I N D O N E S I A S O C I A L A S S I S TA N C E P U B L I C E X P E N D I T U R E R E V I E W U P D AT E Towards a Comprehensive, Integrated, and Effective Social Assistance System in Indonesia The World Bank Office Jakarta Indonesia’s Social Assistance For any questions regarding this Printed in October 2017 Public Expenditure Review report, please contact: Indonesia Stock Exchange Update is a product of the staff Photo Credits World Bank Building Tower II/12th Floor Jl of the World Bank. The findings, Changqing Sun Jend Sudirman Kav 52-53 Jakarta interpretations, and conclusions (csun1@worldbank.org) 12910 expressed herein do not necessarily reflect the views of and Juul Pinxten P (6221) 5299-3000 the Board of Executive Directors (jpinxten@worldbank.org). F (6221) 5299-3111 of the World Bank or the W www.worldbank.org/id Government they represent. The World Bank does not guarantee The World Bank the accuracy of the data included in this work. The boundaries, 1818 H Street NW Washington, colors, denominations, and other DC 20433, USA information shown on any map in this work do not imply any T (202) 458-1876 judgment on the part of the F (202) 522-1557/1560 World Bank concerning the legal W www.worldbank.org status of any territory or the endorsement or acceptance of such boundaries. Copyright Statement for Reconstruction and please send a request with including subsidiary rights, should Development/ The World Bank complete information to the be addressed to the Office of the The material in this publication encourages dissemination Copyright Clearance Center, Inc., Publisher, The World Bank, 1818 is copyrighted. Copying and/or of its work and will normally 222 Rosewood Drive, Danvers, MA H Street NW, Washington, DC transmitting portions or all of grant permission to reproduce 01923, USA, telephone 978-750- 20433, USA, fax 202-522-2422, this work without permission portions of the work promptly. 8400, fax 978-750-4470, http:// e-mail pubrights@ worldbank.org. may be a violation of applicable For permission to photocopy or www.copyright.com/. All other law. The International Bank reprint any part of this work, queries on rights and licenses, Foreword Rodrigo A. Chaves Country Director, World Bank Indonesia In the last decade, Indonesia has reduced its Indonesia aspires to reach high-income status poverty headcount rate from 16.6 % in 2007 by 2030. It is nonetheless facing a number of to 10.6 % by early 2017. This is an impressive significant challenges. Not only is an accel- accomplishment that deserves to be celebrated. erated growth of about 8-9 percent annually Recently, however, the pace of poverty reduction over the next 15 years needed, but this growth has slowed down. In addition, high income needs to become more inclusive and pro-poor. inequality remains a major challenge. Reducing Furthermore, Indonesia, like several other poverty further and improving equality there- middle-income Asian countries, must address fore requires sustained effort – more and better the looming aging population challenge, i.e. spending as well as further improvements in ‘to prosper before getting old.’ A comprehen- the effectiveness of government programs. sive, integrated, and effective social assistance system can continue to protect the poor and It is heartening to see that in early 2017, the vulnerable from suffering under destitution Government of Indonesia renewed its com- and various shocks while, at the same time, mitment to address inequality and financial support their upward mobility so that they can exclusion. The government decided to use lead productive lives through better human social assistance programs as important tools development and more sustainable livelihoods. to reduce inequality, both in terms of income and opportunities. Recent improvements in We at the World Bank Group stand ready to fiscal management have also enabled higher continue working alongside with the Govern- overall budget allocations for social assistance. ment of Indonesia, using all of our tools and At the same time, the application of a uniform expertise, to bring in the human and physical targeting mechanism has improved the tar- investment that Indonesia needs to become a geting of social assistance benefits towards high-income country. I know that this is pos- the poor and vulnerable. sible and we are eager to provide our support in making that vision a reality. This report assesses the strengths and weak- nesses of Indonesia’s main social assistance This work is the result of strong partnerships programs, which currently benefiting close between many government agencies and the to 100 million people. As an update to the World Bank. We are especially grateful for World Bank’s 2012 report Protecting Poor and the support and cooperation of the Ministry Vulnerable Households in Indonesia, this docu- of Finance (MoF), National Development ment reviews the progress achieved from 2011 Planning Agency (Bappenas), the Ministry until 2016 and proposes options for feasible of Social Affairs, and the National Team for the reform and policy planning. We at the World Acceleration of Poverty Reduction (TNP2K). Bank believe that Indonesia’s social assistance We would also like to thank the Australian system could be further developed to provide Government that, through the Department of an effective suite of support to poor households Foreign Affairs and Trade, generously provided that can address lingering risks and gaps in financial support for the production of this opportunities faced by its people. report, and in partnering with the World Bank to provide technical advice and support to the Government of Indonesia towards achieving further poverty and inequality reduction. Acknowledgements This report has been prepared by members lyst), Gracia Hadi­wijaya (Consultant), Mi- The report was edited by Peter Milne (Con­ of the Social Protection and Jobs Global chaelino Mervisia­no (Research Analyst), Ba- sultant), Juul Pinxten (Research Analyst), Practice Team of the World Bank Office in gus Arya Wirapati (Research Analyst), Ririn Steisianasari Mileiva (Operations Officer), Jakarta. The core team is led by Pablo Acosta Salwa Purnamasari (Senior Economist), Pu- Gedsiri Suhartono (Consultant), Ilsa Mei­ (Senior Economist), Changqing Sun (Senior guh Imanto (Energy Specialist), Rythia Afkar dina (Social Protection Specialist), Maria Economist) and Juul Pinxten (Research An­ (Economist), Fran­cis Addeah Darko (Econ- Jenny Puspitawati (Executive Assistant) and alyst), it includes as team members Luisa omist), and Indra Budi Sumantoro (Consul- Atin Parihatin (Consultant). The design of Fernandez (Senior Social Protection Spe­ tant) for comments and technical support. this report was prepared by the Bentuk team: cialist), Jon Jellema (Consultant), Talitha Muhammad Kamal and Philip Tanaka. Chairunissa (Consultant), and Kara Parahita The team would also like to thank Yus Me­ Monica (Research Analyst). dina Pakpahan (Research Analyst), Ahya The team is also grateful to the Ministry of Ihsan (Senior Economist) and Ratih Dwi Social Affairs (MoSA), Ministry of Finance The team is grateful to Rodrigo A. Chaves Rahmadanti (Research Analyst) of the Mac­ (MoF), National Development Planning (Country Director), Jehan Arulpragasam roeconomics & Fiscal Management Glob­ Agency (Bappenas), TNP2K, the Presidents (Practice Manager), Truman Packard (Lead al Practice team for preparing the public Staff Office (KSP) and the Australian Gov- Econ­omist), Iene Muliati (Senior Social expenditure summary, and Raditia Wahyu ernment’s Department of Foreign Affairs Protec­tion Specialist), Edgar Janz (Senior (Staff, National Development Planning and Trade (DFAT) for providing use­ ful ma- Poverty Specialist), Vivi Alatas (Lead Econo- Agency, or Bappenas) and Adi Nugroho ( terials on social assistance programs and mist), Hans Anand Beck (Lead Economist), Ministry of Finance) for providing budget comments on the report. Support for this re- Ruslan Yemtsov (Lead Economist), Matthew data and Dyah Larasati (Social Protection port has been generously funded by the Aus- Wai-Poi (Senior Economist), Cindy Pal- al Team for Policy Specialist in The Nation­ tralia’s Department of Foreign Affairs and adines (Young Professional), Pandu Harim- the Acceleration of Poverty Re­ duction or Trade through the Partnership for Knowl­ urti (Senior Health Specialist),Imam Seti- TNP2K) for providing information on Pro- edge-based Poverty Reduction Trust Fund. awan (Research Analyst), Daim Syukriyah gram Indonesia Pintar. (Economist), Lau­ ra Wijaya (Research Ana- Abbreviations & Acronyms ASLUT Social Assistance for the Elderly Program, Asistensi Sosial Lanjut Usia Terlantar Bappenas State Ministry of National Development Planning, Menteri Negara Perencanaan Pembangunan Nasional BLSM Unconditional Cash Transfer Program, Bantuan Lansung Sementara Masyarakat BPNT E-voucher, non-cash, component of Rastra, Bantuan Pangan Non Tunai BSM Poor Students’ Support Program, Bantuan Siswa Miskin Bulog National Logistics Agency, Badan Urusan Logistik Nasional CMRS Crisis Monitoring and Response System DJSN National Social Security Council, Dewan Jaminan Sosial Nasional ECED Early Childhood Education and Development FDS Family Development Sessions GoI Government of Indonesia IDR Indonesian Rupiah JKN-PBI Recipient of Government Paid Health Insurance Premium - Penerima Bantuan Iuran within the National Health Insurance Program, Jaminan Kesehatan Nasional - KKS Family Welfare Card, Kartu Keluarga Sejahtera KPS Social Security Card, Kartu Perlindungan Sosial M&E Monitoring and Evaluation MIS Management Information System MoEC Ministry of Education and Culture MoRA Ministry of Religious Affairs MoSA Ministry of Social Affairs NER Net Enrolment Rate ODA On-Demand Application OJK Financial Regulatory Authority, Otoritas Jasa Keuangan OOP Out-of-pocket Spending P2B Sustainable Livelihoods Program, Pengembangan Penghidupan Berkelanjutan PBI Recipient of Government Paid JKN Health Insurance Premium, Penerima Bantuan Iuran PIP Smart Indonesia Program, Program Indonesia Pintar PIS Healthy Indonesia Program, Program Indonesia Sehat PKH Conditional Cash Transfer Program, Program Keluarga Harapan Raskin (Former) Subsidized Rice Program, Beras Miskin Rastra Subsidized Rice Program, Beras Sejahtera SA Social Assistance SJSN National Health Insurance System, Sistem Jaminan Sosial Nasional SLRT Integrated Referral System, Sistem Layanan Rujukan Terpadu SRIS Social Registry Information System TNP2K National Team for the Acceleration of Poverty Reduction, Tim Nasional Percepatan Penanggulangan Kemiskinan UDB Unified Database WFP World Food Programme P. 01 CONTENTS P. 19 CHA PT E R 1 EXEC UTIV E SOCI A L ASSI STA N C E S U MMA RY E XPE N DI TURE A N A LYSI S P. 29 CHA PT E R 2 MA I N PORT FOL I O O F SOCI A L ASSI STA N C E PROGRA MS 2.1 UNCONDITIONAL CASH TRANSFER Bantuan Langsung Sementara Masyarakat (BLSM).................................................... 31 2.2 SUBSIDIZED RICE FOR THE POOR Subsidi Beras Sejahtera (Rastra)......................................................................................... 37 2.3 SUBSIDIZED SOCIAL HEALTH INSURANCE Jaminan Kesehatan Nasional-Penerima Bantuan Iuran (JKN-PBI)........................ 47 2.4 CASH TRANSFER FOR POOR & AT-RISK STUDENTS Program Indonesia Pintar (PIP)........................................................................................ 55 2.5 CONDITIONAL CASH TRANSFER Program Keluarga Harapan (PKH)...................................................................................63 P.73 P. 83 C HA PTER 3 CHA PT E R 4 THE “LIF E-CYC LE” I N T E GRAT E D SO C I A L APPRO AC H TO S O C I A L ASSI STA N CE : ASS ISTANC E & S A FE T Y POSSI B I L I T I E S & NETS B E N E FI TS P.89 P. 91 C HA PTER 5 S U MMA RY & MA IN RE FE RE N C E S REC O MMEND ATIO N S Executive Summary Indonesia has committed to developing a comprehensive & effective social assistance system for poor & vulnerable households S ince 2010, the Government of Indonesia (GoI) has set official poverty reduction targets and emphasized the importance of a well-func- tioning social assistance system in continued poverty reduction. The Government has exe- cuted several consequential social assistance reforms, while P. 0 1 spending on permanent social assistance programs rose in real terms between 2010 and 2016. Several rounds of reduc- tion in expensive and untargeted subsidies were achieved, and the expenditures saved reallocated to the social assis- tance sector through: (i) temporary, emergency, uncondi- tional cash transfers targeted to poor and vulnerable house- holds; (ii) benefit and coverage increases for Indonesia’s education cash transfer program; and (iii) conditional cash transfers. Standardized procedures for targeting and iden- tifying potential beneficiaries, drawing on a newly updated national registry of around 26 million poor and vulnera- ble households (the Unified Database, or UDB), were put in place for all implementing agencies to adopt. Most recently, the Government has released the National Financial Inclu- sion Strategy, which calls for achieving greater financial inclusion by transforming cash-based social assistance pay- ment systems into a cashless system using one single card (Kartu Keluarga Sejahtera, KKS). In 2017, the Government has decided to reduce electricity and LPG subsidies, and is plan- ning to redistribute the LPG subsidy to poor and vulnerable households via the KKS card in 2018. Moreover, the condi- tional cash transfer program (PKH) has expanded from 3.5 million households in 2015, to 6 million by end of 2016, and expected to reach 10 million in 2018. 3 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 With Indonesia’s economic development facing new challenges, now is a good time to review and update social assistance reform strategies. However, despite the efforts in expanding ening inequality is also evident in non-income Poverty rate: 10.6% coverage of the social protection system, the poverty indicators, such as education, health, pace of poverty reduction in Indonesia has and labor-market outcomes. While all Indo- slowed significantly in recent years, while nesian households experience good outcomes both chronic poverty and vulnerability have more often than they did 10 years ago, the gap persisted. The average annual reduction in the in achievement between poor and non-poor headcount poverty rate fell from 1.2 percent- households has widened for some indicators. age points between 2007 and 2010, to just 0.5 The administration that took office in late 2014 of a percentage point between 2011 and 2014. added a focus on reducing inequality, identi- From 2014 to 2017, poverty reduction contin- fying social assistance as a means of reducing From 2014 to 2017, poverty reduction con- ued at the same slower rate, falling by just 0.6 inequality in both incomes and opportunities. tinued at the same slower rate, falling by just of a percentage point to reach 10.6 percent as 0.6 of a percentage point to reach 10.6 per- of March 2017. Key reasons for this slowdown This report reassesses the strengths and weak- cent as of March 2017 in the pace of poverty reduction are that those nesses of Indonesia’s main social assistance living in poverty are increasingly further away programs, and proposes feasible reform op- from the poverty line, and so require greater tions, both program-by-program and for the effort—better spending, targeting and integra- social assistance system as a whole. This 2017 tion—to lift them out of poverty. Moreover, the update, following an earlier World Bank (World 24 percent of Indonesians living between the Bank, 2012j) report seeks to provide evidence of poverty line and 1.5 times the poverty line are the progress made between 2011 and 2017, togeth- still highly vulnerable to falling back into pov- er with relevant benchmarks for future reforms erty1 if they experience a shock, such as illness, and policy planning. The review presents analyt- a natural disaster, or any other interruption to ical evidence on salient program features and is- their regular earnings and livelihood. sues, and proposes additional efforts and options toward a truly integrated system. Moreover, income inequality is on the rise, while access to opportunities remain unequal. While inequality in Indonesia by end-2004 was on par with its level in 1980, the Gini coefficient rose by about 6 percentage points in the peri- od 2005-12, and declined subsequently by 1.7 points to 39.3 Gini points by March 2017. Wors- 1 Susenas (2016) and World Bank staff calculations. 4 Executive Summary The path toward a FIGURE ES1 SA convergence in the poorest 10 percent of households comprehensive and effective Indonesian social assistance system is clear RASTRA 78% 43% PBI 53% PIP 16 % PK H 6 .5 % ALL FOUR 2.2% PIP,PKH 2.6% Source Susenas 2014 RASTRA, PBI, PIP 11% RASTRA,PIP 14% and World Bank staff calculations RASTRA,PIP,PKH 2.5% RASKIN, PKH 6% PBI,PIP,PKH 2.3% Persistent poverty, vulnerability, low mobility, simulations indicate that an integrated social plaints—can accrue from the elimination of and inequality can be ameliorated by a more assistance system—bringing the existing set of the duplication that is currently pervasive. effective social assistance system. A compre- independently operating programs and their As this report details, these processes (except hensive social assistance system can provide implementing agencies together via common targeting) are carried out independently and the basic necessities that poor households do minimum standards—could provide a boost without regard to a common standard by many not access frequently enough, providing an im- to consumption expenditure equal to between different agencies, although some convergence mediate, direct impact on extreme poverty. It 14 and 21 percent of an average targeted house- with regard to benefit delivery systems was can simultaneously assist poor and vulnerable hold’s budget and would have an immediate begun in late 2016. However, this confusing households to mitigate risks by encouraging impact on poverty. The “overnight” reduction landscape is difficult for actual or potential larger or more consistent investments in mem- in the headcount poverty rate that would result beneficiaries to access, and is responsible for a bers’ human and financial capital, as well as re- from benefit integration is expected to be 2 to 4 de facto, ground-level separation of programs ducing reliance on negative coping behaviors, percentage points, depending on the coverage and initiatives targeting the same poor and vul- which can sacrifice productive investments for rate chosen being 10, 25 or 40 percent, all else nerable populations. For instance, Figure ES1 the sake of maintaining minimum consump- being equal.2 Even partially integrated social shows that in 2014 no more than 2.2 percent of tion. This helps households to absorb and mit- assistance could slow down the rate of increase the poorest 10 percent of households received igate negative shocks in the most flexible ways in inequality that Indonesia has experienced all four of Indonesia’s main social assistance such that welfare losses are less severe and not recently. For example, extending a social assis- programs.3 This reality has not changed signifi- compounded. Finally, a robust social assistance tance package to the 10 million poorest house- cantly when referring to 2016 data. The cur- system can make government-driven policy holds that combines the three current direct rent administration urgently needs to develop reform more palatable, thereby encouraging cash or near-direct cash transfers into one, a broadly agreed integrated social assistance more sustainable economic growth. would create benefit with a magnitude similar system operation plan, which includes clari- to that in countries where direct transfers re- fication of the roles and responsibilities of all Beneficiary households would be well-served duce poverty without distorting labor-market the agencies involved in the provision of social by better coordination and integration of the decisions. Renewed efforts and consistent at- assistance. existing social assistance programs. While the tention are required to broaden and deepen the main programs are more effective and efficient work begun by the previous administration. All else remaining equal. For reference, if actual headcount than they were just a few years ago, and coordi- 2 poverty continues to fall at the rate experienced between nation is more prevalent than before, institu- Efficiency gains from integration in the provi- 2013 and 2014 (about 0.1 to 0.2 of a percentage point per year), it would take about 10 years to achieve the “overnight” tional “silos” still exist within the social assis- sion of social assistance program sub-process- reduction that the least expensive integration scenario could achieve immediately. tance sector and most activities are carried out es—outreach, targeting, enrolment, beneficia- 3 The conditional cash transfer program, PKH, has both the with limited coordination among programs, ry verification, benefit transfer, Monitoring lowest coverage targets and the strictest means cutoffs. Therefore, any households receiving PKH would automatically implementers, and stakeholders. Results from and Evaluation (M&E), grievance and com- qualify (in principle) for the other three programs. 5 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 Additional reforms can help social assistance Of equal importance, Indonesia should be providing an empirical review of the impact of programs reduce poverty more efficiently at prepared to spend more to safeguard progress social assistance reforms that have taken place any level of expenditure, but new programs already made through the establishment of over the past six years, this update indicates ar- for uncovered risks will raise anticipated ex- pre-planned social assistance responses to the eas in which further reforms are needed, with penditures. As currently less than one-quar- natural disasters and macroeconomic instability a focus on the next generation of reforms and ter of social assistance benefits reach poor that are part-and-parcel of Indonesia’s geographic making progress toward an integrated social households, further program and system-wide location, and economic and financial openness. assistance system.4 The recommendations pro- reforms discussed in this report would allow vided here fall into three broad categories: social assistance expenditures (at any level) This report provides an updated assistance to reduce poverty more efficiently. However, review to support the current Government’s overcoming persistent poverty requires new social assistance reforms and poverty and programs or more generous benefits (or both). inequality reduction efforts. In addition to i ii ii i Increase accessibility to social assistance pro- Create new programs and innovate within Boost integration of the social assistance sys- grams by poor and vulnerable households in existing programs to provide solutions for tem by: (i) continuing to develop a suite of order to achieve poverty reduction targets. In- key Indonesian life-cycle risks and vulnera- social assistance delivery processes that are creased coverage may naturally result from sys- bilities. For instance, these could include early integrated across, and serve all, existing and fu- tem integration, but should be pursued rapidly childhood education, retirement savings, and ture social assistance initiatives; (ii) improving and independently. macroeconomic or natural disaster crisis re- coordination at the regional and local levels sponses, which Indonesia’s current social assis- among implementation partners, local gov- tance programming do not address adequately. ernments and public service agencies; and (iii) Without further innovation attuned to the In- reforming and revising operations within in- donesian context, the social assistance system dividual social assistance programs to prepare will, no matter how well-integrated, remain in- them for incorporation into an integrated and complete. more effective system. Table ES1 below summarizes suggested ac- tions within these three overarching social assistance system-wide goals for the current administration: The Government should intensify social assistance reform efforts to ensure that Indonesia remains on the right path of poverty and inequality As the 2012 review made clear, social assistance cannot 4 bear the entire burden for improving household welfare reduction and reducing inequality; rather social assistance should be complementary to policies and programs that improve access to high-quality, low-cost public goods and services and access to high-quality, secure jobs. 6 Executive Summary TABLE ES1 Building a comprehensive & integrated social assistance system GOAL SHORT-TERM ACTIONS (NEXT YEAR) MEDIUM-TERM ACTIONS LONG-TERM ACTIONS (NEXT 2 YEARS) (NEXT 4 YEARS) Increase All social assistance initiatives incorporate A. New programs (proposed Omnibus funding—rather accessibility by the beneficiary eligibility criteria and select to cover key uncovered risks) than program-by-program poor and vulnerable beneficiaries from one common targeting eligibility criteria incorporated negotiations—for all eligible households database (based on the current Unified into the SRIS functionality. beneficiaries (as determined Database). Each program will also need by the SRIS) in all programs to incorporate two-way updating—from B. Outreach for, and registration officially part of the “One the targeting database to program-based of, potential beneficiaries through System” coordinated social beneficiary lists, and from program-based a dynamic SRIS conforms to assistance framework. information to the targeting database, the a jointly agreed structure, initial step toward constituting a dynamic, principles, and implementation two-way, social registry information system arrangements. (SRIS) to ensure progress inclusion of all poor and vulnerable households. Address key Agencies responsible for health, education, A.Agencies responsible for health, Approved programs and uncovered risks and social insurance, development planning, education, and social insurance operational arrangements (for vulnerabilities poverty, and crisis monitoring and response propose, with cost estimates, crisis response) are piloted identify sector-specific uncovered risks faced social assistance programs within the newly created “One by Indonesian households. covering as yet uncovered sector- System” coordinated social specific risks to development assistance framework. planning and poverty agencies B.The national crisis response framework strengthens the roles of existing social assistance delivery platforms as part of “crisis response” strategy and negotiates with social assistance- executing agencies for joint planning, deployment, and monitoring . Boost “One Bring all institutions, agencies, and initiatives Formalize the division of roles, Formally establish Indonesia’s System” framework providing social assistance under common responsibilities, financing, “One System” social for social assistance standards and procedures for targeting, authority, and accountability assistance framework and beneficiary selection, payment systems, between central and local formalize the roles and grievances and complaints, and performance agencies involved in social purview of each of the M&E. assistance. system’s partner executing agencies in health, education, social insurance, planning, poverty, and crisis monitoring and response. 