very small transfer, making Raskin the least cost-effective program when considering actual benefits delivered. Spending on the sector as a whole is mildly pro-poor: around 60 percent of total benefits from the four largest programs go to poor and vulnerable households (roughly equivalent to the bottom four deciles) and the remaining 40 percent of benefits went to households in the top six deciles. BLT's higher coverage of the bottom 10 percent of households is notable, as is BSM's higher coverage of the top 30, 20 and 10 percent of households. Overall, the current collection of SA programs in Indonesia does not constitute a true social safety net: many gaps still remain. There is currently no program that anticipates risks from, and prevents negative coping behaviors during, household-idiosyncratic risks such as temporary unemployment. Indonesia also does not have an automatic safety net that kicks in to protect households in response to global, macro, regional or micro shocks. Large numbers of those from especially vulnerable groups such as destitute elderly and disabled remain unprotected. Promotion on a large scale is also underprovided. PKH is a relative success story but is confined to a small subset of very poor households. BSM serves a larger proportion of the population with a valuable protection-and-promotion benefit, but is struggling to be effective. Early childhood interventions in education, nutrition, and vaccination are not yet national in coverage. Lastly, with respect to reform, Indonesia has a proven program in BLT. However, BLT has only been used on an ad hoc basis and has not been institutionalized for political reasons. The effectiveness of the system as a whole is constrained by fragmentation, lack of coordination and duplication. Programs operate in isolation of each other creating a fragmented approach to social protection. The eight major programs are spread across five different implementing agencies and many other institutions are involved in support operations, disbursing and delivering benefit packages, and policy planning. Fragmentation also occurs within agencies: the scholarships program is actually comprised of 10 different independent initiatives spread across Kemdikbud (Kementerian Pendidikan dan Kebudayaan, Ministry of Education and Culture) and Kemenag (Kementerian Agama, Ministry of Religious Affairs) with little inter-connectivity between them. The PKH, JSLU, JSPACA, and PKSA programs are run independently out of four different administrative clusters within Kemensos, virtually guaranteeing the duplication of many common processes. This also prevents households from being inducted into the entire array of initiatives available and prevents implementing agencies from realizing economies of scale or scope in their operations. These issues are mirrored in budget formulation for the social assistance sector. Budgets are fragmented across and within agencies and overall budget formulation for the sector is not supported by existing budget classifications. 7. Recommendations for an Indonesian Social Safety Net Indonesian SA programs have proliferated, but much work remains to turn the loose collection of programs into a true social safety net. Each of the major programs faces design and delivery challenges and there are significant gaps in both risk and population coverage, leaving many vulnerable households exposed to poverty. Fortunately, the country is in a strong position both fiscally and macroeconomically (trends which are projected to continue). It has the will and creativity necessary to meet the challenge of developing a true social safety net which reliably protects the poor and vulnerable from the risks they face and promotes investment in productive and poverty-reducing behaviors. The following recommendations outline some crucial steps in creation of such a system. 22 Executive Summary U A. First, spend public money better by reforming and re-engineering programs and implementation to achieve a better mix of welfare-improving programs Scale up PKH while revising benefit levels to continue delivering better health and education outcomes for poor households. Make PKH a national program by expanding coverage to all very poor households. Increase PKH benefit levels to ensure they are appropriate for education costs and include transition bonuses (for basic to junior secondary and junior to senior secondary). PKH has one of the only comprehensive MIS systems in Indonesia and should continue to refine the processes by which MIS-generated information is incorporated into a continuous reform and improvement cycle. In addition to further refinement of the PKH conditionalities and the MIS system monitoring all subprocesses, PKH will benefit from a redesign to its collaboration with service providers and local governments as well as capacity and quality upgrading in its facilitator corps. PKSA, JSPACA and JSLU have the potential to help especially vulnerable groups, but lack capacity and resources for needed facilitation and outreach, appropriate safeguarding, and effective delivery. These programs should start with a redesign of the mix of cash and facilitated services that make up the benefit package as well as the outreach, intake, and triage processes that could direct beneficiaries with highly specialized needs to service providers in other sectors able to provide the care that is immediately necessary. In parallel, the programs should begin a guided upgrade to safeguarding activities (socialization, targeting and prioritization, facilitator capacity and services delivered, monitoring and evaluation, and complaints and grievances) and devote more financial and human resources for this purpose. Consolidating the three cash transfers and instituting a common systems approach for all support operations will save time and help realize greater economies of scale in operations. Here the programs can learn from PKH and can make arrangements to share implementation processes and systems, especially PKH's MIS system. When reforms have momentum, begin considering increasing coverage beyond current levels based on soundly-estimated regional needs. Reform and re-engineer BSIVI to remedy its current ineffectiveness for poor and vulnerable households and then expand availability to all poor and vulnerable households. BSM benefits should be recalculated to be commensurate with the total costs of education and cash transfers should be delivered when needed. BSM design should be revised so that the program is able to provide reliable relief for students and households during the riskiest periods of an educational career; a "graduation bonus" or "transition bonus" will encourage students to continue across transitions and provide funds for education before school expenditures ramp up again. The administration of the BSM program in Kemdikbud must be re-designed so that the BSM can follow students across schooling levels (from basic to junior secondary, junior to senior secondary, and senior secondary to university). Consider consolidating the 10 independent BSM initiatives across agencies and across school levels so that the program can follow a student along his/her educational career and establish a single coordination unit in Kemdikbud (or another agency) to implement the unified program, including more thoughtful and effective socialization and better targeting using a national database of poor students. Revisions to Jamkesmas are essential as it currently struggles to increase utilization among needy beneficiaries who are either unaware of the program or cannot afford the costs of access. Three major revisions to Jamkesmas' overall benefit package are necessary for effectiveness and sustainability: a revised mix of free medical services and facilitation and outreach would be more effective for poor households; a revised mix of free medical services and benefits for general access costs would also increase utilization among poor households; and the medical benefit package itself should be revised as it is currently more generous than most other schemes available in Indonesia and internationally. To ensure that beneficiaries get the quality care they need will require increased monitoring of service providers, the establishment of a complaints and grievance system, and better socialization of Jamkesmas benefits, goals, and rights and responsibilities. As Indonesia has struggled to keep pace with the rest of the region in maternal and child mortality and malnutrition, the revised and re-engineered Jamkesmas program should be extended to the bottom 40 percent of the Indonesian population while Jamkesmas needs to develop medium and long-term scenarios that are scientifically costed (and not based on current supply-side limitations and beneficiary underutilization) to ensure the program's longevity. Jamkesmas should also develop plans to ensure that Jamkesmas beneficiaries retain coverage during the transition to any upcoming universal health insurance scheme. Raskin delivers very little at unknown cost and would benefit from process re-engineering and rationalization. If Raskin is going to continue providing SA benefits with public monies, a thorough reorganization is necessary. Business process analysis may indicate where, why, and how so much Raskin rice is lost; may determine where, why, and precisely how much government agencies spend to achieve Raskin delivery; and can suggest technologies and processes to economize on those costs. Lastly, household rice purchases will have to be monitored and controlled more tightly in order for the Raskin program to deliver full benefits to only poor and vulnerable households. If Raskin cannot improve in these three areas, it should cease using public money to deliver SA products. 23 E Protecting Poor and Vulnerable Households in Indonesia Past reforms have demonstrated the usefulness of a quickly-deployed but temporary emergency income support. BLT worked to protect incomes and safeguard good behaviors partly because it was deployed rapidly and valuable benefit packages arrived just in time. Cash benefits also proved useful as households were able to immediately apply benefits to whatever expenditures were necessary and normal. When the next crisis or policy reform package hits Indonesia, social safety net providers should have a temporary cash-for-service initiative ready to be deployed, so developing protocols, procedures, and institutional authority for an automatic BLT will ensure timely disbursement. Before the next crisis, both the evidence on BLT effectiveness and procedures for initiating a BLT (as a response to crisis) should be codified and automated so that BLT becomes an apolitical, technical tool for combating the stresses and difficulties that households experiencing crisis face. The current array of programs could consistently reach the same poor, near-poor, and vulnerable populations by developing a common targeting standard based on the PPLS11 survey. Targeting in each program should identify the chronically poor, the near poor, and the especially vulnerable (but not currently poor) across Indonesia. The PPLS 11 survey - which represents a massive increase in data collection as well as an improvement in data collection methodology - will be able to produce such national lists of poor, near-poor, and vulnerable households. Benefit allocation based on the PPLS1 1 survey (and corresponding list of eligible households) is expected to result in significant targeting improvements over previous methods. Moving all programs to a common standard and eliminating idiosyncratic approaches and duplication in data gathering will also cut down on administrative expenditures and will reduce the risk that households fail to have reliable access to all programs for which they are eligible. A common targeting standard may also serve a demonstration effect to implementing agencies and service providers and may encourage the development of minimum service standards in other program areas. B. Cover the most important risks while extending at least basic coverage to all poor and vulnerable households. Social safety nets should target all chronically poor households with greater assistance and be able to provide basic protection to the 40 percent of all households that are most at risk of becoming poor in any given year. The current range of SA programs does not go far enough in protecting income and promoting healthy behaviors in chronically poor households, nor do current programs protect all households that are highly vulnerable to shocks. To cover all vulnerable households with some basic protection, the social safety net needs a broader reach. A core component of a future social safety net for Indonesia is protecting households from risks to their heath. Illness, work accidents, and long-term debilitating health setbacks are inherently unpredictable. Treatment can be costly and difficult to plan for, while those whose work is interrupted pay twice: once for medical care and again in foregone income. All poor and vulnerable households need permanent and easy-to-use programs that provide low- or no-cost access to health care providers. Households with more specialized needs and costs will require extra support. Expand the coverage of Jamkesmas to all vulnerable households, offering a basic benefits package that is fiscally sustainable. In addition, provide PKH to all chronically poor households that experience greater burdens, but lighten the conditionalities in areas where health services are still limited. Expand coverage and facilitated health services of programs that cover the especially vulnerable elderly and those living with serious disabilities. Poor and vulnerable households need access to permanent and easy-to-use programs that provide low- or no- cost access to all levels of public education. Education is a key to helping families break the intergenerational transfer of poverty. With higher levels of education, youth are more likely to find good jobs and benefit from high wage premiums and earn their way out of poverty and vulnerability. The social safety net, however, must ensure that children and youth from disadvantaged families can continuously stay in school for as long as possible. Interrupting education at any point in a child's life can open up gaps that persist for a lifetime. The BSM program, once consolidated and re-engineered, can provide much needed assistance to students who are most at risk of dropping out. PKH students should automatically be linked to the BSM program and PKH households should face lighter conditions that are possible to achieve in areas where school availability is limited. At the same time, expand coverage of PKSA that reaches out to youth who are at greater risk. To fill the gap in the critical early years, pilot and test a program that provides effective and affordable early childhood development (ECD) services for poor families, including parental education. Social safety nets should ensure a minimum level of income so that vulnerable households are not forced to make difficult choices. Persistently poor households have difficulty generating sufficient income to lift themselves out of poverty. Vulnerable households are likely to turn to negative coping mechanism - sending more members to work and pulling more members out of school, switching consumption to less nutritious but cheaper foods, and foregoing health care - precisely when their incomes are threatened. Indonesia needs income support initiatives that reliably address 24 Executive Summary U both difficulties. The cash transfers to severely disadvantaged households - PKH, JSPACA, PKSA, and JSLU - should be expanded to national coverage. Raskin should provide additional in-kind permanent income support to poor households only, but this will require a major reform to operating procedures and operating costs. Pilot a national workfare program so that all vulnerable households can rely on a guaranteed number of working days when difficult times occur. Vulnerable households may not face income risk every month, but sudden unemployment, illness, bad harvest, or other idiosyncratic shocks can interrupt regular earnings or regular productive activities. With a workfare program that vulnerable households can opt into when stipulated wages become attractive, the ever-present risk to income generation is partly addressed. A workfare program is also a good time and place for contact by a facilitator who could enroll eligible households in Jamkesmas and BSM (if applicable). Well-designed workfare programs set wages below the prevailing market wages so only households with no better outside opportunities apply. A coordinated and authorized list of projects and sites where labor is needed must be available at all levels of government. A quickly-deployable and automatic emergency income support facility will be useful in the face of future crises or difficult policy reform. Current SA programs focus on long-term poverty and vulnerability. These programs must be folded into a system that includes a crisis monitoring and response mechanism that addresses short-term, acute shocks and that focuses on providing income and basic necessities to all households at risk of curtailing human capital investments in health, nutrition, childcare and education. The national development planning agency (Badan Perencanaan dan Pembangunan Nasional or Bappenas) should reinvigorate its collaboration with BPS in order to ensure the timely processing and release of high-quality and highly-relevant data that is amenable for near-real-time monitoring of household conditions. Then, a successful vulnerability mitigation tool should be developed that can respond precisely when a crisis forces vulnerable households into negative coping strategies. Some of the response might include temporary scaling up of social safety net programs, but the GOI should develop protocols and cement the legal basis for the automatic and rapid disbursement of a pre-identified social assistance package (and associated targeting procedures) before the next crisis or downturn hits. C. Explore a longer-term transition to an integrated safety net hub architecture To prevent vulnerable households from falling through the cracks and to economize on implementation costs, current fragmentation and duplication must be eliminated. A single agency should be in charge of developing plans for implementation, monitoring, evaluation, and reform of all SA initiatives. The same agency should have the power to delegate implementation tasks, either to already existing government agencies or external contractors. The quickest way to jumpstart SA integration is through the National Targeting System that is already in development. The National Targeting System will construct a unified targeting registry of potential beneficiaries, based on the PPLS1 1 survey (see above) and with improved targeting methods. With this single source of quality-controlled data, programs can improve targeting outcomes. Moreover, programs with the same target population will have consistent beneficiary lists, leading to more complete coverage and more effective realization of program complementarities. In addition to targeting, the rest of SA support operations should be brought under a "minimum service standards" framework through which each program is monitored, evaluated, and reformed. In order to harmonize both the quality and effectiveness of all social safety net initiatives, a single agency or body should develop minimum service standards and indicators that reliably track performance in each program. The implementation steps that will need to be brought under this common framework are: socialization and outreach procedures; monitoring and a common Management Information System; evaluation activities (these may benefit the most from participation by external, independent agencies); complaint, grievance, and appeals procedures; and finally promotion and public relations for the social assistance initiatives. Another quickly achievable integration step is through rationalization of the social safety net budget development and budget reporting processes. Seamlessly protecting poor and vulnerable households from diverse risks over their lifetimes may ultimately require the consolidation of the current programs and agencies into a "single window". In Indonesia, the collection of social assistance initiatives is not aligned along a household's life cycle, meaning missed opportunities to protect and promote productive behaviors as new risks arise. In order to reduce these missed opportunities, some middle-income countries have established a single coordinating hub, single agency, or even a single program, targeting many vulnerable groups and risks. With a coordinated social safety net operation, households can access the entire array of services for which they are eligible by making a single visit or through a single facilitator. 25 Introduction Indonesia Vulnerability Profile Despite strong economic growth and falling poverty over the last decade, many households continue to live on the edge of poverty. The last decade in Indonesia has seen a return to strong economic growth, and the poverty rate has fallen from 23.4 percent (1999) to 12.5 percent (2011). The falling overall poverty rate, however, partially masks a high degree of vulnerability: much of Indonesia's population is clustered just above the poverty line, consuming approximately Rp 233,000 per month in 2011 (about US$ 27 at 2011 nominal exchange rates). Around 24 percent of Indonesians live below the official near-poor line (with consumption of approximately 1.2 times the poverty line) while 38 percent of the population lives below 1.5 times the poverty line and is almost equally vulnerable (Figure 6 and Table 3). Even relatively small shocks to these vulnerable households can be enough to push them into poverty. Vulnerable households have less income security and frequently fall into and out of poverty. Approximately half of poor households in any given year are chronically poor: over the past three years, they have always been poor. The rest of the poor (in any given year) are households that are highly likely to move both into and out of poverty. For example, 12.6 million people who were not poor in 2009 had fallen below the poverty line by 2010, thereby making up over half of all poor in the latter year (Figure 7). This high rate of churning into and out of poverty is a persistent feature: in the last three years, a quarter of all Indonesians have been in poverty at least once while about 43 percent fell below the official near-poor line at least once (Figure 8).6 Furthermore, over 80 percent of the households who fall into poverty from non- poor status the year before are households with expenditure levels no higher than 1.5 times the poverty line; in other words, it is the bottom 40 percent of vulnerable households from which the "newly poor" are overwhelmingly drawn each year. 6 Statistics Indonesia (BPS) defines the poverty line as the amount required to obtain 2,100 calories per day from local food commodities and a small amount for other basic necessities, such as clothing, housing, and transportation. In 2010, the poverty line was approximately Rp 211,000 per month or Rp 7,033 per day. Near-poor is defined as 1.2 times the poverty line. In 2010, the near poor line was around Rp 250,000 per month, or Rp 8,400 per day. 26 A large portion of Figure 6. Indonesia Per Capita Consumption Distribution, 2011 Indonesians are clustered near the official poverty line, meaning - - headcount- poverty rates fail 38% EBob. 1.5 PL to summarize the vulnerability of m any individuals........................ and households. 24% Below 1 2L mPCverty Line (PL) 0 200,000 400,000 600,000 800,000 Monthly Household Per Capita Consumption (Rp.) Table 3. Poverty and Vulnerability Headcount Rates, 2008-2011 Poverty Rate (%) Poverty Line Multiple 2008 2009 2010 2011 NoesTenaioalpvetyl N i onl RPT 233$700 per0 perso per month in2011 1.2 PL -$PP 1.2) 7.8 5.624.4227 E Protecting Poor and Vulnerable Households in Indonesia Moving into and Figure 7. Poor, Near-Poor and Newly Figure 8. Exposure to Poverty, 2008- out of poverty Poor Individuals, 2010 2010 is common for most vulnerable households, oo a woo - reflected in large numbers of poor l s - so - households who os - so are newly poor in any given no s year and the J20 2 20- approximately 0 25 percent of Poor in 2010 Near Poor in 2010 Number of Times Poor Number of Times Near Indonesians who 2008-2010 Poor 2008-2010 have been poor Newlyin Poor (ota Poor in 2009) Never Poor Poor Once at least once mnNwyPo No ori 09 Poor Twice Poor Three Times the past three years. Sources: Susenas and World Bank calculations. In addition, poor households still have stubbornly low secondary enrolment rates and high malnutrition rates, indicating lingering gaps in productive human capital. In 2010, the rate of attaining at least a 1011 grade level of education among students from poor households was 50 percentage points lower than students from non-poor households, indicating a widening of the education achievement gap from 2000 (Figure 9). Trends in infant mortality, malnutrition, and health service utilization show a similar pattern of increasing gaps between rich and poor households. Educational Figure 9. Education Attainment by consumption quintile, 2000 and 2010 attainment has increased 2000 2010 only slightly10 0 for the poorest 900 quintiles over the 2 past 10 years,700 meaning that the I achievement gap 4 between poor 30 and non-poor200 households has in-1 not narrowed 1 2a 0 1 12 122 3 0 11 12 1 during those Yearoedcai Yxearsofeducato.n years. - -Quintilel1 Quintile 2 QunO l Qu ntfle4 Quintlde5 Source: Susenas 2000 & 2010 and World Bank staff calculations. Such gaps inevitably worsen when households are threatened by shocks as poor or newly impoverished families resort to negative coping strategies that have long-term impacts on younger generations. During the Asia Financial Crisis (1997-99), newly poor households primarily relied on two strategies for reducing crisis impacts. The first was to cut household expenditures by consuming lower quality food and reducing spending on education and health services. These cuts allowed sufficient provision of daily calories to all household members. The second was to boost 28 Introduction U household income by sending mothers and school-age children into the workforce where they typically contributed to informal sector activities.' Both behaviors created long-run gaps in investment in nutrition, health, education, and other human capital and social skills, resulting in negative, long-lived impacts especially among younger generations. It is possible that such crisis-era behaviors are partly responsible for elevated rates of non-enrollment and malnutrition still observed among younger generations today. Smaller shocks can be equally stressful to vulnerable households and produce equally negative responses. During 2008 and 2009, households experienced difficulties due to acute inflation that resulted from fuel subsidy reductions as well as international crises in food prices and financial markets. Poor and vulnerable households again reported difficulty in meeting consumption needs and adopted negative coping behaviors, like substituting lower quality or lower cost food for more expensive proteins, just to keep up. Expenditure cuts directly affected nutrition status, particularly the micronutrient intake, of younger household members; it is these younger members who most depend on proper nutrition for physical, mental, and social development. Forced choices like these can reduce both current and future earnings and impair productive potential. Role of Social Safety Nets Indonesia faces a double challenge: helping poor households escape impoverishment while protecting the 40 percent of Indonesians who remain highly vulnerable. Policies and programs must be tailored to fit the Indonesian context, which is one characterized by a high level of vulnerability and churning near the poverty line and only marginal improvements to social indicators for poor households. Social safety nets, which typically consist of non-contributory cash or in-kind transfer programs targeting the poor and vulnerable, are designed to directly respond to such challenges. They are one component in the social protection suite, which typically also includes social insurance, active labor market programs, and provision of high-quality, low-cost education and health services accessible to all. The first function of safety nets is to protect households from destitution and unnecessary loss of human capital. Social safety nets can provide direct income support, providing an immediate impact on reducing extreme poverty and inequality. This is often achieved via direct transfers of cash to households but also includes transfers of in-kind goods or services (like food or jobs) and fee waivers or no-cost insurance for crucial services like health and education. As households have no choice over how to use in-kind goods and services, waivers, and insurance, these instruments are not as general as direct income support through cash.9 For those vulnerable to poverty, safety nets also provide ex-ante protection from the poverty-inducing effects of shocks like acute inflation, job loss, macroeconomic or financial crisis, natural disaster, or an expensive bout of ill health. When such shocks occur, vulnerable households protected by a safety net may not have to resort to negative coping strategies. Direct income support and fee waivers can keep all members from vulnerable households consuming and investing at non-crisis rates. This keeps households closer to their productive potential and smoothes consumption and investment across difficult stretches. 