7 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 8 Executive Summary Continue expanding & deepen programmatic reforms for existing programs Households would benefit from a more inte- the UDB, is not dynamically updated and does FIGURE ES2 grated and navigable social assistance land- not currently allow for non-included house- scape. Indonesia’s current social assistance holds to request inclusion. A second reason Coverage of major active household programs correspond logically to the import- is that, the use of proxy means testing, in the targeted social assistance programs ant risks faced by poor and vulnerable house- absence of more accurate means testing, within (%), 2016 holds (see Section 3). Integration of these Indonesia’s targeting database means that not all programs through common standard setting; poor and vulnerable households can be correctly R AST R A through central-level oversight and regulation identified, even with the best possible methodol- Poor 70.4 of locally-implemented (and locally relevant) ogy and implementation of a targeting system. Vulnerable 64.2 initiatives; and through the provision of holis- The Rest 31.8 tic outreach, beneficiary selection, and facili- Significant increases in allocative efficien- P B I/J KN tation, could make a significant difference for cy via system-level improvements is within Poor 30.4 poor and vulnerable households that currently reach. With improved targeting through a more Vulnerable 25.2 do not have access to every program and initia- dynamic targeting database, through the gov- The Rest 12.1 tive for which they are eligible. ernments SLRT and ODA initiatives, a future SA system in Indonesia could continue to reduce P IP Three out of the four major and active SA pro- allocation of benefits to the non-poor, non-vul- Poor 22.70 grams are at or nearing the right coverage nerable populations (the grey bars in figure ES2). Vulnerable 18.70 level. Rastra and PIP are at the right coverage The Rest 7.13 level of 25 percent of the population and PBI/ Improvements within existing programs P KH JKN nearing its target of 40 percent coverage, will lead to a more effective system. Many Poor 30.50 the main program that remains low in cover- programs are not providing an entire benefit Vulnerable 12.60 age, at just about 10 percent, is PKH, covering 6 package to those to whom it was promised: in The Rest 3.20 million families at the end of 2016, PKH is rel- other words, these programs are not providing atively low in coverage. The government has re- the right benefits. Other social assistance pro- cently decided to expand the program up to 10 grams are not providing benefits at the right Source Susenas 2016. Note: for PKH 2014 data is used to million families in 2018 bringing the program time: benefits are distributed either too early impute allocation of beneficiaries with 2016 coverage level of 6 million families. closer to a coverage level (14 percent) compara- or too late related to some necessary purchas- ble to other similar countries with a maturing es or investments that households make. And, CCT program: Brazil’s Bolsa Familia covering with few exceptions, social assistance programs 14 million families or 25 percent of the popu- do not always reach the right people: over half lation; Phillipines’ Pantawid Pamilya covering of the benefits available end up with non-poor, 4.4 million families or 20 percent of the popu- non-vulnerable households. Most program lation; Mexico’s Prospera covering 5.8 million level performance M&E mechanisms have not families or 20 percent of the population. been effectively assessing the gaps between the program design and actual implementation Despite having expanded significantly, the and therefore have not supported adaptation of current array of SA programs still do not program design and implementation arrange- cover large shares of their target population. ments to achieve better results. Viewing coverage by a welfare disaggregation of poor, vulnerable and non-poor and vulner- The table ES2 and and the remainder of this able (Figure ES2), considerable coverage gaps section summarize the priority operational re- persist due to inclusion errors. While targeting forms recommended for each existing program. errors can be improved, and have improved for some of the programs over the years, they can only improve further to some degree. A prime reason is that the current targeting database, 9 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 rASTrA JKN–PBi Subsidized Rice Program Subsidized Social Health Insurance Previously called Raskin (Beras Miskin), Indonesia’s Rastra (Beras Se- The PBI component of JKN (recipients of JKN health insurance fee jahtera) has strong potential but has been failing operationally to achieve waiver) has accomplished major coverage increases and has successful- fundamental social assistance goals. The consistent provision of a basic ly been merged into the National Health Insurance (JKN) system. JKN- food package could protect poor households from food-price volatility, calo- PBI is the largest single source of health insurance coverage in Indonesia, rie scarcity, and malnutrition. However, Rastra suffers from dilution of ben- covering over 92 million individuals in 2016. JKN-PBI’s value to house- efits and coverage errors, missing rice, and hidden financing burdens, all of holds is significant as it promises a nearly unlimited-in-value health which reduce the transfer values provided to target households. benefit to poor and vulnerable households. In other countries health insurance coverage expansion is often accompanied by falling rates of Poor targeting, dilution of benefits, and missing rice are long-standing out-of-pocket expenditures, while in Indonesia out-of-pocket expenditures and well-known Rastra issues. It is the least well-targeted of any of Indo- for health have only fallen slightly, indicating systemic issues with JKN-PBI nesia’s social assistance programs and the average benefit package is signifi- (World Bank, 2016b). cantly diluted when the “right” to buy Rastra rice is re-allocated at the local level to include many non-poor households. In addition, large quantities of First, JKN-PBI outreach and facilitation need major improvement. JKN- rice procured for Rastra do not reach localities and no extra effort is made to PBI households, for example, often do not know which treatments, pro- put Rastra rice in targeted households when total supplies are low. Finally, a cedures, providers, and medicines are covered and which are excluded. lack of clarity concerning responsibilities and financing at the “last mile” of This lack of knowledge reduces utilization rates and the value of the JKN- Rastra distribution means that Rastra-purchasing households—especially PBI program (World Bank, 2016b). Establishing common information those in remote areas—receive a lower per-kilogram benefit than promised. standards, and verifying that the standard has been met, is essential for delivering valuable healthcare services to poor households. Nevertheless, Rastra is the largest social assistance initiative in terms of coverage and second-largest in terms of budget, making reforms that Second, JKN-PBI’s targeting and beneficiary selection procedures need much more urgent. Only with the incorporation of the PBI component further reform to minimize exclusion of eligible households. For ex- into the much larger Universal Health Coverage initiative under JKN has ample, previous local-level variation in eligibility determination and Rastra fallen from first into second place in terms of the size of its program targeting practices (World Bank, 2013d)5 may have been reduced through budget. But it is still huge, and as such Rastra reforms have the potential to JKN-PBI (named Jamkesmas at the time) joining the UDB-based target- make the greatest impact for both government and households. ing system, but as yet there is no mechanism for local-to-central regis- try updating that would keep local-level JKN-PBI allocations current. In Rastra should revise practices to achieve its social assistance mandate. Ra- order to mitigate this risk, JKN-PBI should develop a robust grievance, stra quotas and actual household allocations should be based on a dynami- and reporting system that runs in parallel to the UDB-based grievance re- cally updated national registry. The “last mile” of Rastra allocations should porting system, so that households unfairly excluded from JKN-PBI ben- be monitored to ensure that a full allocation reaches all eligible households eficiary status can be reinstated when they most need it. To further empower first. Rastra socialization should be re-enforced with performance incen- the poor and vulnerable, such a grievance system could be complemented tives so local communities can develop grassroots monitoring and provide by efforts to raise beneficiary awareness of program entitlements. feedback to implementers. It has been also suggested that Rastra increase local-level transparency by listing eligible beneficiaries at the village level, Third, M&E systems for JKN-PBI should be upgraded to monitor health, and formalize and regularize the process of recipient replacement that is healthcare usage, financial protection and cost (from the household currently often achieved unilaterally by the village administration (Perdana side), and supply-side performance and readiness. As there is signifi- et al., 2015). cant regional disparity in the availability and quality of healthcare ser- vices, access to health care could be an issue due to weak supply. An issue In response to long standing delivery issues, reform is underway as the underlying this is the lack of full clarity on the roles, responsibilities and Government has introduced an e-voucher initiative. Rastra reform has be- capacities for overseeing administrator performance. JKN-PBI should ex- gun shifting towards cashing out the Rastra benefit, initially in areas with plicitly recognize that mutual assistance and support (especially through functioning rice markets, while the current operational model will likely information sharing) between program administrators and healthcare remain in the more remote areas of Indonesia. Starting in 2017, 1.4 million providers will only improve the healthcare service options delivered to Rastra beneficiary families in 44 cities will be able to purchase rice and in low-income beneficiaries and will help JKN-PBI to make good on its un- some cases other pre-specified food items from a network of e-Warongs, limited benefit package promise. which are operated by various entities, including small traders and chain stores. These e-Warongs are supported by a participating bank and equipped Lastly, JKN generally must make strong efforts to serve poor house- with EDC/Pos devices for processing transactions using the combo KKS holds more effectively while addressing other priorities that do not card. To be successful in delivering the Rastra benefits (as well as PKH and necessarily regard the poor and vulnerable population segment. A re- potentially PIP, and even selected targeted subsidies), the initiative needs port (World Bank, 2015b) indicated that JKN’s current priorities are the close M&E of implementation processes expansion of membership to the private and informal sector; increasing contribution collections from those not currently contributing; improv- 5 For example, the criteria summarizing household characteristics varied across districts; ing financial and fiscal sustainability; and enhancing the JKN adminis- in some districts, midwives and health center officials distributed PBI / Jamkesmas cards trator’s overall governance structure. While crucial for JKN’s future and the according to their own criteria, regardless of economic status; there were no incentives in the system to either maximize PBI enrolment or minimize targeting errors, while the list of eligible consistent availability of all JKN services for all households, these items do beneficiaries compiled by district officials was not subject to validation by higher levels of program administrators. not provide immediate improvements for poor and vulnerable households. 10 Executive Summary PKH PiP conditional cash transfer Cash Transfer for Poor and at Risk Students 92 miL. PKH’s positive impacts in welfare, in PIP has begun to demonstrate its full poten- health-seeking behavior, and in education tial but can still deliver more to those most in can be extended if the program scales up. need. With recent increases in coverage and Two impact evaluations have shown that PKH reforms to implementation, PIP is now making families have greater access to health and ed- significant positive contributions to welfare in ucation. They show that PKH households have poor and near-poor households (with students) a 2.7-percentage-point decrease in severe stunt- and to the Government’s drive to provide uni- ing and an 8.8-percentage-point increase in the versal basic education. PIP should focus on “JKN-PBI is the largest single source of health rate of transition from primary to secondary continuous and coordinated monitoring, eval- insurance coverage in Indonesia, covering over school for children in beneficiary families (TN- uation, and improvements in delivery: most 92 million individuals in 2016.” P2K, 2015a; World Bank, 2011a). These impacts importantly, benefit-level updating should were estimated most recently in 2013 before the occur more frequently in order for the PIP program was expanded to the current size of 6 transfer to remain relevant. PIP should be at million families. In addition, the program has the forefront of positive outreach to poor stu- also demonstrated that it can be flexible with its dents, especially those approaching the senior operational protocols and varied in its approach secondary or university levels, and facing the to service provider coordination and assistance. highest out-of-pocket and opportunity costs. Further innovations in its facilitation ap- PIP’s biggest hurdle may be its current institu- proach can help PKH to serve more house- tional form. PIP is split among two ministries holds in need. For example, since 2013 “Family and several directorates, each of which carries Development Sessions” (FDS) were introduced out most program functions independently. through PKH to provide group-level training While some effort has been made since 2013 by in early childhood education, parenting, health TNP2K and Kemenko PMK to ensure a great- and nutrition, household finances, small busi- er degree of coordination within policy and ness development, and entrepreneurship. PKH planning, some aspects are still lingering. For could take the lead in facilitating access to so- instance, there is currently no mechanism to cial assistance and publicly provided services jointly provide (and jointly review the effec- more generally by using its own resources to tiveness of), for example, management perfor- mobilize local governments, service providers, mance reviews, M&E efforts, socialization cam- and other stakeholders to provide access for paigns, a grievance redress platform, or a policy poor and vulnerable households to all locally review of the suitability of a “transition bonus” available resources. for eligible students making the leap from one school level to the next. There remain many PKH needs to continue strengthening its ad- opportunities for better program integration ministration capacity, information manage- that can in turn provide a better experience for ment, and HR systems, as well as the capaci- students and households. ty of affiliated service providers. Continuous enhancement of core program functions is essential for efficient delivery of benefits and effective access for households: timely verifica- tion of beneficiaries’ status and conditionality fulfillment; regular Management Information System (MIS) updating, adjustment of benefit levels and timely disbursements; determina- tion of local-level capacity for distributing benefits and implementation support; and suggestions for remediation of local supply in- adequacies in health, education and program 6 All of these PKH processes socialization, are some of the aspects that need (as well as some others) were found to be not oper- strengthening.6 As the program has been ex- ational or only sporadically panded significantly and is expected to under- operational in a first round of implementation “spot go further expansion, especially towards more checks” completed over 2008 and 2009 (Centre of remote regions, it is critical to strengthen the Health Research, 2010) and delivery systems to keep up with the needs of more recently throughout 2016 by the World Bank the program. Social Assistance team through several spot checks. 11 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 TABLE ES2 Priorities for currently active social assistance programs PROGRAM ISSUE/ CONSEQUENCE SUGGESTED CHANGE Rastra Large inclusion errors; dilution of Select beneficiaries based exclusively on the common targeting database and put in benefits place a two-way updating mechanism between the common targeting database and Rastra beneficiary records. Monitor and evaluate the e-voucher initiative started in February 2017 (Non-cash Food Assistance or BPNT) as a possible alternative delivery system to Rastra. As- sess the implementation and make adjustments to ensure the expected program outcomes can be achieved as well monitoring on actual benefits to beneficiaries in terms of convenience, quantity and quality of food items purchased. Further adapt and improve the e-voucher program design to ensure its applicability in rural areas and inclusion of nutritious food options in addition to rice. JKN-PBI Access in some places not Improve outreach, facilitation, and beneficiary support so that intended beneficiaries assured; low utilization. Uneven are aware of what services are covered and what a “best practice” healthcare service access to quality health services, schedule looks like for all household members. low level of knowledge of JKN-PBI entitlements and persistent mis- M&E systems for JKN-PBI should be upgraded to monitor health, healthcare usage, targeting. financial protection and cost issues (from the household side), and supply-side pre- paredness indicators and outcomes. PKH Unequal coverage; low benefit Continue expanding while strengthen implementation capacity, revamp IT systems, levels; inadequate training and improve HR management and ensure adequate training of facilitators, and expand support to facilitators; limited family development sessions for all families. Increase benefit levels. Improve informa- coordination with health and tion sharing with service providers and service provision planning authorities. education service providers at national and sub-national levels PIP Benefit levels incommensurate Adjust benefit levels annually to ensure they are in line with actual costs of attending with education costs; low uptake each level of school. at advanced education levels; weak monitoring Develop outreach facilitation, and beneficiary support modules for senior secondary and university school-dropouts. Consider delegating the responsibility of outreach and enrollment to MoSA Suitability Not all important household risks Facilitate greater incorporation into the social protection system livelihoods and la- of Program to well-being are covered bor market activation initiatives for instance under the Pengembangan Penghidupan Composition Berkelanjutan (P2B) initiative. A crisis Monitoring and Response Continue refining the crisis-data-collection-and-monitoring system while planning System (CMRS) exists but is not for a range of social assistance initiatives that can be flexibly and quickly deployed at yet actively used the household level when social or economic crises strike. 12 Executive Summary OBSTACLES Allocations and “last mile” not currently controlled by local governments nor the Rastra administrator. The National Health Insurance system (SJSN)—of which JKN-PBI is now a part—is focusing more on financial sustainability, which may potentially divert attention away from improving JKN-PBI service in the short term. The budget for expansion and system strengthening needs to be guaranteed. Better coordination with health and education service providers would require enhanced and formalized coordination mechanisms at national and sub-national levels. PIP is fragmented internally; policy and planning proceed independently for regular and madrasah schools. Uncertainty over which of the many small livelihoods projects currently active in Indonesia will be effective and which can be scaled. Monitoring requires coordinated, timely inputs from many government agencies; response requires flexible, just-in-time expenditures difficult to include in regular budget negotiations. 14 Executive Summary A future social assistance system should be able to respond to as of yet uncovered risks Besides strengthening and successfully completing planned coverage expansions of existing programs, new programs covering important risks that are currently not being addressed would make an integrated social assistance system more effective. There are some life-cycle risks not adequately addressed by the current collection of social assistance programs; see section 3 of this report for more detail. Currently, poor and vulnerable elderly, very young chil- by Bappenas centers mostly on the household and, after dren and disabled receive very little social assistance in identifying a household’s particular need, develops a strat- line with their needs. Nearly 70 percent of pre-school age egy centered on skills training, professional coaching, and children from poor households are not enrolled in any pre- training and thoughtful sequencing of cash grants, cred- school initiative. For children who are 5 to 6 years old and it and the transfer of assets. Several approaches under the their parents, there are no national programs or initiatives P2B framework had been piloted in six sub-districts as of that provide low-cost access to Early Childhood Education late 2016, with positive outcomes. Future prospects for a co- and Development (ECED) activities, or outreach and infor- ordinated and wider P2B roll-out, however, remain unclear mation to parents who might not yet fully understand the (Bappenas, 2017). value of such activities. To begin addressing this risk, the government could consider a fee waiver to access PAUD ear- In addition, Indonesia’s Crisis Monitoring and Response ly child hood learning centers for 40 percent of the poor System (CMRS) should be operationally linked and lever- and vulnerable children aged 5 to 6 years old. Such a pro- age the social assistance system when needed. Households gram would cost just over IDR 6.1 trillion per year and would in Indonesia are vulnerable to stresses that the internation- bring myriad benefits such as reduced malnutrition, greater al and national economies, as well as the environment, inev- cognitive development and more time for mothers to work. itably produce, and there is as yet no pre-planned response For the elderly, income security remains a dire issue; while mechanism providing social and economic support to mit- poverty and vulnerability rise with age, a third of the elderly igate against large negative shocks to a household’s welfare. are either living alone or with one other person while 40 per- A functioning monitoring system is already in place, man- cent do not have health insurance. Compounding these risks aged and analyzed by TNP2K. The system makes use of time- is the lack of a social pension for those who are elderly and no ly, high-quality data inputs from across several government longer working. To address this risk in the short term, a social agencies. It is also currently focused on social disasters and pension could be set up targeting the poor and vulnerable el- does not include a natural disaster component. Response derly. Covering 40 percent, or approximately 7 million of those protocols for both sorts of crises at different levels of sever- aged 64 and older, the cost to provide a minimum level of pro- ity are needed so that programs under different ministries tection would be just about IDR 27 trillion.7 can be automatically funded, activated, and implemented when needed, and so that budgetary and parliamentary The disabled are also more likely to be or become poor or procedures do not prevent timely assistance from being vulnerable as they are often limited in their opportunity released. Indonesia should also develop programs that can to generate income. In addition, they may face above aver- be deployed rapidly and counter-cyclically, such as public age expenditures in health. Recent survey data on the dis- works. Certainly such a set of protocols would require inten- abled does not exist and so no new program simulation is sive institutional coordination across several government introduced; the Ministry of Social Affairs however, plans to agencies that should be part of the response system. While subsume both the disabled and elderly programs (ASODKB some ideas and protocols have been discussed there has been and ASLUT both currently at very low coverage) within the little take up of these ideas in the current administration. PKH CCT opening up possibilities to scale up protection for both the elderly and disabled that are currently not covered If the current suite of programs remains over the next de- by corresponding social assistance programs. cade or so, rather than further coverage expansion beyond planned targets by 2018, the next steps would be to pursue The social assistance system should provide active support greater integration, improved delivery systems and com- to poor and vulnerable individuals and households mov- mon standards, better targeting and the development of new ing from a state of dependence and vulnerability to one of programs to address uncovered risks in line with the main independence and resilience. For example, students from recommendations of this report. poor and vulnerable households getting ready to enter the labor market or under-skilled individuals already working would benefit from labor-market activation programs, or “livelihoods” initiatives such as job training (or re-training), skills enhancement and second-chance education. To some 7 Both estimates assuming 10% administration costs. For the ECED fee waiv- extent, the Government is responding to this unmet need er: unit cost per child per year of about IDR 1,000,000 (adjusted for inflation to 2019 and based on 2013 World Bank and Unicef estimates of IDR 800,000 under the Sustainable Livelihoods Program (P2B) launched per year per child) and planning for a gradual scale up to reach a 40 percent in 2015 by the State Ministry of National Development coverage level by 2019. For the Social pension: the minimum pension payout is modelled to follow the BPJS Labor pension programs current value of IDR Planning (Bappenas). The current P2B strategy developed 300,000 and adjusted for inflation to 2019. 15 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 Indonesian social assistance programs are and agencies with social assistance expendi- ture oversight or M&E responsibilities, span- more united under common, minimum ning sectors and levels of government, with standards for delivery than ever before, but different spheres of influence and only partial- ly overlapping information needs producing a continued effort should be made to achieve confusing landscape that does not yet produce effective integration. authoritative regulations or guidance for social assistance providers. Variability in local execution exacerbates the negative effects of a fragmented central architecture. Qualitative and quantitative ev- Social assistance execution has historically (M&E); in outreach, socialization, and aware- idence suggests that idiosyncratic and varied been highly fragmented across ministries and ness; and in grievance procedures. This is due eligibility determination procedures, targeting agencies, but the Government’s push for great- to institutional fragmentation and the lack of frameworks, and benefit ownership and con- er financial inclusion may consolidate social a common, authoritative standard, without trol rights, negatively affect Social Assistance assistance delivery in an important way. At which these sub-processes are still needlessly program integration at the household level. For the central level, the execution of the major duplicated and delivered with varying quality. example, the allocation of household “rights” to social assistance programs is still shared by six Likewise, though Rastra and PKH began using purchase subsidized Rastra rice is done by village central institutions (World Bank, 2012j). This the UDB in 2012, and while PBI and PIP joined heads and sub-village administrations; a portion is likely to continue and need not necessarily in 2013, only PKH and PIP, since 2013, have ful- of PIP benefits (and the right to distribute them) change radically in order for social assistance ly adopted the UDB standard for use in quo- is still controlled by schools; and the distribu- delivery to improve. However, the President’s ta-generating, eligibility determination, and tion of the fee-waiver component of subsidized Decree on a Financial Inclusion road map in beneficiary selection, and only PKH has im- health insurance coverage has not previously 2016 has made an important push to begin plemented a two-way updating procedure that been monitored or evaluated. The fact that such achieving fully integrated digital social assis- works in concert with the UDB. practices persist and are tolerated indicates a low tance payments by 2022 (MoSA, 2016d). level of coordination between central-govern- The Government has launched two initiatives ment-level policy planners and funding authori- Indonesian social assistance programs are to build common platforms: (i) an Integrated ties, and the regional- and local-level administra- more united under common, minimum stan- Service and Referral System as the citizen inter- tions that have been delegated crucial portions of dards for delivery than ever before, but con- face for multiple government services (SLRT) important social assistance processes. 8 tinued effort should be made to achieve effec- and the On Demand Application (ODA) as a tive integration. Efforts at integration have way to update household information with the Indonesia’s “One System” social assistance been made: an identity-card-based system; involvement of the local government; and (ii) a framework would mean all institutions, “bilateral” automatic eligibility efforts, where delivery gateway for social assistance cash and agencies, and initiatives providing social as- receipt of one transfer makes a household auto- in-kind transfers (such as for PKH, PIP, Rastra sistance would be bound by, and evaluated matically eligible for another; and an integrat- and some selected subsidies) using one inte- according to, common standards. This ap- ed outreach effort to “enroll” households in all grated social assistance card (a “combo” KKS proach achieves economies of scale, reducing programs for which they are eligible, were all card) (MoSA, 2016d). Both initiatives bring In- duplication of crucial social assistance delivery piloted recently. However, very little progress donesia closer to its “One System” framework. processes, such as targeting and beneficiary se- has been made regarding common standards However, beyond the two pilots, there still exist lection, payment systems, grievance and com- and processes in Monitoring and Evaluation many overlapping government-wide systems plaint recording and monitoring, and M&E, 8 Unlike most other public social expenditures which are in large part executed by regional governments (primarily district-level governments), 85 percent (on average) of SA programming expendi- tures are centrally executed. See the Expenditure Summary report in this series or World Bank (2012j) for more detail on the history and contemporaneous particulars of this arrangement. BOX ES1 International experience on Social Assistance integration International experience suggests that an integrated social assis- well as local-level unified outreach, facilitation, and planning for the tance system can be achieved in a variety of ways. Brazil has suc- achievement of better outcomes with beneficiaries. While in both cessfully merged multiple cash, in-kind, and indirect social assistance countries central-level bodies regulate the use of the national tar- transfers into one single conditional cash transfer program, called geting database and are uniquely authorized to determine eligibility, Bolsa Familia (“Family Grant”).9 While it is nominally executed by each registry’s “implementation” is decentralized, as municipal-level the Ministry of Social Development, payments and the management administrators maintain and locally update it through the receipt information system (MIS) are “outsourced” to the Caixa Economica and processing of applications and grievances. In addition, both Chile Federal (a public bank), while regional governments are tasked with and Colombia use locally sourced social workers and program facil- beneficiary selection, updating, complaints, compliance verification, itators who, in concert with local government, determine poor and and facilitating connections to complementary public and private vulnerable household needs. They then match households’ profiles services. with existing national and local social programs, for example in child care, youth training, micro-credit, scholarships, or housing subsidies. Colombia and Chile have instead left the expanding program collec- tion alone and pursued integration through common standards, as 9 Created in October, 2003. Law No. 10.836, 2003. 16 Executive Summary The Government has had key initial that are currently pursued differently (if at all) by most agencies with a social assistance deliv- successes in uniting portions of the social ery mandate. It can also help the Government assistance portfolio under a common in rationally allocating limited resources based on: the risks potential beneficiaries face; their targeting and beneficiary selection system unmet needs; and the social assistance pro- (today, the UDB). But the ultimate gram mix that can remedy these two shortfalls. Finally, it encourages individual agencies to goal for this system is to put in place a work as “one government” that has an increase in beneficiary welfare and a reduction in bene- dynamic, two-way updating ap­ proach, ficiary risk as the ultimate goal. Indonesia will require clarity regarding the division of roles, responsibilities, financing, and authority between central and local agen- cies responsible for social assistance. While BOX ES2 Poverty databases as a basis for designing social assistance transfers are executed by the central government, many support functions— social protection systems: experience from beneficiary socialization and outreach; M&E; some Latin American countries1 complaint and grievance-handling—are dele- gated to regional and local governments. These same governments also determine to some ex- A common feature in social assistance beneficiary targeting in Chile, Colombia, Mexico, and tent policy, planning, and service schedules in Brazil is the use of national poverty databases as authoritative registries of poor and vulner- front-line health and education providers, and able households. Common design elements and implementation choices in these targeting can execute their own policies and schedules systems as follows:* even when they contradict, for example, na- tional social assistance guidelines or objectives. At the same time, enterprising districts have de- A. Efficient data collection. D. Clearly-defined institutional roles veloped their own social-assistance initiatives Following sequential steps of collection and have proven essential for the suc- to complement or enhance national initiatives. processing under strict supervision proce- cess of household targeting systems. These areas will need technical support to im- dures contributes to the quality of informa- Cross-country comparisons reveal some prove, expand, and harmonize these local ini- tion gathered. important advantages of centralized design, tiatives within the national framework. In par- administration, database management, and B. Simple, user-friendly management ticular, village funds allocated according to the eligibility determination while day-to-day information systems (MIS). Village Law could be harnessed to support both operations are delegated to the local levels. Information management procedures However, the most effective organization national-, provincial-, and district-level social remain crucial in the construction of a re- of an authoritative beneficiary targeting assistance program implementation. liable and always-current database. Unique and selection institution will depend to a country-wide individual identification is great extent on capacities and comparative The establishment of common platforms often used to avoid duplication and to link advantages. and further collaborative refinements will registry information and beneficiaries with encourage ongoing integration of individual other systems and programs. E. Transparent monitoring and over- programs under one roof. For example, the sight mechanisms ensure credibility and Government has had key initial successes10 in C. Standardized household assessment. can help control fraud, malfeasance, and uniting portions of the social assistance port- In Chile and Mexico, which use proxy-means corruption. folio under a common targeting and beneficia- tests to assess household vulnerability, When multiple checks—such as supervision ry selection system (today, the UDB). But the to 90 percent of program benefits are re- of household assessment interviews, veri- ceived by the poorest 40 percent of house- ultimate goal for this system is to put in place fication of information, automated checks, holds, while costs are relatively low—from a dynamic, two-way updating approach, where comparing registries with other data, US$2 to 8 per interview on average in Latin random-sample quality control reviews, program-level data on household characteris- America—and administrative requirements and citizen or ombudsman oversight —are tics and current trajectory can inform the over- are manageable. These countries have built into the system, potential beneficiaries all targeting system, and data can help program found that providing a household assess- and non-beneficiaries alike can be confident implementers update their operational prior- ment within a broader geographic target- that eligibility determination is undertaken ities. This two-way participation encourages ing framework greatly improves targeting the same way everywhere anonymously and implementing agencies to invest in common accuracy. decentralized data collection is more likely standards they are also bound to uphold. The to conform to a common standard. Author- next wave of integration platforms—in M&E; itative, common standards for the imple- socialization, outreach, and induction; and mentation and use of rigorously designed grievance complaint and reporting, for exam- targeting systems that query single, unique, ple—should likewise focus on participatory and authoritative registries of poor house- improvements and refinements. holds have over time served as an institu- tional coordination mechanism and have standardized access of poor and vulnerable 10 While all Social assistance programs are now using the UDB household to a larger set of social programs to at least generate initial beneficiary quotas, not all programs * Source: Castañeda, et al., 2005. and services. use the UDB to determine eligibility and select beneficiaries, meaning SA integration through a common targeting standard is far from complete. 17 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 A “single- window” framework could generate cost-savings for the Government. Picture source SLRT 2017 Regional or staggered “pilot” programs can little experience specific to delivering public ble households access an integrated package efficiently test approaches to integration and awareness or outreach to poor, marginalized, of benefits and complementary services, and provide a baseline for further refinements. vulnerable, or difficult-to-reach populations. would allow them to more quickly exit poverty For example, a “multi-channel” complaint and Clearly, the less effectively disseminated are and vulnerability. grievance system—where there are several ways public awareness strategies, the more serious of reporting and different actors who might be will be the gaps in eligible or potentially eligi- A “single-window” framework could generate “first responders” depending on when, where, ble household access to benefits. Adherence to cost-savings for the Government. For example, and how an individual complaint is received— a common socialization standard, where com- extending a combined PKH, PIP and mone- can be tested alongside the traditional system pliance is measured by awareness surveys com- tized-Rastra package to 10 million households, in a few representative areas. Similarly, social- pleted by the standard-setting agency would with an average benefit of about 20 percent of ization, outreach, and active induction strate- improve adherence to a common targeting the value of annual household expenditures in gies—of which there are an incredible variety standard, reduce variability in benefit access, the target population (or IDR 3.3 million per that are potentially effective—can likewise be and generate a minimum level of performance. family per year) would create a single transfer tested in various forms in different regions. In with a magnitude similar to that in countries addition, technology pilots—linking standard Also critical is the effective provision of ser- where direct transfers reduce poverty without identity cards to the updated national registry vice-provider links. Indonesia’s decentralized distorting labor market decisions.12 The fiscal and using electronic identification technolo- administrative and public expenditure frame- cost of the transfers alone—at about 0.3 percent gies to reduce error and fraud, for example; or work makes local governments responsible for of GDP—is slightly less than the cumulative switching to automated, electronic, or deper- the majority of social expenditures in health cost budgeted for these three transfers (deliv- sonalized payment mechanisms to minimize and education. Furthermore, in addition to ered independently) in 2016 (just over 0.4 per- leakage or corruption—will be necessary to providing support services for social assistance cent of GDP in 2016). In addition, participating keep expanded programs (and an expanded programming, local and regional administra- government agencies would see lower benefit social assistance sector) working efficiently for tions have begun experimenting with social as- delivery and oversight costs. The savings could the Government. sistance programming for residents.11 To help be channeled back into social assistance to cre- households take advantage of this diverse pro- ate, for example, the cadre of locally placed fa- Integration for households can also be grammatic landscape and better understand a cilitators mentioned above. achieved by delivering socialization and pub- diverse set of operating principles, Indonesia lic awareness of common standards, regard- should train a cadre of knowledgeable facil- 11 This is most noticeable in health insurance: many districts less of location or the manner of a household’s itators who would be tasked with counseling (and some provinces) have developed their own health insur- ance offerings for poor and vulnerable households that com- first contact with social assistance programs. and providing strategies for vulnerable house- plement the national health insurance initiative (formerly Currently, much of the social assistance social- holds to take advantage of all locally available Jamkesmas, now JKN-PBI). See the JKN-PBI chapter in this report. ization is delegated to the Ministry of Com- programs. Locally placed resources for “tying 12 In Mexico and Colombia the transfers of conditional cash transfer programs range between 21 to 25 percent of average munications and Information, which has very it all together” would help poor and vulnera- consumption of target groups. 18 Executive Summary P.19 Social Assistance Expenditure Analysis CHAPTEr 1 ublic expenditure on social assis- P and 2016. However, these permanent programs tance programs has roughly kept still receive relatively small budgets or output pace with increases in national out- shares when compared with other sectors. For put and total public expenditures example, in 2006 permanent social assistance (Figure in Box 1.1). Social assistance programs accounted for about 1.9 percent of expenditure magnitudes have been increasing total national expenditures, or 0.9 percent of steadily, with total central and sub-national GDP. By 2016, the analogous numbers were 3.8 spending on permanent social assistance13 ris- percent (out of 2016 total national expendi- ing by 128 percent in real terms between 2009 tures) and 0.7 percent (out of GDP). 14 13 Excluding BLT/BLSM; see the following paragraph in this note. 14 Sub-national expenditure data for 2016 use planned budget. 21 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 BOX 1.1 Defining social assistance spending in Indonesia In this note, social assistance spending follows the definition devel- Public spending on social assistance by various oped in the previous Social Assistance PER (World Bank, 2012j). So- definitions (Idr Trillion) cial assistance is defined as non-contributory cash or in-kind transfer programs targeted in some manner to the poor or vulnerable. Indo- nesia budget composition does not have a specific budget line that Social Social Household Social aid protection protection social economic includes the social assistance sector. Since 2009-14, the GoI has de- function sector (WB assistance class signed the Master Plan for the Acceleration and Expansion of Indo- (Govt def.) def.) nesia's Poverty Reduction (MP3KI). The GoI articulates its poverty alleviation strategy around three “clusters” (where households, com- 2008 2016 munities, and micro-enterprises are targeted); the first pillar (house- holds) is roughly equivalent to the definition of SA used in this note. No official budget category meets either the SA definition used here or the definition of the GoI’s first poverty reduction cluster. 144.4 Economic classifications in Indonesia’s budget expenditures include a “social assistance” category, which is used broadly and includes a wide array of social spending in areas such as education, health, agri- 92.1 97.9 97.1 90.9 culture, industry, and disaster relief. Functional classifications of In- 87.1 75.6 donesia’s budget expenditures include a “social protection” category, 73.8 68.6 71.1 68.8 66.4 which up until 2016 was used narrowly and consisted mainly of initia- 57.7 54.9 tives at MoSA (Ministry of Social Affairs). In 2016, however, the social 38.1 44.8 35.7 34.8 protection function was reclassified to include components that were 31.5 30.3 28.6 24.7 31.2 31.6 previously mapped under the “General Government Administration 8.7 10.1 11.1 14.1 25.0 21.8 7.6 function” including food / housing subsidies and social contributions. 5.8 7.2 8.4 9.0 11.2 This note aggregates identifiable social assistance expenditures and examines the total as if it were a standalone sector and budget item. At the central government level, 10 major social assistance programs, as well as remaining MoSA and minor social protection expenditures, are aggregated to create total social assistance expenditure. At the sub-national level, where budget data are more limited, the functional classification “social-protection” expenditures are used as a proxy for aggregate social assistance expenditures. Source MoF, Bappenas, and World Bank staff calculations. Public expenditure on SA has increased markedly, but remains low as share of GDP and of national expenditure. Indonesia spends less as share of GDP in social assistance (at 0.7 percent of GDP) than the average of lower middle-income countries. The spending is less than half of the average spending of that group, which is about 1.5 percent of GDP (Figure 1.1).15 When Indonesia is compared with some of its regional peers in the East Asia and Pacific region (EAP), the share of GDP for SA is similar to countries such as Vietnam and Thailand, but it is lower than most Lat- in American (LAC) and East Europe and Central Asia (ECA) countries. Global evidence shows that aggregate spending of social assistance (so- cial safety nets) rises as countries become richer, but still averages at just 1.6 percent of GDP. The average for richer countries is about 1.9 percent of GDP, while lower-income countries spend on average about 1.1 percent of GDP (World Bank, 2014b). Likewise, when comparing social assistance spending with other types of spending in the national budget, as shown in Figure 1.4, social assistance expenditure is one of the lowest shares, similar to agriculture spending. 15 Countries data refer to different years. 22 Chapter 1— Social Assistance Expenditure Analysis 0.7% IDR 24.8 TriLLiON FIGURE 1.3 On average, higher spending is allocated Previously known as Jamkesmas, and now as PBI for Rastra, JKN-PBI Indonesia spends less as share of GDP in social under the National Health Insurance program & PIP (central government major SA HH assistance (at 0.7 percent of GDP) than the JKN within SJSN, the social assistance sector’s programs, percent) health insurance program saw its allocation average of lower middle-income countries. increase significantly to IDR 24.8 trillion in 2016 (from IDR 8 trillion in 2013) FIGURE 1.1 BLT/BLSM/ 38.2 KKS&KSKS 9.3 FIGURE 1.2 Social assistance spending as a share Raskin/ 33.2 of GDP (%) – regions, income levels and Health insurance for the poor and Rastra 37.3 selected countries unconditional cash transfer (BLSM) have Jamkesmas 14.2 been the main drivers of recent increase /SJSN/KIS 26.9 Europe & Central Asia 2.3 in central government spending on HH 9.2 social assistance programs BSM/PIP 16.8 Sub-Saharan Africa 1.7 (central government expenditure on 3.5 Middle East & North Africa 1.4 major SA HH programs, per capita/poor PKH 8.5 Latin America & Carribbean 1.3 headcount, 2010 prices), IDR million) 0.9 East Asia & Pacific 1.2 PKSA IDR million per capita 0.6 South Asia 1.1 3.0 0.6 JSPACA 0.4 Upper middle income 1.6 0.2 World 1.6 JSLU 0.2 Low income 1.5 Lower middle income 1.5 Average 2004—2010 South Africa 2015 3 Average 2011—2016 Kenya 2014 2.7 Source MoF, Bappenas, and World Bank staff calculations. Russian Federation 2015 1.8 Mexico 2014 1.6 Peru 2015 1.4 enrolment of an additional 16 million benefi- Nepal 2014 1.3 ciaries, as well as an increase in the per-capita premium calculated by the GoI. At 34 percent Brazil 2015 1.3 of all social assistance expenditures on perma- Turkey 2013 1.3 nent programming in 2016, PBI now ranks just Vietnam 2015 1 higher than the subsidized rice program, Ras- Bangladesh 2014 0.7 tra, in terms of SA allocation magnitudes.16 Indonesia 2016 0.7 2005 2016 Phillipines 2014 0.5 Permanent programs providing larger pro- Unconditional cash transfer (BLT/BLSM/ portions of benefits to poor and near-poor KKS&KSKS) households—PBI, PIP, PKH—have seen their Subsidized rice (Rastra) Health insurance for the poor (PBI/KIS) social assistance expenditure shares rise re- Source World Bank Aspire 2017 and World Bank staff calculations. Scholarship for poor students (PIP) cently. While the rapid recent rise in enrolled Note Selection based on data availability and being recent Conditional cash transfer (PKH) PBI beneficiaries in the JKN account for the enough. For the categories of regions and income levels, the Child social services (PKSA) value shown represents a 2008-14 average, for the regions lion’s share of this pro-poor increase, both category, all income levels are considered in the value shown. Disabled social services (JSPACA/ASODKB) Countries shown are neighboring countries and several other Elderly social services (ASLUT) PKH and PIP have been expanding coverage as lower middle-income countries to provide perspective on well: PIP quadrupled the number of beneficia- Indonesia’s position. Source MoF, Bappenas, BPS, and World Bank staff calculations. ries between 2010 and 2016, while PKH more than doubled the number of covered house- holds during the same period. Together with the slight reduction in the number of Rastra Among permanent programs, the health in- beneficiaries, this has led to a majority of per- surance fee waiver program targeted to poor manent-program expenditures being directed and near-poor households attracts the larg- to the set of programs emphasizing pro-poor est social assistance budget allocation. Pre- coverage.17 When viewing social assistance ex- viously known as Jamkesmas, and now as PBI under the National Health Insurance program 16 Prior to the 2014 budget year, Rastra was consistently awarded a majority share of SA allocations for permanent JKN within SJSN, the social assistance sector’s programming; Rastra’s share approached 60 percent in most years before 2013. Spending on Rastra declined in 2013 when health insurance program saw its allocation in- budgeted coverage fell by 2.5 million households. crease significantly to IDR 24.8 trillion in 2016 17 See the Program Notes in this report for additional details on poor and near-poor coverage in the permanent social as- (from IDR 8 trillion in 2013), following the sistance programs. 23 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 penditures of the earlier, pre 2011, government energy subsidy policy, driving subsidized fuel administration (including defense spend- administrations, Raskin/Rastra expenditures and electricity prices higher instantaneously ing) has reliably accounted for just less than were higher, while Jamkesmas/PBI, BSM/PIP and economy-wide prices higher over the short one-quarter of national expenditures, energy and PKH were lower overall, especially in the past to medium term. In each instance, the GoI subsidy spending has been more volatile (tied two years of the current administration’s tenure. distributed a temporary unconditional cash as it is to international energy price fluctua- transfer to about 30 percent of the Indonesian tions). However, on average, it also accounted Social assistance spending measured on a population as compensation for the negative for just less than one-quarter of national public per-poor or vulnerable individual basis has impacts on household purchasing power from expenditures between 2004 and 2014, although risen. This result has been driven automati- these policy revisions. The GoI’s use of these through 2015 and 2016 it was significantly re- cally by coverage increases in most programs energy-subsidy-adjustment periods to also ex- duced to just 7 and 8 percent of national ex- (and resulting increases in spending on social pand permanent social assistance programs penditure. Since 2009, education spending assistance programs) that occurred simulta- in recent instances (July 2013 and November (excluding social assistance delivered via Min- neously with a decline in the number of poor 2014) were no exception. So while the pro- istry of Education programs) has accounted for and vulnerable households. Total spending in- nounced spikes in social assistance spending just under one-fifth of national expenditures. creased in 2016 almost three times the level in in 2005-06, 2008-09, 2013, 2014, and 2015 were Infrastructure spending saw a large uptick in 2010, while the poverty headcount poverty fell temporary, a much smaller increase in perma- 2015, from 11 to 14 percent of national expen- by about 3 million people. nent SA-program spending was also generated diture. With about two-thirds of an average an- during those periods. nual budget accounted for by those four items, In certain years, emergency unconditional increases in other sectors’ shares have been limit- cash transfers drive social assistance expendi- Government administration, education, and ed. Spending on development priorities, such as ture temporarily higher. On several separate energy subsidies and infrastructure have health and social assistance spending, has risen occasions—during the 2005-06, 2008-09, 2013, remained priority items in annual budgets over the years but in relative terms received only 2014, and 2015 fiscal years—the GoI revised its through 2016 (Figure 1.4). While government marginally more in 2016 than they did in 2004. FIGURE 1.4 Sectoral composition of national expenditure, 2004—16 (Percent of national expenditure, %) Government General Admin Interest payments Education (Exc.SA) Health (Exc.SA) Infrastructure HH Social Assistance (SA) Subsidies (Exc.SA) Agriculture 30 25 20 15 10 5 0 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Note: National expenditure is the sum of central and sub-national government’s actual expenditure including subsidies and interest payment. 2015 sub-national and 2016 use budget data. Sources: World Bank staff calculations based on MoF data. 24 Chapter 1— Social Assistance Expenditure Analysis Indonesia does not have policy instruments The current administration has instituted an executed social assistance sector received IDR 75 triggered by business cycle events. Fiscal expenditure-stabilizing energy subsidy reform trillion in 2015 and an IDR 73 trillion in 2016, rules, the structure of public revenues and ex- accompanied by significant reallocation of re- compared with IDR 56 trillion in 2014. penditures, and the format of the budget for- maining expenditures to development prior- mulation and revision procedures in Indone- ities including social assistance. The subsidy At the same time, Indonesia’s unaddressed sia, all constrain the disbursement of public policy revision (effective from January 1, 2015) challenges in revenue generation could par- expenditures. These include disbursements on eliminates subsidies for low-octane gasoline and tially eliminate the fiscal savings from energy social assistance transfers and other social pro- introduced a fixed per-liter subsidy for diesel, so subsidy reform. While expenditures immedi- tection instruments, such that they cannot be net-of-subsidy prices for both fuels now track ately became less volatile upon the most recent “conditional” on non-budgetary events (such international oil price movements (adjusted by energy subsidy reform, Indonesia’s debt levels as an increase in prices or an increase in lay- nominal exchange rates). As a result, and in con- have stabilized at low levels, despite a sharp offs), as they are in other countries that have, trast to previous energy subsidy policy revisions uptick in 2016. Furthermore, while overall for example, unemployment insurance. Once a that kept domestic energy prices insulated from debt- and fiscal-management practices have re- budget has been agreed (or revised and agreed), international oil price fluctuations, the fuel por- mained prudent (Figure 1.5), domestic revenue no events external to the legislative-budgetary tion of the GoI’s subsidy bill fell sharply to IDR growth continues to decelerate. For example, cycle can then determine expenditure magni- 43.6 trillion (0.35 percent of GDP) in 2016’s revised domestic revenue growth was down slightly to tudes for any program, initiative, or transfer. budget, from the IDR 391 trillion (3.8 percent 7.6 percent in 2015 from 8.0 percent in 2013. In While social assistance spending has often in- of GDP) allocated in 2014 (World Bank, 2017a). 2015, the trend reversed and revenues fell by 0.5 creased upon the enactment of subsidy reform, Although, the implementation of the new fuel percent. For 2015, the decline in overall reve- the connection was purely political and there- pricing system has been uneven so far, the 2016 nue growth was due to a range of factors: slow- fore negotiated and uncertain—not automat- budget sustained the 2015 reforms. The recent er nominal GDP growth; declining commodity ic—as was the subsidy reform itself. Further- removal of the electricity subsidy to 18.9 million prices (particularly crude oil prices); and lower more, beneficiaries for these compensatory non-poor households with 900 VA connections oil lifting. In addition, various tax policy revisions social assistance transfers were pre-determined is estimated to save IDR 15 trillion in 2017.18 The and the implementation of a mineral export ban and receipts of the transfer did not depend on fiscal space unlocked by these reforms has al- (effective January 2014) contributed to the decline the beneficiary experiencing an event (as in lowed expenditure reallocation toward the GoI’s in revenue growth (World Bank, 2015b). unemployment). development priorities, including infrastructure, agriculture, and social programs. The centrally Based on recent discussions, the government may revise the 18 target and add 2.4 million new customers as eligible for the subsidy FIGURE 1.5 Indonesia’s public debt has stabilized at low levels (Percent of GDP, %) Domestic debt External debt 48.3 41.2 34.3 31.9 30.6 22.0 43.6 37.5 17.0 33.9 15.4 16.0 30.2 11.8 11.9 26.7 12.5 11.2 10.2 11.1 10.1 9.8 22.3 19.5 17.5 16.3 14.8 14.8 14.0 13.6 13.1 12.8 12.7 12.7 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Source world bank staff calculations 25 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 87% Central government ministries and agencies a confusing landscape. Many overlapping gov- remain primarily accountable for deliver- ernment-wide systems and agencies with M&E ing social assistance programs. Central gov- responsibilities, spanning sectors and levels of ernment spending accounts for more than government, have different spheres of influ- 87percent of total social assistance spending. ence and only partially overlapping informa- Essentially, all social assistance programming tion needs (World Bank, 2012j). covered in this report is planned, executed, and While the central government spends the most, implemented by the central government. Provinces and districts allocate a small districts have been increasing their spending amount of their own resources on social as- and allocate so­ cial assistance expenditures for program ad­ ministration, including civil servant Social assistance program implementation sistance spending. Social assistance spend- salaries, in support of central government remains highly fragmented across ministries ing by sub-national governments accounted programs and agencies. At the central level, the respon- for around 1.2 percent of total sub-national sibility for executing these major programs has spending over the past five years. Case studies been shared by six central institutions, while from the Social Assistance Public Expenditure the remaining central social assistance expen- Review 2012 indicate that districts—which ex- ditures were distributed across 12 ministries, 12 ecute the majority of the sub-national social programs, and more than 87 activities (World assistance expenditures (Figure 1.6) but which Bank, 2012j). More recently, the government have few discretionary resources—allocate so- has begun efforts to slim down the number cial assistance expenditures for program ad- of Ministries executing the main household ministration, including civil servant salaries, targeted social assistance programs to MoSA in support of central government programs.19 (PKH, Rastra, ASLUT and ASODKB), MoEC & Recent analysis shows that significant increas- MoRA (PIP), MoH (PBI-JKN) and Kementeri- es in spending by local governments over the an ESDM for the energy subsidy). Similarly, past decade are not associated with any im- M&E of social assistance programming, as well provement in outcomes (see Development Pol- Based on recent discussions, the government may revise 19 the target and add 2.4 million new customers as eligible for as expenditures oversight in general, remains icy Review 2014), measured broadly. the subsidy FIGURE 1.6 The majority of social assistance spending is mostly Central (IDR TLN) implemented by central government (national Province (IDR TLN) expenditure on social assistance by level of District (IDR TLN) government, nominal IDR trillion) % Central SA Exp. of National SA Exp. (RHS) 100 TLN IDR 100% 93.1 92.5 89.2 89.7 87.1 88.6 88.1 87.1 86.2 86.4 87.0 86.7 85.9 85.8 85.0 68.8 66.4 44.8 38.1 35.7 31.6 31.5 30.3 15.6 16.7 7.9 2004 2006 2009 2008 2005 2007 2010 2014 2016 2015 2013 2012 2011 Source MoF, Bappenas, and World Bank staff calculations. Note 2015 sub-national and 2016 uses budget data. 26 Chapter 1— Social Assistance Expenditure Analysis FIGURE 1.7 Social protection expenditures by intervention, 2006—16 12.3 15.4 1.4 1.4 12.6 12.1 12.3 11.93 1.5 1.27 11.3 1.4 99.6 1.48 10.5 1.1 10.5 10.7 1.39 1.18 1.25 98.4 82 90.1 8.27 0.99 HH Social Assistance 69.2 (Cluster I+BLT/BLSM) (IDR 61.9 trillion) 76.7 78.3 37.3 48.5 52.7 Social Insurance (IDR 60.9 57.8 trillion) 23 Total Central Govt 27.4 Exp.