7 See Sumarto et al. (2010) fora full report. Frankenbergetal (1999) demonstrated the following AFC impacts: poor children were 5 times more likelyto be out of school; health service utilization rates and preventive health behaviors decreased for poor households; and nutritional status and micronutrient concentration decreased within poor households. Additional studies focusing on impacts of crisis on poor households in Indonesia include Levinsohn et al. (1999), Thomas et al. (2001), Cameron (2002), Block et al. (2003), and Giles and Satriawan (2010). 8 Another common behavior signaling consumption difficulty in Indonesia is the pawning or sale (at reduced prices) of productive assets like motorcycles and livestock. 9 There may be a more-or-less liquid secondary market for in-kind goods, like food staples; other in-kind goods, like nutritional supplements or vaccinations administered at a health post, may not actually be transferrable. 29 E Protecting Poor and Vulnerable Households in Indonesia Socia safey Tabeo4.wo Mshuldsafetycil netswcer? Tablet 5ataaft eto Thet scofuncinoldeynt st rmt potntes ieiod n etrjb hog oshl inestmesind ma aunerale Eapitle Safet netprogramsat otndsintrmtePrseofteccssromt tro oivGre antdatidems bt e vulnerableruou expliiticentveso pinreas cosupinr fo nrulsritin, (health)n dcto htriespoutvt,icms n wel-beng.Houehods mhicotae advangBelopriiaerrqetl Poine rglreonmcadsoilaciiis whichpromte cnfidnwandinvovmn irnsfr commuityeerld dvelo mnt.Ofcoe,m uaecptlinetensas ledoros bofo, wHelp implement Special Transfers Faciliation and Social Care reforms (e.g., subsidy b groups removal) The second function of safety nets is to promote opportunities, livelihoods and better jobs through household investments in human and social capital. Safety net programs are often designed to promote use of and access to nutrition, preventative healthcare, and education. All household members, but especially younger generations, are given explicit incentives to increase consumption - of nutrition, health, and education - that raises productivity, incomes, and well-being. Households which take advantage also participate more frequently in regular economic and social activities, which promote confidence and involvement in community-l evedlopment. Of course, human capital investments also lead to productivity gains in tomorrow's labor force, which is good for overall growth as well as government revenues. Finally, safety nets can help make macroeconomic reform plans palatable while enhancing balanced growth. Safety nets may help government replace inefficient redistributive policies from other areas, or secure change in macroeconomic policy and structure to improve growth. By reducing risks, alleviating burdens, and compensating those most at risk under the new policy regime, safety nets may even leave poor households in a better post-reform position. In Indonesia's case, the last major adjustment to costly and regressive fuel subsidies was packaged with direct income support and fee waivers to households (as well as subsidies for schools and village improvement grants). Reorienting spending towards progressive transfers and providing consumption support during the acute inflationary environment that followed the subsidy reduction prevented reforms from being reversed. Social safety etts are often provided to vulnerable subgroups which benefit fof t irom fe otection strategies. The elderly poor, the disabled, homeless or orphaned youth, or informal workers bearing the brunt of government reforms also benefit from safety net protection. Shocks may present unique problems for these populations which have difficulty raising income, and rely on the support of others, even in normal times. All groups vulnerable to uninsured risk should confidently expect a safety net to automatically prevent the difficult choice between food on the table and investments in human and physical capital that modern Indonesian society depends on. Box 1 below describes approaches to social safety nets in four different middle income countries; many of the issues in social protection faced by other countries, as well as their solutions in Indonesia, are discussed in the rest of this report. 30 Box 1. The Minimalist intervention with simple means-testing and benefits Role of In China, the revocation of employment guarantees has left low-skilled, chronically ill, and disabled Safety Nets citizens vulnerable to destitution. China's national urban social safety net policy, Di Bao (sometimes in Middle referred to as the "minimum livelihood guarantee scheme"), grew out of a 1993 Shanghai city program. Income The program covered as many as 23 million residents in 2008; it provides a single cash transfer to all Countries means-tested recipients to bring them exactly to the Di Bao line, which is a minimum expenditure level determined by the government. Once eligible for Di Bao, households may also be eligible for subsidies in education, health care, utilities, and housing. Using centralized power, thorough coordination, and documented success to achieve further success Mexico spends approximately 50 percent of the federal anti-poverty budget on just one conditional cash transfer program, Oportunidlades (formerly Progresa). Enlargement and refinement of the program was made possible by a thorough and objective monitoring and evaluation initiative. Oportunidades is centrally run by an oversight, policy, and coordination body with the power to secure cooperation and outcomes from all affiliated agencies and ministries. The program was first conceived by officials from several different ministries and with strong presidential support, which set the right collaborative tone from the beginning and has made subsequent cooperation between the health, finance, education, and social security ministries much more efficient. Conditions include better health, nutrition, and education behaviors while benefits are provided directly to households. Oportunidlades has matured with very low administrative costs. Much effort has been made to objectively evaluate the effects of Oportunidades on beneficiary households and then to publicize results to a full range of stakeholders. Through prudent governance (including a ban on signing up new beneficiaries within six months of national elections) program staff and offices have established Oportunidlades as an independent institution outside of politics; it has proceeded intact through two political transitions and several rigorous impact evaluation reports. Oportunidlades has added new components - financial rewards for high school graduates ("Jovenes con Oportunidades") and additional transfers during food price spikes - during recent presidential administrations. Mexico's experience shows that it is feasible to carry out a national, targeted program even within poor, isolated areas with few services, and in a country with a limited welfare state. Crisis-motivated reforms leading to an integrated and comprehensive safety net In Colombia, a macroeconomic crisis in the late 1990s demonstrated the inability of the social safety net system to respond effectively to households in need. Poor targeting, institutional inflexibility, fragmentation and duplication, low levels of spending, lack of transparency and the lack of a strategic focus (within implementing agencies) on safety nets all contributed to ineffective social assistance. System-wide reform led to the Red de Apoyo Social (RAS) (Social Support Network) which was housed in the President's office and bypassed the inflexibility and infighting of other agencies. At first, the Network had three main interventions: a Conditional Cash Transfer (CCT) (based on health, nutrition, and education), a youth training program, and local workfare programs; it thereby covered all poor individuals from birth to the end of their productive lives. Much as in Mexico, rigorous evaluation and dissemination of program results and cost and benefits to all stakeholders, as well as the relative independence of the agency, has led to program durability. RAS has grown in coverage while continuous reform and improvement have led to a more sophisticated program. Juntos, a newer initiative based on Chile's Solidario program, is focusing on the abilities and livelihoods of the extreme poor by providing not only remediation at the household level but by encouraging and incentivizing the surrounding communities and political bodies to promote pro-poor growth and make sure there is adequate access to and adequate supply of social services in areas with extremely poor households. In addition to standard areas of concern in social safety nets - income and work, health, nutrition, and education - Juntos addresses identification (getting beneficiaries the proper and useful government forms and certificates, including land titles) living conditions, family dynamics and psycho-social wellbeing, credit and access to banking and savings, and justice and the rule of law. The early childhood interventions in health, nutrition, and education are linked in a comprehensive package while beneficiary families consult with a social worker on strategies for exiting extreme poverty. The long-term objective is to move toward a social safety net that provides bundles of programs that are tailored to meet the specific needs of hard-to-reach households. E Protecting Poor and Vulnerable Households in Indonesia Box 1: continued South Africa's categorically targeted benefits provide major income transfers When the system of apartheid was dismantled in South Africa, many social programs, including categorical grants and transfers, that were previously reserved for whites were left in place and extended to all citizens. In the post-apartheid era, coverage of grants and transfers in particular has increased substantially while additional grants and transfers have been added to the socia assistance suite. As a result, social assistance spending in South Africa has reached 3.5 percent of GDP in 2010; this is more than doube the median for developing and transition countries of 1.4 percent of GDP The major social assistance grants include an old age pension; a disability grant, which includes a transfer made in the event of job loss due to ill health; a child support grant for under-1 8s residing with low-income caregivers; and a foster care grant given to children paced with foster parents. One or more of these grants covers over 28 percent of South African individuals (14 million covered from a population of 49 million) with the child support grant alone in the hands of 18 percent of the population (which is equivalent to 60 to 70 percent of children 18 or under) and about 6 percent of working age adults receiving the disability grant. Each grant is generous: the old age, disability, child support and foster caregrats pre transfers equal to 1.75, 1.75. 0.4, and 1.15 times the median per-capita income in South Africa. At least for the child support grant, the means test has been repeatedly relaxed and is now approximately equivalent to 10 times (20 times) the value of the grant for single (married) caregivers. As a result, two-thirds of income in the bottom 20 percent of households comes from social assistance grants whie over half of households report receiving income from one (or more) of these grants (Woolard and Leibbrandt, 2010). n fact, the old age pension and the disability grant were large enough to lift all but the largest households out of the owest income quintile (Leibbrandt et aL, 2010). The South African grants system is truly unique in both the level of public spending and the amount of income transferred to qualifying households, but the same historical path that led to generous benefits has also led to a focus on children and elderly persons, rather than the prime-age adults in the labor force. Brief History of Social Assistance in Indonesia" 2 Indonesia's first generation of SA programs began in the wake of the 1997-98 Asian Financial Crisis (AFC). Prior to the crisis, social spending was general: supply-side improvements in health, education and infrastructure (water, sanitation, electrification) as well as large fuel subsidies dated back to the late 1 960s. In response to the AFC, Indonesia's central government established a broad safety net (Jaringan Pengaman Sosial, JPS) and introduced large food subsidies (Operasi Pasar Khusus, OPK) including for rice. The JPS consisted of temporary; short-term programs including public works, scholarships and funding for health services. The purpose of this first generation of SA programs was to ensure that the poor would maintain access to affordable food, and health and education services during the crisis period. Of these major crisis-era initiatives, only the rice subsidy (renamed Beras untuk Rakyat Miskin, or Raskin, in 2002) continued to receive significant budget allocations and it has become a permanent SA program. In 2005, following the partial removal of fuel subsidies, a second generation of SA programs was introduced. A portion of the savings from the subsidy cuts was reallocated to three SA programs: an expanded Raskin, health insurance for the poor (Asuransi Kesehatan Miskin, Askeskin) and a large-scale temporary unconditional cash transfer (Bantuan Langsung Tunai, BILT). BILT helped over 19 million poor and near-poor households cope with the inflationary shock from the increase in fuel prices. After the fuel-price crisis had passed both Askeskin and Raskin continued to receive budget allocations while BILT shut down as planned. Askeskin was re-named Jamkesmas (Jaminan Kesehatan Masyarakat) in 2008 and now covers as many as 18 million households. Raskin continues to distribute rice to millions of families. BILT was re-deployed in 2008-09 following another fuel price adjustment, and again shut down (as planned) after the crisis had passed. More recently, the government has introduced more sophisticated SA programs intended to break the generational transmission of poverty. In 2007 the Government of Indonesia (GOI) began piloting a conditional cash transfer program (Program Keluarga Harapan, PKH) which transfers assistance only when families obtain preventive basic health and nutrition services and send their children to school. In 2008, the government refocused general scholarship programs towards students from poor households (Bantuan Siswa Miskin, B3SM). These programs joined an 10 See "Social Assistance Program and Public Expenditure Review 8: History and Evolution of Social Assistance in Indonesia" in Volume 2 for a more detailed treatment. 32 Introduction U array of small social welfare programs targeting especially vulnerable groups including at-risk children, the disabled and vulnerable elderly. The GOI has also made considerable progress in expanding coverage of its major community-driven development programs, collectively called Program Nasional Pemberdayaan Masyarakat (National Program for Community Empowerment) or PNPM, as well as in expanding operational grants to social service providers such as schools, which benefited from the Bantuan Operational Sekolah (School operation funds) or BOS program." Current Portfolio of Household-based Social Assistance Programs Today, the majority of SA spending is accounted for by eight major programs. While Indonesia does not have a social safety net sector per se, in recent years the government has articulated its poverty alleviation strategy around three "clusters" of programs. Cluster I consists of household- and family- focused programs, Cluster II includes community- based programs, and Cluster II are mostly enterprise-based, providing assistance for micro- and small-sized enterprises. Cluster I programs form the foundation of an emerging, permanent social safety net system which supports poor and near-poor households. BLT (when active), Raskin and Jamkesmas are the three biggest programs and each covers almost a third of the population. Following its rapid expansion in recent years, BSM is the fourth largest program in terms of coverage, followed by four smaller pilot programs: PKH, PKSA, JSPACA and JSLU. These programs are described in more detail below and their key characteristics are summarized in Table 6. Responsibility is spread across six key central agencies responsible for policy formulation, implementation and monitoring. The National Team for Accelerating Poverty Reduction (Tim Nasional Percepatan Penanggulangan Kemiskinan, TNP2K), housed in the Office of the Vice-President, is responsible for developing SA policies, and provides leadership in program targeting. TNP2K, together with Bappenas (Badan Perencanaan dan Pembangunan Nasional, the National Development Planning Agency), oversee implementation of SA programs and achievement of program goals. Kemensos (Kementerian Sosial, Ministry of Social Affairs) is responsible for the largest number of Cluster 1 SA initiatives reviewed here while providing additional services and in-kind assistance through other programs not specific to Cluster 1. The three largest permanent programs, by either coverage or expenditure - Raskin, Jamkesmas and BSM - are the responsibility of sectoral agencies with only a secondary focus (at best) on providing SA. Existing service providers such as schools, hospitals, health centers, and local governments also play a major role in the implementation of the three largest programs while a host of agencies play important supporting roles across programs. BPS (Badan Pusat Statistik, Statistics Indonesia) plays a critical role in targeting; Kemenkominfo (Kementrian Komunikasi dan Informatika, Ministry of Communications and Information Technology) in socialization of programs; PT Pos (the Postal Service) in the distribution of cash benefits; and local governments again in socialization and monitoring and evaluation. BLT, launched twice in the past, was Indonesia's largest SA program and transferred cash to nearly one third of Indonesian households. BLT transfers continued long enough for over 19 million households to adjust smoothly to the new fuel price schedules and general increases in the cost of living. Households in all provinces, regions, and villages received BLT transfers equivalent to Rp 100,000 per household per month. In 2005, transfers began in October and continued (in quarterly tranches) for a full year before the BLT window closed; BLT transferred a total of Rp 1.2 million per household. The re-launched 2008 BLT program had only three quarterly tranches and the window closed after nine months and having delivered a total of Rp 900,000 per household. BLT's administrative footprint was relatively light because most of the support functions for regularly-delivered, permanent programs (beneficiary tracking, program monitoring, complaint evaluation, information management) were not performed before the BLT window closed. Kemensos was identified as the implementing agency, but in practice the funds preceded directly from Kemenkeu (Kementerian Keuangan, Ministry of Finance) to households through post offices, while targeting and socialization activities were delegated to BPS and Kemenkomlnfo respectively. Raskin is the longest-serving and largest of the permanent SA programs. Raskin is a national program that provides subsidized rice to help households fulfill food consumption needs at reduced cost. The state-owned Bulog (Badan Urusan Logistik, National Logistics Agency) purchases wholesale quantities of rice from domestic producers at fixed prices and then delivers agreed quotas of low-quality rice to regional distribution points where eligible households may make below- market-price purchases. In 2010, Raskin aimed to supply over 3 million tons of rice to 17.5 million poor and near-poor 11 See "BOS Review", World Bank (2010a) for further detail on BOS program, and "PNPM-Rural Baseline Report", World Bank (2008), "Indonesia's PNPM Generasi Program: Interim Impact Evaluation Report", World Bank (2010b), and Smeru (2008b) for PNPM. 33 * Protecting Poor and Vulnerable Households in Indonesia households. According to regulations, each targeted household could purchase a maximum of 14 kilograms of rice per every month at Rp 1,600 per kilogram (compared to that year's market price of Rp 5,060 per kilogram). While Bulog is the key executing agency, Kemenkokesra (Kementrian Koordinator Kesejahteraan Rakyat, Coordinating Ministry for Social Welfare) has been the key policy agency for Raskin since 2007 and final-stage delivery to households is the responsibility of local governments. Jamkesmas is a tax-financed healthcare fee waiver entitling members to free in- and out-patient care at hospitals and primary health centers. The program provides poor and near-poor households with a comprehensive package of free health services and benefits. The implicit value of benefits received by each beneficiary varies depending on their actual utilization of health services, but the benefits (on paper) are potentially unlimited. The program is financed by the central government from general revenues and does not require insurance contributions or cost-sharing on the part of beneficiaries or local governments. According to official data, 18.2 million households consisting of 76.4 million people - about a third of the population - are covered by the Jamkesmas program making it the largest health service scheme in Indonesia. Kemenkes (Kementerian Kesehatan, Ministry of Health) is the key policy and executing agency for Jamkesmas. Hospitals and local health centers deliver free services to beneficiaries and later submit claims for reimbursement. The BSM program provides cash transfers to current public school students from poor households. BSM programs exist at all public schools - secular and religious - across all levels of education. BSM provides currently enrolled students from poor households with an annual cash transfer in one lump sum installment to alleviate financial barriers to education access. BSM funds are intended for use on education fees and other non-fee costs of attending school, such as transportation to school and uniforms. BSM benefits rise with school level, from Rp 360,000 per year for primary school students to Rp 1,200,000 per year for university students. BSM programs assisted over 4.5 million poor students in 2010. BSM consists in practice of 10 independent initiatives with implementation responsibilities delineated by type and level of education. BSM does not have a central coordinating unit. Scholarships for students attending secular public schools are managed by Kemdikbud (Kementerian Pendidikan dan Kebudayaan, Ministry of Education and Culture), while those for students attending public religious schools are managed by Kemenag (Kementerian Agama, Ministry of Religious Affairs). Within both agencies, implementation is further fragmented by level and type of education. For example, in Kemdikbud there is one subdirectorate each for administering the BSM program for primary, junior secondary and senior secondary school students while university scholarships are administered by another Directorate General altogether. Within Kemenag, primary and secondary school scholarships are administered by the Secretariat General" while university scholarships are administered by the Directorate representing each major religion (Muslim, Christian, Catholic, Hindu and Buddhist). Schools play a key role in implementation, especially in the selection of beneficiaries as well as in the eventual distribution of BSM transfers. PKH provides direct cash benefits conditional on household participation in locally-provided health and education services. PKH is a conditional cash transfer with two main components - a cash transfer and monitored conditionalities - that provide an immediate impact on household vulnerability while encouraging investments in long- term household productivity. Cash transfers, which range from Rp 600,000 to Rp 2.2 million per year depending on the number of qualifying dependents in the household, are delivered four times per year. Continued cash delivery, however, depends on a mother's verified attendance at pre- and post-natal checkups, a professionally-attended birth, newborn and toddler weighings and health checks, or after verification that a PKH household's school-aged children have good attendance records at their schools. In 2011, PKH reached over 800,000 "extremely poor" households in 25 out of 33 provinces (and 118 out of 497 districts). PKH is executed by Kemensos; funds are disbursed to households through PT Pos. Finally, Kemensos has been providing new cash transfers to especially vulnerable groups like the elderly (JSLU), the disabled (JSPACA), and at-risk youth (PKSA). JSLU and JSPACA were established in 2006 and PKSA in 2009. JSLU and JSPACA provide a monthly transfer of Rp 300,000 per month (Rp 3,600,000 per year) while PKSA transfers range from Rp, 1,300,000 to Rp 1,800,000 per year. Outreach and facilitation are also provided by locally recruited facilitators. In 2010, there were less than 20,000 JSPACA beneficiaries, less than 15,000 JSLU beneficiaries, and less than 5,000 PKSA beneficiaries. Each of the three programs is managed by a separate directorate within Kemensos, although all fall under the same Deputy. 12 A majority of primary and secondary BSM resources managed by Kemenag go to Muslim schools. 34 Introduction * 000 co0 o . - o o oe ca-),) 8 5 $3N 5 Nä N Äs 82 0 -2 2 2w 2 c, c 1 - E ca a o €.$ --a® E c E c cE 8 E cc E . o sc ome s® e dol o Es 3 ~ ~ ~ - 3 gå å å â E 00L cu E E E E E o c ..7c . E &Ê 3 -2 E 8 E - E 8 -2 13 . > u co- -um 00 0 9,6 S0. 0 o 800a 0, -c 0ho £·8seà O --Ds d8Ooi,= mm6m t om |. > a al E c E0 L U Uc:) cD - cl .t Qt c) C) W [m gë- d c)cccDcý c cD areare ag* 23 ..3 a e a å, ....- n Q c - -cý ý o0, c-c c E 0a c) c >. 0.o cu 2 a) ~35 c,n c) oLDwI)GlL-w ý l lI cLI cD 0 d 02 1) cc u ~ u c) o 2 224 c c c - 2 ). c n c 3) ý QLd m LG) c ccD D( c).~ 2o c) a) G) c D Dc (NL~L c3OLfLDc o -o o o0 35 * Protecting Poor and Vulnerable Households in Indonesia Purpose of Report Accelerating poverty reduction remains a key objective of the Indonesian government. Indonesia has included poverty reduction as a national development goal since the mid-1990s and has signed various laws, treaties, and international agreements committing itself to equal treatment and pro-poor development. Most recently, the RPJM (Rencana Pembangunan Jangka Menengah, the Medium-Term Development Plan) for 2010 to 2014 sets goals of reducing the poverty rate to between 8 and 10 percent by 2014 and reducing income inequality. Strengthening the Cluster 1 SA programs is a key component of this poverty- and inequality-reduction strategy; the government is currently developing a roadmap to this end. The GOI is working together with the World Bank and other development partners on all three program clusters, with a strong emphasis on reforming SA. The GOI is increasingly committed to poverty reduction through a comprehensive social safety net program. World Bank assistance has focused on providing technical assistance in the design and implementation of individual programs, including BLT, Jamkesmas, and PKH. The nature of the engagement is maturing into a sector-wide approach with a focus on the development of comprehensive poverty reduction and social safety net strategies. Fragmentation among SA programs, however, has so far limited the ability of the government to review and possibly reform the overall SA system. This report provides the government with an analytical base to pursue improvements in program delivery and sector-wide reform. To do so, the government requires a better understanding of how current SA programs are funded and delivered, whether they are effective in protecting their targeted beneficiaries, and the institutional arrangements to support delivery of a SA system. Although some studies of individual programs have been carried out, a comprehensive review of current SA programs is necessary to support efforts towards the creation of a third generation of SA programs that are integrated within a comprehensive system. The report has four main objectives. The first is to report, describe, and ultimately understand the size and composition of public expenditure on SA. An expenditure tracking exercise, undertaken through repeated interaction with budget report preparers, will allow all stakeholders to observe the resources dedicated to the various social safety net initiatives. This exercise also includes a decomposition of social safety net spending across initiatives and agencies but also compared to other categories of social spending. This will allow stakeholders to understand which initiatives are receiving the most government resources and whether those allocations are logical. The second is to assess the appropriateness, efficiency, sustainability and effectiveness of the current SA programs. Policymakers will want to know whether programs receiving funds are effective for beneficiaries; whether they are delivering more or less than they promise; how much they costs in terms of salaries, overhead, and other operational and delivery support; and whether the initiatives, as currently run, will be an increasing burden on government resources. This exercise relies heavily on in-depth program-by-program review of each of the social safety net initiatives. Each program review consults nationally representative household surveys, proprietary surveys, administrative data, qualitative data, and on-the-ground reporting in order to present a comprehensive picture of the effectiveness of the design, implementation, and impact of each program. The exercise also makes use of public expenditure and budget data to consider the efficiency and sustainability of the initiatives and implementing agencies. The third is to provide inputs that support continuous improvements in program implementation. To that end, background reports were prepared for each of the Cluster 1 programs; the series of reports provide suggestions for policy makers and program implementing agencies on how to improve the design and implementation of each respective program. In addition, separate background reports were prepared on the history of SA interventions in Indonesia, and a summary Cluster 1 expenditure review. They also provide "nuts-and-bolts" details of the specific successes and failures that initiatives have experienced and feasible solutions for improving outcomes, including the likely cost to the government of such reforms. 36 Introduction U Finally, this report can inform policy dialogue on the development of a new generation of SA programs and an integrated social safety net. In order to move beyond a collection of initiatives to a comprehensive and effective social safety net system, this report contains recommendations for the integration of existing and future initiatives under a common hub that maintains common standards, common processes, and a common recruitment and facilitation system so all households receive all available assistance or benefits for which they are eligible. The report also includes a discussion of international examples of these kinds of reforms - from a collection of programs to a safety net under one umbrella agency - as well as the cost savings and efficiency gains from that consolidation. An assessment of social safety net initiatives can be summarized in six questions. Social safety nets are effective when they meet the objectives discussed earlier: protecting vulnerable households in times of difficulty; promoting the behaviors and investments that enhance long-term well-being and reduce vulnerability; and enabling reforms, all in the most efficient way possible. In order to assess whether those goals are being met, the report asks the following key research questions": 1 Does Indonesia allocate the right level of resources to household SA? 2 Do programs provide the right benefits? 3 Are benefits reaching the right people? 4 Do people receive the benefits at the right time? 5 Are programs implemented in the right way? 6 Does Indonesia have the right programs and system in place? 13 Throughout this report "right" is used as shorthand to indicate effectiveness or efficiency and is not meant as a normative indicator of "correct", "proper", or "meeting a defined standard". For example, the "right" time to deliver benefits is when they are needed and when they can and will be used as intended. Similarly the "right" benefits are not a certain percentage of median incomes; rather "right" characterizes benefits that allow households to achieve what the program intends for them to achieve. The report will clarify this usage in each section. 37 Does Indonesia allocate the right level of resources to household social assistance? Cluster 1 SA expenditures are "right" when they are sufficient to achieve the poverty reduction, social protection, and promotional goals the GOI has formulated and when expenditures allow the achievement of those goals for all of members of the intended target population. Comparisons of SA expenditures to expenditures in other sectors (and for other goals) help set the Indonesian context while international comparisons may serve to demonstrate what other countries with similar SA goals have determined is necessary. Box 2 below details the methods by which household-targeted, Cluster 1 SA can be extracted for analysis from standard Indonesian budget reporting documents. Trends in Social Assistance Expenditures Public expenditures on Cluster 1 SA have increased significantly since 2005 and equal approximately 0.5 percent of GDP today. From a low base in the early 2000s, Indonesia's aggregate national public expenditures - combining central, provincial and district sub-totals - on Cluster 1 SA nearly doubled in 2005 in either nominal or real terms. This increase partially reflects the reallocation of fuel subsidy savings to SA initiatives (Figure 10 and Table 7). Spending remained permanently higher thereafter, reaching peaks in 2006 (1.0 percent of GDP) and 2008 (0.7 percent of GDP) reflecting large BLT disbursements those years. Excluding BLT, SA expenditures are modest but have risen steadily in real terms and as a share of total expenditures. In 2010 (a year without BILT) national expenditures on SA are estimated at almost Rp 30,000 billion (around US$ 3.3 billion), equivalent to 2.6 percent of total national expenditures and 0.5 percent of G D P 38 Public spending Fiue1.PbiExedtrsoFiue1.EpniueSasbylvlf on household HoshlSoilAssac,20200 gvrmn,01 social assistance (SA) R ilo ecn has increased 50005 1% significantly since 2005 and 4,0 0 is dominated by central 3,0 0 government Fiue1ePbi xpenditures on002 0 30,000 3 0 expenditures.r LH) xenitr epedtue Share~~Fiur 11. Exp xpniue RSentr Prve ha ryle elo Shgovernment,(2010 Sorcs ndNoes Krnnku,BP ad ord an 5af. Sorcs ndNoes Knnnku ndWol100% taf 80%9 E Protecting Poor and Vulnerable Households in Indonesia Spending since 2005 reflects greater investment in initiatives by the central government, which accounts for the bulk of expenditures. Indonesia embarked on an ambitious (but still incomplete) decentralization program beginning in 2001 and sub-national governments - especially districts - have taken on greater responsibility for delivering public services and executing public expenditures.14 Today, sub-national governments execute around 40 percent of total national expenditures, and over half of expenditures in key sectors such as education, health, infrastructure and agriculture (Figure 11). In the SA sector, however, the central government dominates, accounting for 88 percent of total estimated expenditure in 2010. PuliS ale 7. tlnational Publi Exenitre on 7 Socia Asisane 2004-201 Iexpenditures ons havee increasedN 2004* 2005 2006 2007 2008 2009 2010 commniiaty rve tiveomnt fou.CutrSAfteG oet leito taeg scmrsdo omnt sicC05 xentres 67935 14,41 31,6848 16,23 36,089273,689 29,17 Share of national SA 85 89 93 87 92 89 88 expenditures (%) Inoei a loicesexpenditures on849 other 8 poverty reductio program ,11 specially those3with 3 community driven development focus. Cluster I1 of the GOl's poverty alleviation strategy is comprised of community development and empowerment programs and is dominated by Indonesia's national poverty reduction program (Program Nasional Pemberdayaan Masyarakat Mandiri, PNPM-Mandiri). This Cluster has also seen an increase in resources in recent years as PNPM has been scaled up (Figure 12). The central government has also been allocating more resources in recent years to Cluster Ill, which targets especially vulnerable communities and small enterprises, though the increases are on top of a relatively low base. In total, the government spent around Rp 42,000 billion on poverty alleviation in 2010, equivalent to 6 percent of total government expenditures and 0.6 percent of GDP. Of this, around two-thirds went to Cluster I and one-third went to Cluster II. 14 See Box 2 as well as "Social Assistance Program and Pubic Expenditure Review 1: Public Expenditure Summary" and "Social Assistance Program and Pubic Expenditure Review 8: History and Evolution of Social Assistance in Indonesia" in Volume 2 for more detail on Indonesian decentralization and social service delivery, including social assistance. 40 The GOI Figure 12. Poverty Alleviation Expenditure Figure 13. Social Protection Expenditure by has broadly by cluster, 2007-2010 type, 2007-2010 increased 44,000 84,000 spending on 7 initiatives, E programs, 33,000 6300 and agencies for poverty 22,000 42,000- alleviation, social assistance, 11,000 21,000 and social mnsurance. 0 "' 0 2007 2008 2009 2010 2007 2008 2009 2010 Cluster I Cluster II Social Household SA insurance (Cluster I + BLT) BLT Cluster Ill Sources and Notes: Kemenkeu and World Bank staff calculation. E xe nture in g Figure 14. Expenditures on HouseholdIn o ei ' ov rll v l ofh u h ld Houehd Soial SA (report definition) versus GOI-dlefined cnee Aepniue antb Assistel Sin l Social Assistance and Social Protection, dermndbrfrigtofica Aindonei 2005-2010 bde ouet rcasfctos IndonsiIndonesia's goverallmevel of houresehld- undertaking its planning and budget 1.6 formulation for SA in a coordinated manner, there is no official budget category for SA. Also, the limited accessibility of data at the sub-national evel makes it impossible to reliably 0.8 determine expenditures on SA executed at this level. This presents significant challenges in conducting expenditure 0A analysis on SA. oo,C) As defined in the Indonesian budget 2005 2006 2007 2008 2009 2010 " social assistance" is significantly Household SA Social Assistance broader than standard international Social Protection definitions. internationally, "social Sources and Notes: Kemenkeu and World Bank staff calculation. assa c "rfr o"o c nrb tr transfer programs targeted in some manner to the poor or vulnerable", a definition that excludes broader social spending on education and health (Grosh et al 2008). Under this definition, benefits are provided directly to households and classified according to the type of payment scheme (i.e. social assistance or social insurance schemes). Further, they exclude transfers made in response to events such as natural disasters. Indonesia uses the social assistance budget category both more broadly and more narrowly. It encompasses: (1) direct transfers to households; (2) transfers to social institutions such as schools and religious institutions; and (3) spending on education and poverty alleviation in addition to SA. Of the expenditures in the social assistance category in the GOI budget, for example, around two thirds go to the large School Operational Assistance (Bantuan Operasiona/ Sekolah, BOS) program while another significant share is allocated to PNPIM, a community block grants program. At the same time, the Indonesian social assistance category excludes some Cluster 1 programs like BLT and Raskin (which are categorized elsewhere) and therefore does not give an accurate estimate of SA. The social assistance category in the GOI budget does not match standard international definitions and in practice over-estimates (at 1. 1 percent of GDP in 2010) the level of expenditures in the Cluster 1 SA programs. This limits its usefulness for analytical and international comparison purposes. E Protecting Poor and Vulnerable Households in Indonesia Box 2: continued At the same time the functional classification category "social protection" is used very narrowly in Indonesia and underestimates Cluster 1 expenditures. Social protection is an umbrela term that encompasses both SA and social insurance programs (such as pensions and unemployment benefits). To the extent that social insurance schemes deliver benefits based on member contributions, they are not considered SA or safety net programs (Grosh et al 2008) Since government spending on contributory social security schemes in Indonesia is negligible, the social protection function should in theory be a reasonable upper-bound estimate of SA spending. A detailed examination of spending under the social protection classification, however, indicates that the category is largely confined to activities executed by Kemensos and thus only captures the smaller Cluster 1 SA programs such as PKH. The remaining social protection budget is spread out across smaller programs managed by minor agencies, namely: BKKBN (Badan Koordinasi Keluarga Berencana Nasional, the National Family Planning Coordinating Agency); KPP & PA (Kementerian Pembe dayaan eran rempuan dan Perlindungan Anak, the Ministry for Women's Empowerment and Chil Protection); and Kemenkokesra. The category excludes most of the large househod SA programs (BLT, Raskin, Jamkesmas and BSM). Moreover, since 2008, the majority of government administrative and salary costs associated with the delivery of social protection activities have been mapped to the "general public services" function, rather than the function for which they were incurred. For these reasons, the social protection function in the Indonesian budget is significantly narrower than standard international definitions and grossly understates (at 0.1 percent of GDP in 2010) spending on Custer 1 household SA. This limits its usefulness for analytical and international companison purposes. These limitations necessitate tracking exercises to identify potential expenditures. Such investigation is challenging because programs and budgets at the central government level are spread across multiple agencies and often across multiple administrative clusters within agencies. Moreover, detailed budget data at the sub-national level remains largely inaccessible. This report sums identifiable SA expenditures and examines the total as if it were a stand-alone sector and budget item. At the central government level, this consists of the eight major SA programs as well as remaining Kemensos and social protection (SP) function expenditures. At the sub-national evel, SP function expenditures are used to approximate SA expenditures." Using this approach, expenditure levels for Cluster 1 SA lie between official "social assistance" and "social protection" budget categories. Indonesia has also increased expenditures on social insurance, though expenditures mainly cover civil servant pensions. In 2004, Indonesia passed the National Social Security System Act which laid out plans for universal social insurance coverage. Some legislative progress has been made - for instance in formalizing the institutions that will manage and deliver the different insurances - but many implementation details remain to be decided. Currently, public expenditures on social insurance are confined to civil servants and these have been growing quickly in real terms and exceed SA expenditures (Figure 13). In 2010, the government spent an estimated Rp 54,000 billion on social insurance, equivalent to 8 percent of total government expenditures (0.8 percent of GDP);- these funds mainly covered civil servant pensions (provided through PT Taspen, Tabungan dan Asuransi Pegawai Negenl). Since 2009, 100 percent of pension funds have been covered by the central government. A portion of this expenditure is also used to cover health insurance premiums for current and retired civil servants through PT Askes (Asuransi Kesehatan). Overall, the central government spent 1.3 percent of GDP on social protection in 2010, of which around one-third went to household SA and two-thirds to social insurance. While interest payments have declined dramatically, education and government administration expenditures have consistently increased and health and Cluster 1 SA expenditures remain at low relative levels. A decade of political stability, sustained economic growth and sound fiscal management has expanded Indonesia's public resources. Moreover, sustained debt reduction has led to a declining interest payment burden from 25 percent of total expenditures in 2001 to under 9 percent in 2010, significantly expanding Indonesia's fiscal space (Figure 15). Reflecting national priorities, education spending has increased significantly and now accounts for over 20 percent of total national expenditures (up from 12.6 percent in 2001). Infrastructure and health have been increasing slowly since 2001. At the 15 No breakdown of these high-level budget classifications is currently available at the sub-national level. At the sub-national level, expenditures classified as Social Protection may exclude household social assistance spending classified under the education and health functions. On the other hand, it may also include broader social and poverty reduction expenditures. 42 Does Indonesia allocate the right level of resources to U household social assistance? same time over 19 percent of total expenditures are now consumed by government administration (up from 11 percent in 2001) and energy and other subsidies continue to consume a large portion of the fiscal envelope (over 20 percent in some years). In comparison, household social assistance - which includes programs normally classified as education, health and subsidy expenditures - accounted for 2.9 percent of total national expenditures in 2010, up from 1 percent in 2001. Declining Figure 15. Sectoral Composition of National Public Expenditures, 2001-2010 debt and debt-service so so3 burdens have allowed 2s 25 ucpublic spending in 20F-20 20 most other sectors to is-cl increase. 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Agriculture Education (exc, SA) Household social assistance (SA) Health (exca SA) Government Administration Energy and other subsidies (exc. SA) Infrastructure Interest payments Sources and Notes: Keenkeu and World Bank staff calculation. Energy and other subsidies includes: fuel, electricity and tax subsidies. Indonesia's impressive fiscal consolidation has opened up capacity to increase SA expenditures. Indonesia has reduced its public debt quite notably over the past decade: debt-to-GDP ratios have declined from around 90 percent in 2000 to just over 25 percent in 2010 (Figure 16). Indonesia's fiscal consolidation is notable relative to other Asian economies as well as other emerging economies that have experienced debt crises (Figure 17) as well as when compared to the rising indebtedness of many economies following the recent global downturn. Looking forward, Indonesia's debt sustainability appears favorable: the strong starting position and baseline outlook for growth relative to real interest rates points to a continued downward trajectory across different scenarios, with the baseline indicating a debt-to-GDP ratio of around 18 percent by 2014. This strong fiscal position affords Indonesia the option to increase expenditures on key development objectives such as poverty reduction through higher borrowing without raising debt ratios. Under a baseline outlook, higher fiscal deficits of up to 3 percent of GDP would still be consistent with stable debt-to-GDP ratios. 43 E Protecting Poor and Vulnerable Households in Indonesia After Figure 16. Indonesia Debt to GDP Figure 17. Regional Debt to GDP Ratios, ballooning Ratio, 2000-2010 1991-2010 during the Asian Financial 120 120 120 120 Crisis (AFC) the debt-to- 100 1oo 1oo 1oo GDP ratio in so so sO s o Indonesia o so o a e o has declined G smoothly, 2140 40 40 40 which is 20 20 20 20 noticeably a 0 different npoca-wi g o D e grs 1991 1994 1997 2000 2003 2006 2009m t from other - - - I I I AFC-affected External debt Domestic debt economies. Source: Kemenkeu, BI, BPS, World Bank staff Source: Kemnenkeu, BPS, World Bank staff calculations, IMF calculations. public debt database. Note: Debt-to-GDP ratios may differ with Kemenkeu Note: Indonesia is central government debt. Definition for figures due to exchange rate used to convert external other countries varies. IMF database used for Indonesia debt to local currency. pre-2000. However, regressive energy subsidies continue to consume fiscal resources and dwarf SA expenditures. While the 2005 fuel subsidy reforms and subsequent adjustments helped reduce the burden of energy subsidies, they remain costly - especially during periods of elevated global oil prices (Figure 18). For example, at the height of global oil prices in 2008, Indonesia's annual spending on energy subsidies reached Rp 223 trillion (US$ 23 billion), equivalent to 4.5 percent of GDP and 32 percent of total central government expendit us two-thirds of this spending went to fuel subsidies alone. Energy subsidies in ndonesia do not target poorer segments of the population, as most benefits accrue to commercial users and non-poor car-owning households. Despite their regressive nature, spending on energy subsidies was more than seven times greater than spending on SA in 2008 and more than five times greater in 2010 (Figure 19).16 Energy Figure 18. Public Expenditures on Energy Figure 19. Public Expenditures on subsidy Subsidies, 2004-2010 Energy Subsidies and Household Social spending has Assistance, 2008 and 2010 been volatile recently and 240 40 continues to consumes a much 1so 3o larger share 4 Electri- of public 5 ct 120 20 expenditures than SA. 2 - Electri- 60 10 Fuel city Fuel 0 - 0 0 1 2004 2006 2008 2010 Energy Household Energy Household Energy subsidy spending (Rp trillion, LHS) subsidies SA subsidies SA % of central gov. expd (RHS) % of GDP (RHS) 2008 2010 Source: Kemenkeu, BPS and World Bank staff. Source: Kemenkeu and World Bank staff calculations 16 See March 2011 Indonesia Economic Quarterly (World Bank 2011) for fuller discussion of current issues with Indonesia's fuel subsidy system. 44 Does Indonesia allocate the right level of resources to U household social assistance? The fiscal space for SA expenditures could be enlarged with additional subsidy reform. Citing their large costs and poor targeting, the government has in recent years signaled its intention to reduce spending on energy subsidies. However, fuel subsidy reform has subsequently stalled, creating uncertainty about the future path of subsidy spending. The RPJM for 2010 to 2014 specifies a sharp decline in nominal annual spending on fuel subsidies, with spending totaling just Rp 211 trillion between 2011 and 2014. If reform plans remained stalled, however, spending could rise by an additional Rp 250 to 350 trillion, even under RPJM-assumed low oil prices of US$ 80 per barrel. If oil prices remain closer to US$ 100 per barrel, subsidy spending could be as much as Rp 500 trillion higher. For example, unofficial calculations of fuel subsidy spending in 2011 indicate that it rose to over Rp 160 trillion, which is nearly double 2010 levels and over Rp 30 trillion more than than planned for in the 2011 budget. Such unplanned additional spending on fuel subsidies could jeopardize the government's plans to increase spending on other key development plans, such as increased spending on infrastructure, health insurance and SA programs. Composition of Social Assistance Expenditures Raskin dominates SA expenditures, accounting for over half of total expenditures, while Jamkesmas and BSM also consume significant shares of spending. In years when it has been deployed, BLT consumed a large share of SA expenditures, peaking at over 60 percent in 2006 and 40 percent in 2008 (Figure 20 and Table 8). BLT has been delivered as a temporary program, however, and in routine years, Raskin - which delivers subsidized rice to enhance food security - typically accounts for over half of total expenditures, including an estimated 53 percent in 2010 (Figure 21). Jamkesmas is the second largest initiative in terms of expenditure (accounting for almost a fifth of expenditures in 2010), followed by BSM (14 percent). In contrast, spending on pilot cash transfers targeting the most vulnerable groups collectively account for less than 10 percent of total expenditures. PKH, which targets very poor families, accounted for 4 percent of total SA expenditures in 2010. The other programs implemented by Kemensos are allocated a small share Cluster 1 SA resources. PKSA (abandoned children) received 1 percent, JSPACA (disabled) received 0.8 percent and JSLU (vulnerable elderly) received 0.3 percent of total expenditures. SA spending is Figure 20. Household Social Assistance Figure 21. Household Social Assistance spread across Expenditure Level and Composition, Expenditure Shares, 2010 eight major 2004-2010 programs but is dominated Rpblo,nmnlBLT by the in-kind 35,000 Rbilo,nm alOther SP Disabled* Elderly* Other SP Other rice subsidy, 30,000 as81 0.% F .-- Kemensos Raskin. The 0Other Children* theelaget 25,000 MOSA 10% programs 20,000 Disabled* 4% together BSMI-- account for 15,000 Children* 14% approximately 10,000 PKH "akn 85 percent 3SM of all SA 5,000 lme as Jamkesmnasl_ expenditures. oC 18% 2004 2005 2006 2007 2008 2009 2010 Raskin Sources and Notes: Kemenkeu and World Bank staff. * Refers to all spending in the diectorate implementing these programs. 45 E Protecting Poor and Vulnerable Households in Indonesia The remaining SA expenditures are highly fragmented and are distributed across 12 ministries, 12 programs and 87 activities. Of the remaining SA expenditures, the bulk (over 80 percent) are executed by Kemensos and consist of 37 small social protection activities, grouped under 8 programs, plus all salary and administrative costs. The remaining expenditures consist of an additional 50 small social protection activities, grouped under 14 programs and spread across 11 ministries; they are focused mainly on family, gender and children's issues.17 The majority of sub-national SA expenditure appears to go to staff salaries and general administration in support of centrally-delivered programs. While sub-national governments (provinces and districts) officially have the primary responsibility for social welfare, they are estimated to have executed only 12 percent of total national expenditures on SA in 2010.1 Case studies and field visits indicate that districts have little discretion over their SA budgets and that the majority of expenditures are absorbed by staff salaries and general administration in support of central and provincial government programs. Even though discretionary funds are limited, small programs for vulnerable groups and poor families appear to be common, as well as more general social programs for natural disasters and community development and empowerment. Local health insurance programs for the poor have proliferated in recent years to provide a complement to, or address gaps in, the Jamkesmas program (Box 3).1 There are also local scholarship programs. 17 ee 5ocal ssitaCePoaman Publi Epeites Rve 13,9 ublic Expenitur 1ummary f8,7r more6 deal6nths,malr1oil2sisac % total central government pendiexpenditures 2.3 3.6 6.7 2.8 4.8 4.4 3.7 % Total central government expenditures 2.3 2.3 2.5 2.8 2.8 3.8 3.7 2 Raskin 4,831 5,218 5,570 6,584 9,926 12,987 13,925 4. BSM - 1,238 2,562 3,607 6. Child 5ervices (inc. PKA n/a 104 211 187 311 296 254 8. Elderly Services (inc. JSLU) n/a 26 53 57 69 82 75 Other Social Protection (SP) 1,899 180 197 295 297 302 352 17 See "Social Assistance Program and Public Expenditure Review 1: Public Expenditure Summary" for more detail on these smaller social assistance activities. 18 Information and data on program spending at sub-national levels, including impacts and expenditure, is limited and incomplete so this is a rough estimate only. 19 In addition to Box 3, see also "Social Assistance Program and Public Expenditure Review 4: Jamnkesmas" and the references therein for more detail on the locally-developed health insurance programs. 46 Does Indonesia allocate the right level of resources to U household social assistance? Internationally, more mature social safety nets allocate larger resource shares to cash transfers. Partly this stems from different target groups (like war veterans) and partly from chosen obligations (for example, delivering a minimum old-age pension to all citizens), but also a preference for cash rather than in-kind transfers.20 Developing countries in Europe and Central Asia allocate most of their SA expenditures to cash transfer programs; little is spent on in-kind transfers. The largest share of expenditures, on average, goes towards benefits for war veterans (24 percent), followed by vulnerable families and children (20 percent) and the disabled (16 percent) (Figure 22). Considerable shares are also allocated to last-resort income support programs, which include conditional cash transfer programs in some countries (Turkey, Macedonia and Tajikistan) and to elderly benefits. In addition, a number of Latin America countries also spend heavily on CCT programs. For example, both Brazil and Mexico spend around 0.4 percent of GDP on their flagship CCT programs (Bolsa Familia in Brazil and Opportunidades in Mexico) that cover around a quarter of total population (46 million people in Brazil and 25 million in Mexico).21 Finally, a large number of developing countries also spend heavily on benefits for poor elderly. For example, one report examining expenditures for ten such programs around the world found that spending ranged from 0.2 to 2 percent of GDP with a simple average of 0.8 percent of GDP22 Brazil's rural old- age support scheme, for example, spent 1.0 percent of GDP to cover 6 million beneficiaries. In contrast Figure 22. Social Assistance Expenditure Composition, ECA Averages to Indonesia, more mature Last-resort SA systems income support in Eastern 12% Europe and Central Asia (ECA) Other allocate19 22~~~~aml aonadadHnz(05 the bulk of Elderly Famid expenditures benefits allowances to cash 9 0 transfers. Veteran Disability Benefits benefits 24% 16% Sources and Notes: Europe and Central Asia Social Protection Database, World Bank. Most data from 2008 or 2009. Last- resort income resort includes CCTs in some countries. 