(%) 38.7 29.4 35.2 30.6 as % of Nominal 27.4 26.1 GDP(%) 14.2 Source MoF and World Bank staff calculations and estimates. Note Based on data available at the time of 2006 2016 writing. The current administration be- poverty rate is falling, those that social assistance expenditures (in- As part of the revised 2015 gan to modify access to social as- remain impoverished are by defi- cluding non-permanent spend- budget, a card-based beneficiary- sistance programs. As part of the nition the most difficult to bring ing on temporary cash transfers identification system was instituted within the existing social revised 2015 budget, a card-based out of poverty. Entrenched, per- accompanying energy subsidy assistance programs. beneficiary-identification system sistent poverty may require addi- reductions). On the whole, central was instituted within the existing tional programs or more generous government expenditures on so- social assistance programs. This benefits (or both) to overcome.20 cial protection—social insurance 01 KKS included: (i) a Family Welfare In addition, natural disasters and plus social assistance—has re- Card (KKS), which identifies 15.8 macroeconomic instability come mained relatively low, accounting million households eligible for part and parcel with Indonesia’s for 1.52 percent of GDP in 2015 receiving an unconditional cash geographic location, and econom- (Figure 1.7). Family Welfare Card transfer (BLSM); (ii) the School ic and financial openness. Such Cash transfer Program (PIP), crises can instantaneously wipe Design and operational reforms which identifies 19.5 million chil- out years of poverty-reduction within programs can help social 02 PIP dren eligible to receive cash trans- progress. Indonesia needs to safe- assistance benefits reduce pover- fers to cover education costs; and guard that progress by planning ty more efficiently regardless of (iii) the Healthy Indonesia Pro- for an SA response when crises the level of spending. It is clear gram (PIS), which identifies over become acute. Both stubborn pov- that, with less than one-quarter of School Cash transfer Program 92 million individuals eligible erty and susceptibility to natural social assistance benefits reaching to receive premium fee waivers disasters and international insta- poor households, social assistance for the JKN health insurance pro- bility, therefore suggest that social expenditures could reduce pov- gram. The increase in social assis- assistance expenditures should erty more effectively if program- 03 PIS tance spending accompanying the rise beyond current levels, even af- and system-wide reforms resulted 2015 energy subsidy revisions (see ter current programs are revised, in more poor households receiv- above) was directed primarily to so that they deliver more of bene- ing available benefits. The follow- PIP, PIS, and PKH. The first was in- fits to more poor households. ing sections in this report offer Healthy Indonesia Program tended to nearly double coverage greater detail on how and why by adding about 10.5 million new Social insurance expenditures increased coverage has only led student beneficiaries; the second still exceed social assistance ex- to increased shares (of total social added another 2 million benefi- penditures, though far fewer assistance resources available) in ciaries (after having just added 10 households and individuals are some programs. They also suggest million new beneficiaries between covered by these programs. Al- how integration and better pro- 2013 and 2014); and the third though social assistance expendi- gram design can remedy current increased coverage by another tures grew faster (at a 29 percent shortcomings to make social assis- 200,000 households. cumulative growth rate) than tance expenditure a more efficient social insurance spending (12 tool in poverty reduction. New programs for currently un- percent) between 2011 and 2015, Section 3 of this report provides more de- 20 covered key risks will raise an- social insurance expenditures are tail on, and potential solutions for, key risks and vulnerabilities not currently covered by ticipated expenditures. When the still 30 percent larger than total social assistance programs. 27 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 ANNEX 1 Central government expenditure on social assistance programs, 2004—16 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Unconditional - 4,487 18,619 - 13,966 3,733 - - - 9,300 6,200 9,470 - cash transfer (BLT/BLSM/ KKS&KSKS) Subsidized rice 4,831 6,400 5,300 6,600 12,100 13,000 13,925 15,270 20,926 21,497 18,165 21,846 22,077 (Rastra) Health insurance - 1,300 3,074 4,567 4,448 4,620 4,763 6,300 7,300 8,100 19,900 19,884 24,815 for the poor (Jamkesmas/JKN- PBI/SJSN/KIS) Cash transfer for - - - - 1,238 2,562 3,607 4,700 5,400 14,100 6,600 6,388 10,572 poor and at risk students (BSM/ PIP) Conditional cash - - - 605 946 1,068 1,123 1,600 1,900 3,600 5,200 6,324 8,542 transfer (PKH) Child social n/a 104 211 187 311 296 254 256 306 339 345 462 294 services (PKSA) Disabled social n/a 65 130 152 190 217 209 70 79 79 79 531 351 services (JSPACA/ ASODKB) Elderly social n/a 26 53 57 69 82 75 48 64 64 64 201 142 services (JSLU/ ASLUT) Other social 1,899 180 197 295 297 302 352 1,944 2,743 3,871 1,268 3,879 8,591 protection (SP) Total Central 4,831 12,382 27,387 12,169 33,269 25,578 23,956 28,244 35,975 57,079 56,553 72,062 72,908 Social Assistance by major programs Total Central 6,730 14,028 29,411 14,228 35,263 27,472 26,127 30,646 38,718 60,950 57,821 76,738 77,356 Social Assistance (nominal)* Total Central 13,227 16,578 16,398 19,574 25,156 25,650 26,127 28,517 34,724 52,077 46,878 56,960 58,815 Social Assistance (real) National 7,919 15,644 31,575 16,560 38,125 31,536 30,298 35,736 44,817 68,822 66,423 87,110 89,249 expenditure on HH SA (nominal) National 15,654 27,181 47,980 22,782 45,034 34,074 30,298 33,253 40,195 58,803 53,852 67,758 67,857 expenditure on HH SA (real, 2010 prices) National 0.32 0.53 0.89 0.39 0.73 0.53 0.44 0.46 0.52 0.72 0.63 0.75 0.72 expenditure on HH SA (% of GDP) National 1.77 2.96 4.68 2.08 3.65 3.11 2.72 2.60 2.82 3.73 3.41 4.19 3.81 expenditure on HH SA (% of total central + SNG) * total central social assistance by major programs includes "other MoSA" social assistance expenditure, Source 2004 – 2011: World Bank 2012h. 2012- 2016: Ministry of Finance, and World Bank staff calculations. Note 2015 sub-national and 2016 (both central & sub-national) use realized data. 28 Chapter 1— Social Assistance Expenditure Analysis ANNEX 2 Targeted beneficiaries of five major social assistance programs, 2008—16 2008 2009 2010 2011 2012 2013 2014 2015 2016 Unconditional Cash 18.7 18.7 - - - 15.5 15.5 15.8 - Transfer (BLT/BLSM) —households (millions) Rice for the poor (Raskin/ 17.5 17.5 17.5 17.5 17.5 15.5 15.5 15.5 15.5 Rastra) —households (millions) Health insurance for the 76.4 76.4 76.4 76.4 76.4 76.4 86.4 88.2 92.4 poor (Jamkesmas/JKN- PBI/SJSN/KIS) —People (millions) Cash transfer for poor 4.6 4.9 5.8 8.2 9.5 16.6 11.2 20.37 19.7 students (PIP) —students (millions) Conditional Cash Transfer 0.72 0.72 0.81 1.11 1.51 2.4 2.8 3.5 6 (PKH) —poor families (millions) Source Ministry of Finance (2008-13; 2015-16), Bappenas (2014). P.29 2 2 .1 U N C O N D IT IO N A L C AS H T RA N S FER ( BLS M ) P.31 C H A P T E R 2.2 S U BS ID IZ ED R IC E FO R T HE PO O R ( RAST RA ) P.37 Main 2 .3 Portfolio S U BS ID IZ ED S O C IA L HEA LT H IN S U RA N C E of Social ( J KN -PBI) P.47 Assistance Programs 2 .4 C AS H T RA N S FER FO R PO O R & AT -R IS K STU D EN TS ( PIP) P.5 5 n Indonesia, targeted social assistance i interventions (non-contributory), which transfer resources (in-kind, cash, or ser- vices) to particular at-risk groups in- clude the following main programs: (i) This section describes 21 2 .5 the main social assistance the unconditional cash transfer program programs described in (BLSM); (ii) the largest in-kind transfer program (Rastra); that represent 99 percent of total Social assistance C O N D IT IO N A L C AS H (iii) the health insurance fee waiver program (JKN-PBI); (iv) budget (See Annex Tables). The latter two cash T RA N S FER ( PKH) the educational cash transfer for poor and at risk students transfers are not discussed P.6 3 specifically in this review program (Program Indonesia Pintar/PIP) directed to poor as they are very low in and low-education individuals in primary (SD), junior sec- coverage and have not been planned for future ondary (SMP) and senior secondary (SMA) or equivalent scale up or transformation. Importantly, the old age education; (v) the conditional cash transfer program (PKH); cash transfer (ASLUT) may and (vi) small cash transfer programs for vulnerable chil- however be included in PKH, the extent to which is as of dren, disabled and vulnerable elderly.21 yet unclear. 31 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 IDr 6.5k Unconditional Cash Transfer (BLSM) BLSM has a clear objective, namely to In 2013, it was estimated that an increase in temporarily protect welfare in times of the fuel price of premium gas to IDR 6,500 (US$0.50) per liter would increase the poverty anticipated macroeconomic stress, which headcount rate by 1.5 percentage points it achieves efficiently. Current BLSM without any BLSM compensation operations have improved slightly on previous iterations by delivering larger shares of available benefits to targeted households. However, non-targeted households still receive substantial bene- fits from BLSM, while from a household- and community-based perspective BLSM accessibility still remains difficult. BLSM provided cash assistance to households affected by an economic shock. BLSM (2013) added cash amounts to a (25-percent-poorest) household’s budget equal to about 11 percent OVERVIEW of regular expenditures.23 Average fuel prices in June 2013 increased by 33 percent. While BLSM-targeted households consume little fuel directly, fuel price increases are passed on to other economic sectors, especially food The unconditional cash transfer program and transport, which account for significant (Bantuan Langsung Sementara Masyarakat, shares of expenditure. It was estimated that or BLSM) has a clear objective: to supplement an increase in the fuel price of premium gas consumption for poor households facing an- to IDR 6,500 (US$0.50) per liter (which would ticipated, policy-based price increases. In have represented a 44 percent increase for that late June 2013, and again in November 2014 fuel type at that time) would increase the pov- and 2015, the Government reduced existing erty headcount rate by 1.5 percentage points fuel subsidies and compensated poor and without any BLSM compensation (World Bank, near-poor households for the subsequent rise 2012j). in fuel, food, and transport prices with a tem- porary unconditional cash transfer. It was ex- Positive experiences with BLT/BLSM continue pected that the BLSM transfers would be large to outweigh negative experiences. Indone- enough (in terms of both coverage and amount sia has several years of experience with direct transferred) that the “regular” pace of poverty emergency cash transfers: in 2005, subsidy reduction would not slow when fuel and econo- cuts raised household fuel prices by an aver- my-wide prices spiked as a result of the subsidy age of over 125 percent and the Government reduction. responded with a BLSM-like transfer (Bantuan Langsung Tunai, or BLT). Again, in 2008, when BLSM reached households everywhere in In- international crises in both financial markets donesia. In 2013, about 15.5 million were tar- and in food prices combined with another do- geted to receive IDR 600,000 (about US$53) in mestic reduction to fuel subsidies, the Govern- two phases for a total cost to government of ment released another emergency direct cash IDR 9.3 trillion (US$864 million).22 BLSM was transfer (again called BLT in that year) (World in theory funded partially from the implied Bank, 2012c). While the political and social de- budgetary savings from subsidy reductions. It bate over the suitability of unconditional cash transfers for Indonesian households has con- 2.1 was targeted to the poorest 25 percent of Indo- nesian households that, because of consump- tinued to be lively, the incidence of negative tion patterns, were receiving only small shares social impacts associated with BLT/BLSM has of resources transferred via the Government’s been on the decline. Judicious local interven- energy subsidy program and were therefore tion—village leaders actively re-allocating BLSM most at risk from the negative impacts on con- benefit pools to defuse protests and negative dis- sumption from price increases. ruptions—may be responsible for this decline. For Indonesian Rupiah conversion into US$ the October exchange rate value of each year is used (except 2015). 22 Bank (2016) shows that in 2012, fuel and electricity subsidies received by an average poor or near-poor household were valued 23 at slightly less than 10 percent of total consumption expenditure. So BLSM 2013 provided a direct transfer which was not smaller than the transfer previously received indirectly through purchases of subsidized energy. Additionally, energy remained subsidized (though at a lower rate) after the BLSM cash transfer was distributed. 32 Chapter 2 – BLSM <50% “… the total transfer executed under BLSM FIGURE 2.2 (2013) had a magnitude of less than half of the total BLT I transfer” Institutional responsibility & flow of funds (as of 2015) FIGURE 2.1 BLT/BLSM targeted Expenditure (LHS)(IDR Billion) Target no. of beneficiary 1 MoSA manages targeting coverage & expenditure households (Million) and creates registry of families IDR Billion Million 19 19 20000 18 15.5 15.5 16 20 2 Ministry of Finance 15.4 releases funds to PT. Pos 15000 15 3 18619 PT. Pos cooperates with 9470 sub-district authorities 9300 10000 10 and facilitators to 13966 organise distribution 6200 3733 4487 5000 5 2005 2006 2008 2009 2013 2014 2015 Source Ministry of Finance, Bappenas. 0 SUB cuted by the Ministry of Finance. Subsequent DI ST RI CT FACI L I TATO R funding distribution to recipient households PROGRAM SIZE, is likely to continue to be achieved via the na- INSTITUTIONAL SET- tional postal service’s branch network (PT. Pos) as well as the banking system. Coordination at ( T KSK ) the provincial and district levels is facilitated Support Verification UP, ELIGIBILITY, & by provincial and district governments, as well as BLSM facilitators (Bappenas, 2014). Figure Coordination BENEFITS 2.2 below summarizes BLSM flow of funds and task management. SUB DIST RICT BLSM (2013 and 2014) covered households Roughly the poorest 25 percent of Indone- AUT HORIT IE S Coordinate in all provinces and districts, though it was sian households qualify for BLSM. Eligible Villages smaller in scope than previous BLT programs households use their Kartu Perlindungan Sosial (Figure 2.1). The 2005/6 BLT program provid- (KPS) or Kartu Keluarga Sejahtera (KKS) (see Box ed a per-household transfer of IDR 1.2 million 2.1 below) social protection cards to prove el- V I L L AG E (US$122) to about 17 million households; BLT igibility for BLSM and other social programs. Households receive KKS cards via the national AUT H O R I T I E S II (2008/9) provided IDR 900,000 (US$80) to Verify Targeting about 19 million households; and the BLSM postal service (PT. Pos) after having been veri- Support Delivery program provided IDR 600,000 (US$53) to fied as poor or vulnerable by the national reg- about 15.5 million households. In other words, istry (UDB). KPS/KKS holders should retrieve the total transfer executed under BLSM (2013) their BLSM transfers at the nearest post of- had a magnitude of less than half of the total fice. Should there be another BLSM launched, BLT I transfer. households will need either a KPS card or a “Developing Productive Families” KKS card. BLSM is a diffuse program with nearly com- As of 2017 and in coming years, an increasing plete delegation of sub-processes. The Min- share of households may be able to receive ben- istry of Social Affairs (MoSA) is the key pol- icy and executing agency for BLSM, with that efits electronically. P O ST agency’s sub-district social welfare workers (Tenaga Kerja Sosial Kecamatan) facilitating BLSM (2013) benefits provided a boost to bene- ficiary household in the target group of about OFFICE Distribute Cards the distribution process. The dissemination 11 percent; average monthly household expen- Disburse cash of information materials is undertaken by ditures in the poorest 25 percent of households the Ministry of Communications and Infor- in late 2013 were equal to about IDR 1.3 million mation, while funding disbursement is exe- (US$123). 33 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 BOX 2.1 “Developing Productive Families”, Kartu Keluarga Sejahtera (KKS) & Kartu Simpanan Keluarga Sejahtera (KSKS) The KKS and KSKS introduced by the current Government as part of the new SA scheme are meant to give households access to social assistance programs such as Rastra, a rice subsidization program, and BLSM, an unconditional cash transfer (UCT) program that was STEP 1 launched in November/December 2014 in response to a reduction in Take the KSKS SIM the fuel subsidy. The KKS and KPS have been used to disburse BLSM card off the card transfers electronically and in cash, using the post office (PT. Pos) and a state-owned bank, Bank Mandiri, respectively. Coverage & Eligibility By November 2014, 1 million KKS and KSKS cards had been disbursed to families targeted in the first phase of the program’s implementa- tion. These two cards are intended to cover 15.8 million households, STEP 2 and will be replacing the existing social protection card, the KPS. The KSKS SIM card is KSKS card contains a cellular phone SIM card to which UCT benefits put into the phone are transferred for some households (Bappenas, 2014). Targeting for the entire range of cards is done using the Unified Database (UDB), a registry of poor households employing a proxy means test meth- odology to rank households in terms of predicted expenditure. To be eligi- ble for the KKS, a household must be considered poor or near-poor, name- ly, being in the poorest 25 percent of households according to the UDB. STEP 3 Program Flow & Benefit Structure Total savings status is communicated via Through November and December in 2014, with the coordination of pro- SMS vincial and district level authorities, the 1 million cards were sent to PT. Pos offices at the village level where households were to pick up their cards, exchanging their KPS for the KKS. The remaining 14.8 million households were to continue using the KPS card to access KIP and KIS until they re- ceived a KKS and KSKS card STEP 4 olders of the KKS or KPS card received the BLSM UCT transfer in No- The SMS and the housing vember/December 2014 to account for higher prices experienced by of the KSKS SIM card are households due to the reduction in the fuel subsidy. Over those 2 months, to be presented at the households received IDR 200,000 per month (US$16). Out of 15.8 mil- nearest post office to lion household recipients of KPS and BLSM, 1 million were to receive the withdraw funds from the payments via a program called E-money which was accommodated by the savings account. KSKS card, detailed above.24 The other 14.8 million households received the UCT by presenting their KPS card at the nearest post office. As of early February 2015, 93 percent of BLSM benefits had been claimed by beneficiaries. To date, there is no way of telling how many of those benefits were claimed through the use of the SIM card itself, since benefits could also be claimed by scanning a bar-code on the KSKS package at the post office. By late 2015, local media reports were suggesting that around one-quarter of the recip- ients were leaving some money in their account rather than withdrawing all of it. Holders of the KSKS could access their e-money account by withdrawing it in cash at the post office, but they could potentially use applications on their phone to do transfers, bill payments, and even ATM withdrawals. For rural areas where banks or post offices are far away, this capability is particularly innovative and marks an important step in moving toward greater financial inclusion in Indonesia (TNP2K, 2014k). In 2017, the KKS is being equipped with a magnetic strip to allow for cash withdrawal from E-money was facilitated by Bank Mandiri and three major phone network operators. 24 ATMs, bank agents, and pre-determined locations for specific social as- As opposed to a PKH e-money pilot in 2014, the SIM cards are pre-activated and are valid until December 2019, indicating the SIM card could be used to accommodate future sistance transfers. For others, the KKS can be used to purchase subsi- e-money initiatives. dized Rastra rice at pre-determined locations under the Governments’ Source TNP2K, 2014k and local media reports e-Warong program. 34 Chapter 2 – BLSM FIGURE 2.3 FIGURE 2.4 BLT/BLSM coverage by 2005 BLT/BLSM incidence by 2005 COVERAGE, 2008 2008 expenditure quintile (percent) expenditure quintile (percent) 2013 2013 TA RG E T I N G , & I M PA C T S POOREST RICHEST POOREST RICHEST 20% 20% 20% 20% BLSM (2013) covered fewer people overall but 50 50 has improved on the share of resources dis- tributed to poor and near-poor households. Overall coverage in previous emergency un- 41 40 conditional cash transfer programs (2005/6 or 37 2008/9) was higher at about 27 percent com- 36 35 35 pared with about 21 percent in BLSM (2013). While all households regardless of income 28 rank were covered at lower rates in 2013, the 27 27 26 26 poorest 20 percent “lost” the fewest BLSM 25 households—there were 17 (41) percent fewer 20 poorest (richest) quintile households covered 19 19 18 18 17 in 2013 relative to the 2008/9 BLT (Figure 2.3). 13 13 12 12 As the number of covered households in the poorest quintile fell the least (in between 7 7 2008/9 and 2013 iteration of the BLT/BLSM 5 5 4 4 program), this has led to further increases in Source: Susenas (various years) and World Bank staff calculations. the share of available benefits received by that FIGURE 2.5 same poorest quintile (Figure 2.4). While the total benefit pool is smaller in 2013 (than either BLT/BLSM coverage by 2005 2005/6 or 2008/9), the poorest 20 percent of 2008 expenditure quintile (percent) 2013 the population have seen their share increase by about 10 percent (from 2005/6). Coverage (% of households Incidence (% of total receiving benefits) beneficiaries by group Leakage to non-targeted populations is still significant in BLSM. In order to facilitate com- parisons between BLSM and other social assis- POOREST RICHEST POOREST RICHEST 20% 20% 20% 20% tance transfers (which may have slightly dif- 53 51 ferent target groups) Figure 2.5 shows coverage 46 44 and incidence for the “Poor”, the “Next 30” per- 41 38 cent, and “The rest”.25 The figure illustrates that, 36 35 33 32 32 31 while the poverty headcount rate fell by about 30 26 24 4 percentage points between 2008 and 2013, the share of BLT/BLSM benefits accounted for 16 15 by the “Poor” group fell by only 2 percentage 11 points. The “Next 30” group, which contains Source: Susenas (various years) and World Bank staff calculations the same proportion (30 percent) of near-poor households in every year, gained the most in terms of share of available BLSM benefits. Fi- nally, while “The rest” grew larger (by the same 4 percentage points that the poverty headcount fell), this group’s share of BLSM benefits stayed BLSM in 2013 covered fewer roughly constant. Together, these patterns indicate that over the years it has been exe- people overall than in previous cuted, about two-thirds of the BLT/BSLM ben- efits available have gone to the poor and near- years but has improved on poor populations. However, BLSM has not yet the share of resources distrib­ stemmed leakage to non-targeted populations. uted to poor and near-poor 25 The “Poor” are all households with per-capita expenditure below the relevant year’s poverty-line expenditure; the “Next 30” are the 30 percent of Indonesian households with the households. lowest per-capita expenditure levels who are not counted as poor; and “The rest” are those households not “Poor” or part of the “Next 30”. As headcount poverty rates have been declining in Indonesia—from nearly 16 percent in 2005 to just over 11 percent in 2013—the number of households in the “Poor” and “The rest” groups will change (while the “Next 30” is always the 30 percent with the lowest per-capita expenditure amounts who are nonetheless not poor). This definition will be used in the analysis of all main SAP described in this section. 35 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 TABLE 2.1 On average, BLSM is distributed to house- holds that exhibit correlates of income pov- Characteristics within erty. BLSM has a large overall coverage target Indonesian populations, 2014 (25 percent of the population) but still fails to distribute benefits to over half of the very poor % of all Indonesians who: % poor population who: % of BLSM recipients households in the first decile. However, on av- who erage, BLSM transfers are being delivered to households that exhibit most of the non-in- Do not have access to 18 31 27 bottled, tap or well water come correlates of poverty, such as relatively low access to clean water, sanitation, and ed- Do not have access to 27 53 54 ucation (Table 2.1). The distribution of BLSM private sanitation shows no significant differences (in incidence) for rural versus urban households (not shown). Live in rural areas 50 62 66 Live with more than 5 25 44 31 Previous BLSM-like transfers have protected household members the worst-off households.. BLT prevented con- sumption expenditures in poor households Have not completed 10 13 15 from being negatively affected by fuel subsi- primary education dy reductions. BLT transfers were used to buy Are illiterate 8 14 14 basic necessities (especially rice); on one-time costs such as school fees or clothes for Idul Work in the agriculture 34 58 50 Fitri holidays; or on transportation. In addi- sector tion, there was a spillover effect: the BLT pro- Source Susenas 2014 (to capture late 2013 into early 2014 transfer) and World Bank staff calculations. ‘Work in…’ refers to gram actually helped to stimulate an increase share of working individuals, not all Indonesians. in spending among non-recipient households. Heads of households who received BLT were not more likely to leave work. On the contrary, BLT households were more likely (by 10 per- centage points) to report that they had found FIGURE 2.6 new jobs and moved into employment, perhaps using the BLT money for job-specific transpor- Social disharmony in Violent incidents tation or childcare. Deaths different eras Injuries Damaged building BLSM saw in 2013 further reductions in so- cial disharmony from the introduction of a BLT I BLT II BBM PROTEST OTHER BLSM valuable cash transfer. Only nine incidents 2005 —06 2008 —09 JAN. —OCT. PROGRAM JUL. —OCT. of non-fatal violence were reported (Figure JUL. 2.6), while cumulative social disharmony (of 2013 2013 —OCT. any type) continued to fall from peak levels 2013 during BLT 2005/06. Nonetheless, about 80 193 percent of a set of villages studied during and after BLSM 2013 reported some unfair exclu- sion from BLSM. Protests and negative impacts on relationships between leaders and citizens occurred in one-third of these villages. In such villages, local authorities were often blamed for BLSM’s inclusion and exclusion errors. Pro- tests and conflict tended to arise in areas where relations between inhabitants and local lead- ers were already strained prior to the launch of BLSM. Local leaders opted to manage such tensions by either not participating in the pro- gram, or by sharing out BLSM benefits.26 74 67 Decreases in BLSM-related tensions may be 52 due to local management and control rights. 45 45 Over one-quarter of BLSM recipients received 13 less than the stipulated amount at least once, 10 8 9 6 and one-fifth reported transfer reductions in 6 3 4 0 both tranches.27 Of the 27 percent that expe- Source World Bank 2015a rienced at least one reduced BLSM transfer, 26 Based on data from a Susenas-based Social Protection Module from March 2014 where households are asked about their participation in the 2013 BLSM rounds in June – August and September – December. 27 During two previous instances of BLT, there were reports that after cash transfers were collected by recipients some PT. Pos officials and community officials charged “fees.” This occurred 10 percent of the time in 2005, and 46-54 percent of the time in 2008-09. This was typically done to re-distribute to households that were not included in the list (due to miss-targeting or otherwise) and to subsidize collective transportation and identity card costs. 36 Chapter 2 – BLSM about three-quarters (73 percent) noted that Recent studies reveal shortcomings in BLSM cifically with respect to BLSM targets likely un- the reason given by those reducing the amount information dissemination and awareness dermined the program’s overall effectiveness, (most often village or sub-village heads) was campaigns. A nationwide information cam- as village heads frequently re-allocated BLSM for the purpose of sharing the BLSM benefits paign was produced and delivered before the transfers (World Bank, 2015a).30 Village leaders among those in the village who were not target- first BLSM transfers were made. However, also chose not to support BLSM in some cas- ed but deemed eligible.28 That is, village leaders research has revealed that BLSM facilitators es. Increasing the scope for community-based actively re-allocated BLSM benefit pools to pro- had little effect on raising awareness of the targeting methods to complement to the use of vide reduced transfers to those who believed program’s design, aims, beneficiary selection, the national registry prior to the launching of they were eligible but did not receive BLSM, in or beneficiary responsibilities. Community BLSM would likely help reduce targeting errors order to defuse protests and negative disrup- members questioned the transparency of the and direct dissatisfaction with the program tions. This practice is referred to as “bagi rata” beneficiary selection process and would lodge away from local authorities. (equal sharing) and it is done at the district lev- their complaints about wrongful inclusion or el through reallocating quotas, rotating access exclusion with village leaders. However, village While information campaigns were launched to various programs over time and, based on leaders often knew little about BLSM or benefi- and information materials were distributed, survey data, and reducing the benefits received ciary selection logic, and they generally preferred the qualitative and quantitative findings in- by targeted beneficiaries (World Bank, 2015a).29 not to become involved (World Bank, 2015a). dicate that the information provided and the Linking anecdotal with survey evidence, the manner in which it was provided were inad- dilution of the BLSM benefits is not a hidden BLSM’s accessibility has not improved over equate. In future rounds of BLSM, more time activity, nor is it perceived negatively. It seems previous emergency cash transfer itera- and resources should be used for socialization to be an accepted practice anchored in local tions. Despite a new card-based ID system for with clear structures of accountability commu- conceptions of social justice; one village offi- BLSM-eligible beneficiaries, the pathways by nicated to all actors. This process should occur cial explained: “people here have a saying: ‘no which households and individuals learn about well in advance of the launching of BLSM. This one’s above and no one’s below, everyone’s the and access the BSLM process remain the same. would likely have a positive spillover on tar- same”. Bagi rata is seen as a legitimate response Recipients are still informed of their eligibil- geting: the village-level targeting-error redress to community expectations, as according to ity through, and must retrieve BLSM funds process would be used more effectively if re- a sub-district official: “If they hadn’t shared from, PT. Pos. While the rate of “deductions” sponsibilities and program logic are agreed on, out BLSM, village heads here would have been from BLSM packages has retreated from its and absorbed by all stakeholders. finished”. By managing potential conflict this BLT 2008/9 peak (when nearly half of transfer way, negative social impacts may have been packages had fees deducted or were re-distrib- reduced, while targeting accuracy and benefit uted), beneficiary control over transferred re- delivery suffered instead. BLSM funds reached sources remains weaker in 2013 than in 2005/6. almost all recipients on the beneficiary lists, Roughly one-quarter of BLSM beneficiaries in but local officials increasingly extracted “fees” the 2013 round indicated that deductions were from beneficiaries (World Bank, 2015a). taken from their transfers (see above). There is no evidence indicating BLSM-like transfers undermine social capital. Critics in Indonesia have argued that significant, non-universal cash transfers erode social cap- CHALLENGES, 28 World Bank staff calculations, 2015. While the most common intent, namely to redistribute benefits more equally, of BLSM deductions and the most common identity of those ital (as proxied by, for example, semi-voluntary community improvement projects). However, R E C O M M E N D AT I O N S making deductions, namely the village or sub-village head, were the same in previous BLSM iterations in 2005 and 2008/9, the deduction rate, at 27 percent, is higher than the 2005 BLT but lower than the 2008/9 BLT; see World Bank no research to date has clearly proven that the effects of cash transfers undermine social capital. & MOVING FORWARD (2012c). 29 The first two trends were found in 19 out of 24 districts surveyed in a forthcoming qualitative study. 30 Such that more local households received a lower-valued transfer.. Compensatory UCTs have been shown to be ef- fective in protecting poor Indonesian house- ACCESSIBILITY holds from anticipated price shocks. From March 2013 to March 2014, the headcount pov- erty rate decreased from 11.4 to 11.3 percent. BLSM does not track program processes and If there had not been a BLSM (with a transfer outcomes; potentially eligible but exclud- budget of about IDR 9.4 trillion), the headcount ed households do not have a clear recourse. poverty rate would likely have remained flat or Beyond a little-used community verification increased. At least in the short run, use of BLSM process that helps village authorities address has protected the purchasing power of many targeting errors, BLSM has no mechanism, poor and vulnerable households (World Bank, tool, or protocol to report on its own perfor- 2014; World Bank, 2013b; World Bank, 2012c). mance (World Bank, 2015a). Households may lodge BLSM-related complaints via SMS to the Though the program’s ability to find priority “Lapor!” hotline, but evidence concerning this households has improved from previous iter- hotline’s use or the links from “Lapor!” reports ations (such as BLT I and BLT II), BLSM (2013) or management to BLSM-executing agencies or exhibits both exclusion and inclusion tar- partners is unavailable. geting errors. There are inherent difficulties with the UDB but inconsistent use of the local verification system to redress errors made spe- 37 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 Subsidized Rice and near-poor households are net consumers (rather than net producers) of rice, while rice for the Poor consumption accounts for about two-thirds (Rastra) of all food expenditures.31 Food prices are OVERVIEW more unstable than other economy-wide pric- es. Over a 15-year period, the average annual percentage change in the economy-wide and food-only price indices was 7.5 and 8.2 percent, Rastra has positive potential: the Food availability is an important issue for respectively. However, measures of price vola- consistent provision of a basic food poor households in Indonesia, as are food pric- tility were nearly twice as large in the food-only es. For poor Indonesian households, food expen- index. In other words, in Indonesia food price package could protect poor households ditures represent two-thirds to three-quarters highs are higher (and lows are lower) than from food-price volatility, calorie of the household budget. The majority of poor the general price level. In welfare terms, such scarcity, and malnutrition. However, in its operation, Rastra fails to achieve most of these fundamental social assistance goals. Dilution of benefits, FIGURE 2.7 missing rice, and hidden financing Vulnerability to food insecurity in Indonesia, 2015 burdens all reduce the transfer values provided to target households. Rastra reform has begun shifting towards cashing out the Rastra benefit, initially in areas with functioning rice markets, while the current operational model will likely remain in the more remote areas of Indonesia. 2.2 Source: adapted from World Food Programme. 2015. Food Security and Vulnerability Atlas of Indonesia.33 38 Chapter 2 – Rastra heightened volatility is riskier for households Sumatra, and remote districts in Sulawesi. The Indonesia. Rastra rice is purchased wholesale with consumption baskets weighted more most food insecure districts (Priority 1) are all by the state-owned National Bureau of Logis- heavily with food items. still found in Papua, and the majority of Prior- tics, Bulog, which then delivers rice to over ity 2 districts are in Papua and East Nusa Teng- 50,000 regional distribution points where it is Some regions in Indonesia do not produce gara (NTT).32 sold at below market prices. Through these op- enough calories for their local populations. erations, Bulog intends to stabilize the domes- The United Nations’ World Food Programme Rastra, previously called Raskin, allows tic rice price and to protect households from (WFP) estimates that about 12 percent of Indo- households to purchase rice at a subsidized food insecurity. By design, Rastra’s transfers of nesia’s 514 districts are food insecure. Figure rate. Originally developed to provide food important basic commodities may significant- 2.7 below provides a summary showing higher consumption assistance to households during ly increase household welfare, especially in food levels of food insecurity (light yellow, pink, and the Asian financial crisis (1997/98), Rastra insecure areas where regular markets cannot be red areas) in eastern Indonesia (especially Pap- was until very recently the largest permanent relied upon for a consistent supply of reasonably ua), the island districts off the western coast of household-based social assistance transfer in priced foodstuffs. Priority 1 Districts Priority 2 Districts Priority 3 Districts Priority 4 Districts Priority 5 Districts Priority 6 Districts In contrast, high-protein items like meat, fish, and diary represent about 10 percent of food expenditures. 31 As compared with the previous estimate of food insecurity in 2009, 67 percent of all districts saw an improvement in food availability. See WFP (2015) for more. 32 33 The WFP analysis divides 13 indicators into two sets: chronic food and nutrition insecurity and transitory food insecurity, these indicators are combines into a single composite indicator that ranks the priority level of districts. 39 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 FIGURE 2.8 PROGRA M SI Z E , Rastra targeted coverage & expenditure Expenditure (LHS) Household target (RHS) IN STITUTI ON A L S E T - IDR Trillion Millions UP , EL I GIBI LI T Y , & 25 20 BENEF I TS 19.1 18.5 17.5 17.5 17.5 15.5 15.5 15.5 15.5 While Rastra’s total target coverage has fallen recently, its nominal expenditures per target- ed beneficiary continue to rise (Figure 2.8). 12.1 13.0 15.2 16.5 19.1 21.5 18.8 21.8 22.1 From 2013 to 2016, Rastra’s policy authority (see below) assumed 15.5 million households would purchase the full Rastra subsidized rice pack- age and in those 4 years the Government allo- 2008 2009 2010 2011 2012 2013 2014 2015 2016 cated about IDR 21.5, 18.8, 21.8 and 22.1 trillion, respectively, for the program.34 While Rastra’s Source MoF financial note on the national budget 2008-12. 2015 and 2016 are realised budget. share of the social assistance budget has fallen from over half in 200935 to just about one-third in 2016, per-beneficiary spending has risen by about 60 percent over the same period. Rastra’s delivery procedures are complicated by the granting of meaningful managerial authority to sub-national actors. Figure 2.9 presents a stylized, simplified, and optimistic FIGURE 2.9. flowchart describing Rastra in operation: a list of eligible households (by name and address) Rastra targeted coverage & expenditure is generated by the unified database (UDB); and the list is then given to the national postal service (PT. Pos), which is responsible for dis- tributing social assistance program beneficiary cards (Kartu Keluarga Sejahtera [KKS], or Kartu Perlindungan Sosial [KPS]) to those on the list. Local governments also receive the list and PT. POS Pre–printed list of generate verifications of households together name and address UNIFIED of targeted with an official request for Bulog to distribute PT Pos sends households the statutory Rastra-rice amounts to local dis- KKS to targeted tribution points. Bulog complies by delivering the requested amounts of Rastra rice (for sale DATABASE households at a stipulated below-market price) to these dis- tribution points. In order to buy Rastra rice, (UDB) Quota & data households must possess their KPS/KKS card on targeted List of and proceed to their local distribution point households names and to complete the sale. As part of Rastra reform addresses of initiatives, after several pilot schemes in 2016, targeted households in 2017, 1.4 million beneficiaries in 44 cities are able to purchase rice using digital cash by way BULOG Distribute is published at the village office or distribution of the KKS card based e-wallet at e-Warong. The Rastra to all points distribution program is likely to continue undergoing sig- points nificant reform beyond 2017 (MoSA, 2017). However, officially and in practice, the final allocation and local distribution of Rastra LOCAL rice—everything occurring at and after the very bottom-most arrows in Figure 2.9—de- GOVERNMENT pends on the involvement of sub-village, vil- Issue letters to allocate Raskin DISTRIBUTION POINTS according to the quota for each village Households show 34 For Indonesian Rupiah conversion into US dollars the October exchange rate value of each year is used (except KKS to purchase Receive & check 2015). Rastra 35 In 2009, Rastra targeted over 18 million households with a rice quality before budget of IDR 13 trillion (US$1.35 billion) and accounted for 55 distributing Rastra percent of the total assistance budget in that year. Source: adapted from TNP2K 2014b 40 Chapter 2 – Rastra FIGURE 2.10 lage, sub-district, or possibly district, admin- 114 kilograms per year) (TNP2K, 2014b). There- istrations. At the furthest remove from local fore, a full Rastra package purchased every Urban/rural Rastra coverage by delivery, and before the UDB stage, Rastra’s cen- month could mean a welfare gain of as much expenditure decile, 2016 (Percent) tral-level policy group, the “Tikor Rastra Pusat” as 32 percent to a family of four, by providing generates total Rastra quotas, and sale and pur- about 40 percent of desired rice consumption Urban POOR chase prices for rice. Bulog then determines its at 80 percent below market price.37 Rural own cost of distribution. Local areas to which rice is delivered are expected to pay (with pub- 69.5 1 lic or private funds) all or part of this cost. Ra- C OVE RAG E, TARGET ING, & 70.5 stra rice may incur additional costs in remote areas when it is transferred from the distribu- I MPACTS 59.3 tion point to the actual point of sale.36 Some 2 local governments use general public revenues Rastra actual coverage38 is higher than tar- 68.8 to cover these additional costs, whereas in oth- geted coverage. At around 44 percent of the er areas households are expected to pay these total population covered in 2016 Susenas data, 51.9 transport costs. Finally, local authorities are re- Rastra coverage still dwarfs the next largest 3 66.3 sponsible for independently preparing a sales program, JKN-PBI. While coverage reaches 70 report, which means those same authorities percent in 2017 for the poorest decile, coverage 44.1 are free to distribute the “right” to buy Rastra to in the richest deciles is still high especially for 4 anyone regardless of whether they are officially rural areas (Figure 2.10) (World Bank, 2012d).39 62.1 eligible. Rastra in practice is not an income-pover- 37.6 Poor and vulnerable households are eligible ty-targeted program. In order to facilitate for Rastra. Since 2012, TNP2K had been us- comparisons between Rastra and other social DECILE 58.1 ing the UDB in order to generate a list of the assistance transfers (which may have slightly poorest 25 percent of households that are eli- different target groups), Figure 2.11 illustrates 35.9 gible to purchase Rastra benefit packages. The coverage and incidence for the “Poor”, the 52.9 Rastra beneficiary list is updated every year at “Next 30” percent, and “The rest”. This figure the local level through village or community demonstrates that while the poverty headcount 29.7 meeting schedules (the musyawarah desa/kelu- rate fell by 3.3 percentage points between 2009 7 rahan series). As with most of the other social and 2016, the share of Rastra benefits account- 47.8 assistance programs, Rastra has now switched ed for by the “Poor” group fell by almost 4 per- to KKS/KPS-based eligibility for beneficiaries. centage points. The “Next 30” group, which 23.4 However, as indicated above, village authorities contains the same proportion (30 percent) of 8 44.4 still have final authority when it comes to Ras- near-poor households in every year, has a Ra- tra beneficiary selection. stra share that fell by 2 percentage points over 14.0 the same time period. Finally, “The rest” has 9 On paper, the Rastra benefit package is com- a Rastra share that increased by 6 percentage 36.5 mensurate with needs. Eligible households have the right to purchase 15 kilograms of rice 36 In non-remote areas, the distribution point is also the point 3.8 (per month) at a price roughly 80 percent below of sale so no additional transport costs are incurred. 37 For 10 details on average rice volumes produced and sold, see World market price. Poor households dedicate (on Bank, 2009 24.3 37 For details on average rice volumes produced and sold, see average) two-thirds of their food expenditure World Bank, 2009 budget to rice alone, while mean per-capita rice According to households represented in Susenas. RICHEST 38 39 Poor and vulnerable households not purchasing Rastra are consumption is 9.5 kilograms per month (or more likely to live in urban areas. FIGURE 2.11 Poor The rest Next 30 Total Source Susenas 2016 and World Bank staff calculations. Rastra coverage and incidence by poverty groups Coverage (% of households Incidence (% of total receiving benefits) beneficiaries by group) 81 79 76 72 65 59 54 51 49 48 46 42 42 40 40 39 39 29 18 14 14 Source Susenas (various years) and World Bank staff calculations. 41 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 TABLE 2.2 FIGURE 2.12 Characteristics within Indonesia’s populations, 2016 Incidence of Rastra beneficiaries & average Rastra purchases by expenditure decile, 2016 Rural (%) % of all % poor % of Rastra Urban (%) Indonesians population recipients who Purchased Rice (kg) who: who: Coverage (%) Purchased Rice (kg) Do not have access to bottled, tap or well water 16 28 21 16 14 Do not have access to private sanitation 22 41 31 12 12 10 Live in rural areas 49 63 62 8 Live with more than 5 household members 22 38 23 8 6 Have not completed primary education 9 13 12 4 4 2 Are illiterate 7 13 9 0 0 Work in the agriculture sector 33 56 43 1 2 3 4 5 6 7 8 9 10 DECILE Source Susenas 2016 and World Bank staff calculations. ‘Work in…’ refers to share of working individuals, not all Indonesians. Source: Susenas 2016 and World Bank staff calculations. The value of actual Rastra transfers is low because of discrepancies between total Rastra rice procured & total Rastra rice purchased; between total benefit promised & total benefit received; & between total number of beneficiaries targeted & actual beneficiaries. points, showing a deterioration in targeting Subsidized rice volumes purchased are rough- ment tardiness in producing and sending dis- outcomes. This pattern indicates that as the ly equal across deciles (Figure 2.12). Rastra tribution-time proposals. micro-level poverty situation changes—with households in 2016 purchased about 5 kilo- many households exiting poverty year to year, grams per month, up by 1.5 kilograms from the The value of actual Rastra transfers is low be- while fewer enter—Rastra does not adapt by 2009 average. Above-median households have cause of discrepancies between total Rastra making changes to micro-level allocations.40 declining average purchase volumes; nonethe- rice procured and total Rastra rice purchased; Indeed, when the centrally decided Rastra quo- less these households captured about one-third between total benefit promised and total ben- tas were generated in 2012 (for the 2013 Rastra of the Rastra benefits available in 2015.42 efit received; and between total number of distribution) (TNP2K, 2014e), total coverage beneficiaries targeted and actual beneficia- was reduced (by about 2 million households), Rastra allocation and distribution differs by ries. Records show that of the Rastra rice pro- while the non-poor’s share of this reduced ben- area. Some areas, for example, West Sumatra, cured to deliver promised benefits, only about efit pool increased (Figure 2.11). distribute the entire Rastra package amount half of the procured kilograms (in recent years) to targeted households, while other areas, East are actually purchased by households (Figure Rastra rice ends up in many households that Java and Southeast Sulawesi, for example, are 2.13). It is not clear at which stage of the deliv- do not exhibit correlates of income poverty. prone to disregard “official” targets and share ery process rice goes missing.45 In most years, Rastra’s overall coverage, between about 40 and Rastra rice more evenly within local commu- 40 Previous studies have indicated that households rarely leave the 54 percent of the population in most years, is nity. Some areas distribute Rastra every month, program or are denied benefits once they have begun purchasing much larger than other social assistance pro- while others distribute less frequently (once in Rastra rice, even when they have exited poverty. See World Bank (2012e). grams, while Rastra distributes benefits to a 2-, 3-, or 4-month period).43 These variations 41 In addition to only weak correlations with poverty characteristics, World Bank (2012e) found that Rastra allocations are not larger three-quarters of poor households. However, are mainly determined by each community where food insecurity is greater. on average, Rastra rice is purchased by house- and at the local level.44 Based on monitoring 42 World Bank (2012f) demonstrated that when more Rastra is available on the ground, this can lead to larger purchases by poor holds that appear to be better-off than poor throughout 2012, only 46 percent of 220 vil- households but just as frequently leads to smaller purchases by poor households. households according to most of the non-in- lages received their entire Rastra quota on 43 TNP2K (2014f) come correlates of poverty listed in Table 2.2 schedule (TNP2K, 2015b). When Rastra was not 44 ibid 45 Earlier estimates (between November 2003 and January 2004) (for example, access to clean water or sanita- delivered on time, officials cited transportation found that up to 30 percent of Raskin allocations went “missing” in between the distribution points and Raskin-buying households; in tion, and working in agriculture).41 problems, payment arrears and local govern- some areas the estimated rate was as high as 75 percent. 42 Chapter 2 – Rastra FIGURE 2.13 FIGURE 2.14 Rastra procurement & rice purchases (billion kg) Rastra official & actual benefit (kg of rice per-month) Total Rastra rice procured Total rice consumption Budgeted Benefit (Kg/HH/mo) Total Rastra HH purchases (poor & near poor HH) Actual Benefit (Kg/HH/mo) Total rice consumption (all HH) 20 25 20 15 13 15 10 10 6 5 4.6 4 3.5 0 2009 2010 2011 2012 2013 2014 2015 2016 2004 2007 2010 2016 Source MoF and World Bank staff calculations. Source Susenas various years and World Bank staff calculations. 15 kg /mONTH “Rastra should have made 15 kilograms of rice per month available to poor and near-poor households at a subsidized price of IDR 1,600 (US$0.10) per kilogram. ” while procured rice for distribution through ry households have the right to reject and re- paid from public revenues. These markup Rastra represents about 15 percent of the total turn sub-“medium” standard rice for exchange. amounts are not necessarily commensurate rice market (by volume), actual Rastra purchas- However, monitoring throughout 2012, indicated with actual transportation costs, however. For es represent only 5 to 9 percent of market vol- that only 37 percent of villages received “medi- example, in a province where market costs for ume.46 um”-quality standard rice (or above). 47 taking goods the distances in between Rastra distribution points to household clusters were Rastra should have made 15 kilograms of rice Higher prices also drive a wedge between about IDR 44 to 125 per kilogram, Rastra rice per month available to poor and near-poor promised and actual benefit levels. As seen included a transport surcharge of between IDR households at a subsidized price of IDR 1,600 in Figure 2.14, there is a significant difference 200 and IDR 300 per kilogram.49 (US$0.10) per kilogram. This package would between budgeted benefit of Rastra rice, 15kg have translated into about 10 percent of pover- per month comprising around 8 to 9 percent ty line expenditure, and between 30 to 40 per- of poor household monthly expenditure, ver- cent of an Indonesian household’s rice needs sus actual average benefit received, 5kg per (see above). Actual purchases as reported by month comprising around 2 to 3 percent of households, however, were far less (Figure 2.14). poor household monthly expenditure. In 2004 Households purchased (per month on aver- and 2007, Rastra buyers paid about IDR 1,160 46 The figure indirectly demonstrates the importance of age) 3.5, 4.0, and 4.6 kilograms in 2007, 2010, (US$0.10) and 1,689 (US$0.20) per kilogram, rice in the food basket for poor and near-poor households: though they represent about 25 percent of households, rice and 2016, respectively. As a result, the value of respectively, while the official Rastra price was purchases within these households represents about 40 the benefit actually received in those years was IDR 1,000. In 2016, while the official price was percent of the total purchases in most years. 47 Villages receiving sub-“medium”-standard rice found Rastra closer to just 2 percent of poor households’ ex- IDR 1,600 (US$0.10) per kilogram, Susenas re- rice with weevils; brown-colored rice; and rice with unpleasant smells (TNP2K, 2015b). penditure. spondents reported paying about IDR 2,054 48 Reported price paid for Rastra rice varies significantly in all (US$0.20) per kilogram on average.48 years of Susenas data. 49 SMERU (2008) found a province where market costs for Rastra rice does not always meet Bulog’s own taking goods the distance in between Rastra distribution points and points of sale were about IDR 44 to 125 per quality standards. Rastra rice is expected As mentioned above, a higher-than-stipulated kilogram while the Rastra program charged IDR 200 to 300 to meet a “medium” quality standard (good Rastra price often contains additional trans- per kilogram. TNP2K (2015b) found that “collection costs” were reaching IDR 445 per kilogram in Java and as high as rice condition; free of pests) and beneficia- port costs that have not, but could have, been IDR 483 per kilogram, outside Java. 43 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 TABLE 2.3 Rastra-affiliated agencies & their roles ACCESI B I LI T Y MINISTRY/AGENCY RASTRA-BASED TASKS AND FUNCTIONS Coordinating Ministry for Social Welfare Coordinates implementation Bappenas Policy agency (overal poverty/social protection policy function) Rastra policy and oversight functions include many central agencies, as well as their local TNP2K Creates list of eligible beneficiaries instances (Table 2.3). To increase effectiveness and accountability, a central-level Rastra Coor- Central Bureau of Statistics Collects and hosts Rastra beneficiary data dination Team (Tikor Rastra Pusat) was formed Ministry of Finance Executes Rastra budget provisions in 2013.50 Tikor Rastra includes the Coordinat- ing Ministries for People's Welfare and for Ministry of Home Affairs Manages Complaint-Handling Unit Economic Affairs, the National Development Planning Agency, the Ministries of Finance, In- Ministry of Social Affairs Budget Holder/Formal executing agency terior, Social Affairs, and Agriculture, the Cen- Bulog Implementing agency (until the “last mile”) tral Bureau of Statistics (BPS), the State Audit Agency (BPK), and Bulog. Local Government Responsible authority for Rastra’s “last mile” Rastra generates official beneficiary quotas Source World Bank, 2015c. with the UDB, but actual beneficiaries are still determined at the local level.51 In 2012, TN- P2K began using the UDB to generate a list of TABLE 2.4 beneficiaries.52 Rastra teams at the local level were asked to update these lists via the regular Tikor Rastra endorsements community meeting series (Musyawarah Desa/ Kelurahan community meetings). Upon the BUSINESS PROCESS AREA CURRENT CHALLENGES PROPOSED SOLUTIONS switch to UDB-based targeting with commu- nity verification and updating, it was conclud- Transportation Rastra rice delivered by one trans- Undertake regular tender process to ed that any reallocation of benefits achieved port company (JPL) contract transport provider through community meetings could improve Quality Inspection Rastra rice quality inspected by one Undertake regular tender process to program targeting, including cross-village tar- QA provider (Jastasma) contract a quality control company geting. However, more and better socialization is needed to improve community meetings’ ca- Monitoring & Evaluation Diffused responsibilities undertaken New M&E Organization pacity for determining an efficient local distri- by several parties bution of social assistance transfers. Last Mile distribution (rural) Local government undertakes dis- Use local government offices, post tribution office, coops, and military where Rastra’s outreach and socialization procedures possible may be improving. TNP2K launched an aware- ness campaign to better inform program im- Last mile distribution (urban) Local government undertakes dis- Partner with retail outlets to develop tribution a government ‘debit card’ for Rastra plementers at all subnational levels about Ra- and other nutritious food purchases stra program logic, the importance of finding targeted households, and the use of KPS identi- Source World Bank, 2015c. fication to confirm eligibility.53 This campaign reached 106 cities (or urban districts) and 1,114 duce the Raskin purchase by about IDR 250 per the Tikor Rastra endorsed in 2014 and which villages (or rural hamlets) across Indonesia by kilogram.55 The findings for the “Raskin card” Bappenas has instructed local governments to 2016, but impacts haven not yet been reviewed provided impetus for the KPS card in June 2013. pilot, included adapting Rastra in urban areas (TNP2K, 2014b).54 In 2014, PT Deloitte in tandem with the World as a “smartcard”-access food distribution pro- Bank and Raskin stakeholders, mapped the gram,56 and an improved distribution supply Rastra implementers have drawn up strate- Raskin supply chain in order to identify areas chain in rural areas (Table 2.4). Subsequently, TN- gies to more effectively distribute full Ras- where re-engineered solutions could increase P2K implemented a pilot to test the suggested new tra benefits to targeted households (TNP2K, delivery efficiency (also known as a “Business delivery mechanism and informed the develop- 2014e). TNP2K, jointly with Jameel Abdul Latif Process Review”). Recommendations, which ment of implementation guidance (Box 2.2). Poverty Action Lab office in Indonesia, con- ducted pilot tests for a “Raskin card” in 2012. The program distributed 1.3 million eligibility 50 Corresponding coordination teams were also formed at the district and village implementation level. identification cards in 53 cities (or urban dis- 51 Ibid. 52 The list is called Daftar Penerima Manfaat (DPM). tricts) in 7 provinces. The research team con- 53 Materials, including posters and leaflets, were sent to around 78,000 villages/kelurahan. TNP2K has also arranged for media cluded that a “Raskin card” could increase the briefings, talk shows, and informative broadcast with around 150 local and national media outlets in major capital cities. 54 While program implementer knowledge has increased, beneficiary awareness has not responded to the expanded socialization amount of Raskin rice received by poor house- efforts (TNP2K, 2015b). 55 Positive impacts were larger when the “Raskin card” was accompanied by a socialization module. holds by about 2 kilograms per month, and re- 56 Similar to the USA’s Supplemental Nutritional Assistance Program (colloquially known as “food stamps”). 