20 See also Box 1 in the Introduction for more examples of the variety of social assistance instruments, amounts, and coverage profiles that other middle income countries in Asia, Latin America, and Africa have chosen. 21 Grosh et at. (2008) 22 Holzman and Hinz (2005) 47 E Protecting Poor and Vulnerable Households in Indonesia Box 3. Since the introduction of Jamkesmas (previously Askeskin), local governments have been Community- establishing complementary or "top-up" health insurance schemes for the poor. Collectively, based these scheme reos tre Jamkesda (Jaminan Kesehatan Daerah); they provide some coverage development and to those classified as poor or near poor but who are not covered by Jamkesmas or, in fewer cases, PNPM provide additional benefits not covered by Jamkesmas. Some local governments have also opted to provide free health services for the entire population. Gani et al. (2009) recorded that in 2008 there were 65 districts providing 'free health care' and 38 providing Jamkesda schemes in 2008. By 2009, the number of districts providing 'free health care' schemes is estimated to have more than doubed. 2010 estimates indicate as many as 335 (or 67 percent of all 498 districts in Indonesia) are providing Jamkesda to approximately 27.5 million household S.1 Local governments with Jamkesda schemes claim to be responding to noticeable gaps in Jamkesmas coverage of poor and near-poor households. In the absence of coordinated regulation and a clear distribution of roles and responsibilities between the central and local governments for health care insurance provision and coverage, Jamkesda schemes have proliferated. They vary in terms of population group covered,benefitteackageucsfered isnadequae f ert, member ctontgo.n, and the way the schemes operate and are organized. Though Jamkesda are not required to report their coverage, health service utilization, or financial or reated information to the central government, Kemenkes has been collecting information on a voluntary basis for the ast three years. Typically, the information submitted is limited to budget and spending reports. In 2009, 142 districts submitted reports to Kemenkes, which was more than double the 2007 number; however fewer than 60 districts reported two consecutive years in a row. The size of Jamkesda yearly budgets ranges widely, from less than Rp 10 million (around US$ 1000) to more than Rp 30 billion (US$ 3 million), consistent with the variation mentioned above. In the absence of utilization information it is difficult to estimate the contribution of Jamkesda schemes to risk prevention or healthcare promotion, or their impact on overall health spending, among poor and vulnerable households. Overall Adequacy The current level of SA spending is not sufficient to cover all eligible beneficiaries, let alone all vulnerable households that may need protection. BSM, PKH and the cash transfers for marginal groups are still in their pilot phase and do not yet have the mandate or resources to reach all eligible beneficiaries. None of the current programs are intended for the large number of vulnerable households who are in the bottom 40 percent but nonetheless lie above the near-poor cutoff. For example, PKH is only aimed at very poor households (0.8 times the national poverty line), while the three big programs (BLT, Raskin and Jamkesmas) are only aimed at the poor and near-poor (1.2 times the poverty line). Reaching all eligible households and extending coverage of some programs to vulnerable households that lie above the near-poor cutoff would require significantly more resources than currently allocated. The benefits currently transferred by the major programs represent at best 60 percent of the cumulative income gap of poor and near-poor households or 10 percent of the cumulative income gap of all vulnerable households. The cumulative income gap of the poor and near-poor is the amount of expenditure that, if transferred to them, could theoretically raise their consumption level to 120 percent of the poverty line and thereby technically remove them from both poor and near-poor categories. Comparing this gap with actual current benefits transferred gives one indication of whether the level of resources allocated is adequate for poverty reduction goals. Using household survey data, the cumulative annual income gap of Indonesia's poor and near-poor is estimated at around Rp 35,000 billion or 0.5 percent of GDP in 2010 (Figure 23). The cumulative income gap of all vulnerable households (the bottom 40 percent, or those with expenditure of approximately 1.5 times the poverty line or less) is estimated at around Rp 170,000 billion or 2.6 percent of GDP for the same year. Though Indonesia spent around Rp 30,000 billion (or nearly 0.5 percent of GDP) on SA in 2010, according to official and promised benefit levels, only around two-thirds of this (approximately 23 Soewondo(2011) 48 Does Indonesia allocate the right level of resources to U household social assistance? Rp 20,000 billion or 0.3 percent of GDP) is transfers to households from the major programs; the rest is either administration or sub-national expenditure. So, official benefits transferred represent just 60 percent of the cumulative income gap of the poor and near poor or just 10 percent of the cumulative income gap of all vulnerable households.24 In comparison, Indonesia spent 2.2 percent of GDP on energy subsidies in the same year (4.5 percent in 2008), the majority of which benefit non-vulnerable households. Total household Figure 23. Income Gaps and Total Social Assistance Benefits Transferred, 2010 benefits from the major SA programs 3.0 are much lower than cumulative 2.5 income gaps of 2 0 poor and vulnerable f populations as well as t1.5lnerall lower than all energy subsidy spending.1. 0.5 - Near Poor Admin and others n.n Poor Benefits Cumuative income gap Household SA expenditures Energy subsidy expenditures Sources and Notes: Kemenkeu, Susenas 2010 and World Bank staff calculation. According to medium-term planning documents, Indonesia has only modest expansion plans for most Cluster 1 SA programs; expenditures are projected to flat-line at around 0.5 percent of GDP. Plans outlined in the RPiM monitoring and evaluation indicators developed by Bappenas are, for the most part, bunea usehual. Overall, total national expenditures on household SA are projected to rise in nominal terms to around Rp 40,000 billion by 2014 which is equivalent to a 0.5 share of projected GDP (Figure 24). Moreover, the composition of household SA expenditures remains largely unchanged. No mention is made of BLT (which is assumed inactive) or Raskin (which is assumed to remain unchanged), while the indicative Jamkesmas allocation rises only modestly in nominal terms. In contrast, the RPJM outlines ambitious expansion plans for BSM: the target number of beneficiaries is planned to roughly double to almost 8 million students by 2011, enough scholarships to potentially reach all poor school-aged children in Indonesia (based on household survey estimates). RPJM plans for PKH see an expansion from just over 800,000 households today to 1.5 million by 2012, but it remains a relatively small program with low coverage and a small budget."" Similarly, the programs for especially vulnerable groups are planned to expand modestly, but will remain relatively small. 24 Due to necessarily imperfect targeting technologies, only a portion of actual benefit expenditure reaches vulnerable households. For example, in the major national program with perhaps the most accurate targeting (BLT), approximately 64 percent of total benefits were received by households in the bottom 40 percent, meaning approximately 36 percent of benefit spending went to non-poor and non-vulnerable households. This suggests that (given the current targeting technology) only Rp 12,700 billion (or less) of the Rp 20,000 billion in direct transfers are ending up with vulnerable households. This in turn suggests that current benefits transferred to vulnerable households represent (at best) 38 percent of the cumulative income gap of the poor and near poor or 6 to 7 percent of the cumulative income gap of all vulnerable households. This does not yet account for the sometimes sizeable gap between officially promised benefits and benefits actually received; accounting for this gap would reduce the relative size of the transfers provided to poor and near-poor households. See Section 3 below as well as 'Targeting Poor and Vulnerable Households in Indonesia' for additional detail on targeting outcomes. 25 More recently, Kemensos officials, TNP2K officials, and TNP2K national planning documents have indicated that 2014 intended coverage has been raised to 3 million households in all Indonesian districts. 49 E Protecting Poor and Vulnerable Households in Indonesia Current medium- Fgr 4 rjce oilAssac xedtrs 0121 term planning has SA expenditures flat other dtn lining at their current real levels (0.5 a f, cce percent of GDP) in the coming years. 30,000 s t 06 500 10,000 -- 0,2 0 0,0 2010 2011 2012 2013 2014 Nominal budget (LHS) Share of GDP (RHS) Sources and Notes: Kemenkeu, Bappenas and World Bank staff calculation. Indonesia's SA expenditures are low in comparison to other developing countries which spend on average of 1.5 percent of GDP it can be difficult to compare spending on SA across countries due to differing definitions and scope, but best-practice estimates indicate that most developing countries spend between I and 2 percent of GDP (Figure 25). Indonesia's spending is also noticeably low relative to other countries in East Asia and the Pacific, which spend 1 percent of GDP on average. In Latin American, where data coverage is more complete and SA programs are considered to be relatively well developed and efficient, countries spend 1.3 percent of GDP. For reference, the average for OECD developed countries is around 2.5 percent of GDP In comparisonFiue2.ScaAsitneEpniueLvlinItrainlCm rso to a wide range of developing 1 countries, 2 Indonesia SA l spending is at the 8 low end. 2 Sources and Notes: Weigand and Grosh 2008. Data taken from World Bank public expenditure reviews or other similar work. *Data for Indonesia and the Philippines are World Bank Staff estimates. N=number of countries in group with data available. Indonesia appears to be allocating very low levels of resources to SA. Indonesia's current and projected level of spending, at 0.5 percent of GDP, is not sufficient to protect all eligible households, let alone all vulnerable households. Moreover, spending is low in comparison to other developing countries, which spend 1.5 percent of GDP on average. Given its strong fiscal position, Indonesia could afford to spend more. In addition, the current spending mix is unbalanced and dominated by Raskin, an in-kind subsidized rice program. 50 Does Indonesia allocate the right level of resources to U household social assistance? 51 Do programs provide the right benefits? SA benefits are "right" when they allow households or individuals to meet objectives that the transfer program encourages. Indonesia, through the Cluster 1 programs and elsewhere, attempts to bring cash, in-kind goods, and services and facilitation to the disadvantaged to protect them from destitution, to prevent the adoption of unproductive coping strategies when income is scarce, and to encourage the acquisition of the social services and human capital that will keep them less vulnerable in this generation and the next. This report does not attempt to define a minimum (or maximum) standard for benefits; instead the discussion focuses on whether benefits received can facilitate the achievement of a program's goals. SA programs in Indonesia provide benefits that vary significantly in their generosity. BLT and PKH deliver benefits equivalent to around 10 percent of the poverty line (for total household expenditure), while the pilot cash transfers for marginal groups deliver benefits worth between 17 and 34 percent of household poverty line expenditure. Raskin is the least generous, delivering benefits valued at just over 2 percent of the household poverty line, followed by BSM (around 5 percent on average). The total benefit packages (Table 9) for all programs with the possible exception of Raskin include facilitation, outreach, and socialization, and as detailed below, these services too can vary from adequately- to inadequately-provided. 52 1国園 ［コ ー53 E Protecting Poor and Vulnerable Households in Indonesia Some of these benefits are inadequate by design. BSM benefits look adequate at between 15 and 30 percent of per- capita poverty line expenditure, but not when compared to actual poor household education expenditures. Yearly out-of- pocket expenditures for secondary education, for example, range from Rp 1.5 million to Rp 2.2 million (Figure 26). As a percent of overall expenditure, secondary education can be prohibitively expensive for the poorest households, consuming approximately 20 percent of overall expenditure, while they are a smaller 5 to 10 percent of a rich household's overall expenditure. The opportunity costs for a poor household when it sends a child to school are also larger (as a percent of total household income) than for wealthier households, making education, and especially secondary education, doubly expensive. This inadequacy of benefits when measured relative to the actual cost of the promoted service also affects PKH and may be part of the reason that conditional cash transfers have not had a noticeable impact on drawing more children into upper schooling levels. The BSM Figure 26. Household Education Expenditures and Benefit Amounts by school level, scholarship 2009 program provides benefits 3,000,000 - Rupiah that are not commensurate 2,500,000 - with total education 2,000,000 - expenditures; this gap 1,500,000 - increases in higher schooling 1,000,000 - levels. 500,000 0 SID SMID SMU Poor hh, education costs 3SM amount Non-poor hh, education costs PKH amount Source: Susenas (2009) and World Bank staff calculations Note: PKH amounts reflect only the amounts given for school-aged children when they attend school; see "Social Ass stance Program and Public Expenditure Review 6: PKH" for more detail. Total PKH transfers are larger (on average) but are based on a larger number of conclitionalities that pregnant mothers and under-5s must meet. The Raskin program delivers far smaller benefits because of noticeable discrepancies between total Raskin rice procured and total Raskin rice purchased (Figure 27); between total benefit promised and total benefit received (Figure 28); and between total number of beneficiaries targeted and total number of actual Raskin beneficiaries (see Section 3 below). In 2010, based on public announcements, Raskin should have delivered to poor and near-poor households at least 13 to 16 kilograms of rice per month at a subsidized price of Rp 1,600 per kilogram. These amounts would have translated into approximately 11 percent of poor household expenditure and between 30 to 40 percent of estimates of an Indonesian household's rice needs (between 35 and 45 kilograms per month, according to GOI press reports). Actual purchases as reported by households, however, were estimated to be far less (Figure 28). Ministry of Finance budget reporting shows that of the Raskin rice procured to deliver promised benefits, only about half of the procured kilograms (in recent years) are actually purchased by eligible households (Figure 27). It is not clear at which stage in the Raskin delivery process - procurement, storage, transport, or local-level storage and sale - the bulk of the procured Raskin rice goes missing, but the situation on the ground indicates that distribution was not as planned. Furthermore, given actual average Raskin purchases (between 3 and 4 kilograms) and actual average Raskin prices (approximately 60 percent higher than the stipulated price of Rp 1,600 price), actual benefit values were between 2 and 3 percent of poor household expenditure, and only 6 to 11 percent of a household's monthly rice needs. 54 Do programs provide the right benefits? U There are as Rise bu an mrgn Rstibuinises ntbe drsd significant discrepancies and 2)ecn between total amounts of Raskin e H , a tre rice procured for distribution and the total400- amount of Raskin3,0 purchased by households. This combined with local p erences ,000 ( 1,000 per for equitable distribution leads were a cmiao ortlt to a small benefit n l transferred0 through the Raskin2060720809 Figure 27. Raskin ra Procu are and intay sys ic Purchas no6s2009 2000 I 4100 5 355 32000 2006741 Ougtdbnft(gh/o rmsdbnfta fpo 2ou000l 1xpn625r Ttual skni rocuredo otaasi household purchasres also ~ ~ ~~ Toa rcos2 0/6t 0 89, h rq ece conspione por anvnar-pousehuo s Tota in r e comping al o houelds uetin from~~~~~ gurueassres i 09te 28.raskinc PomiedcBenes aavnctuale Ras Purch as 5 es cn o lh place. a 1055 M Protecting Poor and Vulnerable Households in Indonesia Jamkesmas benefits are potentially very generous, but knowledge, health service supply, and access costs limit utilization and therefore actual transfers. The value of a Jamkesmas card depends on whether it is used to acquire health care and if so, which health care services are acquired. As mentioned above, essentially all health care services are covered by Jamkesmas, so households could be transferred a significant implicit value when using a Jamkesmas card. However, reports from the field and impact evaluation studies indicate that households not already familiar with the modern healthcare system (most poor and near-poor households) do not understand how to use the Jamkesmas card, including what services are covered nor if and when household members are eligible. Poorer households also mention that health service provider access costs - a combination of lost wages, transportation, lodging, and childcare or chaperone care (and daily expenses for those individuals) - are prohibitive and cannot be addressed by Jamkesmas. Finally, supply-side constraints and differential treatment of Jamkesmas patients means that actual healthcare services provided are limited for most households who do not request treatments explicitly.26 Cash benefits are typically not inflation-indexed and therefore are eroding in real terms. Benefit amounts for all programs have not been adjusted to take into account increases in the cost of living and have remained unchanged at their initially-set levels, meaning their real value to beneficiaries has declined significantly over time. For example, the JSPACA and JSLU benefits of Rp 300,000 per month have remained unchanged since 2006, and are now worth almost 30 percent less at current 2010 prices. The only program unaffected by inflation is Jamkesmas, which provides fee waivers instead of direct cash transfers. Facilitation, outreach, and in-kind services, which are a significant component of some benefit packages, vary in quality. For example, in the PKSA, JSPACA, and JSLU programs, the cash transfers are larger than all other Cluster 1 transfers, but targeted individuals face an array of difficulties - lack of access to household, kin, and community support networks; difficult access to regularly-provided social services like education, health, and sanitation; psychological, psychosocial, and emotional issues that may be both a cause and product of the individual's disadvantaged status - not all of which are best addressed by cash alone. The BSM scholarship program does not explicitly include outreach for those contemplating leaving school nor is it designed to accept individual application by interested students or households. As mentioned above, the lack of facilitation and outreach in the Jamkesmas program is likely limiting the use of the card among poor households. For all Cluster 1 programs with an "increase in access to and utilization of social services" objective, facilitation and in-kind services may be equally important as cash, but effective delivery of these elements has so far lagged cash transfer disbursement. Only some programs have most elements of an effective benefit package. BSM and Raskin do not provide the right benefits because of both design and implementation issues. PKH's education transfer is too low, while for reducing inequality and for health behaviors the transfers have been appropriate. Jamkesmas fee waivers are very generous, but the Jamkesmas benefit package does not include facilitation or outreach and like BSM, it cannot address the total real costs of access. All of the SA programs with an explicit focus on promotion through behavior change and use of social services - PKH, Jamkesmas, BSM, and the cash transfers for vulnerable populations - would benefit from more high-quality facilitation and outreach but currently cannot or do not provide enough of this valuable benefit; see Section 5 below for greater detail. 26 Even when explicitly requested, some Jamkesmas-covered services remain unavailable if local health service staff do not know about, or are unwilling to provide, the Jamkesmas-covered referral option. 56 Do programs provide the right benefits? U 57 Are benefits reaching the right people? A social assistance transfer reaches the "right" people when a significant majority of the benefits provided reach the households and individuals that the program targets. Most of the programs discussed in this report have target or priority populations that are described in operational manuals and national regulations. With the aid of plentiful Indonesian household survey data, it is a relatively straightforward exercise to evaluate the ability of programs to reach the right people according to a program's own objectives.27 The major programs target poor and near-poor households, while the smaller pilot programs aim to reach the extreme poor and specific marginalized groups. The largest national programs - Raskin, Jamkesmas, and BLT (when active) - offer benefits to poor and near-poor households (approximately 25 percent of all households) and are at least designed to reach a portion of the population vulnerable to impoverishment. The BSM initiative is meant to cover all students or school-aged children from poor households. PKH is still a pilot program that provides benefits to extremely poor and demographically eligible households. The smaller programs - JSLU, JSPACA, and PKSA - cover even narrower groups or marginalized individuals, but are generally prioritized to poor individuals; for example, poor rather than non- poor disabled individuals have priority over JSPACA funds. A significant number of poor households are excluded from beneficiary lists. The poorest households are the most likely to receive program benefits, but less than half of the poorest 40 percent of households receive BLT and Jamkesmas (for example), while 20 to 25 percent of total benefits from both programs goes to the richest 40 percent of households. Over 70 percent of the vulnerable (or the poorest two-fifths of households) receive Raskin, but Raskin also has high coverage of the non-vulnerable, a result of local-level Raskin sharing among all households; as mentioned above in Section 2, this sharing also dilutes each poor and near-poor household's Raskin benefits. In a comparison of targeting 27 Internationally, there are many examples of social assistance benefits with explicit targeting and priority rules as well as many examples of social assistance benefits that are essentially entitlements that any citizen has a right to receive; see Targeting Poor and Vulnerable Households in Indonesia for international perspectives on targeting. 58 - --- - ------ outcomes, and with 100 percent representing perfect targeting according to program design, BLT performs the best at 24 percent better than random, with Jamkesmas and Raskin at 16 and 13 percent respectively.28 BSM performs quite poorly: the poorest 30 percent of students receive less than double the share of benefits than the share received by the richest 30 percent receive; a BSM transfer is nearly as likely to be received by a student from a poor or vulnerable household as by a student in a non-vulnerable household. Overall, Indonesian program targeting, as measured by coverage of the poor, is in line with international benchmarks; however, as measured by leakage to the richest households, Indonesian targeting is much worse than elsewhere.29 This means that, in Indonesia, the percentage of benefits enjoyed by the poorest 40 percent lags behind international benchmarks, while the percentage enjoyed by the richest 20 percent is higher than in other countries (Table 10). 28 That is, targeting outcomes under BLT (Jamkesmas, Raskin) are 24 (16, 13) percent better than if the same number of benefits had been distributed randomly 29 See Targeting Poor and Vulnerable Households in Indonesia (World Bank, 2012a). 59 * Protecting Poor and Vulnerable Households in Indonesia BT* near-poor 46 51 5.0 PSEO5 (2005) households HH update moves/deaths Poor/ 17.5 mn switch to 2006 quota Raskin near-poor 1 74 45 4.1 BKKBN (1999) St5 uta households HPE5 ajsmn Poor/ Jamkesmas near-poor 76.4 mn 40 48 4.2 PSE05 (2005) append add near- householdsPPS8 pot BSM* Students 4 6 mn 3 39 1.8 PSE05 (2005) from poor HH PKH** Very poor 810,000 n/a n/a n/a PSE05 append 2008/9 households HH PPLS08 expansion' JSLU Vulnerable 10,000 elderly JSPACA ve ,000 n/a n/a n/a proprietary n/a PKSA Vulnerable 4,187 children Each program has developed its own eligibility rules while targeting in practice has often strayed from official guidelines." For example, BLT was meant to use a mix of data collection methods, but each step in the data collection procedure was carried out with significant revisions: statistical assessment of poverty status was not in-line with international best practice while community-based assessment was in most cases not consultative or transparent. Revisions to beneficiary lists for the 2008 BLT produced no significant differences from the 2005 beneficiary lists, so poor household exclusion did not improve. Raskin is meant to use official lists of the poor to select beneficiaries, but in practice communities distribute the rice as they see fit, often sharing equally among all or nearly all households. Jamkesmas is also meant to use official lists of the poor but there is considerable variation in beneficiary identities at the local level, with local health officials sometimes choosing beneficiaries, or households selecting themselves based on previous healthcare use." Different targeting approaches mean different beneficiaries for each program, even though all target the same households. Each of the programs approaches targeting in a different way and has a different database of beneficiaries. As a consequence, even though all three major programs target the same target population (the near-poor, or bottom 25 percent of households), less than one third of target households receive all three programs, while nearly half receive one or no program (Table 11). At the same time, over 10 percent of non-target households receive all three, including many of those in the richest half of the distribution. Poor households have an approximately equal chance as non-poor households of receiving one of the three major programs. In general, household eligibility as determined by one program does not seem to matter for eligibility in another, as programs with the same eligible population do not have consistent or common ways of identification and classification. Comprehensive coverage appears a matter of luck rather than design. 30 Refer to Targeting Poor and Vulnerable Households in Indonesia (World Bank, 2012a) for a detailed review and discussion of targeting practices in Indonesia. 31 PKH is not discussed in this section due to insufficient quantitative data. PKH is still a pilot program without dense national coverage; consequently PKH beneficiaries are not sampled with enough statistical precision for calculation of nationwide or program-wide targeting performance. When PKH was first piloted in 2007, it used BPS' 2005 list of the poor (developed for BLT). Households identified as very poor on this list were eligible. By BPS definition a very poor household is a household that has less-than-poverty line expenditure overall; spends a large portion of available income on basic staple food; cannot afford medical treatment (except at the community health clinic or other public health facilities subsided by the government); and cannot afford sufficient new or replacement clothing. In practice, households meeting these standards have per-capita expenditure levels of approximately 0.8 times the BPS-defined poverty line. All households below the very poor cut-off with the right demographic composition (identified with a supplementary survey) were eligible for the PKH program, but the PKH implementing units in Kemensos (UPPKH) chose only some of the eligible households to receive PKH transfers after (UPPKH) together with BPS. 60 Are benefits reaching the right people? U Though eligibility Table 11. Social Assistance Programs Received by poverty category, 2009 rulesocadiatonud mPercentage of Each Poverty Classification by Number of Programs Received to determine quotas Programs are substantially the Received Very Poor Near- AI 25-50 51-80to 81- 100be Non- Total nism a ackofranspal aoor psoilzt oor percntlene electile percentile poor maintains a proprietary 0 1 14 19 16 a 51 81 49 41 beneficiary list: a 1 2 7 3 8 3 71 6 2 household that is a beneficiary under one SA program will not necessarily be a Tota 100 100 100 100 100 100 100 100 100 recipient of another SA Sources: Susenas and World Bank calculations. program. Poor socialization and mistargeting may have undermined support for SA programs. The percent of communities experiencin an buin . Targtintroduction ranged from 25 percent for Jamnkesmas (when it was called Askeskin) to 56 percent for BLT, with those not receiving assistance being the most likely to complain (Table 12). Mistargeting, nepotism and a lack of transparency, as well as poor socialization related to beneficiary selection were the sources of the overwhelming majority of complaints (Table 13).32 Communities witnessed a number households considered deserving and poor not receiving BLT while simultaneously observing non-poor households receiving BLT when they should not have. BLT provided a significant-sized cash transfer while Jamkesmas was a fee waiver and Raskin's in-kind benefits have been diluted and shared equally for some time therefore it appears complaints may be linked to the benefit size, type, and overall population coverage. Initially at least, B LT experienced less redistribution than, for example, Raskin, making BLT distribution more obviously prioritized and not universal. This may have been more controversial upon introduction, especially if program goals were not well-socialized. The nature of the community protests suggests that improved targeting of programs would improve satisfaction and buy-in. Targeting is essential in esultsgfromrtended beneficiaries receive full program benefits, which in turn safeguards effective program performance. In addition, accurate targeting is an important driver of community satisfaction, at least among a significant part of the community. Indonesia represents a complex targeting environment and new data collection can enhance outcomes in all the household-based prograMS.34 Nearly 240 million individuals are dispersed across at least 18,000 islands and over 500 districts (each of which has considerable ownership and operational control of public spending and social sector programs since decentralization). Targeting should be able to identify the chronically poor, the near poor, and the especially vulnerable (but not currently poor) in all these localities and across a relatively equal consumption distribution. In 2011, a large survey - which collected data from over 40 percent of Indonesian households - has allowed BPS to meaningfully update its list of poor, near-poor and vulnerable households and families and it is hoped it can serve as the foundation for a unified Cluster 1 beneficiary registry and an initial eligibility database. This massive improvement in data collection (called the PPLS1 1 survey), which combined results from previous lists of poor households with 2010 population census results and community nomination, is expected to result in significant targeting improvements over previous methods. 32 Those who consider themselves knowledgeable generally accept program goals and objectives, including targeting. 33 See "Social Assistance Program and Public Expenditure Review 2: BLT"7 in Volume 2 for more detail. 34 See Targeting Poor and Vulnerable Households in Indonesia (World Bank, 2012a) for more detail on all points raised in this paragraph. 61 E Protecting Poor and Vulnerable Households in Indonesia Those who didn't receive assistance 81 67 76 Those who did receive assistance 7 16 10 Community leader 7 7 3 Village officials 2 2 5 Others 3 87 The listing and selection was not transparent 32 21 25 Nepotism practice in the selection 10 9 12 The amount received was not as specified 5 13 6 Assistance was late 2 3 3 Unfair distribution 24 23 26 Practice of illegal fee in the program implementation 1 3 2 Assistance was given to those not eligible 20 16 17 Non-transparent implementation of the program 3 3 3 Other 4 9 6 62 Are benefits reaching the right people? U 63 Do people receive the benefits at the right time? Putting benefits in households' hands at the "right" time means distributing benefits when they will be used as the program intends, for acquiring the goods and services the program wishes to encourage. There could be many "right" times for disbursement to households throughout the course of a year, depending on the monthly or weekly profile of costs and expenditures that a household faces. Some benefits, like a health fee waiver, are automatically delivered at the right time because they can be used any time; others, like facilitated services, may only be right if they can be delivered frequently and on demand. Two of the three largest SA programs have delivered benefits to households when needed. BLT was well timed, reaching households during the month when the largest increases in fuel prices occurred; the cash transfers received were quickly spent on necessities, including on education expenses and preparation for religious holidays if either happened to occur during disbursement periods. Jamkesmas is always available to households; however beneficiaries need to cover the costs of access. Raskin is also continuous, with subsidized rice designed to be delivered monthly. However, local-level implementation practices - with rotation and sharing of rice amongst households regardless of strict eligibility - negatively impact Raskin's dependability for poor or vulnerable households. 64 Raskin Continuous dlelivery, but often "rotating "schemes are established, which is not effective for smoothing consumption Jamkesmas Always available BLT Yes (2005 and 2008) BSM No - arrives 1 + year late PKH Partially - timeliness of education benefits has been addressed and improved JSLU JSPACA Bunched in 2nd-half of year; ineffective for consumption smoothing. PKSA 65 I Protecting Poor and Vulnerable Households in Indonesia The smaller programs have experienced implementation bottlenecks preventing benefits from reaching beneficiaries at the right time. PKH faced bottlenecks because of only partially-available and slow MIS systems, as well as delays in the processing of information feeding into the MIS systems; both issues resulted in mis-timed and bunched payments. For example, PKH payments often arrived after the start of the school year, reducing households' ability to use PKH transfers to send their children to school. These problems have since been addressed, with PKH now making smoother payments throughout the year. The smaller pilot cash transfers - PKSA, JSLU and JSPACA - are currently facing similar bottlenecks. Slow bottom-up beneficiary identification and verification processes mean beneficiary lists are usually not finalized until May of each year, with payments reaching beneficiaries only in the second half of the year. This reduces the effectiveness of the programs at promoting better quality of life for marginal groups by making it harder for them to smooth their consumption. Lessons from the PKH experience can be useful in improving the delivery of benefits from these other programs. BSM and PKH Figure 29. Quarterly HH Education Expenditures per Student versus BSM and PKH benefits have Benefit Amounts, by school level not been timed to1,000 meet larger education 1,000,000. expenditures at the soo,000 beginning s o of a school 600,000. year or the beginning 400,000 of a new schooling 200,000 - level. Q15D BSM-s PKH -SD Q45D Q1SMP BSMt-5 MP PKH -MP Q45 isP Q15MU BSM - MU 1 sti g rade 6th grade 1st grade 3rdgra erst sude class j SD (class 6 class o SMP clas tr a ss 1 sMu n Sources and Notes: Susenas 2009 & World Bank staff calculations. BSM scholarship disbursements are usually delivered as one lump-sum payment, not in instalments. In the figure above, a BSM disbursement is imagined to be divided evenly into four quarter y payments to better iustrate the difference between BSM amounts and household schooling expenditures at different points throughout the school year and across the regular transition path from primary to junior secondary to senior secondary. BSM timeliness, and therefore effectiveness, is undermined by a lengthy verification and disbursement process... Children in Indonesia enroll in school in July. BSM then spends the following January through May period collecting beneficiary data and verifying beneficiary attendance." BSM disbursements to households do not begin until the following July and the majority of students do not receive funds before August, or more than one year after enrollment. BSM is therefore ineffective in encouraging enrollment through reducing the burden of school fees when they occur. ...and BSM is unavailable to transitioning students who are most likely to dropout. Owing to the same bottlenecks in beneficiary identification and verification, BSM is also not available during primary-to-secondary or within- secondary transition years;- as noted previously this isprecisely when the greatest poor student-dropout and the sharpest increase in the costs of education occur The primary school (SD) BSM program only delivers transfers to students actually sitting in an SD classroom;- likewise, the junior secondary (SMP) BSM programs only deliver cash transfers to students actually sitting in an SMIP classroom. BSM is therefore not available for 6th graders making the transition to 7th grade, or for 9th graders making the transition to 10Oth grade. All told, fragmentation across schooling levels, illogical rules, and slow disbursement issues combine to make BSM ineffective for the transitioning students from poor households who could benefit most (Figure 29). 35 Actually, each of the ten BSM initiatives undertakes this process separately. 66 Do people receive the benefits at the right time? U Some programs deliver benefits at the right time while some face self-imposed constraints that hamper the timeliness of transfers. Several issues in BSM design and implementation reinforce each other negatively and produce ill-timed benefits. BLT was timed well (during the month of the largest subsidy reductions) and households spent all funds rapidly. PKH faced bottlenecks because of slow MIS systems. The symptom (badly-timed payments) and cause (MIS) have both been addressed. JSLU, JSPACA, and PKSA can learn from the PKH experience and piggyback on the PKH MIS systems to improve timeliness. Local-level implementation practices in Raskin - rotation and sharing - negatively impact Raskin's dependability. 67 Are programs implemented in the right way? Having the "right" implementation is a question of both engineering design and efficiency. A well-implemented program ensures that all subprocesses - from targeting and beneficiary selection and socialization and outreach on to monitoring and evaluation and complaints and grievances - are designed thoughtfully, delegated clearly and with authority, and are provided the human resource and financial inputs necessary for their completion. Likewise, the overall environment in which the program is produced should encourage all implementers to be aware of constraints, bottlenecks and verified successes so that program reforms and enhancements are taken up quickly by those on the front line. Spending on Support Operations BSM, Jamkesmas and, to a lesser extent, BLT likely spend too little on support operations to ensure effective performance. The primary school BSM program at Kemdikbud spends less than Rp 5,000 per beneficiary per year on administration, equivalent to just 1 percent of the program's total budget (Figure 30 and Table 15). Similarly, Jamkesmas spends around Rp 10,000 per beneficiary household per year (around 4 percent of total costs). These figures likely underestimate the full cost of administration as both programs depend on service providers (schools, health posts, hospitals, and affiliated oversight, regulatory, and implementation partners at the local level) to undertake much of the implementation while no specific funds are provided to local partners for these activities. As a result, service providers are forced to absorb an unknown level of administration costs. During its most recent (2008/9) deployment, BLT spent around Rp 50,000 per beneficiary on administration or 5.2 percent of its total budget. This administrative overhead ratio is also somewhat low compared to the international benchmark of 8 percent for cash transfers and partially reflects the absence of many supporting activities - socialization, outreach, monitoring and evaluation, complaints and grievances - which 68 were underprovided, most likely because BLT was designed as a temporary program. The Raskin program is allocated the most for non-benefit spending, and the bulk of this spending goes to physical transportation, distribution and packaging of rice rather than on support operations for beneficiaries. In 2010, budget allocations to Raskin built in a transfer of just over Rp 200,000 per beneficiary for administration and operation (Figure 30). This amount of non-benefit spending (derived from the budget iloutions) may actually be higher or lower than the net amount of non-benefit spending that occurs in the Raskin program. However, according to the formulaic amount Raskin has the highest administrative overhead ratio of all the SA programs (26 percent), despite its large scale of more than 17 million households.6 However, most of this spending is absorbed by the physical transportation, packaging and distribution of rice. Little or no budget is allocated for support operations such as outreach, targeting or socialization, which would enhance the effectiveness of the program for the intended beneficiaries. These responsibilities are instead delegated to local governments, with very mixed results. 36 When comparing the total amount allocated by the budget formula for administration and operation on a per-actual-beneficiary basis with the total actual per-beneficiary benefit received, Raskin per-beneficiary administrative costs (relative to actual benefits received) are higher; see "Social Assistance Program and Public Expenditure Review 3: Raskin" in Volume 2 for more detail. 69 E Protecting Poor and Vulnerable Households in Indonesia Some programs the sal cash transe Administrative Costs appear to spend too little, and er yci some too much, on support operations amount on g to ensure efficient performance across traiingactiitis(wichin gur has houshodial cnseqnces;inseeate) Tese iitaiertisaeraoal Ameicawhee aminstatie oerhad atis Reup too1 percn (al15.Prbeneficiary aLS dministrative costea ando(RS Ioraamstatve ovalercahea mraysdecln urterams he pilots exadndtrelgrater ectsatonomitese oft ale. themdkSAd D)ectdent reaonbl Pie hpltsausadsl s75,63 HH 149 237,777s 25-enfcir 17% orPH JPCadJL are 00n0a 15t7034% 1500 Pilot 5,000 3 335 67 8% 0 0 * - , - , 0 %0 7BSM (MOEC SD Jamkemas BLT (20 9 Raskint PK 2SPC S U n Raskin PKH sends signiicantamountof Ruiahistapereeficiar (LHS Ainitative, oreadri g (RHS)on a dtrinn exenitre ae irct orcvies tandNtes emoenkeutcandsWor BanksafeDthor 200InLes aed RSAskin nmbeors ardeisrive resouces g to trgetig, mchaniclly fro bdets aellocn e scaiation,formulasrannnotnfromaobervedn,itemiedwspending Iotrast, athtesmallcas turasferusprogramsve haveuse higher. amnsrTests althoughrathse ctos are aoal reasonpablesgiven the iottatu aendmall sorcaeho thsers prormst) Per-beneir CT foram JSAAatndJL perbeiciar pher yeinsartile JSPACAead SLU sen cls to pn 350,000 15Perbeneficiary Howeverte dinstativ overheladmisratiofte latera twmramshbe declinefrhrateingt oerxtime and erlie undaer 1ecntins 2010 Unalike RSkein KTpnsabl iniicn Amuoadministrative retsinSoces oscialization,amontrn,eauto,adtann whil thepaount Sotarein as f sNcme r he pr oa' into dc in istv in d ii stativ Administrative epedtrsaredieteoonivts pthatrimprovefoutcomes forfi hoseolst n r JSLU and JSAAaorha mnsrative reures goge top tagtn,w ihrslsibeseelfne oilzto m nioinand (p eevalu ation fol o-p,endfts in l c opa rvi ote inentoaCec mr o ash transfers(8pretan towl-u C Tpog msiLtn Amroiawere d iitaieoeha aisaeu o1 ecn Tbe1) e-eeiir d iitaiecs n conderalilo)(S$ oa ugt andrthose ta rpei fim call i. Cashin transfers (Kure dbu : mnkud at SD) studBn t f siae.Rsi u br rdrvdmcaial rmbde Blloctio 2008-09) antfona 18seve7 imze HHp98n4d8585g5 PK7io07,66H 4 37772 7 Are programs implemented in the right way? U Socialization and Information Dissemination Program sustainability and behavior change both require effective socialization. Local governments, other implementation partners, communities, and beneficiary households or individuals should all understand what a program is meant to achieve, who the priority recipients are (and why those recipients are being prioritized), what beneficiaries will receive and what will be required of them, and how the program will be targeted. Each stakeholder may require a different information package and effective socialization may be achieved through different outlets. Detailed information on program strategy provides policy makers and politicians clear justification for whether programs are desirable expenditures. Potential beneficiaries need to know the program purpose and be aware of their rights in order to actively participate and to ensure benefits are not diluted. Local governments and other implementation partners will need to know the extent to which they can adjust policies to reflect local preferences as well as coordination and reporting requirements (with or to the central government and implementing agencies). When the general public is made aware of program goals and priorities, it is less likely to divert program benefits or change local implementation and allows it to act as bottom-up monitors. Socialization to all stakeholders, including on eligibility criteria, has been weakly provided. Information about eligibility rules, program objectives, and beneficiary rights and responsibilities is typically spread thinly among beneficiaries, eligible households, communities, and local-level program implementers. The large majority of Jamkesmas beneficiaries, for example, do not know which of the common services and medicines are covered by the program (Figure 31). While most PKSA recipients (abandoned or vulnerable children) knew the purpose of the program, they do not know about their rights and responsibilities and were unaware of the conditionalities and penalties associated with the program (Figure 31). Research on socialization practices for the BLT, Raskin, and Jamkesmas programs found weak socialization of program objectives, beneficiary rights and benefit amounts to all levels of government and community." For BLT, researchers concluded "It could be said that socialization to communities essentially did not take place." Weak socialization has contributed to conflict and protest activity surrounding SA programs while failing to contribute to the production of bottom-up monitoring by well-informed communities. Underprovided socialization has resulted in inconsistent information and lack of buy-in from communities and local politicians and authorities (who are often also implementers). Lack of transparency in, and knowledge regarding, targeting and beneficiary selection in particular has led to intra-community jealousy and often a re-distribution of program benefits to greater numbers of households. Without comprehensive and frequently updated program information, including on goals and objectives, bottom-up monitoring of the targeting and benefit distribution processes is precluded (see the "Monitoring and Evaluation" section just below).8 Even local government and authorities who are front-line implementers may not receive enough information to evaluate a program's performance vis- -vis households. 37 SMERU (2006, 2008a, 2009, 2010) 38 An impact evaluation report from Uganda notes that "household knowledge on how to report inappropriate behavior by bureaucrats and unsatisfactory quality of services does help to not only reduce the incidence of corruption but is also associated with significant improvements in service quality." (Deinenger & Mgupa, 2005) 71 E Protecting Poor and Vulnerable Households in Indonesia Results from Figure 31. Beneficiary Knowledge: Jamkesmas and PKSA both larger and smaller Perception of extent ofJamnkesrnas benefits* surveys A uksa,IptetSrie indicate that A uksa,IptetSrie detailed Laboratory knowledge of Medicines SA eligibilityInait rules, benefits,Inait and beneficiary Birth delivery responsibilities Emergency unit response is not widespread. 0% 5o% 100% At Puskesmas, Outpatient Services Laboratory Dental treatment Pre -nrataI ca res Family planning services 0% 50% 100% At Hospital, Inpatient Services 3rd class of Inpatient services Medium and Major surgery ICU/ICCU Birth delivery Emergency unit response 0% 50% 100% At Hospital, Outpatient Services Medical services Radiology & electromedic Dental treatment Medicine Blood services Pre -natal cares 0% 50% 100% Percent answering "covered" mPercent answering "not covered" Percent answering "do not know" 72 Are programs implemented in the right way? U Figure 31. Beneficiary perception of PKSA features continued Complaint mecharism Aid time period Penalty of the program Program evaluation Rights and responsibilities of LKSA Rights and responsibilities of beneficiaries Disbursement location Amount of assistance Disbursement time Purpose of PKSA 0% 20% 40% 60% 80% 100% Percent with information regarding feature o Percent without information regarding feature Percent unaware of feature Sources and Notes : Indonesia Corruption Watch for Jamkesmas, PUSKA PA UI for PKSA. *Aln treatments, services, diagnostics, and medicines listed above are officially covered according to Jamkesmas regulations. Facilitation and Outreach There is a facilitation and outreach gap in the initiatives promoting access to social services like health and education. When poor and vulnerable households are meant to use SA transfers to enhance their own human and productive capital by acquiring locally-provided services, those unfamiliar with services and providers will need to be recruited into and familiarized with both the service and the system(s) that produces it.it For example, in the JSLU, JSPACA, and PKSA cash transfers, facilitators are responsible for making regular instea provisic ded o th de constructive assistance such as access to basic social services like free health care, birth certificate or identity registry, and education and monitor the client's condition and utilization of cash benefits. Facilitators and eble to give motivational support to increa se encaus' ad tieste em and facilitators are th re groun e ram between a beneficiary and a program, so any obstacles, shortcomings, or malfeasance that occurs is usually reported to facilitators first. Neither BSMV nor Jamkesmas include an explicit facilitation or outreach process and this limits beneficiary demand for both the transfers themselves and the services that BSM and Jamkesmas promote. In addition to lower utilization and demand by poor households, non-poor households are given an implicit advantage (through familiarity with services and service providers) which may increase non-poor demand for and capture of benefits. Jamkesmas again provides an illustrative example: impact analyses show that Jamkesmas is generally used more often for inpatient services by the non-poor and or those with previous exposure to inpatient care providers (whether poor or nonpoor). Programs with explicit provision of facilitation and outreach services suffer from low capacity. Of the initiatives promoting services and productive investments - Jamkesmas, BSIM, PKH, and JSLU, JSPACA, PKSA - only PKH and the vulnerable group programs provide significant budget amounts and human resources for facilitation. Several studies have 39 Current socialization efforts in Indonesia are not meant to serve this role, but instead provide basic details on the transfers and program objectives themselves. The unconditional cash transfer (BLT) and the in-kind transfer (Raskin) provide direct income support only and therefore may need less facilitation. However, socialization and facilitation that strengthens beneficiaries understanding of rights and responsibilities as well as technical details like disbursement amounts and times remains the only way to encourage ground-level program monitoring by beneficiaries and communities; see the previous section on Socialization and Information Dissemination. 73 * Protecting Poor and Vulnerable Households in Indonesia noted that the PKH facilitators are crucial for encouraging and monitoring the behavior change the program demands.40 JSLU, JSPACA, and PKSA rely heavily on facilitators as well to both monitor beneficiary outcomes but also to organize additional necessary services. Of those programs, only PKSA has an initiative to recruit and pay for highly-skilled social workers as facilitators. The quality and effectiveness of facilitator networks accompanying cash transfers is highly variable. While field research and interviews with Yanrehsos (Pelayanan dan Rehabilitasi Sosial, Social Rehabilitation & Services) officials managing the JSLU, JSPACA, and PKSA programs indicate that the number of well-trained facilitators is growing, and that facilitators themselves acknowledge their lack of adequate training, the quality and frequency of facilitated services still varies widely. These programs often rely on the efforts of local-level volunteers to deliver both cash and facilitation. Facilitators (volunteer or otherwise) do not always have desired minimum education levels or training, and as a result facilitators are not always equipped to address the complex physical, mental, and social difficulties that beneficiaries present. An additional complication that also affects the PKH program is that several beneficiaries under one facilitator's care may be spread across great geographic distance, which means a facilitator may spend most of his or her time traveling to beneficiary households and performing only a perfunctory check before having to begin travel to the next site.41 Overall few resources are devoted to quality upgrading in facilitation or facilitators. JSLU and JSPACA facilitators earn approximately Rp 167,000 per month, which is below the average wage for domestic help or childcare (for example) in urban areas in Indonesia. PKSA facilitators employed by Kemensos can earn as much as Rp 1.4 million per month, which higher salaries are meant to attract facilitators with higher education levels.42 Facilitators must be mobile and cover large areas, but neither the transportation allowance nor overall salary varies by ground covered or with the number and distance to beneficiaries. Facilitator training is meant to happen as a vertical cascade, with senior facilitators delivering training material and any program updates to cadres in their regions, and with refresher courses available (but usually to newly-hired facilitators, although not necessarily available to even new hires at the time they are hired). Observers have noted that as a result of their many responsibilities, facilitators in the PKSA program are likely to become gatekeepers of cash transfers rather than providers of more comprehensive support and advocacy.43 Monitoring and Evaluation In most cases, monitoring and evaluation efforts do not focus on program performance or effectiveness while programs provide little ex ante investment in monitoring and data gathering activities. While M&E activities are explicitly covered in most Cluster 1 program guidelines, M&E in practice focuses on disbursement and financial performance indicators - see also Box 4 below for a general description of M&E activities across several Indonesian agencies, some of which provide Cluster 1 SA programs. For example, in the BSM program, monitoring reports are concerned with funds flow and whether or not beneficiary verification was done; results from beneficiary verification are not included in monitoring reports. While all programs mention socialization activities and schedules, monitoring and reporting typically records only whether these activities were held and who participated rather than what outcomes were. Cluster 1 programs - with the exception of PKH - do not use monitoring information to address programs on the ground. PKH, the conditional cash transfer, has a dedicated management information system that tracks beneficiaries, updates their demographic profiles, notes their progress attending health and education services, and adjusts payments or applies penalties based on this information. The PKH MIS is beginning to incorporate complaints and grievances reported by beneficiaries or their facilitators. With such information readily available, program implementers can identify frequently occurring problems or weaknesses in verification and deploy resources to remedy the problem. 