44 Chapter 2 – Rastra BOX 2.2 05 Raising the effectiveness of digital payment implemen- TNP2K— Short report of 2016 pilots R E C O M M E N D AT I O N S tation on the distribution of non-cash in-kind 01 Preparation of data on targeted recipients It should be clearly defined which staple foods should be avail- social assistance able at merchants/agents/shops with suitable price and quality If ample time is available, conduct checks to see if potential standards. The registration mechanism should be independent beneficiaries can be found and ensure there is an application to and open for all entrepreneurs or existing shop owners that allow for data updating at the village level. If time is short, the would like to become merchants/agents/shops. To maintain an re-checking of potential beneficiaries should be done togeth- adequate flow of payments, there should be liquidity support BACKGROUND er with the registration step at the village level done together with a village facilitator and bank employee by opening of a spe- for merchants/agents/shops to provide cash to the recipients. cialized desk during a town hall. 06 Trading the food voucher and disbursement of digital Raskin has historically showed poor performance in terms of payments benefit adequacy due to sharing out of benefits, low quality of 02 Sending of invitation letter to recipients as initial rice, higher costs than stipulated due to transport and oth- In trading the electronic voucher to get goods, the merchant/ socialization and education about the programs. er costs being levied upon the purchasing household among agent/shop must give 1) proof of transaction and information of others. In March 2016, The President instructed Rastra be remaining balance to the recipient 2) clear information about Socialization should contain complete information about the reformed into a voucher system and be implemented in early the price of rice, eggs and milk sold. In distributing cash, the program and distribution mechanism including location of 2017. By use of a voucher it is meant that Rastra distribution Bank must ensure recipients can withdraw cash via the mer- registration and transactions. Education to recipients and can be monitored and recipients can receive better quality rice chants/agents/shops by ensuring adequate liquidity and the merchants/agents/shops as banking agents should be done in- of their choosing. Besides rice, the voucher should also be able proximity of ATMs at low cost to the beneficiary. tensively before the program launches to raise understanding. to be used for other staple goods. In April 2016, the President also instructed that social assistance and subsidies should be 07 Services for the recipients 03 Mode / tools of transaction, registration and activation. transferred digitally. Distribution should occur via the banking system to lead to better oversight and monitoring. The use of Ensure there is a grievance redress mechanism. Banks and Debit cards or cellular phones methods both have advantages the banking system is expected to support productive behav- local government should provide a special mechanism for the and disadvantages such that choosing to just use one will en- iors, increase financial inclusion as well as encourage savings. elderly and special needs recipients on all aspects of the pro- gender risks to hinder benefit distribution rather than help it. The president also instructed that the various social assistance grams operation Besides that, banks that participate in the program need to be schemes become integrated into one card and one account. able to manage risks and innovation related to the mode/tools of transaction. It must be remembered that opening a bank 08 Involvement of local governments accounts requires citizenship documents such as the KTP and DESIGN OF PILOTS KK to fulfill bank requirements. At the moment not all families Local government should be actively engaged in the programs have such documents so it should be ensured that the local implementation, in particular with validation and verification TNP2K conducted a digital payment reallocation of in kind as- government can fulfill this important need by providing such of recipient data and socialization. Together with banks, local sistance between September and October 2016. The purpose documents governments should help identify potential merchants/agents/ was to test and design mechanisms overseeing the entire pro- shops to support implementation of the program. Local gov- cess as below. 04 Availability and spread of merchants/agents/shops ernments have the role of providing information and receiving complaints as well as checking merchants/agents/shops stipu- 1 Preparation of data on targeted recipients and The ratio of merchants/agents/shops to recipients is ideally lated prices. They should also provide necessary administra- preparation of merchants. 1:150. For that reason, there still will need to be more agents tion and provide information related to problems and solutions added to reach the ideal ratio. The bank should have at least in the implementation of the program and conduct periodical 2 Sending of invitation letter to recipients as initial two merchants/agents/shops per village to ensure adequate monitoring and evaluation. socialization and education about the programs. choice is available for the recipients and to avoid monopolies in price or procurement leading to suboptimal quality of goods. 09 Raising systems and infrastructures of transaction at 3 Registration and activation by recipients via bank The local bank should ensure technical issues relating to the the agent/merchant level outlets or merchants (via phone or using cards). tools of transaction can be resolved. The bank should take into account OJK regulations related to the recruitment of Laku Program implementers should coordinate with the Ministry 4 Distribution of benefits to savings accounts of Pandai agents and there should be involvement of merchants of Information and telecommunication to minimize the risk of recipients. that are already bank agents so that the ratio is reached. Im- poor network signal to ensure electronic transactions succeed portantly, banks need to conduct adequate training for mer- in each area. Ensure interoperability of banking agents such 5 Use of benefits and trading of electronic chants/agents/shops and should provide a technical support that recipients can visit multiple merchants/agents/shops. On vouchers for goods at a merchant. team in each are to provide facilitation to the merchants/ sustainability, the agent must have specified staple goods and banks/shops as needed. must have the financial incentives to conduct their own invest- ments to procure these goods. Source (TNP2K, 2017) In mid-2016, a program called e-Warong was In February 2017, the government launched and are willing to participate as authorized launched by MoSA and other government a new “non-cash food assistance” (Bantuan merchants to the BPNT program. The move agencies in collaboration with Bank Indone- Pangan Non Tunai, BPNT*) program, which towards e-vouchers captures several important sia, the Financial Regulatory Authority (OJK), aims to eventually replace Rastra’s subsidized objectives: improved targeting, greater access to Bulog, and multiple national banks. These in-kind provisions with a more flexible and nurtritious food and a higher degree of choice e-Warong are owned by groups of PKH benefi- accountable e-voucher based delivery sys- for the beneficiary, increase financial inclusion ciary families under the KUBE-PKH program tem. Under BPNT, each beneficiary household and encourage the development of small scale and are supposed to sell Rastra goods (rice, would receive a voucher worth IDR 110,000 per local businesses.57 The implementation perfor- cooking oil, etc) to PKH and Rastra beneficia- month and can purchase rice and a number of mance of this new program is yet to be evalu- ries. Cardholders can exchange e-vouchers pre-determined food items from authorized ated fully while its design most likely needs to only for pre-determined goods such as Rastra providers (E-Warong). The first phase of BPNT be adapted significantly before scaling up in rural rice, while other SA program benefits may be involves 1.4 million beneficiary households areas and to include other food items such as eggs. withdrawn in cash at the e-Warong or other ap- in 44 cities (MoSA, 2016d) and will be scaled proved locations via agents of multiple banks. up nationally by 2020. Also in response to * Thisprogram is in line and related with the Presidental the BPNT, the definition of e-Warong was ex- Regulation No. 63 in 2017 on the Distribution of Non Cash panded to include market vendors that have Social Assistance and Non Cash Food Support. 57 Timmer, P. et al (2017) Evolution and implementation of the already established their food related business Raskin Program in Indonesia. Forthcoming book chapter. 45 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 CHALLENGES, RECOMMENDATIONS & MOVING FORWARD While Rastra’s policy relevance is still sound: Rastra should continue reforms, such as the TNP2K has also suggested Rastra reforms poor household welfare is negatively affect- ones mentioned in the previous section, de- (in addition to improved targeting) based ed by food scarcity and food price volatility, signed to enhance its efficiency as a social as- on recent comprehensive analytical studies it is evident that both the design and imple- sistance transfer. These reforms include: (Perdana et al., 2015): mentation performance of Rastra have failed to achieve its objectives. Were the program to expand beyond just rice—for example, by in- cluding meat (or other high-protein items) and 1 3 1 3 fruit and vegetables in its benefit package—the Rastra quotas and Rastra socializa- Increased transpar- Formalize the program could also help reduce the high rates actual household tion should be ency by listing ben- process of recip- of micronutrient malnourishment, a precursor allocations should re-enforced with eficiaries by name ient replacement to stunting, which is still affecting Indonesian be based on the performance at village level. done by the village communities, especially those in the poorer community-updat- incentives so local administration. eastern half of the archipelago (World Bank, 2013c). The poor targeting performance and ed national registry. communities can develop grass-roots 2 Adjust quota ceil- weak accountability of the distribution chain have reduced Rastra’s effectiveness in terms of 2 monitoring and provide feedback to ings. its actual impact on poverty and food security. Below the distribu- implementers. tion points, Rastra The ongoing BPNT initiative on the other hand holds great promise in addressing Ra- allocations should 4 be monitored to stra’s weakness. It is expected to better target ensure that a full Monitor, evaluate the bottom 25 percent households using the allocation reaches and adapt alter- UDB, provide better access to nutritious food all eligible house- native delivery by design, allow beneficiaries to choose and holds first. systems such as the control on when, what type, and how much they BPNT program, buy rice and other eligible food commodities, particularly explor- encourage local retail businesses to participate, ing comparative and finally cost savings in public spending cost-effectiveness of stemming from better efficiency of service de- benefit modalities livery (World Bank, 2017b). One distinguishing (in-kind, voucher, feature of the BPNT design is that it is possible and cash) in differ- to closely monitor the program transactions, ent settings. which are carried out electronically using beneficiaries’ KKS cards and service providers’ EDC/Pos devices and hence help hold the ser- vice providers accountable if appropriate mon- itoring and audit mechanisms are put in place as in the case of United States’ SNAP program. 46 Chapter 2 – Rastra While Rastra’s policy relevance is still sound (poor household welfare is negatively affected by food scarcity and food price volatility), it is evident that both the design and implementation performance of Rastra have failed to achieve its objectives. 47 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 Subsidized Social FIGURE 2.15 Health Insurance OVERVIEW Health insurance program coverage by insurance type (percent) (JKN-PBI) No Insurance PBI–JKN Jamkesmas* JKN (non–PBI) JKN-PBI’s potential is vast: on paper it Healthcare access, healthy behaviors, and Jamsostek Askes Private promises to provide a theoretically in- healthy outcomes for all citizens, are a focus of GoI social policy. Towards the end of the 20th value, in-kind health benefit to all poor 9 9 1 century, the GoI began targeting healthcare ser- and vulnerable households regardless vices facilitated by public expenditures specif- 4 5 of location. Unevenly distributed ically to poor households: kartu sehat or health 5 9 26 healthcare facilities and personnel, as card (circa 1994) and its Asian-financial-crisis cousin, Jaring Pengaman Sosial Bidang Kesehatan well as an uneven distribution of quality 13 (JPS-BK), provided poor households with free within the provision of medical services, curative health care at community health cen- 26 mean that for most eligible households ters and referral care at district hospitals. Then, 70 PBI potential is not in proportion to throughout the 2000s, multiple compensatory transfers to poor and near-poor households— 18 what it actually delivers. As JKN-PBI is provided during periods when fuel subsidies part of the National Health Insurance were adjusted—contained a health services Scheme and more beneficiaries are component that again provided free care (in- 52 covered due to mandatory universal patient or outpatient, as well as preventative services and pharmaceuticals). coverage, better M&E, accessibility, and 46 outreach will be key to providing an While the household trend in all these areas effective in-kind benefit for poor and has been positive, progress has been slow. Fig- vulnerable households. ure 2.15 demonstrates that the share of house- holds with no health insurance coverage of any 2005 2010 2016 kind has fallen from 70 to 46 percent in the pe- riod 2005-2016. Source Susenas 2005, 2010, 2016. PBI-JKN coverage in the PBI, previously referred to as Jamkesmas, is Susenas survey is lower than administrative coverage. a component of the Jaminan Kesehatan Nasi- Notes: In 2005, a household is considered "covered" if at least one of the household members reports having (a particular onal (JKN) program (hereafter referred to as type of) insurance. *In 2010, a household is considered "covered" by Jamkesmas if the respondent reports that JKN-PBI). Jamkesmas was a subsidized public the household can access Jamkesmas. In 2016, households healthcare insurance program intended for can report whether they are covered by PBI/Jamkesmas, Jamkesda and others. Households are considered "covered poor and near-poor households that would by PBI" if they received assistance from PBI. They might be covered by Jamkesda and other types of insurance as well. otherwise not be covered by health insurance.59 JKN includes BPJS Kesehatan and other programs that it has Recognizing that poor and vulnerable house- absorbed since 2014 but are still asked in Susenas data as separate categories. Households are considered "covered by holds have higher rates of non-utilization, Jamkesda" if they received assistance only from Jamkesda. However, as of late 2016, 73 percent of districts existing higher rates of preventable conditions, and Jamkesda had already become integrated with JKN and so more frequent income losses due to adverse any mentioning of receiving Jamkesda is taken to be part of JKN (non-PBI). Since there has been some changes in the health events (World Bank, 2013d), Jamkesmas naming of programs 2016 health insurance data is presented with different categories from 2005 and 2010. was developed (circa 2005) to improve utiliza- tion by reducing the costs of services. Today, 46% Jamkesmas is called Penerima Bantuan Iuran (PBI), which is not a program in itself but is part of JKN and it is still targeted to poor and vulnerable households. JKN-PBI makes free the use of all available healthcare services and facilities in accordance to JKN-PBI regulations. “While the household trend in all these areas 2.3 PBI is meant to produce social, as well as indi- has been positive, progress has been slow. vidual, benefits: by promoting healthy house- Figure 2.15 demonstrates that the share of holds, keeping students active and alert, and re- households with no health insurance coverage turning adults to work sooner, all Indonesians of any kind has fallen from 70 to 46 percent in benefit from a more productive population.60 the period 2005-2016. ” 58 Presidential Regulation No. 32/2014 on the Management and Use of Capitation Grants for Quality Improvement in Front Line Services, Minister of Health Regulation No. 19/2014 and Minister of Home Affairs Circulation Letter No. 990/2280/SJ. One regulation provides rules to improve the management and use of capitation grants at non-BLUD Puskesmas; the other two technical regulations enable BPJS Kesehatan to pay capitation grants directly to Puskesmas through a designated account for each center. BLUD (Badan Layanan Umum Daerah) is a designation attached to a public service provider meaning that though the provider’s legal status is attached to local government and the provider must remain non-profit oriented, it is nonetheless able to execute its activities—including pricing schedules and staffing—in order to improve service delivery to the community. In other words, a BLUD service provider is semi-autonomous, while a non-BLUD provider is not. 59 Jamkesmas/JKN-PBI was known as Askeskin when it was established in 2005. Now, Jamkesmas has undergone a transition from a stand-alone program managed by the Ministry of Health to a targeted, subsidized component (PBI) of Indonesia’s National Health Insurance program (JKN), which is itself under the National Social Security System (SJSN). 48 Chapter 2 – JKN–PBI BOX 2.3 amounts households would otherwise be tonomy to these health service providers expected to pay. In other words, when Jam- even when they are accountable to, and have Askeskin Jamkesmas kesmas was eliminated, Indonesia also elimi- their operational funding disbursed by, local JKN–PBI nated its healthcare fee-waiver program for governments (APBD refers local government poor households; at the same time, Jamkes- budget, APBN refers to national government da (the local government health insurance budget). However, there are significant shifts Following the establishment of the managing fee waiver program mirroring Jamkesmas) taking place: Puskemas continue to receive bodies for Indonesia’s National Social Secu- began merging with JKN-PBI. In its place, the general subsidies from the government rity System (SJSN) in early 2014, based on JKN insurance program and an initiative, re- budget to finance operational expenses the 2011 BPJS Law, the subsidized national ferred to as PBI Penerima Bantuan Iuran or and medicine but are at the same time also health program for the poor and near poor, “recipient of government fee support”, was receiving payments from the JKN program; then called Jamkesmas, completed a merger launched to cover JKN premiums for those this shift and the interplay between these to become part of National Health Insurance who qualify for such relief. funding sources and resulting provider (JKN) affiliate of the SJSN. As of 2016 JKN behavior and impact on user experience war- covers around 166 million individuals includ- When Jamkesmas and Jamkesda (were still rants further study. ing previous Jamkesmas beneficiaries who implemented, beneficiary selection was have automatically become JKN members ultimately determined in a highly decentral- Providing premium subsidies to poor house- through PBI, those who have their health in- ized setting, with local governments helping holds through the JKN-PBI initiative will see surance fee paid for by the government. JKN to identify both pre-listed and additional the subsidized national health program for is managed by the Social Health Insurance eligible beneficiaries. As JKN administrators the poor and near poor move away from a Agency (BPJS Kesehatan),. Including locally absorb the Jamkesmas and Jamkesda bene- vertically integrated program, where service financed PBI, the PBI component of JKN, ficiaries and expand JKN coverage, they will providers, those determining eligibility, to be referred to as JKN-PBI, will comprise instead use the national registry for target- and those determining policy for both the around 64 percent of JKN by the end of 2016 ing, identifying, and contacting beneficiaries. program and the service providers are all (MoH, 2016). employed by the same ministry. With BPJS Other recent regulatory changes which Kesehatan as the national health insurance As the Jamkesmas program, as well as clarify fund flow channels and the permissi- provider, the UDB and its managers as the the participants, completes absorption by ble uses of funds sourced in these channels card provider and eligibility gatekeeper, and JKN-PBI, it will cease being a stand-alone may also enhance the ability of the commu- the Ministry of Health as the manager, policy program operating with its own budget. nity health center (or Puskesmas), which formulator, and regulator of the health Instead, PBI’s target group—poor and near- is usually the first-line provider of medical service providers, it is unclear whether poor households—and eligibility criteria will services, to provide accessible and reliable potential beneficiary access to the program remain while JKN will include a subsidized services to JKN-PBI members58 In essence and the services provided will improve. program that covers nominal premium these regulations will provide greater au- Source (TNP2K, 2014c) & (World Bank (2016b) FIGURE 2.16 PBI expenditure and absolute vs. relative coverage61 PBI EXPENDITURE ABSOLUTE VS. RELATIVE COVERAGE (IDR BILLIONS) P RO G RA M S I Z E , Absolute coverage (individuals) Relative coverage IN ST I TUT I ON A L S E T U P, 2005 1,300 60 60 76 76 76 76 76 76 76 86 86 92 EL I G I B I LI T Y , & B E N E FI TS 2006 3,074 34 34 34 34 36 2007 4,567 2008 4,488 33 33 28 32 27 31 31 2009 4,620 2010 4,763 As Jamkesmas gave way to JKN-PBI in 2014, both the number of ben- 2011 6,300 eficiaries and the program’s budget expanded (Figure 2.16). In 2016, 2012 7,200 the Government significantly raised central government expenditures on JKN-PBI from earlier years to IDR 24.8 trillion, giving the program 2013 8,100 a 35 and 39 percent share of total central government social assistance expenditures and total central government health expenditures, respec- 2014 19,900 tively, in 2016. From 2013 to 2014, 10 million beneficiaries were added, 2015 20,347 marking the first major program expansion since 2007. From 2014 to 2005 2006 2009 2007 2008 2010 2011 2012 2013 2014 2015 2016 2016, an additional 6 million beneficiaries were added to reach a targeted 2016 24,814 total of 92 million beneficiaries. Taking the locally financed PBI recip- Source: MoF and World Bank staff calculations. ients (Jamkesda) into account, the total increased to 105 million (MoH, 2016). This leaves JKN-PBI as the largest of the social assistance transfers 61 Absolute coverage is shown as the total population receiving PBI while relative coverage by budgeted expenditure and coverage today. places this number in the perspective of the growing number of the total population by year. 49 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 The JKN-PBI initiative will provide house- based on the Jamkesmas set up from which it BPJS Ketenagakerjaan). In the JKN-PBI initia- holds with a true insurance program; Jam- transitioned. BPJS Kesehatan receives annual tive, JKN premiums are fully covered by the kesmas was a fee waiver/supply-side subsidy budgeted transfers for PBI based on a per-capi- Government; public providers and some pri- hybrid. Prior to 2014, the Ministry of Health ta monthly “premium” and the number of poor vate hospitals can provide services; there will operated Jamkesmas as a fee-waiver program and near-poor beneficiaries targeted, based on be no co-payment, co-insurance, extra-billing and as a supply-side top-up through provision the UDB. In addition, service providers re- or balance-billing allowed; and there are no of capitation grants and claim-based reim- ceived general operational and capital-cost limits on benefits provided to beneficiaries bursement. Local health facilities were allotted budgets from central, provincial, and district (including prescribed pharmaceuticals). The Jamkesmas funds based on the size of the pop- governments to fund Jamkesda, but as Jamkes- supply-side constraints that effectively limited ulation they served, as well as the activities pro- da programs further complete their merge with Jamkesmas benefits, however, will not neces- grammed. Hospitals also received operating JKN-PBI, the program will be entirely centrally sarily be addressed by the JKN-PBI program, so grants tied to population magnitudes and ser- executed. Reimbursement rates to public and JKN-PBI’s generous benefits on paper may be vices offered, and in addition made claims to private hospitals for PBI coverage are largely of less value in the field (World Bank, 2013d). the Jamkesmas administrator for re-imburse- the same, varying only by the degree of special- ment. (TNP2K, 2014d). ization of the hospital. Jamkesmas was accepted at both public Poor and near-poor households are eligible to and private providers and therefore some have their JKN premiums paid by the PBI ini- BOX 2.4 risk-sharing between the Government and the tiative. PBI-JKN will use the Unified Database, Developing Productive Families, private providers was accomplished. However, which will have a complete ranking of the poor- Kartu Indonesia Sehat (KIS) the fee-waiver portion of Jamkesmas was rarely est 40 percent of households in Indonesia, to enforced and out-of-pocket costs were not nec- receive a list (with name and address) of eligi- The KIS as part of the new SA scheme introduced by the essarily reduced for Jamkesmas cardholders, ble beneficiaries to be contacted and verified.63 Government provides a fee waiver for first level health- so from a household perspective the insurance All members from eligible households are con- care costs and was initiated to help establish comprehen- value of Jamkesmas was reduced. While the po- sidered JKN-PBI members. In the Jamkesmas sive health insurance. tential value of a Jamkesmas card was very high, program however, though an initial quota was Coverage and eligibility as nearly all services available were covered by generated by querying a household list con- By late 2014, 4.6 million KIS cards were sent to individu- the fee waiver, in practice the value was consid- taining some socio-economic and demograph- als in the one million poor households targeted in the first erably less because of both remaining out-of- ic information, those given Jamkesmas cards phase of implementation. The KIS card was meant to be pocket expenses, as well as a limited supply of were locally identified by locally-based Minis- given to over 92 million PBI recipients in 2016. Once regis- tered at a public primary health facility, the KIS provides services62 in most areas. try of Health staff, service provider staff, and the user with a fee waiver for first level healthcare ser- local government. When the number of local- vices, including preventative care and early detection.64 For JKN-PBI, the insurance value of the pro- ly-identified households exceeded the Jamkes- Targeting is, as Jamkesmas was, also achieved through gram for poor households should increase. mas quota, the remaining households would be the UDB, now provided via MoSA. Individuals that live in households considered to be in the poorest 40 percent Services acquired by JKN-PBI beneficiaries will encouraged to enroll in Jamkesda, if available are eligible to receive the KIS card. Households that do not be billed according to JKN standard operating (World Bank, 2013d). New eligibility cards, part have the KIS card yet continue to use their Badan Penye- procedures, while complicated funding ar- of the “Developing Productive Families” initia- lenggara Jaminan Sosial (BPJS) Health card to access rangements within the public service provid- tive, will be issued (eventually) to all JKN-PBI health care at a subsidized price (TNP2K, 2014k; collected news stories). ers should cease to be a constraint on the set of beneficiaries (see Box 2.4). services available. However, it is unclear how Program flow and Benefit structure or if the service provider’s ability to charge fees While JKN-PBI offers essentially the same Just like the KKS, the KIS card was sent to the targeted directly to households will be diminished: pre- comprehensive benefits package as Jamkes- household via the national post office (PT. Pos). Users make use of the card as shown below. The KIS card taps mium rates (calculated by the GoI) are not cur- mas, it is considered more generous and inclu- into an existing health insurance network called the Jam- rently based on actual cost of services provided. sive than that of the civil servants (previously inan Kesehatan Nasional (JKN) which provides the recip- Askes, a program now renamed and absorbed ient with health insurance for the monthly premium of BPJS Kesehatan in coordination with the Min- within BPJS Kesehatan management of JKN) 23,000 IDR (US$1.70) per person. BPJS Kesehatan man- ages the card and its implementation while the budget istry of Health executes the JKN-PBI program. and formal sector health insurance programs comes from the JKN-PBI budget. Figure 2.17 describes the flow of funds for PBI, (previously Jamsostek, now absorbed within Dissemination of information materials Dissemination of KKS/KSKS, KIP and KIS program infor- mation to all stakeholders is managed by BPJS Keseha- FIGURE 2.17 tan with the help of the Ministry of Information. In 2014, support was provided by TNP2K as well. For instance, in Flow of funds for JKN–PBI Operational & cost budget to service providers. 2014, 20,000 posters had been set up in strategic loca- tions across 500 villages in 19 districts within 9 provinces. Quarterly budgeted transfer. Input-based government budget. Town hall meetings had been held and 30 radio stations and 10 local TV stations had facilitated media campaigns. Source: TNP2K, 2014k and local media reports. SOURCES OF FUNDS AGENTS PROVIDERS Central Goverment Ministry of health BPJS Kesehatan Public hospitals 62 In other words, the grants and negotiated claims payments did not fully cover the costs of the services at the amounts demanded of those services by Jamkesmas-card-holding Provincial Goverment Province health office Empanelled private hospitals households. 63 The UDB used to be managed by TNP2K but is officially managed by MoSA since 2017. 64 KIS card holders / PBI-JKN recipients are entitled to fee District Goverment District health office Puskesmas waivers for any costs incurred at health centers and are entitled to the whole range of possible referrals from public health centers that are registered with BPJS Health. 