40 In Volume 2 of this report, see "Social Assistance Program and Public Expenditure Review 6: PKH" and the references therein. 41 PKH Facilitators indicate that the quality and intensity of services they have time to provide is limited when large distances separate beneficiaries while all stakeholders note the quality and frequency of facilitators efforts are the best predictors of household success under the PKH program; in Volume 2 of this report, see "Social Assistance Program and Public Expenditure Review 6: PKH" and the references therein. 42 The minimum level of education for a PKSA facilitator employed by Kemensos is a college degree in social welfare science; JSLU and JSPACA facilitators are required to have graduated from senior high school. 43 In Volume 2 of this report, see "Social Assistance Program and Public Expenditure Review 7: JSLU, JSPACA, and PKSA" and the references therein. 74 Are programs implemented in the right way? U Currently, no other programs have this information management capacity which means they are unable to quickly address emerging issues. For example, Raskin's implementing agency, Bulog, does not monitor activities past the Raskin distribution points and so has no ability to address the redistribution of benefits that weakens the impact of the program (see above). Generally in Cluster 1 programs, whatever monitoring information there is does not enter an information gathering-evaluation-feedback loop that could help make programs more effective or troubleshoot incipient or longstanding problems on the ground. Furthermore, a lack of incentives undermines both quantity and quality in these M&E activities (Box 4). 75 E Protecting Poor and Vulnerable Households in Indonesia The historical Figure 32. Indonesia's Monitoring and Evaluation Landscape development of monitoring Level at which M&E takes place and evaluation activities has Whole Entire Sector Individual Loca I resulted in an Activity Government Ministries Government architecture Level that is diffuse, costly in terms PlnigBAPPENAS of staff time, and ineffective in producing inputs to social ueting service reform. utinBAPPENAS Performance BPEA Management o m pema gra BSMKemnagandPKH exibiedlowbudet xectio raes n hei ealy ear ofopeatin m ibendn st Accountability __ .P Complaints and grievances reporting also exist bu tsral bugt oeffective. As with socialization and M&E activities, every Cluster 1 SA program manual contains an explicit provision for a complaints and grievances reporting system. However, because of ls of socialization to the agency or group responsible for providing this support operation and a lack of follow-up or monitoring of outcomes, there is very little incentive to effectively implement the complaints and grievances system as designed. Beneficiaries are most often not aware of how or where to submit appeals and grievances. Those who do are usually disappointed. An observer of the BILT program put it succinctly: "The poor members of the community had given up and were resigned to the fact that no matter how long they struggled with ...BILT, they never succeeded. In their opinions, complaints or any form of protest ... had no impact whatsoever, because things were not decided by the village-level officials." As for the monitoring and evaluation reporting, the complaints and grievances that are reviewed are not consolidated and they do not become the basis of remedial action. Budget Execution Budget execution has improved for most programs, with the exception of Jamkesmas. While many programs (like BSM-Kemenag and PKH) exhibited low budget execution rates in their early years of operation, most disbursed close to 100 percent of their originally allocated budgets in 2009 (Figure 33 and 34). Jamkesmas is a notable exception: in its first two years, the program disbursed more than 100 percent of its original budget allocation due to a relatively relaxed "open membership" policy that encouraged utilization (access was granted even to those who could not display Jamkesmas cards). Budget execution subsequently deteriorated and only reached 71 percent in 2009, reflecting a number of factors, including: a tightening of controls over membership, underutilization of services, and growing confusion caused by the proliferation of competing local schemes and their association regulations. Jamkesmas 2010 disbursement improved over 2009, but this was due primarily to reduced planned budget allocations;- in other words, Jamkemas 2010 realized spending was relative to a noticeably lower planned level of expenditure. 76 Are programs implemented in the right way? U Budget Figure 33. Budget Execution Rates Figure 34. Budget Execution Rate in execution for in Social Assistance Programs, 2006- Jamkesmas, 2006-2010 most programs 2010 improved p e reely strn steadily until 125% - rerc t hPecent 2009, but 2010 15 saw difficulties w ooc fear re-emerge lo for some. 75% . 75 Jamkesmas's budget so% . 5o% execution numbers have 25% -25% been the mirror opposite: 0% o0% declining until 2006 2007 2008 2009 2010 2006 2007 2008 2009 2010 2009 and BSM (MoEC) - PKH improvmng BSM (MOA Vulnerable groups thereafter. Sources and Notes: Kemenkeu and World Bank staff calculation However, during a given year disbursements remain slow and skewed towards the second half, largely reflecting long beneficiary identification processes. Some programs (PKH Jamkesmas) exhibit a relatively smooth budget disbursement profile (Figure 35). PK is supported by a relatively strong MIS and uses advance disbursement of funds to PT Pos (followed by reconciliation). Most other programs - especially BSM and Raskin - exhibit very low disbursement rates prior to around June (Figure 36). Such delays are often the result of long beneficiary identification and verification processes which for many programs (especially BSM but also JSPACA and JSLU) run until around May or June of each year. As a result, implementing ministries rarely send payment authorization letters to Kemenkeu before May or June meaning disbursement of funds to intermediaries (including PT Pos) typically begins in June or July. A few Figure 35. Budget Disbursement Profiles: Figure 36. Budget Disbursement Profiles: programs PKH, Jamkesmas, and PKSA, 2009 BSMVI, Raskin, and JSLU, 2009 exhibit relatively 100 100 smooth budget . disbursement 7s 75 profiles, but a similar number so50* so show lumpy disbursement 255 that is 2 bunched towards the o ."" o . .. .. middle of the Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec calendar year. disburstsement,lineJaks sVunrbeld lyR ki Vulnerable children PKH BSM (Ke dkbd BSM (Kemenag) Sources and Notes: Kemenkeu and World Bank staff calculation. 77 * Protecting Poor and Vulnerable Households in Indonesia Additional public financial management issues include a lack of performance-based budgeting and bottom-up funds monitoring. Mirroring the issues in M&E discussed above, budget audit documents focus on budget execution rather than outcomes, indicating a lack of capacity to support performance-based budgeting. Leakage of funds is not yet a major issue in most programs - Raskin may be an exception - but benefit deductions and other fees are common during implementation and there are no efforts at rights and awareness campaigns that could encourage bottom-up funds monitoring. Local Implementation Arrangements Sustainability and minimum standards are threatened by local-level politics and implementation revisions. Local governments, agencies, service providers, and broader communities are asked to support various stages of SA implementation. Targeting, beneficiary verification, socialization, funds channeling, facilitation, monitoring and evaluation, and the complaints and appeals processes are all areas where local actors may be involved. Weak socialization and inconsistent follow-up, however, allows local actors to revise implementation procedures when they determine it is necessary. Minimum service standards are therefore difficult to enforce, and both implementation procedures and program outcomes vary widely from region to region. Also, there is a proliferation of local-level substitutes or alternatives to Jamkesmas (the national health-fee waiver program) and as a result Jamkesmas allocations go unused. Brazil's Bolsa Familia program, the principal social transfer covering 25 percent of the Brazilian population, has dealt with some of these issues and provides potential solutions (Table 16). Effective coordination between the central government and local-level implementers can be improved through regulatory rationalization. Clarity of roles is a prerequisite of good governance and effective service delivery when program tasks and subprocesses are partially delegated to local-level implementers. However in poverty reduction and SA provision at least, many of the current arrangements governing service provision are unclear on what providers are to deliver and how much they are to receive for doing so. The financing and provision of services is based on bureaucratic instructions which provide relatively little autonomy to providers or beneficiaries and may not be commensurate with the true cost of services provided. For example, a typical Puskesmas (local health center) has as many as eight sources of cash income and 34 operational budgets (of which Jamkesmas is one), many of which are provided in kind by central and local governments (World Bank, 2005). Not only does this mean that Puskesmas do not have the autonomy to fully optimize service delivery based on the populations served and their demands, but, as detailed in field research, these overlapping budgets, regulations, and spheres of control make it difficult for Jamkesmas health care providers to fully use all Jamkesmas monies. The continued lack of clarity in roles and responsibilities is a serious constraint to improved service delivery and is resulting in ineffective spending arrangements. Local-level * Require joint management agreements that establish minimum service standlardsbfr implementation of receiving subsidies to cover the program's administrative costs. national programs * Ensure that monitoring, evaluation, and audit activities are included at the local level. Local-level * Develop an index quantifying management capacity (based on a few easy to verify management, indicators) and use it to measure progress. administrative and * Provide performance-based financial incentives for administration based on management implementation index scores. capacities vary * Target training, capacity building, and other remedies to localities with low scores. Potential duplication * Provide for integration of sub-national programs with national programs via joint with sub-national cooperation agreements. initiatives Promoting sharing * Host and fund an innovations award that promotes the program as welI as regional of experiences and sharing. innovations across * Publish case studies of innovations in program management and share with all regions. regions 78 Are programs implemented in the right way? U Governance Quality and effectiveness of Cluster 1 initiatives would be enhanced, and households made better off, if governance issues could be systematically addressed. These issues include (1) agency capacity and the introduction and institutionalization of technologies and processes that can ensure and track minimum standards for service delivery as well as remedy shortcomings (when they occur) at any level or point in the delivery process; (2) the institutional arrangements that best support the delivery of all available SA services to all eligible households; and (3) divisions and overlaps between centrally-funded and locally-implemented SA services and between nationally-developed and locally- developed initiatives. Box 5 below describes in greater detail what governance is and why Indonesia is struggling with these issues. Governance will improve as the subprocesses mentioned above - M&E, budget execution, and local implementation arrangements - improve. For example, in Cluster 1 SA programs today, the rules, roles, and controls governing all implementation steps are incomplete and lack important details. Concerted efforts in M&E reform combined with consistent clarification of regulations and the division of roles between local- and higher-level actors will produce better governance in all these steps. Similarly, today responsibilities for SA are spread across many government agencies (and both local and national levels), with little coordination, minimum standards, or overall accountability across these ministries and agencies. The formulation of common M&E standards and practices, common socialization procedures and benchmarks, or common targeting procedures (for example) will begin to reduce the variability in the quality of services provided and increase pressure on agencies to institute governance systems that permit delivered services to achieve common benchmarks. International experience suggests that better governance and decentralization initiatives can not be treated separately.4 In Latin America, for example, decentralized authority for determining eligibility for SA transfers was gradually replaced by centrally-designed, rule-based procedures in order to reduce the influence of politicians and the political cycle on the allocation and delivery of benefits. The Eastern European countries that decentralized financing for SA programs found that this resulted in greater inequality in the distribution of benefits and outcomes, as poorer areas were less able to generate revenues and less frequently had the capacity to deliver a locally-appropriate benefit package. Finally, Indian experience shows that when SA budgets are included in general budgets (provided to regions for implementation of government functions), such allocations are treated fungibly and diverted to uses other than SA. 44 This paragraph is based on "Social Assistance Program and Public Expenditure Review 8: History and Evolution of Social Assistance in Indonesia", particularly Box 4 and the references therein, in Volume 2 of this report. 79 I ProtectIn Poor. an Vnral Hoshod in Indnei Box 5. For services provided by public entities, governance is best described as the set of incentives and Governance accountability relationships that govern the way in which service providers are held accountable in Indonesia's for their behaviors and their ability to deliver services with quality and efficiency. More simply, Social Assistance governance is: the "rules" that are developed to safeguard the provision of services, the "roles" Programs and responsibilities delegated to all people tasked with providing services, and the "controls" and accountability measures that describe the manner in which rules and roles will be enforced. There is growing evidence that these rules, roles, and controls can either foster or detract from the efficient and effective delivery of quality services. Effective governance arrangements that encourage thoughtful involvement at every stage can enhance outcomes under even the most basic of programs. Governance issues are not unique to SA programs and solutions can often be found in holistic public sector reform, for example through performance-based budgeting combined with civil service capacity enhancement. At the same time, social safety nets may present greater governance risk due to their inherent design features (which often incorporate difficult-to-verify eligibility), the sharing of implementation across government levels and agencies (making responsibility diffuse and weakening accountability), and the fact that beneficiaries do not participate in a market for SA services, are largely unable to "vote with their feet", and as a group have a very limited political voice with which they could provide "bottom up" accountability. Safeguards, Incentives, and Capacity This report (as well as each background chapter) has a section on implementation, where minimum service standards and the quality of GOI provision of SA services is discussed and rated. As is seen there, for all implementation steps - from targeting, verification, and socialization through to payment processing and benefit delivery and later to monitoring and evaluation - each program's implementation manual usually contains at least an outline of what the implementation step is and the necessary processes associated with that step. However, the "rules, roles, and controls" governing these steps are incomplete and often leave out important details concerning who, what, when, where, and how often. Furthermore, there are no sanctions for ineffective or discretionary implementation nor are there positive financial incentives or competitions rewarding effective implementation. In other words, even though most implementation processes are explicitly covered by a manual chapter, there are no positive or negative incentives - in other words, no controls - that would encourage actors with implementation roles to follow the rules laid out in the manual. This leads to confusion on the part of both service providers and beneficiaries; a proliferation of implementation practices and results; a lack of enforceable minimum service standards; and a lack of effective oversight and remediation of all failures in achieving minimum service standards. The report "Strengthening the governance dimension of Social Safety Nets in ASEAN" (Giannozzi & Khan, 2011) has more details on the ambiguous and non-transparent "rules of the game", low implementation capacity, and inadequate monitoring and evaluation in the Jamkesmas health service fee waiver program. The relatively recent development of the social safety net sector in Indonesia partially excuses these weak governance arrangements. New programs (and especially those set up during emergencies) are often run with only rudimentary capacity for processing eligibility and payments and minimal monitoring and auditing systems while mature programs have more sophisticated versions of all these implementation steps as well as outreach, recertification procedures, grievance and appeals, and more sophisticated incentives for performance. For Indonesia to achieve a higher level of effectiveness and service quality, administrative capacity, implementation processes, oversight, and strategies for remediation will all need to be addressed through concrete planning for governance arrangements. Are programs implemented in the right way? * 81 Does Indonesia have the right programs and system in place? SA programs and systems are "right" insofar as they allow implementers and beneficiaries to achieve the objectives programs were developed for. This section takes as given the policy objectives that SA stakeholders in the Indonesian government have elaborated and have committed to pursuing and evaluates the programs that are in place on the basis of whether those goals are being met. Previous sections have examined specific parts, pieces, and actors involved in the production of Cluster 1 SA transfers. Here a holistic view is adopted and discussion centers on whether agencies and households have been provided a tool which, despite possible flaws and as a result of its good features, keeps individuals and households from falling into poverty, pulls households and individuals out of poverty, or encourages households and individuals to acquire the goods and services that enhance productivity and enhance both present and future opportunities (Table 18). Positive Impacts in Current Programs Three of the Cluster 1 SA programs - PKH, BLT, and Jamkesmas - have been the subject of impact evaluation studies.45 The conditional (PKH) and unconditional (BLT) cash transfers have produced positive outcomes for poor households while Jamkesmas's outcomes are positive overall but not particularly positive for the poorer segments of the population (Table 17). The remaining Cluster 1 programs have not benefitted from impact evaluation, but related evidence indicates that impacts are likely to be small. 45 Impact evaluations attempt to answer the following question: are households that receive benefits changing their behaviors or experiencing benefits that they would not otherwise have? As an answer, impact evaluations typically measure a broad range of household behaviors in beneficiary households and compare them with the same behaviors and outcomes in similar households who did not receive benefits. In-depth discussion and technical details of each of the PKH, BLT, and Jamkesmas impact evaluations are included in the accompanying background papers covering each respective program, all of which are collected in Volume 2 of this report. 82 Known impacts from Table 17. Summary of Positive impacts in Social Assistance Programs SA programs are generally positive Consumption Health Education Child Employment for poor households Labor Jamkesmas n/a Yes, but larger n/a n/a n/a for non-poor PKH YeIlre Ye (lrge Fo enole | No n p households. 0 e. see83 83 S Protecting Poor and Vulnerable Households in Indonesia PKH was successful in improving the welfare of extremely poor households and increasing usage of health services. Although confined to a small set of demographically-eligible very poor households, the PKH pilot program has produced positive impacts. Monthly household consumption increased by 10 percent (over and above initial levels); the largest shares of this increase went to food, especially high-protein foods, and health care. PKH also produced a positive impact on helping extremely poor households to increase primary healthcare utilization like pre- and post-natal visits and child weighings. PKH's presence even produced more pre-natal visits and child weighings in non-beneficiary households living in PKH areas. PKH did not have an effect on drawing more children into school (enrollment rates), encouraging them to stay (dropout rates), or encouraging them to continue (transition rates) due to poor timing, relatively small benefits, and lack of outreach to school-leavers.46 That is, benefit delivery was not timed to coincide with school fee due dates; the education component of the PKH benefit package is lower than the true cost of education; and facilitators were not actively bringing drop-outs back into the system. BLT, which was unconditionally disbursed, allowed households to choose positively, but its effects were more modest. Households reported spending BLT benefits (Rp 300,000 every 3 months) immediately on basic necessities like food, clothes, and transport. Even so, impact evaluation reveals that income support during an acute consumption crisis did allow households to make more productive choices than otherwise. Beneficiary households had children who participated in school more frequently and in wage work less frequently. Heads of household found work more frequently with BLT, and households in general visited health facilities more often with BLT than without. 46 Primary and junior secondary school students already in school did spend more hours there as a result of PKH. 84 Does Indonesia have the right programs and system in place? * E 2 E æl 0 .- -CL > UJ g,c c Eo -o umu -) L n U) E .c 3 - E~c.c o J 00 > > fel -c _o -- r- -w .2 --~4d > ) U) ClV Cj) cfl 2- O 2C V) L) t 0~0 a-l or 03 0- ~ oa aæ +- 85 *** w w 0 0) 0 - LLi- o ( -J rJ - -- o. 0 e - o G 0-0 0 0' 00JT+- 85 E Protecting Poor and Vulnerable Households in Indonesia Jamkesmas has increased utilization of health services, but the effects are larger for non-poor households who received Jamkesmas cards contrary to program rules and for those with prior exposure to the healthcare system. For most types of the health service categories - private or public, primary care or secondary (hospital) care - households in the richest quintile saw their utilization rates increase at much higher rates than households in the poorer quintiles. The only exception was at Puskesmas, which richer households do not visit frequently. Increased utilization for non-poor households indicates the Jamkesmas benefit package is valuable, so implementers should focus on why poor households are not taking advantage. Beneficiaries mention access costs - transport, lost wages, childcare, companion or chaperone food and lodging - and a lack of knowledge about their rights and the benefit package offered as impediments to Jamkesmas use. In addition, there are local-level "Jamkesmas-like" services that households often prefer using. Jamkesmas is not yet effective at mitigating the significant risks to well-being from high health care costs that poor households face and will not be until the above issues are addressed. BSM and Raskin are not likely to have large effects because of design and implementation weaknesses. Actually-delivered Raskin benefits, at 2-3 percent of poverty line expenditure,47 are too small to affect household well- being. BSM benefits are also not large enough to meet the total costs of attending school. In addition, BSM is split into independent initiatives across sub-directorates in charge of various schooling levels, so the BSM program for elementary education is run independently from the BSM program for junior secondary education. Each BSM initiative requires verified attendance (and a letter from the school principal) but gives scholarships only to those in their school level. For this reason, students in sixth grade are no longer eligible for BSM-basic scholarships because once their attendance has been verified, they will no longer be sitting in basic education. Likewise, students entering seventh grade (the first year of junior primary) will have to wait until their attendance is verified and will not receive a scholarship until some time after they have already enrolled in and attended seventh grade. In other words, precisely when enrollment is at risk for poor households BSM program benefits are not available. This design flaw is then exacerbated by late payment issues and together they mean that students receive BSM scholarships over one year after they first enroll. Efficiency and Cost-effectiveness BLT and PKH balance effectiveness with efficiency and therefore offer good value for money. These programs spend 5 and 16 percent respectively (per Rupiah of benefits delivered to households) on administration to deliver proven outcomes (Figure 37). In contrast, BSM delivers cash benefits with minimal overhead but cannot be considered effective: it spends just one percent (per Rupiah delivered to students) on administration and is unlikely to have an impact on outcomes given its design and implementation issues. From national budgets, Raskin is allocated 35 percent (per Rupiah of rice delivered) for administration and operation48 but the program produces the smallest-valued transfer to intended beneficiaries and nearly all administrative expenditures are spent on logistics and management of physical stocks, rather than on the safeguarding and support operations that are crucial for effective service to beneficiaries. It is unlikely to significantly help protect households and is unlikely, in its current form, to be a cost-effective program. 47 Additional comparisons: average Raskin purchases represent 6 to 10 percent of a household's total rice needs; total Raskin purchases by all households are approximately one-fifth the size of total rice purchases by poor and near-poor households; total Raskin purchases by poor and near-poor households are approximately one-tenth the size of total rice purchases by poor and near-poor households; total Raskin purchases by all households are approximately one-twentieth the size of total rice purchases by all households (source: Susenas and World Bank staff calculations). 48 Total amounts spent on administration and operation may be higher - see Sections 2 and 5 above. 86 Does Indonesia have the right programs and system in place? U Ratios of non-benef it Figure 37. Social Assistance Non-benefit Costs to benefit spending (or pdt efficiency) vary widely v 40% - across SA programs, and while overallo E 30% benefit spending is pro-poor, about 40 percent of benefitswe 20% n go to households is nce the top 60 percent 18 of the consumption 10%- distribution. 0% BSM BLT JSPACA JSLU PKH Raskin (MoEC SID) (2008-09) Source and notes: Kemnenkeu and World Bank Staff calculations. Data for 2010 unless otherwise stated. Figure 38. Share of Total Social Assistance Benefits by consumption dlecile 25%- S20%- -15% 10%- 5%- 0% 1 2 3 4 5 6 7 8 9 10 Consumption deciles Big 4 BLT Raskin Jamkesmas BSM Source: Kemnenkeu, Susenas February 2010 and World Bank Staff calculations. Overall spending is 'pro poor' but a sizable share of spending goes to the top half of households, reflecting the issues with targeting and implementation (Figure 38). Benefit incidence analysis indicates that the majority of Cluster 1 SA spending benefits poor and vulnerable households. Around 60 percent of the total benefits from the four largest programs went to poor or vulnerable households (roughly equivalent to the bottom four dleciles) in 2010. However, the remaining 40 percent of benefits went to households in the top six deciles. BLT's higher coverage of the bottom 10 percent of households is notable, as is BSM's higher coverage of the top 30, 20 and 10 percent of households. 87 E Protecting Poor and Vulnerable Households in Indonesia Coordination Within and Across Programs Responsibility for the eight major Cluster 1 programs is spread across six key agencies and as yet no single agency can thoroughly coordinate SA efforts. Kemensos is responsible for the largest number of programs, but the three largest permanent programs (Raskin, Jamkesmas and BSM) are mainly the responsibility of sectoral agencies (Kemenkokesra with Bulog, Kemdikbud with Kemenag, and Kemenkes respectively). Outside of the major programs, the remaining central government SA expenditures are distributed across 12 Ministries, 22 programs and 87 activities. There is also fragmentation within some programs and agencies. For example, BSM is actually made up of 10 initiatives, spread across multiple activity clusters in both Kemdikbud and Kemenag, with little coordination or interconnection between them. Within Kemensos, there is little coordination or interconnection between PKH, JSLU, JSPACA, and PKSA programs, though they are similar in design and rely on similar procedures, delivery mechanisms, and skilled personnel. Instead, they are run independently out of four independent activity clusters within two Directorates General. Moreover, numerous other agencies are involved in implementation and support roles. Service providers such as schools, hospitals, health centers, and local governments play a significant role in the implementation of Jamkesmas and BSM. BPS plays a critical role in targeting and PT Pos in distribution of cash benefits. Local governments also play a role in socialization and M&E and in delivery of Raskin rice to households. In practice such fragmentation has created the parallel duplication of many common processes. It has also inadvertently prevented households from being inducted into the entire array of initiatives available and prevents implementing agencies from realizing economies of scale or scope in their operations. Boxes 3, 4, and 5 above all demonstrate lack of coordination in different areas and between different levels of government or agencies providing the same service. These issues are mirrored in budget formulation for the sector, which remains fragmented and unintegrated. Budgets are fragmented across and within agencies and overall budget formulation for the sector is not supported by existing budget classifications. In addition, budget audit documents focus on budget execution rather than outcomes and do not support performance-based budgeting. Finally, the system contains many gaps in coverage and risks. There is currently no program that anticipates risks from, and prevents negative coping behaviors during, household-idiosyncratic risks such as temporary unemployment. Indonesia also does not have an automatic safety net that kicks in to protect households in response to global, macro, regional or micro shocks. Though BLT turned out to be effective and provided a modest protective transfer, it has only been used on an ad hoc basis and has not been institutionalized for political reasons. Large numbers of those from especially vulnerable groups such as destitute elderly and disabled remain unprotected. Promotion on a large scale is also underprovided: PKH has been moderately successful in promoting human capital investment but is confined to a small subset of very poor households. BSM serves a larger proportion of the population with a valuable protection-and- promotion benefit, but is struggling to be effective. Early childhood interventions in education, nutrition, and vaccination are not yet national in coverage. 88 Does Indonesia have the right programs and system in place? U 89 Recommendations for achieving a social safety net in Indonesia Although Indonesia has made progress in launching SA programs, much work remains to be done to develop a true social safety net. The current range of SA programs does not go far enough to protect the 40 percent of the population with the highest risk of falling into poverty. The first step on the way to a dynamic and responsive social safety net should be reform within currently available programs. This will ensure that the emerging social safety net has thoughtfully-engineered and effective components. Re-engineering Implementation The recommendations for within-program reform are based on much more detailed information available in the program reports in Volume 2 of this report. As mentioned previously, all eight Cluster 1 SA programs discussed here are the subject of a program- and agency-level review (of implementation, impacts, and public expenditure); those reviews plus two additional chapters concerning the history and evolution of SA in Indonesia and a public expenditure summary are collected in Volume 2. The recommendations for program reform that follow immediately below summarize the findings in and conclusions from those reviews. Scale up PKH while revising benefit levels to continue delivering better health and education outcomes for poor households. Make PKH a national program by expanding coverage to all extremely poor households. Make sure that students from PKH households are also covered by the BSM program (see below); automatic coverage with additional education instruments will encourage enrollment and transition from basic to junior secondary and junior to senior secondary level. If automatic links to BSM are not feasible, PKH benefit levels for education should be increased to ensure they are appropriate for education costs and PKH should consider providing its own transition bonus for students entering junior secondary school. PKH has one of the only comprehensive MIS systems in Indonesia; operators and the PKH administration should continue to refine the processes by which MIS-generated information is incorporated into a continuous reform and improvement cycle. PKH administrators should also begin socializing their MIS system to other agencies providing or operating SA programs, including concrete examples of how the low-cost MIS system has led to program improvements. 90 Re-engineer BSM to remedy its current ineffectiveness and then expand availability to all poor and vulnerable households. BSM levels should be recalculated (and indexed to inflation) so they deliver benefits commensurate with the real education expenditures that poor households must make. Program administrators should consider the benefits of an actual tuition-and-fees scholarship with a transportation voucher instead of a cash transfer. In addition, in order to pull students across transition periods (currently the period during which most drop-outs occur), the BSM program should consider offering a merit-based award to all BSM recipients who do graduate and enroll in the next level of education, and publicizing the award during the final year of the previous grade. The administration of the BSM program in Kemdikbud must be re-designed so that the BSM can follow students across schooling levels (from basic to junior secondary, junior to senior secondary, and senior secondary to university). Kemdikbud and Kemenag should consider consolidating the 10 independent BSM initiatives across (and within) agencies and across school levels so that the program can follow a student along his/her educational career and establish a single coordination unit to implement the unified program. BSM should not have to rely on school officials to perform targeting and allocation functions; the upcoming PPLS1 1 survey of poor households and the unified registry of SA recipients can instead be utilized to determine initial allocations and beneficiaries for the BSM program. The BSM program should explicitly budget for increased facilitation and outreach, an application procedure, and other recruitment that will help motivate possible school-leavers. In general, the amount of money spent on safeguarding and operations is too low in BSM to achieve effective monitoring so more resources should be allocated to support operations while a continuous monitoring, evaluation, and reform plan is put in place. PKSA, JSPACA and JSLU have the potential to reach vulnerable groups with generous benefits and effective facilitation, but three main shortcomings need to be addressed immediately. Regarding (1) benefit design, the cash transfers portions of the JSLU, JSPACA, and PKSA programs are relatively generous by Indonesian standards, but the facilitated services may be equally valuable for beneficiaries in terms of overall health, mental and social well-being, and inclusion. However, a standard facilitation package has not been developed and there are no guarantees that facilitators will have the training or experience necessary to identify and provide remedies for the most relevant physical, mental, and social difficulties that beneficiaries face. JSLU and JSPACA should follow the PKSA lead in setting higher salaries for facilitators and beginning a recruitment process that will result in a facilitator corps with upgraded skills and capabilities while the facilitator corps should be given every opportunity to upgrade the services that they themselves provide, for example with cross-region forums for facilitators where experts and others can give advice and answer practical concerns. Regarding (2) intervention design and capacity constraints, the cash transfers as designed have no triage system to deliver 91 E Protecting Poor and Vulnerable Households in Indonesia the worst-off beneficiaries to agencies and care institutions that can provide specialized assistance when needed: any beneficiaries identified are given the same basic benefit package regardless of special circumstances. This is less effective for those who would benefit from a front-line emergency response and leads to increased burdens on Yanrehsos staff and local affiliates, who may not be equipped as front-line emergency providers. Finally, all of (3) socialization, targeting and prioritization, service delivery, monitoring and evaluation, and complaints and grievances (i.e., support operations) are not effectively and consistently provided. The large amounts of time and staff resources that JSLU and JSPACA in particular spend verifying eligible beneficiaries could be reduced if all programs rely on the upcoming registry of poor households (PPLS1 1) for identifying their potential beneficiary pools and for determining quotas to regions. The Yanrehsos cash transfers should economize on already scarce resources by pooling socialization, and monitoring and evaluation activities and by exploring the option of introducing its facilitator corps to the PKH MIS systems and monitoring and evaluation apparatus. As progress is made on safeguarding, implementation, process-engineering, and facilitation quality upgrading, PKSA, JSPACA, and JSLU should begin working on increasing meager coverage to reach all eligible poor Indonesians. Jamkesmas struggles to increase utilization among needy beneficiaries who are either unaware of the program or cannot afford the costs of access, thus making revisions to targeting, outreach, and facilitation essential. Benefit package revisions for program sustainability are also necessary. At least three revisions to the overall benefit package are necessary for effectiveness and sustainability. The first revision should be to the mix of free services and facilitation: many cardholders are unaware of the benefits of preventative care and also the extent of Jamkesmas benefits, so provide better socialization, facilitation and outreach in order to increase utilization (and provide a low-cost introduction to modern medical services) among the poor and near-poor households who need it most. The second revision should be to the mix of free services and benefits for other access costs: cardholders indicate that the costs of health facility access (lodging, transportation, lost wages, and others) are a significant impediment to using Jamkesmas, so a Jamkesmas cardholder should be provided with ways to reduce those costs. The PNPM community development program (including PNPM-Generasi) can spearhead initiatives to collectively subsidize necessary services like public transport and childcare. The third revision should be to the medical benefit package itself, which is currently more generous than most other schemes available in Indonesia and internationally. This generosity encourages Jamkesmas capture by the non-vulnerable and contributes to overcrowding (and potentially worse service quality) in public and private health facilities. At the same time, actuarial studies have shown that full utilization of Jamkesmas's currently unlimited services by all eligible beneficiaries would lead to unsustainable cost increases that would require significantly more resources than currently allocated. The most conservative estimates of the costs of delivering an unconstrained Jamkesmas benefit package put monthly "premiums" at around Rp 12,000 per beneficiary per month, or double the current allocation of Rp 6,250, which implies a doubling (at least) of current budget allocations for the program. Jamkesmas's potentially unaffordable benefit levels should be "reverse engineered" based on desired coverage of both access and utilization costs, actuarial projections based on real cost data and different supply constraint scenarios, and proposed rationalization with regional and local health insurance schemes. The Jamkesmas program has very little budget for safeguarding, monitoring and evaluation, and a continuous program reform cycle. Allocate more resources to support operations and safeguarding, including for better socialization, facilitation and outreach. Finally, because shocks from health events are one of the most frequent threats to household well-being and as Indonesia has struggled to keep pace with the rest of the region in maternal and child mortality and malnutrition, the revised and re-engineered Jamkesmas program should be extended to the bottom 40 percent of the Indonesian population through the PPLS1 1 registry of poor households. Revisions to the overall benefit package would help offset the cost of expansion. Raskin delivers very little and needs to be re-engineered or rationalized. Raskin has the highest non-benefit costs (as a share of benefit costs or total costs) and spends the most (of all the SA programs) to deliver a benefit package with a value not too much higher than the amount households spend to acquire it. If Raskin will continue using public funds to provide SA benefits, thorough reorganization is necessary. First, a business process study should indicate where, why, and how so much Raskin is rice is lost in between initial procurement by Bulog and the local-level distribution points and suggest safeguards to prevent future losses as well as appropriate technologies for real-time tracking of rice amounts. The same study should determine where and why Bulog spends so much money to achieve delivery of Raskin rice to distribution points and both technologies and processes to economize on those costs. Finally, the allocation of rice from the distribution points to households is not controlled, monitored, or even observed by any central government agency. This is precisely the point where Raskin rice gets shared more widely than its intended target, and part of the reason why eventual Raskin benefit packages are so diluted. If the Raskin program continues to provide income support and food security to poor households, rice purchases will have to be monitored and controlled more tightly. Once a real-time monitoring system is in place and functional, a system of financial incentives and penalties should be developed. Consider a three year period over which improvements are made, with specific pre-determined targets for progress, after which an independent evaluation is carried out. If sufficient improvement is not evident, the program should no longer use public money to provide a social safety net function. 92 Recommendations for achieving a social safety net in Indonesia U Past reforms have demonstrated the usefulness of a quickly-deployed but temporary emergency income support. Lessons from the BLT experience should be applied in developing the next emergency benefit package. BLT worked to protect incomes and safeguard good behaviors partly because it was deployed rapidly and valuable benefit packages arrived just in time. Cash benefits also proved useful as households were able to immediately apply benefits to whatever expenditures were necessary and normal. When the next crisis or policy reform package hits Indonesia, social safety net providers should have a temporary cash-for-service initiative ready to be deployed. As for BLT, to economize on costs, the registry of poor households being established in 2011 should be used to target the benefits, and the benefits should be delivered through PT Pos directly to beneficiaries. Covering More Risks and More Households Reforming individual programs is not enough; developing a true social safety net requires ensuring that all poor and vulnerable are consistently and reliably protected in the face of risks. The current range of SA programs provides partial and non-guaranteed protection to the poor and vulnerable from some, but not all, of the risks faced. In other words, Indonesia will not be able to create a social safety net via reforms to existing programs alone. However, Indonesia benefits from a strong macroeconomic and fiscal position and an administration committed to poverty reduction and social protection, which makes the development of a comprehensive social safety net feasible. The following recommendations outline steps in the development of such an initiative: a. Ensure that households are protected from health risks and encourage all households to adopt and acquire healthy preventative behaviors and services; b. Promote continuous education so children can find good jobs and earn their way out of poverty; c. Eliminate destitution; d. Guard against shocks that could push households into poverty; while Figures 39, 40, and 41 below summarize the recommendations for the development of an integrated, household- centered social safety net in Indonesia. Social safety nets should target all chronically poor households with greater assistance and be able to provide basic protection to the 40 percent of all households that are most at risk of becoming poor in any given year. The current range of SA programs do not go far enough in protecting income and promoting healthy behaviors in chronically poor households. PKH, for example, currently provides assistance to only a fraction of households that are poor year in and year out. Nor do current programs protect all households that are highly vulnerable to shocks. Some target the near-poor (1.2 times the poverty line), but about 70 percent of the newly poor (in 2009) came from households with per capita consumption below 1.5 times the poverty line, or from the bottom 40 percent of households. To cover all vulnerable households with some basic protection, the social safety net needs a broader reach. 93 9 Protecting Poor and Vulnerable Households in Indonesia E oE o 5cc - -2 .o -oE uo R m , JD o- - :e 5 -¶ _ -) 0 - E 0n m, - 0- c~ > 1) 'o > > E I r> -m o ~ ~ ~ ~ s -E cuc- 0 -20 l_k 2,c - o-0oo ~ so - e~~~ >, _-o- ..3 - o - o c - o = e _ o - o_-¤ --o- -J -0 0 x oI xD ly 94c œœC,-0 C=, s E c oc, 10 c)cco -0 ~ ~~~~ c3 c239m N -a c2 -3 m O Š- - -- -0 0 a) F6 y -- c- u ao -0 c -O - - c2 o5 o m g c 1 oz EE ~ ~ ~ ~ - c)oo c u o E s D o~ ~~~~ ~~ al 3 -c- -.7-6_, - eo --ue > -w o -1 E c a)o. c I -= o a) -0 C C a o - - a) _ . c , ' - = - c .0 - 0 - u O u > - - 0-c ~ ~ E w c a)- - -o oeo ¯ 0 E +-m> ro ao E) oc > cý 100mo e-D -aà · ..- - w o --o E . o3 ýA 85 2 occ-) -a o - 2~~~~ u u r x 2 5, > 2 ,roc. , n- a- o m-- - n_ 0 a-o a- -5 -G - - . 0 Ln ND-o 2 -0 rO > >1 0 232 = .r r i o c -2 5 - 2, J9 oro cp 0 c.2 aa EIO cc Uo -2= cn LL--- o - r 6. -o ->.9 E : c) Er 0 0 0cr 95 E Protecting Poor and Vulnerable Households in Indonesia o -.eo > o omu Ec .-o E Q) E - c-a E, o - ~ - e~~~ ~~ 21 3 - o~ occ oE~~w 0ra o 8 R s E E g o 2 o COc Soo s o- 2 N ×> -¤ c o m o . E· - E o o - - -.0 -c: - - -o E gt o e c - - - o. .l o ' - E CL u ou o Q) G- C)E y O - c O - E .ex - u s> o o E ®-= ,0 2. ¯ E) 3 3 E E 4-c a o E o oo o - o. n: o o a œ ¯E ' o ¯-E oE o 'd 2 z o, o a .-Joa e -- D- ¿-I c0 . -3 8 oa E - >m - --wc oc u c o o - 0 r o E o u r s 960 Recommendations for achieving a social safety net in Indonesia U Protecting Households from Health Risks A core component of a future social safety net for Indonesia is protecting households from risks to their heath. Illness, work accidents, and long-term debilitating health setbacks are inherently unpredictable. Treatment can be costly and difficult to plan for while those whose work is interrupted pay twice: once for medical care and again in foregone income. All poor and vulnerable households need permanent and easy-to-use programs that provide low- or no-cost access to health care providers. Households with more specialized needs and costs require extra support. Expand the coverage of Jamkesmas to all vulnerable households while formulating and delivering a fiscally sustainable benefit package. Jamkesmas does not yet provide adequate population coverage. The newly available PPLS1 1 census of poor and vulnerable households and the upcoming unified registry of poor and vulnerable beneficiaries can be used to facilitate the issuing of cards to all households living below 1.5 times the poverty line. This report notes, however, that there is a trade-off between adequate coverage and the depth of benefits; Jamkesmas currently provides nearly unlimited "on paper" benefits, which in practice are limited by supply-side constraints and underutilization. To ensure the fiscal sustainability of the program, redefine the benefit package to include available preventative health care and an affordable set of facilitated basic services. Jamkesmas' political sustainability is uniquely tied to the still-to- be-developed universal health coverage plan (stipulated in the 2004 SJSN law); program administrators should develop transition and coordination plans for ensuring that all Jamkesmas beneficiaries have smooth and continuous coverage during the development of universal coverage. Until the universal coverage era arrives, Jamkesmas could be offered to non-poor and non-vulnerable groups for a contributory premium. This would enlarge the health-related risk pool; could provide valuable lessons regarding utilization and service mix, appropriate premiums and costs; could begin covering more workers from the informal sector; and will provide an indirect incentive for market provision of high-value health insurance products. Other programs (e.g., regional health schemes) and market-based products may add health service packages not covered by Jamkesmas. Provide PKH to all chronically poor households with greater burdens and less experience with healthcare providers. Lighten the health-related conditionalities in areas where health services are limited. Individuals in PKH households should automatically be granted Jamkesmas cards (i.e., all PKH households should automatically be included in the Jamkesmas beneficiary list), but they still need more. Persistently poor households require additional assistance as they are generally larger, less experienced with, and less aware of the benefits of modern healthcare, especially modern preventative care, and therefore less likely to access healthcare providers. The financial incentive and the facilitation in PKH can provide the extra "push" into the modern health care system that Jamkesmas does not provide. In areas where heath facility availability or quality is limited, PKH beneficiary households should not be expected to comply with the full set of health behavior conditions. In these situations, a modified PKH program can be introduced, using a set of lighter conditions that are achievable in supplying poor areas, while at the same time, introducing households to the concept of conditions that encourage investments in healthy behaviors. in areas with insufficient supply of medical services, PNPM programs (including PNPM-Generasi) can direct community-owned resources to address some of the deficiencies. in areas with sufficient supply, PNPM programs can empower communities to monitor the provision of services and improve quality through bottom-up accountability. Expand coverage of programs that cover the especially vulnerable elderly and those living with serious disabilities. The elderly who are bedridden, living alone with no family assistance and those who are otherwise socially excluded also require additional assistance beyond Jamkesmas. Not only are they less likely to access healthcare service providers, or be "pushed" there by family, kin, community or other social networks, but medical issues in the elderly population are on average more complicated and more costly to treat. Similarly, all severely disabled face both increased medical costs and increased costs of access to all social services including healthcare. In Indonesia, there is not yet a universal old-age pension or affordable old-age insurance, so most elderly face elevated healthcare costs at the same time that their regular income is diminished. The disabled face elevated healthcare costs both because of specialized needs and a lack of curative remedies. The JSLU cash transfer should be expanded to cover all elderly who fit the programs' criteria for extreme vulnerability. The JSPACA cash transfer should be expanded to cover all poor and vulnerable households with a disabled individual while disabled individuals without social standing, a reliable family or extended kin network, or living in communities without low-cost or free assisted living arrangements should be given priority access to JSPACA. For JSPACA and JSLU, progress on safeguarding, implementation, process-engineering, and facilitation quality upgrading, will need to be made first before programs are ready to expand thoughtfully. 