50 Chapter 2 – JKN–PBI since then more recipients of old programs FIGURE 2.19 now phased out have become absorbed within JKN managed by BPJS Kesehatan. Figure 2.18 JKN-PBI coverage and incidence by shows program coverage by decile for PBI and poverty groups Jamkesda as of 2016. PBI coverage in the first decile is relatively low, at 30 percent, while cov- Poor The rest Next 30 Total erage in the richer deciles is also significant: 18 percent of households in the fifth decile are covered by PBI.65 Jamkesda has low coverage for Coverage (% 36.7 COV E RAG E , of households most deciles, but a considerable increase in cov- 30.4 receiving 30.4 erage can be seen between the 5th and 8th deciles benefits) 25.2 TARG E T I N G , & 21.3 18.0 The transformed and expanded JKN-PBI al- 14.9 12.7 IM PACTS 12.1 lowed the program to provide more benefits 9.5 to poor and near-poor households. In order 7.1 4.7 to facilitate comparisons between JKN-PBI and other social assistance transfers (which may have slightly different target groups), Fig- 2009 2012 2016 ure 2.19 shows coverage and incidence for the “Poor”, the “Next 30” percent, and “The rest” for Incidence 44.0 (% of total 42.0 PBI recipients only. The figure demonstrates 40.4 39.7 39.9 38.7 beneficiaries that while the poverty headcount rate fell by by group) less than 1 percentage point between 2012 and 2016, the share of PBI benefits accounted for by 20.9 18.4 16.1 JKN-PBI coverage reached 88 million people the “Poor” (incidence) group grew only slight- in 2015, while expanding to over 92 million ly by 2 percentage points; the “Next 30” group, people in 2016. In 2014, about 7 percent of which contains the same proportion of near- these JKN-PBI beneficiaries “transferred” from poor households in every year, saw a similar in- 2009 2012 2016 the locally-run health insurance programs, crease in its PBI share. All the while, “The rest” Jamkesda, as well as the PJKMU (TNP2K, 2014c), has fallen by 4 percentage points. Together, this 65 Household coverage discussed here and as measured in the Susenas survey may differ from official estimates of the number of cards distributed for at least three reasons: (i) Susenas survey weights may not reflect the correct probability of contacting a JKN-PBI receiving household; (ii) households themselves may be KIS cardholders but mistakenly report coverage by Jamkesda (or any other scheme) or may think they are covered even though they are not cardholders; and (iii) not all distributed cards have actually reached beneficiary households. TABLE 2.5 JKN–PBI coverage and incidence by FIGURE 2.18 poverty groups JKN–PBI and Jamkesda coverage by PBI / JKN Jamkesda expenditure decile, 2016 % of all % poor % of PBI Indonesians population recipients Poorest DECILE Richest who: who: who 1 2 3 4 5 6 7 8 9 10 Do not have 16 28 21 access to bottled, tap or well water 30.2 Do not have 22 41 34 access to private 27 sanitation 23.9 Live in rural 49 63 58 areas 20.9 Live with 22 38 26 17.8 more than 16.7 5 household members 14.1 Have not 9 13 14 11.7 completed 10.5 10.2 primary 9.7 10 9.5 8.8 education 8.3 8.2 6.9 6.6 Are illiterate 7 13 10 4.8 Work in the 33 56 44 2.7 agriculture sector Source: Susenas 2016 and World Bank staff calculations. ‘Work in…’ refers to share of working individuals, not all Source: Susenas 2016 and World Bank staff calculations. Indonesians. 51 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 pattern indicates that as the micro-level pover- FIGURE 2.20 ty situation changes—many households exit poverty year to year, while fewer enter—PBI’s JKN-PBI incidence and outpatient utilization by PBI Incidence merger with JKN and implementation revi- expenditure decile, 2016 (percent) PBI Outpatient Usage sions have allowed it to continue to find, albeit at a slower pace, the remaining eligible poor 57 and near-poor households. 55 55 53 53 53 53 54 54 52 On average, JKN-PBI household exhibit cor- 20 relates of income poverty. JKN-PBI overall coverage (in the 2016 Susenas) at about 18 per- 17 cent of households, is close to triple the 2009 level. These additional PBI benefit funds have 14 on average been to households with most of the 12 non-income correlates of poverty (Table 2.5). 10 9 The switch to UDB-based targeting has de- livered a larger share of benefits to poor and 7 near-poor households. Previous analyses of 5 Jamkesmas targeting found that it did not ef- ficiently distribute benefits to its target popu- 3 1 lation (TNP2K, 2014c; World Bank, 2013d; TN- P2K, 2013). Figure 2.20 illustrates that, by 2016, 1 2 3 4 5 6 7 8 9 10 JKN-PBI benefits are concentrated in the lowest DECILE deciles: the bottom 30 percent of households contains just over half of the JKN-PBI benefi- Source: Susenas 2016 and World Bank staff calculations. ciaries. However, leakage to the non-poor is still evident, with 37 percent of the JKN-PBI benefits going to the top 60 percent of the households. Figure 2.20 also summarizes out- patient usage rates (for those with certain PBI AC C E S S IB IL IT Y coverage): outpatient usage rates in the richest decile of households are generally the same FIGURE 2.21 across all deciles, indicating that it may be dif- Poorer households with PBI do not visit pro- ficult to eliminate leakage entirely when the Rates of utilization by insurance type viders—especially high-value inpatient pro- benefit package available is of such high value. (percent) viders—frequently enough to give them a Outpatient Rate (2012) disproportionate share of benefits available. Healthcare utilization is growing for PBI ben- Inpatient rate (2012) Reviewing the previous trends: Figure 2.20 eficiaries at a rate similar to those who are Outpatient rate (2016) shows that most PBI beneficiaries are concen- Inpatient rate (2016) beneficiaries of other insurance programs trated in poorer households and that PBI use (Figure 2.21). For example, outpatient utiliza- is roughly constant regardless of income level tion rates have grown by about 3 percentage Jamkesmas Askes Private No and Figure 2.21 shows that PBI households in points in between 2012 and 2016 for Jamkes- /PBI /BPJS Insurance general have seen increases in healthcare uti- mas/JKN-PBI beneficiaries; and by about 2 per- Kesehatan lization rates. Yet, Figure 2.22 below shows that centage points for those without formal health PBI benefits—acquisition of healthcare ser- insurance. Inpatient rates have grown also by vices at a healthcare provider—are generally about 2 percentage points for JKN-PBI/Jamkes- 19.2 not concentrated among poorer households. mas beneficiaries; and by about 1.5 percentage 17.9 For example, while the bottom 30 percent of the points for those without formal health insurance. population accounts for just over 50 percent of 16 15.4 the PBI beneficiaries, the same bottom 30 per- Both private outpatient clinics and public cent accounts for only 32, 26, and 12 percent of 15.6 hospitals have accounted for most of the in- 13.9 all outpatient, all inpatient, or private-facility crease in PBI-facilitated utilization. For those inpatient utilization, respectively, accounted 12.6 with no insurance, private outpatient clinics, 14.2 for by PBI recipients or KIS cardholders. and private and public hospitals (in similar amounts) account for the increase in their uti- The variation in access and quality (private vs lization rates. As the increases in utilization public) for poor and vulnerable households in- (for PBI recipients at least) do not come dispro- dicates an interplay of various factors. In other 6.0 portionately from either poor or rich house- 5.6 words, high quality services may not be available 19.2 holds, the preference for private outpatient for poorer segments of the population lowering 3.3 facilities among PBI members, and those with the value of PBI-based access to health care. Based no insurance, suggests that the differences in on demographic and household characteristics 3.1 2.7 2.3 1.5 total household costs between service-provid- alone, poorer households would be expected to er types (see below) do not outweigh the per- prefer more health care than richer households ceived differences in quality. Source Susenas various years and World Bank staff calculations. (World Bank, 2013d). Nonetheless, the richest 52 Chapter 2 – JKN–PBI FIGURE 2.22 Outpatient, inpatient, private benefits, and beneficiary Outpatient incidence incidence of JKN–PBI by expenditure decile, 2016 Private inpatient Inpatient incidence Beneficiary incidence 20 18 16 14 12 10 8 6 4 2 0 1 2 3 4 5 6 7 8 9 10 DECILE The value of PBI is not uniform; it is proportional to the extent & quality of the services actually available. households represented in the Susenas survey in location-based and supply-system-based con- with “socialization” of the JKN program, and 2016 have outpatient, inpatient, and private-facil- straints on poor household access to a full PBI they have agreed to deliver program informa- ity inpatient utilization rates that are in the range benefit package. While these constraints are not tion specifically to JKN-PBI members, the only of 22, 131, and 610 percent higher than the poor- unique to, or produced by, JKN-PBI it nonetheless currently operational grievance and redress est households.66 suggests that an in-kind health transfer to poor system is the UDB-centered system, which cur- households should logically coordinate its activ- rently does not allow for dynamic updating Circumstantial evidence indicates that cost of ities and share its goals with service providers (a of potential beneficiary data. In addition, this access is a constraint: those provinces where la PKH; see the PKH volume in this report series). system is not capable of responding to claims the difference in the amount of private-facil- about, for example, denial of service (including ity outpatient care acquired by rich and poor Regular PBI M&E has catalogued disburse- through long waiting times), low quality ser- households is greatest—mostly provinces in ment and utilization rates. However, it does vice, lack of service, erroneous or disallowed remote eastern Indonesia—is also where the not explicitly monitor or target health or fi- charges, and any other facility- or supply-side- total amount of outpatient care acquired by nancial protection outcomes among beneficia- based deficiencies that reduce the value of ben- poor households is lowest. In other words, ries. The information collected and discussed efits received. in those areas where private-facility care is during a regular M&E cycle has been used in too costly for low-income individuals to ac- premium calculations and to remove con- cess, public facilities are either not providing gestion in the payment and re-imbursement a low-cost alternative, or they are providing a mechanisms.67 It was not used to revise health- low-quality alternative that low-income house- care delivery mechanisms to ensure qualitative holds do not value. In addition, transportation improvements in health outcomes in PBI-tar- and opportunity cost of travelling also contrib- geted households. ute to the cost of access. 66 If instead one first calculates average days of outpatient The UDB system provides the only direct ac- utilization (in the past month) by decile and by province and takes a 33-province average of the difference in province- The value of PBI is not standardized; it is cess to JKN-PBI for potential or actual bene- average outpatient days between richest- and poorest-decile proportional to the extent and quality of the ficiaries. Most preparatory implementation households, the richest households represented in Susenas 2016 outpace the poorest households by only 8 percent. services actually available. The variability in activity to date has focused on accomplishing There is even one province – DKI Jakarta – where the poorest households acquired nearly twice the number of outpatient health facility coverage and costs means, in a JKN program roll-out with regional govern- days (in the past month) as the richest households. essence, that the value of an in-kind PBI trans- ments.68 Currently and in the short to medium 67 For example, in 2011, the Ministry of Health and TNP2K developed a model and guideline to calculate the premium fer received by two similarly-aged individuals term, local governments play no role in pro- for Jamkesmas members. However, the JKN-PBI program administrator has calculated their own premium and cost with similar health status may be quite differ- posing additions or revisions to the benefi- structure for JKN members. ent depending on the service environment in ciary list, as this is done centrally via the UDB. 68 For example, there have been general agreements regarding health infrastructure, human capital, and public health which they are located. So there are pre-existing While regional governments have been tasked awareness campaigning. 53 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 C H A LLE N G E S , R E C O M M E N DAT I O N S & M OV I N G FO RWAR D JKN-PBI should continue to reduce always correlated with geographi- the standard has been reached for all the mis-targeting that has made it cal location, in the availability and households should contribute to an less effective as a social assistance quality of healthcare services. While increase in the value of the JKN-PBI transfer. A large portion of previous it is not JKN-PBI’s responsibility to benefit for poor households. mis-targeting has been due to the achieve meaningful reform in health significant local variation in eligi- care availability, quality, and delivery, While the goal of JKN is to increase bility determination and targeting it should at least explicitly recognize access to, and the quality of, health practices (World Bank, 2013d).69 that the more mutual assistance and services while prioritizing equity, Now, JKN-PBI members will be support there is between program many challenges lie ahead. Before selected according to the national administrators and providers, the 2014,71 organization, payment and targeting systems and procedures, more valuable the transfer will be the health service system in general so an emerging risk (for potential to low-income beneficiaries. were fragmented. The involvement beneficiaries) is that the “central of hospital associations, medical versus local” targeting pendulum JKN-PBI should monitor and at- professionals and academics will be swings too far in the other direction: tempt to remedy the unexpected70 a critical part of achieving Universal if there is no two-way updating to disparities in “benefit uptake”, or Health Coverage by 2019. The plan the JKN-PBI list or no UDB system the acquisition of facility-based to reach full population coverage via that can receive and act on updates health care, between poor and JKN by 2019 will require a not only from the field, the lack of local-level non-poor households. To reduce a tremendous increase in spending information may produce a distri- these disparities, JKN-PBI will need on health care, but also judicious bution of JKN-PBI beneficiaries that to do a much better job of explaining planning for allocating spending op- is unsatisfactory from a local-level to beneficiary households how to timally. Integrating existing health perspective. In order to mitigate this use the benefits for which they are insurance schemes at national, as risk, JKN-PBI should develop a ro- eligible. JKN-PBI households, for well as local, level will prove a great bust grievance and reporting system example, did not know which treat- challenge in the short term and re- that runs in parallel to a UDB-based ments, procedures, providers, and quires continuous monitoring and grievance reporting system. medicines were covered and which evaluation (TNP2K, 2015c). were excluded. This information 69 For example, the criteria summarizing M&E systems for JKN-PBI should be gap naturally reduced utilization household characteristics varied across districts; in some districts, midwives upgraded to monitor membership, rates (at the margin) and therefore and health center officials distributed health, healthcare usage, financial the value of the PBI program (World Jamkesmas cards according to their own criteria, regardless of economic status; protection and cost issues (from the Bank, 2012). While disparities in there were no incentives in the system to either maximize Jamkesmas enrolment or household side), and supply-side utilization are determined by factors minimize targeting errors while the list of preparedness indicators and out- other than information sets, such as eligible beneficiaries compiled by district officials was not subject to validation by comes. Currently, the supply side is the total cost of access to health care higher levels of program administrators. 70 and household composition alone would a weak link in the JKN-PBI transfer services of sufficient quality, estab- predict a higher rate of utilization for poor program: there is significant vari- lishing common information stan- households. 71 OPP payments are above average from a ability, which is usually but not dards for JKN-PBI and verifying that regional perspective. 54 Chapter 2 – JKN–PBI Given the great variation among districts in Indonesia, M&E will be paramount in achieving Universal Health Coverage. Based on study by TNP2K published in 2014, reaching UHC by 2019 most efficiently will require at least (TNP2K, 2015c): 01. Stocktaking of the distribution 04. Strengthening the DJSN (National 07. Synchronizing the provision of and needs of public and private Social Security Council) as the government health funding with health facilities; M&E institution for SJSN imple- this plan; mentation; 02. Promoting preventative medi- 05. 08. Increasing public sector financing cines through population-wide Adapting existing systems and of health care through qualified interventions; practices to encourage village investment strategies that pro- authorities to invest in priority mote supply-side readiness; and 03. health issues; 09. Raising accountability through monitoring and evaluation; 06. Developing a master plan to inte- Assessing payment structures and fund disbursement in coordi- grate the public and private health nation with government health sectors; priorities at multiple levels. Premiums and payments should of 2017, the premium paid by the M&E cycle now feeds into PBI pre- be calculated scientifically in or- Government for PBI recipients is mium calculations, as well as ra- der to encourage healthcare use, IDR 23,000. Likewise, JKN admin- tionalization o f reimbursement not to discourage the provision istrators have demonstrated con- and claims payment procedures. of services. For example, in July cern with health facility quality 2013, the premium for poor mem- improvement through the estab- bers (PBI) was calculated at IDR lishment of capitation grants and The capitation grant is a monthly amount 72 paid in advance to the Primary Health Care 19,225 per capita per month—a the generation of rules regarding Facilities (FKTP) based on the number of beneficiaries regardless the type and amount huge increase from the previ- the rational management and use of medical services provided (Presidential ously calculated IDR 6,000 per of these grants.72 Furthermore, the Regulation No. 32/2014 on the Management and Use of Capitation Grants for Quality capita per month “premium”. As information generated by JKN-PBI Improvement in Front Line Services). 55 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 Cash Transfer for Poor & At-risk Students (PIP) With a major expansion in coverage coupled with significant reforms to implementation—including to targeting, eligibility volatility, benefit size, and payment schedule—PIP, previously OVERVIEW known as BSM (Bantuan Siswa Miskin), has begun to demonstrate its full potential as a social assistance transfer. In order to deliver more, PIP should focus on continuous and coordinated monitoring, evaluation, and improvements in delivery. Most importantly, benefit-level updating should occur more frequently in order Indonesia enshrines education as a basic right for all cit- …while average for the PIP transfer to remain relevant; izens. From shortly after independence, the GoI has been devoting resources to initiatives for expanding enrolment PIP should consider positive outreach to poor students at the SMA level as they in primary, secondary, and tertiary education for all citizens. levels of education In 1970, the gross primary enrolment rate was under 70 per- face the highest out-of-pocket-spending cent; by 1994, when a mandatory 9-year basic education was and opportunity costs, and are at the enshrined, universal primary enrolment was the norm; by are steadily rising, 2016, gross junior secondary enrolment rates were topping greatest risk of non-continuation; 90 percent. Enrolment in tertiary education has risen from and PIP should find ways to pursue about 2,000 students in 1945 to around 5.7 million in 2016. students integration in program functions such as socialization, M&E, policy-setting, But while average levels of education are steadily rising, from poor and grievance redress. students from poor households remain far behind. In households 2016, 94 percent of 26 to 28 year olds73 from all income lev- els attained at a primary level, 6-year education. However, 90 remain far percent of the poorest 26 to 28 year olds attained a complete, behind. 6-year primary education (while for those from the richest households the proportion remained at almost 100 percent). Even a single year of post-primary education is out of reach for many children from poor households: attainment rates of at least a 7th-grade education drop to 51 percent for those in the poorest households (while the number for those in the richest households is about 90 percent). Achievement gaps at the senior secondary level are larger: 2016 attain- ment rates of at least 10th-grade education are 50 percentage points lower in the poorest than in the richest households.74 While the primary school net enrolment rate has been com- fortably over 90 percent since the early 2000s, these good starts do not lead to high educational achievement for poor 2.4 households. Students from poorer households drop out in large numbers during the transitions from primary school to junior secondary school and from junior to secondary school (Figure 2.23). 73 The majority of individuals in this cohort are expected to have finished their educational careers; in other words, their attainment as recorded in the Susenas household survey is expected to be their lifetime educational achievement. Younger cohorts are still enrolled in significant numbers, while older cohorts progressed through an education system that differed in important ways from the current system. 74 In 2010, attainment rates of at least 7 years of education were 44 and 90 percent for 26 to 28 years olds in the poorest and richest quintiles, respectively, while the 2010 rate of achievement of at least 10 years of education was 50 percentage points higher in the richest quintile than in the poorest quintile. 56 Chapter 2 – PIP FIGURE 2.23 A key constraint for poor households is the fi- nancial cost of access. Figure 2.24 below shows Educational attainment, 26–28 year olds by that enrolling a student is costly: out-of-pocket consumption quintile, 2016 (percent) costs (including transportation) range from around IDR 1 million to over IDR 3 million (US$77 to US$230) (depending on school level), 120 with poorer households paying slightly less and richer households slightly more. Secondary 100 education (SMU (Sekolah Menegah Umum), or SMA (Sekolah Menegah Atas) and SMK ((Sekolah Q1 Q2 80 Menegah Kejuruan)) can be prohibitively expen- Q3 sive for the poorest households: regular costs to Q4 send one child to senior secondary school con- Q5 60 sume about 18 percent of a poor household’s overall budget. For poor households, the oppor- 40 tunity costs of education—incurred as foregone income when a child attends school instead of 20 working—will be relatively larger also, making YEARS OF EDUCATION secondary education doubly expensive.75 1 2 3 4 5 6 7 8 9 10 11 12 >12 0 Registration and other school fees such as Source: Susenas 2016 and World Bank staff calculations. tuition and school committee fees make up the bulk of these education expenditures, but FIGURE 2.24 quasi-discretionary items such as uniforms, books, and supplies also account for signifi- Household education expenditures per student, by school level, 2015 (IDR Million) 1,000,000 cant shares, especially at the primary school level. Notwithstanding sizeable government TO mOrE funding streams such as BOS (Bantuan Oper- asional Sekolah)76 and 2008 legislation prohib- iting fees at public education institutions, the THAN total cost of education has risen between 2009 and 2012: average household education expen- 3,000,000 ditures (not including transport) have risen by about 13, 19, and 15 percent at the primary, junior secondary, and senior secondary levels, respectively.77 “…enrolling a student is costly: Program Indonesia Pintar (PIP) lowers the net out-of-pocket cost of access to education by providing cash costs (including transfers directly to poor students. Students transportation) from poor households who are verified to be range from around eligible by their school78 are provided annu- IDR 1 million to al cash transfers of IDR 450,000 (US$35), IDR over IDR 3 million 750,000 (US$58), or IDR 1 million (US$77) for (US$77 to US$230) enrolment costs at the primary, junior second- (depending on school ary, and senior secondary level, respectively. By level), with poorer lowering the enrolment cost hurdle in a target- households paying ed way, the GoI hopes to tackle the low enrol- slightly less and ment rates and high dropout probabilities of richer households poor students, and eliminate the education gap slightly more.” (World Bank, 2012h). PIP also helps the GoI meet its constitutional guarantees by provid- ing incentives for all children to complete at least a 9-year basic education. SMP . NON POOR HH SMA . NON POOR HH SD . NON POOR HH 5, 469, 088 2 ,246, 640 3, 43 5, 454 1 ,0 99,57 5 2 ,1 90, 1 50 SMP . POOR HH SMA . POOR HH 3, 747, 81 7 SD . POOR HH 75 Costs reported in Susenas are considered official payments. There are no qualitative studies investigating informal payments or elite capture. However, the presence of categories such as ‘fees for courses’ and ‘others’ that fall outside of the official fees category show why real costs for school are markedly higher than official estimates. 76 Bantuan Operasi Sekolah, or School Operational Aid. 77 Expenditures on university level education have increased by over 60 percent in nominal terms between 2009 and 2012. 78 The role of schools, school committees, and local education Source Susenas 2015 and World Bank staff calculations. stakeholders in selecting beneficiaries was revised somewhat Note Average transportation costs are calculated over households who indicate use of public when the PIP program agreed to use UDB procedures for or collective transport. preliminary identification of PIP-eligible students. 57 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 FIGURE 2.25 PIP targeted coverage and budget by year Expenditures PROGRA M SI Z E , Target beneficiaries INSTITUTI ONA L S E T - U P , 16,000 25 E LIGIBI LI TY, & B E N E FI TS 20.4 19.5 14,100 TARGET BENEFICIARIES (MILLIONS OF STUDENTS) PIP has expanded every year; from 2012, the pace of expansion has picked up dramatical- ly (Figure 2.25). In the year it began operations 16.6 (2008), Program Bantuan Siswa Miskin (BSM) achieved an expenditure level just over one- BUDGET (IDR BILLION) fifth of the size of the then-largest program, the 10,572 rice subsidy program known as Beras Sejahtera or simply Rastra (see the Rastra program note in this series).79 in 2016 PIP reached about 19.5 11.2 million students while expenditures reached 9.5 IDR 10.5 trillion, or over 8 times the 2008 lev- 8.2 el. While PIP is still only the third-largest (in 6,600 terms of expenditure or coverage) of the per- 6,388 5,400 manent social assistance programs, its average 5.8 share of the social assistance budget has in- 4.9 4,700 creased from 9 percent in 2005-10 to 14 percent 4.6 in 2016 (see the Expenditure Summary report 3,607 in this series).80 2,562 PIP is a national-coverage cash transfer given to enrolled students or school-age children 1,238 from the poorest 25 percent of households, who have either a Kartu Indonesia Pintar (KIP) card or a Kartu Perlindungan Sosial / 0 0 2008 2009 2010 2011 2012 2013 2014 2015 2016 Kartu Keluarga Sejahtera (KPS / KKS) card (see Box 2.5). 81 In 2016, PIP targeted 19.5 mil- lion enrolled students between 6 and 21 years of age. Beneficiaries of other social assistance programs—for example, the conditional cash Source MoF 2008-14 are budgeted totals, 2015 is realised and 2016 are realised budget. transfer called PKH—are automatically PIP-el- igible. PIP provides transfers to students in any school level from primary (Sekolah Dasar, or SD) to secondary (both junior secondary The major changes when or equivalent, SMP, and senior secondary or equivalent, SMA).82 The two ministries respon- sible for education in Indonesia, the Minis- 79 Rastra was the largest of the permanent moving from BSM to PIP are related to eligibility social assistance programs in 2008, but try of Education and Culture (MoEC) and the there was also a temporary cash transfer, Ministry of Religious Affairs (MoRA), both called BLT, in that year; see the BLSM deliver a PIP program for students in regular report in this series for more information on these temporary (i.e., non-repeating) cash criteria: school children not yet receiving PIP in transfers. and madrasah schools, respectively.83 Major 80 In 2014 one of BSM implementing agencies, changes involved in moving to PIP from BSM the Ministry of Education and Culture, are that school children not yet receiving PIP re-directed discretionary expenditures (including BSM expenditures) to curriculum either formal or non-formal education institutions, or reform and the BSM budget was reduced in either formal or non-formal education insti- considerably. It is unknown what happened to tutions, or those who are not attending school, the bulk of BSM beneficiaries whose benefits can reach out and register for PIP cash trans- were cut in between the 2013 and 2014 fiscal years. those who are not attending school, can reach out and 81 World Bank (2014d). From 2014 onwards, fer provided their family has a KPS/KKS card. PIP will encompass the Kartu Indonesia register for PIP cash transfer The BSM program targeted children that were Pintar (KIP) program as well. KIP targets school-age children who are currently not already attending school and were listed in the enrolled or who are enrolled in training provided their family has a courses that are not part of Indonesia’s UDB as poor or vulnerable. With PIP, there are mandatory 9-year curriculum. KPS/KKS card. more mechanisms (most are linked the social 82 SMA is used here to refer to senior high school but when referring to Susenas data, protection cards KPS and KKS) that allow stu- vocational school level, SMK, is also included. 83 In 2016, MoEC manages approximately dents to become enrolled in PIP, the aim being 92 and MoRA 8 percent of all primary, to get out-of-school children back to school and secondary, and technical school scholarships (according to program administration to increase take-up rates (TNP2K, 2016). documents) 58 Chapter 2 – PIP The PIP program continues to operate as sev- Payment disbursement and final PIP nomina- BOX 2.5 eral largely independent initiatives in two tion is announced by the MoEC and MoRA, via Developing Productive Families, separate ministries, with neither a central co- a decree and recipient lists sent to district level Kartu Indonesia Pintar (KIP) ordinating unit nor a unified budget. Within education offices and to payment institutions the MoEC, PIP budgets and administration are BNI and BRI banks under MoEC and MoRA Program Indonesia Pintar (PIP) and the as- managed and implemented separately by the management since 2013.). Then, recipient lists sociated Kartu Indonesia Pintar (KIP) are separate directorates corresponding to school and disbursement times and locations are sent managed by the Ministry of Education and levels. Within the MoRA, the program is frag- to schools directly from the district education Culture and the Ministry of Religion. PIP, or mented by the type of religious school: primary offices. The schools then notify the students KIP, are part of the new scheme for SA imple- to senior high cash transfers are administered or parents about the time and location of their mented by Indonesia’s current Government. by the Secretariat General with a majority of PIP benefit disbursement. Since PIP uptake The goal of PIP is ensure children of school- resources allocated to Muslim schools and a depends on students bringing their KIP, KKS ing age in poor and vulnerable households smaller share for each of the remaining official or KPS to school, since not all targeted students complete high school. This, through providing access to cash transfers to cover the out-of- religions (Christian, Catholic, Hindu and Bud- may make use of PIP, schools and local educa- pocket costs of attending school. dhist). The scholarship program for students at tion officials may nominate PIP students not the tertiary/university level has been renamed on the UDB-generated list only if the PIP quota Coverage & eligibility Bidik Misi and is managed and implemented for the district is not filled yet. The criteria for Since 2014, the Government has started to by the Ministry of Research, Technology, and nomination are meant to include school-aged distributed KIP cards to replace KPS cards Higher Education, separate from PIP. children (6-21) having characteristics such as as the primary means of proving program el- living in a PKH family; having a higher risk of igibility. Distribution was done through three PIP is allocated using the Unified Database non-continuation because of financial difficul- phases. The first phase was to be conducted (UDB). 