97 E Protecting Poor and Vulnerable Households in Indonesia Promoting Continuous Education to Break the Inheritance of Poverty Poor and vulnerable households should have low-cost access to permanent, easy-to-use, high-quality programs at all levels of public education. In addition to healthy behaviors and outcomes from birth to death, education is a key to helping families sever the links between poverty in this generation and the next. With higher levels of education, youth are more likely to find good jobs and benefit from high wage premiums, and will be better placed to apply all acquired human capital to earning their way out of poverty and vulnerability. A social safety net should guarantee that children and youth from disadvantaged families can continue in school for as long as possible for as low a cost as possible. Interrupting education at any point in a child's life can open up gaps that persist for a lifetime. BSM, once consolidated and re-engineered, can provide much needed assistance to students who are most at risk of dropping out. The program should help all poor and vulnerable families cover all education-related expenditures as they occur as well as provide financial incentives to bridge transition periods (from basic to junior secondary, and from junior to senior secondary) when most students from poor households drop out. Benefit levels need to be adjusted upwards for higher levels of schooling, which are more expensive and usually require higher transportation costs, while benefit delivery needs to coincide with higher charges in the very first quarter of the school year. For example, scholarships meant to cover the first year of junior secondary education should be awarded in the last year of primary school and a larger portion of the total benefit should be delivered just before the student makes first-quarter payments to her new junior secondary school. Potentially, benefits might be set at lower levels for households with higher income levels who could contribute more out-of-pocket costs. University scholarships can also be provided through the program and advertised early in a senior-secondary student's career, but scholarships should only be awarded upon acceptance into a state university. In the future, additional instruments may be introduced for senior secondary or university education, such as low-interest education loans or work-study programs. In order to make BSM coverage more pro-poor (as well as to increase coverage), two reforms to allocation could pay immediate benefits. First, make sure all school-age children from PKH households are awarded a BSM scholarship. These students from chronically-poor households will then benefit from a positive cash incentive for attendance (and graduation) while the high costs of access are reduced. Secondly, the newly available PPLS1 1 census of poor and vulnerable households and the upcoming unified registry of poor and vulnerable beneficiaries can be used to facilitate the issuing of BSM cards to all households living below 1.5 times the poverty line. In areas where schools and remedial education activities are limited, PKH may need to be expanded with lighter and easier-to-meet conditions that are locally relevant, while facilitators and program officials everywhere should reorient recruitment towards real or potential school leavers. Persistently poor households are generally larger as well as less experienced with, and less aware of the benefits of, education. The financial incentive and facilitation in PKH can provide the extra "push" into education and will be more effective when costs of access are reliably addressed by the BSM scholarship initiative (see above). As for BSM, PKH should increase benefit levels to match the non-tuition fees that parents face, and the payments need to be timed during the academic calendar year so that parents have additional funds to cover the annual placement fee. In addition, consider a transition bonus as an incentive to move on to the next level of schooling. In areas where schools are still limited, however, PKH beneficiary households cannot be expected to comply with the full set of education behavior conditions. In these situations, a modified PKH program can be introduced, using a set of lighter conditions that are achievable in "supply poor" areas while, at the same time, introducing households to the concept of conditions that encourage investments in education. Both PNPM (including PNPM-Generasi) and BOS should continue to focus on easing supply-side constraints in education, especially in kindergarten and junior secondary schooling. At the same time, expand coverage of PKSA, which reaches out to youth who are at greater risk. At-risk youth face greater costs of access to education as a result of less familiarity with the education system, a more acute awareness of the opportunity costs of education, and a diminished support network (parents, elders, siblings). At the same time, programs like PKH and BSM that will be ready to receive vulnerable youth once they have re-gained the social and behavioral capacity necessary for regular instruction provided through locally-available public schools. The PKSA facilitated cash transfer should be revised so that it addresses both the monetary cost and the social or behavioral cost of access. Where professional human resources are available, the conditionalities in PKSA should be enforced while remedial and graduation plans should be developed. As progress is made on safeguarding, implementation, process-engineering, and especially facilitation quality upgrading, PKSA can begin refining its targeting and recruitment model and then a reasonable scale-up plan. There is growing evidence that a child's very earliest years are critical to lifetime mental and social development. Pilot a program to provide effective and affordable early childhood development and parenting education services for all poor and vulnerable families. Early Childhood Development (ECD) programs including parenting education reach children and their first teachers during the "golden years" of human development and prepare them for long-run success. However, few poor households in Indonesia have access to such comprehensive ECD 98 Recommendations for achieving a social safety net in Indonesia U services. Most government- and privately-funded programs are available only in urban areas and tend to serve well-off households. In the 0 to 5 age group, approximately half of the richest households participate in ECD programs, compared to only 20 percent of children from the poorest households. A National Early Childhood Development Program should be piloted with an eye towards developing a feasible, financially sustainable, and appropriate initiative for mothers, fathers, and children from poor households or communities. Community-based providers can be responsible for delivering ECD services to poor households and families; training, facilitation, and socialization support could be provided by a national agency or group. Standards and initiatives for poor communities should be based on and conform to current Indonesia-specific ECD standards. Initiatives will also need to identify affordable and effective community-based delivery mechanisms with accompanying financing models. Priority regions and communities could be identified by geographic poverty mapping. Protecting Incomes and Preventing Negative Coping Social safety nets should ensure a minimum level of income and provide a cushion during difficult periods so vulnerable households will not be forced to make difficult choices. Persistently poor households have difficulty generating sufficient income to lift themselves out of poverty. Vulnerable households are likely to turn to negative coping mechanisms - sending more members to work and pulling more members out of school, switching consumption to less nutritious but cheaper foods, and foregoing health care - precisely when their incomes are threatened. Indonesia needs income support initiatives that reliably address both difficulties. The cash transfers to severely disadvantaged households - PKH, JSPACA, PKSA, and JSLU - should be expanded to national coverage. In all of these, the value of the cash transfer is significant and households spend virtually the entire transfer on regular necessities.49 PKH should be expanded to cover all persistently poor households while in areas that are not supply ready, only light conditionalities should be imposed; households in such areas can graduate to regular PKH when supply is improved. The JSPACA, PKSA, and JSLU groups will face heightened difficulty generating reliable and livable incomes and the programs addressing such difficulties should all be expanded to reach all poor households or poor and vulnerable households with beneficiaries who meet the additional program-specific criteria. Pilot a national workfare program so that all vulnerable households can rely on a guaranteed number of working days when difficult times occur. The majority of vulnerable households do not face income risk every month, but are likely to enter poverty if they are not appropriately protected. When unemployment, illness, bad harvest, or other idiosyncratic shocks interrupt regular earnings or productive activities, expenditures are also often disrupted. With a workfare program that vulnerable households can opt into when stipulated wages become attractive, the ever-present risk to income generation is partly addressed. A workfare program is also a good time and place for contact by a facilitator who could enroll eligible households in Jamkesmas and BSM (if applicable). Well-designed workfare programs set wages below the prevailing market wages so only households with no better outside opportunities apply. A coordinated and authorized list of projects and sites where labor is needed must be available at all levels of government. Individuals should be alerted very early on that they have a maximum number of days of guaranteed wage labor per year (45 days, for example) while households have at most a higher guaranteed total (90 days, for example), so that workfare serves as a bridge through periods when other income-generation activities have failed (and not a permanent source of wages). Raskin could provide additional in-kind permanent income support to poor households only, but this would require a major reform to operating procedures and operating costs. Reliable delivery of a suitable number of kilograms of rice at reduced prices would be a valuable benefit for all poor households. However, as demonstrated earlier, Raskin currently delivers a thin benefit widely. If the Raskin program wants to begin providing an effective SA package to households that are truly in need, it will have to embark on a program of internal reform and re-engineering with an end goal of stopping leakage and re-targeting that are features of the current program. This will have to include control and monitoring of the distribution of Raskin rice past the Bulog-maintained distribution points. Even with process re- engineering and control and monitoring of actual Raskin rice purchases, Bulog will still be implementing a program with very high operational costs and should be encouraged to bring those costs in line with other programs delivering similarly- valued benefit packages. In the short-term, Bulog should present a reform plan with clear performance goals (agreed to by all stakeholders). At the end of an agreed period of time, if Raskin has not made any progress toward these goals, then it should no longer use so many public resources to provide SA. A quickly-deployable and automatic emergency income support facility, possibly tied to other social service utilization, will be useful to make future policy reforms palatable. Past subsidy reforms were packaged with temporary emergency income support (BLT) that produced positive results for beneficiary households and communities 49 See in particular "Program Keluarga Harapan: Impact Evaluation Report of Indonesia's Household Conditional Cash Transfer Program" (World Bank, 2011) and the other reports and references therein. 99 E Protecting Poor and Vulnerable Households in Indonesia alike by providing a "just in time" cash transfer that was delivered in tranches over a year-long period. The cash was used immediately on regular necessities, but BLT also safeguarded good behaviors in health, education, child labor, and job seeking by preventing households from having to make tough budget choices. However, concerns exist about the social costs of providing cash unconditionally. By adding light and easily monitored conditionalities (community service, labor for neighborhood improvement, spending time at local libraries or assisted living facilities, or simply registering for an eKTP identity card, for example) to the next emergency cash transfer, the same results can be achieved while households receive cash-for-service that helps them smooth their own consumption and human capital investments during rough times. Finally, while the existing social safety net programs focus on long-term poverty and vulnerability, they need to be coordinated with the development of a crisis monitoring and response system for short-term shocks. National and regional economic crises, price shocks, widespread layoffs, natural disasters, poor harvests or weather, and food insecurity can occur rapidly. A permanent crisis monitoring and response system needs to be established which will alert government to how shocks are being transmitted to households, how they are responding, and what the impact on socio-economic outcomes is. In turn, this can guide policy makers in determining which responses are most appropriate, where to deploy them, and when. Bappenas should reinvigorate its collaboration with BPS in order to ensure timely processing and release of high-quality and highly-relevant data that is amenable for near-real-time monitoring of household conditions. Monitoring is only half the story: a successful vulnerability mitigation tool will be one that can respond precisely when a crisis forces vulnerable households into negative coping strategies. Some of the response might include temporary scaling up of social safety net programs, but the GOI should develop protocols and cement the legal basis for the automatic and rapid disbursement of a pre-identified SA package (and associated targeting procedures) before the next crisis or downturn hits. The response should be automatic and not open to political manipulation or bargaining, and should be a coordinated response involving all relevant agencies and ministries and not be subject to further deliberation or politicking when crisis occurs. An emergency response should focus on providing income and basic necessities to all households at risk of curtailing human capital investments in health, nutrition, childcare and education. The workfare program proposed previously could absorb workers and laborers during both idiosyncratic or generalized shocks. To function effectively during times of rapidly developing crisis, the workfare program will have to maintain a national list of projects to which labor, materials, capital and supervision and oversight can be rapidly shifted so that the excess labor can be quickly absorbed and wages paid. The workfare restrictions discussed earlier should be lifted during times of regional or national crisis (and can be re-instated when the crisis has ended). The proposed cash-for-service program can also be used (for a pre-announced duration) during times of crisis to provide households without surplus labor with reliable income support. The last element of the proposed emergency bridge to normal times is an in-kind response through Raskin. Instead of simply scaling up Raskin amounts everywhere, priority allocations should go first to areas identified by a combination of a food insecurity index and the registry of poor households (PPLS1 1). The emergency Raskin response will be coordinated (through Bulog) with the Food Access Program which seeks to stabilize prices and ensure rice availability during times of elevated food prices. The social safety net proposed above and in Figures 39, 40, and 41 implies an increase in resources devoted to SA programs. Current spending on these programs represents approximately 0.5 percent of GDP (2010) and about 3.3 percent of national government revenues (2010). The proposed social safety net would represent a doubling of current SA spending to just less than 1 percent of GDP, and is less than half of all current spending on fuel and electricity subsidies. See Figure 42, Figure 43, and Table 19, which summarize the costs and resources used, program by program, in an SSN system that originates with currently available programs while expanding them to cover more households and adding new initiatives and better safeguarding processes. 100 Recommendations for achieving a social safety net in Indonesia U The Current Figure 42. Proposed Social Assistance Figure 43. Proposed Social Assistance SA spending Expenditure Composipositedaion without BLT mix is weighted Other programs toward Raskin 4% Raskin; the 7% proposed BLT Raskin social 26% 9% safety net Jamnkesmas Public works distributes a 1%PIKSA 3% 14 larger ttotalSPAost2 Other programs JJamkesmas SA envelope 3% JSLU 3% 29% more Public works 11% BSIM PKH equitably 20% g% among PKH currently PKSA 7% BSM existing 2% JSPACA JSILU 27% and new 1% 2% programs. Source: Kemenkeu, Bappenas IMTDIP 2010-14 M&E Indicators and World Bank staff projections Table 19. Proposed Social Safety Net: Key Assumptions PKH JSAA l Raskin Jamnkesmas BSMV JSLU 1101 E Protecting Poor and Vulnerable Households in Indonesia Transitioning to an Integrated Safety Net The current non-unified collection of initiatives may not be the most effective way to protect households. Current initiatives are providing an array of services, but separate administrative structures and implementation procedures are creating inefficiencies, gaps, duplications in procedures and coverage, and some spending that is not effective for assisting households. To prevent some vulnerable households from falling through the cracks and to economize on implementation costs, current fragmentation and duplication must be eliminated. A single agency should be in charge of developing plans for implementation, monitoring, evaluation, and reform of all SA initiatives. The same agency should have the power to delegate implementation tasks, either to already existing government agencies or external contractors. An overarching, authoritative body with responsibility for SA planning, and judged by outcomes and results, is likely the only way around the current institutional inflexibility, lethargy, and lack of coordination that seriously limits current program effectiveness. The quickest way to jumpstart SA integration may be through the National Targeting System that is already in development. The National Targeting System will construct a unified targeting registry of potential beneficiaries with improved targeting methods. With this single source of quality-controlled data, programs can improve targeting outcomes. Moreover, programs with the same target population will have consistent beneficiary lists, leading to more complete coverage and more effective realization of program complementarities. For example, PKH beneficiaries should also automatically be Jamkesmas beneficiaries, allowing them lower-cost access to conditioned health services. One of the main challenges in establishing a targeting system is to ensure that the system is dynamic enough to respond to the highly fluid nature of poverty in Indonesia. To this end, introduce dynamic mechanisms that keep beneficiary lists current, such as: community and self-targeting methods that can be used to update beneficiary lists as household circumstances change over time, or to determine beneficiaries altogether. In addition to targeting, the rest of SA support operations should be brought under a "minimum service standards" framework through which each program is monitored, evaluated, and reformed. In order to harmonize both the quality and effectiveness of all social safety net initiatives, a single agency or body should develop minimum service standards and indicators that reliably track the performance of each program and each and every program administrator at each and every implementation step. The implementation steps that will need to be brought under this common framework are: socialization and outreach procedures; monitoring and a common Management Information System; evaluation activities (these may benefit the most from participation by external, independent agencies); complaint, grievance, and appeals procedures; and finally promotion and public relations for the SA initiatives. Together with common minimum service standards, tracking indicators, and common M&E activities, the social safety net agency should also develop common remediation strategies and a plan for agency interventions when implementation difficulties do arise. Another quickly achievable integration step is through rationalization of the social safety net budget development and budget reporting processes. Improve public financial management for the sector by reconfiguring official budget classifications and promoting performance-based budgeting. The SA and social protection budget classifications could be adjusted to bring them into line with the international standards established in the IMFs Government Finance Statistics. For the social assistance budget classification, this would entail excluding social expenditures that are not transferred directly to households (e.g. spending on programs such as BOS and PNPM) and including those that are (e.g. spending on programs such as BLT and Raskin). For the social protection budget classification, this would entail including all of the Cluster 1 household SA programs that are currently mapped to sectoral or other functions (such as BLT, Raskin, Jamkesmas and BSM) as well as their related administrative and salary expenses. Moreover, in the transition to performance-based budgeting, all programs could be required to report annually their performance against outcomes and minimum service standards for support operations in addition to the regularly- reported budget execution performance. Seamlessly protecting poor and vulnerable households from diverse risks over their lifetimes may ultimately require the consolidation of the current programs and agencies into a "single window". In Indonesia, the collection of SA initiatives is not aligned along a household's life cycle, meaning missed opportunities to protect and promote productive behaviors as new risks arise. In order to reduce these missed opportunities, some middle-income countries have established a single coordinating hub, single agency, or even a single program, targeting many vulnerable groups and risks. Brazil's "Bolsa Familia" program and its reform experience offers an interesting example of how one large, decentralized middle-income country transitioned to a consolidated program (Box 6). 102 References U The single window approach ensures that all eligible beneficiaries are recruited, covered, and aware of all available programs that could protect them against a range of potential risks. Additionally, households will be able to rely on a common and reliable source for socialization of rights, responsibilities, and "know how" for programs and a consistent set of administrators or facilitators who can provide general advice and strategy for vulnerable households hoping to acquire health, education, or income protection services at low cost. For government, one of the benefits of a single window approach is that agencies focus on service delivery while program recruitment, socialization, and general knowledge and general beneficiary management and care (for all initiatives) are delegated to specialized employees with relevant skills. There are also political advantages from an easily recognizable and well-run safety net "brand". BP o B s la Governm tort hediferenas transfer fra strg Socalproiddbsb-ntinalgoenmetinsCT Seaatetadinisntativ reuctue aon procEduresio Assisancecreaed infficencis, rsulteinonsierab ert aps and dulitons cind coveae andmse Architecture:~~ter imotnpyege rmjonl rmtn dctonet adhlth.g heecide fared uchlae d Brazilen In at 203 Bazl auchd hBolsa Famdlia Program ina (BFP) bymrigtee4or am Minisr an effot t imroveficincyadohence of thdefisocals safet ineand tonscaler up atnet GDP.Incoutryevenlarelth Indneia coragety inreaed to il 25 ufph toalpuato BFP hs beter-han-verag taretinaccracy an had demotrated maxisignifcnmato povety nd neqaliy Ths i acievdtrou herapi mehnim annhrecidrmens-etn 94% t the oores two uils (cmare To aboupste prn Se for ther botMintile inIdonsiapemannt artaSNogrm) Throe prora acUnteD fopr a Miniiatsre (20to 5 prcet) fBazl'sreent(and impresseductions ofinoneqalt anFod percen ofefcare thee recent falietreeiovrt The cenra government fo esabise a new 5,564foo muniipaltiessTh 103 E Protecting Poor and Vulnerable Households in Indonesia References Block, S., Keiss, L., Webb, P., Kosen, S., Moench-Pfanner, R., Bloem, M., and Timmer, P. 2003. "Macro Shocks and Micro(scopic) Outcomes: Child Nutrition During Indonesia's Crisis." Nutrition Working Paper no. 1. Journal of Economics and Human Biology. Cameron, L. 2002. "The Impact of the Indonesian Financial Crisis on Children: Data from 100 Villages Survey." World Bank Policy Research Working Paper no. 2799. World Bank. Center for Health Research at the University of Indonesia (CHR-Ul). 2010. "Health Insurance in Indonesia: A Stakeholder Accountability Review of the Jamkesmas Program." Jakarta, Indonesia: University of Indonesia. Deinenger, K & Mgupa, P. 2005. "Does Greater Accountability Improve the Quality of Public Service Delivery? Evidence from Uganda." World Development 33 (1). Frakenberg, E., Thomas, D., & Beegle, K. 1999. "The Real Cost of Indonesia's Economic Crisis: Preliminary Findings from the Indonesia Family Life Surveys." Labor and Population Program Working Paper Series 99-04. RAND. Gani, A., Thabrany, H., Pujianto, F., Yanuar, T., Tachman, A., Siregar, H., Wahyu, S., Soerachmad, Widyastuti, Yulherin, and Nurbaiti. 2009. "Good Practices of Local Health Financing Schemes in Indonesia: Their Contribution Toward Universal Coverage of Health Insurance." Jakarta, Indonesia: AusAID. Gianozzi, Sara, and Asmeen Khan. 2011. "Strengthening the Governance Dimension of Social Safety Nets in ASEAN." Washington, DC: World Bank. Giles, J. & Satriawan, E. 2010. "Protecting Child Nutritional Status in the Aftermath of a Financial Crisis: Evidence from Indonesia." World Bank Policy Research Working Paper no. 5741. World Bank. Grosh, M., del Ninno, C., Tesliuc, E., & Ouerghi, A. 2008. For Protection and Promotion: The Design and Implementation of Effective Safety Nets. Washington, DC: World Bank. Holzman, R., and Hinz, R. 2005. Old-Age Income Support in the 21st Century: An international Perspective on Pension Systems and Reform. Washington, DC: World Bank Indonesia Corruption Watch (ICW). 2008. "Studi Kebijakan Jaminan Kesehatan Masyarakat (Jamkesmas)." Jakarta, Indonesia: Indonesia Corruption Watch. Leibbrandt, M., Woolard, I., Finn, A., and Argent, J. 2010. "Trends in South African Income Distribution and Poverty since the Fall of Apartheid." OECD Social, Employment and Migration Working Papers, No. 101. South Africa: University of Cape Town. Levinsohn, J., Berry, S., Friedman, J. 1999. "Impacts of Indonesian Economic Crisis: Household Evidence." National Bureau of Economic Research. Lindert, K., Linder, A., Hobbs, J., and de la Bri6re, B. 2007. "The Nuts and Bolts of Brazil's Bolsa Familia Program: Implementing Conditional Cash Transfer in a Decentralized Context." Social Protection Discussion Paper. Brazil: World Bank. SMERU. 2006. "A Rapid Appraisal of the Implementation of the 2005 Direct Cash Transfer Program in Indonesia: A Case Study in Five Kabupaten/Kota." Research Report. Jakarta, Indonesia: The SMERU Research Institute. SMERU. 2008a. "The Effectiveness of the Raskin Program." Research Report. Jakarta, Indonesia: The SMERU Research Institute. SMERU. 2008b. "Qualitative Baseline Study for PNPM-Generasi and PKH: The Availability and Use of the Maternal and Child Health Services and Basic Education Services in the Provinces of West Java and East Nusa Tenggara." Research Report. Jakarta, Indonesia: The SMERU Research Institute. 104 References U SMERU. 2009. "A Rapid Appraisal of the Implementation of the 2008 Direct Cash Transfer Program and Beneficiary Assessment of the 2005 Direct Cash Transfer Program in Indonesia." Research Report. Jakarta, Indonesia: The SMERU Research Institute. SMERU. 2010. "Social Health insurance for the Poor: Targeting and Impact of Indonesia's Askeskin Program." Working Paper. Jakarta, Indonesia: The SMERU Research Institute. SMERU. 2011. "Study of the System and Implementation of M&E of Development Programs: Consolidated Report," for the World Bank Project: Quality Assurance and Capacity Building for Monitoring and Evaluation. Jakarta, Indonesia: The SMERU Research Institute. Soewondo, P 2010. "Universal Health Coverage: Cost Estimation." TNP2K Presentation files. Jakarta, Indonesia: Government of Indonesia. Son, H. And Sparrow, R. 2010. "Impact Evaluation of Social Health Insurance for the Poor in Indonesia: A Panel Data Analysis." Unpublished Manuscript. Sparrow, R., Suryahadi, A. and Widyanti, W. 2008. "Public Health Insurance for the Poor: Targeting and Impact of Indonesia's Askeskin Programme." SMERU Working paper. Jakarta, Indonesia: The SMERU Research Institute. Sumarto, Sudarno, and Asep Suryahadi. 2010. "Conclusion: Coping with the Crisis", in Poverty and Social Protection in Indonesia. Singapore: Institute of Southeast Asian Studies. Thomas, D., Beegle, K., Frakenberg, E., Sikoki, B., Strauss, J., Teruel, G. 2001. "Education in a Crisis." California Center for Population Research: University of California. Weigand, C, and Grosh, M. 2008. "Levels and Patterns of Safety Net Spending in Developing and Transition Countries SP Discussion Paper. Washington, DC: World Bank. Woolard, 1. and Leibbrandt, M. 2010. "The Evolution and Impact of Unconditional Cash Transfers in South Africa." Southern Africa Labour and Development Research Unit: University of Cape Town. World Bank. 2005. "Information and Communication Technologies for Rural Development: Issues and Options." Washington, DC: World Bank World Bank. 2008. "PNPM-Rural Baseline Report." Jakarta, Indonesia: World Bank. World Bank. 2010a. "The BOS Review." Jakarta, Indonesia: World Bank. World Bank. 2010b. "Indonesia's PNPM Generasi Program: Interim Impact Evaluation Report." Jakarta, Indonesia: World Bank. World Bank. 2011. "2008 again?" in Indonesia Economic Quarterly March 2011. Jakarta, Indonesia: World Bank. World Bank. 2012a. Targeting Poor and Vulnerable Households in Indonesia. Jakarta, Indonesia: World Bank. World Bank. 2012b. Protecting Poor and Vulnerable Households in Indonesia. Jakarta, Indonesia: World Bank. World Bank. 2012c. "Social Assistance Program and Public Expenditure Review 1: Public Expenditure Review Summary." Jakarta, Indonesia: World Bank. World Bank. 2012d. "Social Assistance Program and Public Expenditure Review 2: BLT." Jakarta, Indonesia: World Bank. World Bank. 2012e. "Social Assistance Program and Public Expenditure Review 3: Raskin." Jakarta, Indonesia: World Bank. World Bank. 2012f. "Social Assistance Program and Public Expenditure Review 4: Jamkesmas." Jakarta, Indonesia: World Bank. World Bank. 2012g. "Social Assistance Program and Public Expenditure Review 5: BSM." Jakarta, Indonesia: World Bank World Bank. 2012h. "Social Assistance Program and Public Expenditure Review 6: PKH." Jakarta, Indonesia: World Bank. World Bank. 2012i. "Social Assistance Program and Public Expenditure Review 7: JSLU, JSPACA, and PKSA." Jakarta, Indonesia: World Bank. World Bank. 201 2j. "Social Assistance Program and Public Expenditure Review 8: History of Social Assistance in Indonesia." Jakarta, Indonesia: World Bank. 105 E Protecting Poor and Vulnerable Households in Indonesia 106  0 THE WORLD BANK | BANK DUNIA THE WORLD BANK, JAKARTA OFFICE ndonesia Stock Exchange Building Tower 11/12th Floor JI. Jend. Sudirman Kay. 52-53 Jakarta 12910 Tel: (6221) 5299-3000 Fax: (6221) 5299-3111 Website: www.worldbank.org/id