84 After the UDB is queried and returns ty; living in an orphanage; being a victim of a on March-April 2015, the second phase was to a list of students from the poorest 25 percent natural disaster; and those that are no longer be conducted on June-August 2015.For the of households, KPS or KIP cards were distrib- going to school. (TNP2K, 2016)86 third phase, in 2016, 19.5 million cards were dis- uted to households (via a delivery specialized tributed by MoEC and MoRA to PIP recipients. firm. Registration to PIP differs by several cat- PIP benefits were raised slightly (Figure 2.26) egories, whether the school is under the MoEC but a large gap remains between the PIP val- Overall eligibility for KIP as of 2016 is for or the MoRA, whether the school is formal or ue and the total out-of-pocket cost for house- school aged children, 6 to 21 years of age: in KPS card holding families that were receiving not, whether the student has a KIP card or not holds. According to official estimates, the costs BSM in 2014, KPS/KKS holders who are not and whether the family has a KKS or KPS card. for one child to attend a full year of school yet receiving PIP benefits, in PKH families, In general, eligible students (or their parents (including transportation, food, uniforms and living in orphanages, that have dropped out or guardians) take their KIP, KKS or KPS iden- materials, and most of the other items listed of school due economic difficulties or natural tification and register with a school in order before) are IDR 450,000 (US$35) for primary, disasters and those that are not yet going to to access PIP (World Bank, 2014d). Upon com- IDR 750,000 (US$58) for junior secondary, IDR school or dropped out based on recapitulat- pleting these registrations and also adding 1 million (US$77) for senior secondary school; ed data as of the second semester of 2015 school-identified PIP beneficiaries not on the and IDR 1.8 million (US$138) for university (TNP2K, 2016). officially-eligible list, the schools send a pro- students (TNP2K, 2016). However, households posed PIP registry to a District Management report greater expenditures (in 2012) at about Program flow and benefit structure Team for verification. All district-level teams IDR 1.0, 1.5, and 2.0 million (US$78, US$115 and Eligible households are sent the card (along then forward the registries to the PIP Central US$154) for a single student in primary, junior with the other cards listed above) via two Management Team.85 secondary, or senior secondary school, respec- banks. Using the existing PIP program, card tively. In other words, official estimates appear holders are entitled to receive benefits as to be too low by about half. There is no ben- shown below the diagram. While in the future, KIP benefits are planned to be disbursed by efit difference between urban and rural areas, both banks and post offices, the payment de- although in general expenses are expected to be livery structure appears to be using the same higher in urban areas. method as the older BSM program: upon con- firmation at the school of registration, funds are released in cash at EDUCATION PRE 2013 PIP REVISED 2013 LEVEL BENEFIT PER AND CURRENT STUDENT PER PIP BENEFIT SEMESTER PER STUDENT PER SEMESTER SD (primary) IDR 190,000 IDR 225,000 (US$15) (US$18) SMP (junior IDR 275,000 IDR 375,000 84 An additional validation – for PIP-eligible beneficiaries determined by the UDB – is completed by matching PIP-suggested high) (US$22) (US$30) beneficiaries with the Dapodik database, which is an integrated database system used by the Ministry of Education and Culture for national education program planning. SMU (senior IDR 375,000 IDR 500,000 85 These procedures describe PIP implementation in both the MoEC and MoRA. high) (US$30) (US$40) 86 Schools may also be able to remove a listed beneficiary from the PIP program by determining that the eligible beneficiary is not poor. Prior to the 2012-2013 academic year, BSM allocation was based exclusively on referrals from schools (in coordination with school committees) and targeting accuracy was low: as little as 10 percent of all BSM beneficiaries at the primary school Source: TNP2K, 2014k and local media reports. level fell into the poorest category (TNP2K, 2014a). Program regulations state that for PIP in the future, proposed beneficiaries will be named beneficiaries only after the Government has determined the KIP take-up rate for each education level. 59 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 FIGURE 2.26 CPI inflation-adjusted PIP benefit levels Senior high school Junior high school BSM/PIP benefit levels adjusted for inflation (IDR) Primary school 780,000 742,455 704,797 700,492 602,125 654,444 583,821 621,004 615,690 550,000 523,526 528,598 493,937 451,594 437,866 465,753 461,467 434,141 342,671 550,000 323,304 270,957 262,720 317,159 302,051 284,165 279,452 2008 2009 2010 2011 2012 2013 2014 2015 2016 Source: MoF, BPS, and World Bank staff calculations. C OVE RAG E , TARG E T I NG, & I M PACTS The PIP transfer value has eroded over time. PIP benefits have been re- PIP covers almost 30 percent of poorest decile households that have vised once in the program’s history and the real value of PIP benefits de- school-age children 89 (Figure 2.27). However, coverage among non-tar- clined by about 27 percent between 2008 and 2012, if adjusted using con- geted households ranges between 13 and 21 percent for median-income sumer price index-based inflation. Benefit levels were increased in 2013 households, and between 1 and 11 percent for the richer households. by 25 to 35 percent (excluding PIP at university level which increased Households with children are more likely to be covered by PIP if they by 50 percent), but regular inflation means that even with these benefit are living in rural areas. Indeed, the rural share of PIP beneficiaries is 63 amount increases households receive as much in real terms in 2015-16 percent in 2016. as they did in 2012. Furthermore, the increase in the total out-of-pocket cost for a household to enroll one student has outpaced general infla- On average, PIP is received by students in households that exhibit cor- tion, increasing by as much as 35 to 60 percent from 2009 to 2012.87 PIP relates of income poverty. PIP’s overall coverage for the poorest (in 2016), benefits are less adequate in both real terms and in terms of education at about 18 percent of households with at least one school-age child, is purchasing power. The failure to adjust for rising costs of living and the around 12 percentage points higher than the 2013 level. These additional rising real cost of education could be undermining PIP’s objectives. PIP funds have, on average, been distributed to students in households sharing most of the non-income correlates of poverty (Table 2.6). PIP transfers are received by households in advance of the period in which school fees are levied. 88 The first of two tranches is received be- Following continued expansion and revisions to targeting practice, PIP tween August and September. The second is received between March and has improved its ability to identify students from near-poor house- April in the next calendar year. This disbursement schedule is expected holds. In order to facilitate comparisons between PIP and other social to reduce dropout rates. Since most PIP beneficiaries are now selected by assistance transfers (which may have slightly different target groups), querying the UDB, the likelihood of an interruption in the PIP transfer— Figure 2.28 shows coverage and incidence for the “Poor”, the “Next 30” that is, receiving PIP in one year but not receiving it the next—has been percent, and “The rest”. Figure 2.29 demonstrates that while the pover- lowered; this is also expected to reduce dropout rates. ty headcount rate fell by less than 1 percentage point between 2012 and 2016, PIP expanded significantly and the share of PIP benefits accounted for by the “Poor” group fell by about 1 percentage point. The “Next 30” group, which contains the same proportion of near-poor households in every year, has a PIP share that grew by about 6 percentage points over 87 World Bank (2012g) notes that expenditures on education the same period. Finally, while PIP coverage among “The rest” has grown for poor households rose in real terms by 20 to 50 percent between 2006 and 2009. With inflation (as measured by the larger (by about 6 percentage points), this group’s share of PIP benefits consumer price index) at about 16 percent in between 2009 and 2012, education expenditures for all households have has fallen by 5 percentage points. While the 2012-16 trend looks positive, risen in real terms by about 20 to 45 percent between 2009 the changes in incidence of the “Poor” and “The rest” fell by 2 percent- and 2012. 88 This payment schedule was implemented beginning age points and increased by 5 percentage points, respectively, between academic year 2013/14. See World Bank (2012g) for more details on the previous mismatch between BSM transfer 2015 and 2016, indicating that the expansion between 2015 and 2016 may receipt and payments for school-related bills. have led to poorer targeting outcomes. That said, over the whole period 89 That is, for households with at least one child from 7 to 22 years of age. since 2012, the pattern indicates that as the micro-level poverty situation 60 Chapter 2 – PIP FIGURE 2.27 FIGURE 2.28 PIP coverage among households with school-age PIP coverage and incidence children by expenditure decile, 2016 (percent) by poverty groups 32 Poor The rest Urban Next 30 30 30 29 Rural 22.8 25 25 COVERAGE (% 18.8 OF HOUSEHOLDS 21 21 RECEIVING 20 BENEFITS) 17 16 7.13 14 13 3.28 2.41 12 1.38 2.3 11 1.5 10 0.6 7 7 4 2009 2012 2016 1 INCIDENCE (% OF TOTAL 1 2 3 4 5 6 7 8 9 10 BENEFICIARIES BY GROUP) DECILE 43.0 44.2 41.8 Source Susenas 2016 and World Bank staff calculations 38.2 38.0 30.6 27.6 19.0 Following continued expansion & 17.7 revisions to targeting practice, PIP has improved its ability to identify students 2009 2012 2016 from near–poor households. Source Susenas (various years) and World Bank staff calculations. changes—many households exit poverty year available for students from near-poor house- C to year, while fewer enter—PIP’s most recent holds. Figure 2.29 illustrates that the bulk of COVERAGE (% OF HOUSEHOLDS RECEIVING BENEFITS) BY EXPENDITURE DECILE implementation and coverage revisions have this increase is due to PIP transfers at the pri- allowed it to continue to find the remaining eli- mary and junior secondary levels.92 However, 40 gible poor and near-poor households. the targeting accuracy of PIP cash transfers 35 for primary (SD) students has not increased 30 Poor and near-poor households receive the markedly since 2012. SD targeting outcomes majority of PIP benefits available (Figure increased slightly in 2014 (not shown): from 68 25 SD SMP 2.29). 90 Students attending SD, SMP and SMA percent of benefits in the poorest 40 percent to 20 or equivalent levels of schooling in the bottom almost 70 percent), but fell back to 68 percent 15 SMA 30 percent of households account for about in 2015 and 2016. For SMP, targeting outcomes 57, 55 and 45 percent, respectively, of all PIP fell by almost 5 percentage points, but appear 10 transfers. Leakage to students in non-targeted to have increased again in early 2016 by 1 per- 5 households is still significant, as such students centage point. That said, the program went in the fifth decile or above captured 31, 32 and through a major expansion in those years and 0 1 2 3 4 5 6 7 8 9 10 43 percent of the total PIP benefits distribut- between 2015, and SMP and SMA PIP targeting ed to each school level in 2016. For example, accuracy actually increased at the margin (both 1 Expenditure decile households with at least one school-age child91 percentage point higher allocation of benefits in in the bottom 4 expenditure deciles accounted the poorest 40 percent). For the “Poor” between D INCIDENCE (% OF TOTAL BENEFICIARIES for almost 70 percent of the PIP benefits avail- 2015 and 2016, incidence to the poor of program BY GROUP) BY EXPENDITURE DECILE able at both the SD and SMP levels. At the SMA benefits in general fell slightly, but by almost 3 level, the analogous number is only 57 percent. percentage points for the SD school level, suggest- 25 SMP Among the “Poor”, “Next 30”, and “The rest” ing a decrease in targeting accuracy. 20 SD groups, the poor’s share of PIP-SMA benefits decreased between 2012 and 2015. More pro-poor allocations of SMA-level cash 15 transfers will require continued coordinated SMA 10 We have seen that even as poverty rates fell, effort. The revised UDB-based targeting and PIP has provided a greater share of benefits KIP/KPS/KKS eligibility determination proce- 5 0 90 Incidence corresponds to the ranking of beneficiaries per decile as proportion of total beneficiaries. Since PIP benefits change a household’s per capita consumption, then to fairly reflect targeting of pre-program household consumption, consumption has 1 2 3 4 5 6 7 8 9 10 been adjusted by the amount the household has reportedly received from PIP. 91 We have taken coverage and incidence only over households with at least one school-age child in the relevant school-level Expenditure decile range, so for primary (SD equivalent), age 7 to 12; for junior secondary (SMP equivalent), age 13-15; and for senior secondary (SMA equivalent), age 16 to 18. 92 “Basic schooling”, which the Indonesian Constitution indicates is the right of every citizen, is defined as 9 years of schooling, Source: Susenas 2016 and World Bank staff calculations. from primary through junior secondary. 61 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 TABLE 2.6 FIGURE 2.29 Characteristics within Net enrolment rates (%) by school level Indonesian populations (2000—2016) 100 % of all Indonesians % poor population % of PIP recipients who: who: who Do not have access 16 28 16 to bottled, tap or well 80 SD water Do not have access to 22 41 32 private sanitation 60 Live in rural areas 49 63 63 SMP Live with more than 5 22 38 35 household members 40 Have not completed 9 13 8 primary education* Are illiterate 7 13 6 SMA /SMK 20 Work in the 33 56 44 agriculture sector 0 2000 2016 Source: Susenas (various years) and World Bank Source: Susenas 2016, and World Bank staff calculations. *For members of a household receiving PIP that are older than 18. staff calculations. dures (see above) have reduced the risk that PIP issues simultaneously, and in coordination FIGURE 2.30. beneficiaries will become ineligible (from any with both schools and potential students, will given school year to the following school year). it be likely that PIP-SMA allocations improve Net enrollment rates (%) by school level In addition, the revised payment schedule (see substantially. and poor / vulnerable welfare status above) is expected to reduce the risk that stu- All dents cannot meet the payments necessary to Significant enrolment rate increases in 2015 Vulnerable begin a school year, thereby reducing drop-outs. suggest a positive role for the much-expanded, Poor reformed PIP program. Figure 2.30 indicates Fewer students enroll for upper-secondary ed- that after 15 years, during which annual average 41.1 SMA 51.1 ucation from poor and near-poor households. net enrolment rate (NER) increases were 0.8 /SMK 59.9 The lower poor- and near-poor household in- and 1.0 percentage points at SMP and SMA lev- cidence at the SMA level is likely due in part els, respectively, the NER increased by 4.1 and 70.0 the significantly higher costs associated with 4.6 percentage points, respectively, in 2015. In SMP 75.2 attendance (and the relatively low PIP benefit other words, the magnitudes of the 2015 NER 77.9 level). In addition, increased PIP nomination increases at SMP and SMA levels were equiv- 96.4 by schools (at the SMA level) is also correlated alent to 25 and 29 percent, respectively, of the SD 96.9 with high drop-out rates and low enrolment total 2000 to 2014 increases in the NER. This 96.8 rates for poor and near-poor households.93 In does not prove conclusively that the signifi- other words, school-based PIP nomination is cant PIP expansion, coupled with operational Source: Susenas 2016 & World Bank staff calculations greater at the SMA level where there are fewer reforms enacted, are responsible for the abnor- UDB-based PIP beneficiaries and higher drop- mally large increase in the NER. However, it out rates and lower enrolment rates are logi- does put into clear relief the lack of NER move- cally tied to expectations about cost. It is two ment in the years following the establishment to three times more expensive for one year of of PIP (2008) or its first large expansion be- SMA education than for one year of SD educa- tween 2010 and 2012 (World Bank, 2012g).95 As tion, while the share of this cost that a PIP trans- of 2016, the increases in the NER were smaller, fer can be expected to cover is low (and decreas- at around only 0.1 percentage points for each ing).94 Only when PIP can address all of these level of schooling. 93 According to Susenas 2016, nearly 45 percent of eligible children from the poorest 20 percent of households were not attending SMA. 94 In addition, the opportunity costs of education (measured as foregone wages and household production) rise as a child acquires more education, so the total cost (out of pocket costs plus opportunity costs) of one year of school at the SMA/SMK level is likely even greater. 95 Most of the operational or implementation-based shortcomings of earlier versions of the BSM program previously thought to be limiting the program’s impacts—the benefit size; the mismatch between disbursement and school fee schedules; and the focus on enrolled students selected by schools—were at least partially addressed in the recent reforms. 62 Chapter 2 – PIP cessing PIP benefits, initial findings suggested students at all levels. In a best-case scenario, under-utilization of, and a lack of socializa- transfer levels would automatically adjust to tion about, the program and the cards, while any increases in the cost of schooling by, for ACCESSI B I LI T Y confusion over local roles and responsibilities caused payment delays and data inaccuracies.98 example, tying benefit-level calculation di- rectly to cost-of-schooling indices reported by In 2014, the working group began develop- households. PIP’s current operation led to only ing guidelines for PIP recipients to report to one transfer level revision in the eight years PIP policy, implementation, and monitoring schools “out of cycle” in order to avoid delays of the program’s operation, while the general are executed by two ministries and several di- in the data recapitulation process, while con- price level has risen by 45 to 50 percent during rectorates with little coordination and few sidering a more integrated monitoring and the same period.99 In 2009 and 2012 (years for common monitoring or feedback mecha- evaluation program—including a shared Man- which there is detailed education expenditure nisms. The internal PIP monitoring that is agement Information System (MIS) that is al- data), PIP transfer magnitudes were less than achieved—for example, over the selection pro- ready in use at the MoEC —to track PIP-reform half of the total per-student cost of education cess, the resulting cash transfer allocation, fund progress (IPC, 2014). as reported by households. distribution realization, or the withdrawal pro- cess—is carried out by the directorate hosting PIP should consider alternative ways of pro- the particular cash transfer program (together moting enrolment, re-enrolment, and con- with provincial and district offices).96 There is tinuation within the group of students most no defined procedure that cumulates the infor- C HAL L E N G E S , likely to be inactive in the education system, mation into a PIP-wide program improvement namely SMA-aged students in poor and near- cycle and no incentive for PIP operators to R E C O M ME N D AT I O N S & poor households. As the gap between house- share results or experiences. hold expenditures for a year of education and MOVI N G F O RWAR D the PIP transfer magnitude is largest at the SMA PIP socialization and complaint-handling level and, not coincidentally, this is also the lev- may be improving. A working group consist- el where PIP struggles to find and keep enrolled ing of TNP2K, and the implementing minis- PIP should continue building on the recent targeted poor and near-poor recipients, PIP im- tries MoEC and MoRA, increased PIP social- successes it has had in program implementa- plementers should develop strategies and prin- ization effort and presence in 2014. Through tion and positive contributions to household ciples specifically for the recruitment of poorer stakeholder coordination meetings, traditional welfare. With the contemporaneous increase students at higher education levels. and social media campaigns both within and in coverage and reforms to implementation, external to the government bureaucracy, the PIP is now making significant positive con- PIP’s biggest hurdle to further improvements production and distribution of new socializa- tributions to welfare in poor and near-poor in delivery is its current institutional form. tion materials, and larger event-based socializa- households (with students) and to the Govern- PIP is split among two ministries and several tion campaigns in large urban areas estimated ment’s drive to provide universal basic educa- directorates, each of which carries out most to have significant shares of potential PIP bene- tion. When coverage and expenditures rose program functions independently. While some ficiaries, the central government has expanded dramatically between 2012 and 2016 and about effort has been made since 2013 by TNP2K and PIP program information breadth and depth. 10 million more beneficiaries were added, the Kemenko PMK coordination is still lacking. In The same working group has also re-examined UDB-based targeting system that PIP adopted addition, there is currently no mechanism to existing complaint- and grievance-handling helped put a significant majority of those trans- jointly provide (and jointly review the effec- mechanisms, including PIP-generated griev- fers in the hands of students from poor and tiveness of), for example, management perfor- ances concerning the UDB and the KPS card near-poor households. A further positive con- mance reviews, M&E efforts, socialization cam- system. New PIP guidelines suggest that access sequence for PIP of switching to the UDB-based paigns, a grievance redress platform, or a policy to these mechanisms should be expanded, per- beneficiary selection and eligibility procedure review of the suitability of a “transition bonus” haps via the support of local governments as is that it has reduced eligibility volatility— for eligible students making the leap from one well as continuing efforts by central program that is, when PIP beneficiary eligibility status school level to the next. In particular, the lack administrators. On the TNP2K website a full list switches from year to year because of an unpre- of a centrally managed M&E and the capacity of contact details for grievances based on the dictable selection process. The revisions made to conduct case-by-case outreach to PIP bene- type of PIP benefit given is provided (IPC, 2014; to the payment schedule increased the accessi- ficiaries will continue to be a weakness for PIP. TNP2K, 2016). bility, and therefore the relevance, of PIP trans- In other words, even though the recent signif- fers among students and households expecting icant reforms have undoubtedly improved the Efforts are being made to improve targeting large enrolment (and continuing) costs. PIP program, there remain many opportunities protocols for the PIP context. When “spot for better program integration and monitoring checks“97 were carried out to assess the effec- Transfer levels should be reviewed and adjust- that can in turn provide a better experience for tiveness of the new card-based system for ac- ed more frequently if PIP continues serving students and households. 96 PIP students receive transfers upon enrollment but there is no further verification or monitoring of school attendance. 97 Carried out by interviewing 632 households with at least one grade-7-age child from 15 districts in 8 provinces. 98 For example, only 22 percent of the cards delivered were utilized during the study period. 99 Inflation in household-reported education costs has been about twice the level of inflation (in the general price level) in the years it has been measured. 63 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 Conditional Cash Transfer (PKH) PKH has demonstrated positive impacts in consumption and health- seeking behavior (and minor impacts OVERVIEW in education) for poor families and the communities in which they live. Its M&E system has been able to provide internal management indicators, as well as indicators of household experience that feed into a program improvement cycle. PKH continues to experiment Continuous improvements in education and FIGURE 2.31 health outcomes for all citizens have long with program guidelines and benefit been a focus of GoI social policy. The 1945 Educational and health attainment packages in order to remain relevant for Constitution establishes the right of Indone- at a glance all eligible beneficiaries but will need to sian citizens to quality education and health services. In the post-independence and Suharto ensure compliance verification functions Have attained a degree eras, economic development strategies focused well and redesign the program to on financing capital investment in education, have finished basic education (SD + SMP) have completed higher education operate in remote areas. While PKH does health, and related social services. Even as 87 not get everything right, it should be the Asian financial crisis unfolded in the late 83 1990s—with the headcount poverty rate dou- 73 encouraged to continue innovating and 69 64 bling, real economic activity contracting by growing in capacity to better support over 13 percent, and the Suharto regime even- 51 49 beneficiary families as it expands. tually being removed from power—spending 26 28 20 on health and education did not fall from pre- vious levels. To the contrary, a constitutional 0.2 0.9 amendment in 2000 reaffirmed the rights for all citizens to education and medical care, and Poorest Richest Poorest Richest decile decile decile decile legislation in 2003 obligated the nation to pro- vide education for all children 7 to 15 years of age. Indonesia has made great strides in these ar- 2007 2 01 6 eas, but poor households continue to lag be- hind. For example, while educational achieve- Used doctor or midwife for childbirth ment continues to rise for all groups (Figure Recipient of health insurance 2.31), in 2016, 85 percent of people living in poor households will not have completed SMP 98 or equivalent level of school and virtually none 83 of them will have completed higher education. In 2016, about 9 percent of children from the 45 poorest households did not receive a single 40 39 26 immunization and there has been only slight improvement in this indicator since 2007. 6 7 Program Keluarga Harapan (PKH), launched in Poorest Richest Poorest Richest decile decile decile decile 2007, is a conditional cash transfer for poor households100 meant to alleviate short-term 2.5 poverty and increase investments in educa- tion and health. PKH households receive cash 2007 2 01 6 transfers when individuals meet specified health or education requirements. The cash transfers provide welfare in the short term and also reduces the opportunity cost of acquiring those services, while the requirements them- selves should lead to improvements in the lon- ger term. PKH is targeted to families rather than households, for the 100 Source: Susenas 2016 and World Bank staff purposes of this review the terms are used interchangeably. calculations. 64 Chapter 2 – PKH PRO G RAM S I ZE , I N ST I TUT I O N AL S E T U P, E L I G I BI L I T Y, & BE N E F I TS PKH’s steady expansion has led to benefit PKH’s total budget has increased by nearly the PKH is executed by the Ministry of Social Af- availability in 34 of Indonesia’s provinces and same factor as has household coverage: from fairs (MoSA) with funds disbursed to house- coverage of nearly 6 million families (Figure under IDR 1 trillion in 2007 to over IDR 8 tril- holds through a collection of state owned 2.32). PKH, which was rolled out in seven prov- lion in 2016 (Figure 2.32). Even though it has banks (2.34). Before 2017, payments were man- inces and to just under half a million families expanded significantly, PKH remains the small- aged by the postal system (PT. Pos). A central- in 2007, had by 2016 expanded coverage six est of the national social assistance transfers ized program implementation office within times over (to over 3.5 million families) in al- (Figure 2.33) with less than 50 staff supporting MoSA102 oversees all stages of program imple- most all provinces, including those in eastern its implementation at the central level. Rastra mentation. In the first stage of the program, Indonesia.101 The Government has planned to had an expenditure share about 2.5 times as province- and district-level quotas are negoti- expand PKH to 10 million families in 2018. large as PKH’s in 2016, but PKH estimated to be ated and agreed. Then the UDB is queried to far more effective at reducing inequality and extract a list of eligible beneficiaries (eligibil- poverty, than Rastra (World Bank, 2016a). ity requirements are discussed below). MoSA then distributes that list to its local offices, which are responsible for confirming eligibil- ity. Upon verification of compliance, payments FIGURE 2.32 FIGURE 2.33 are authorized by MoSA and budgeted funds are PKH coverage & budget Central government spending on disbursed to payment service providers, which in social assistance programs turns transfers funds to regional branches. The PKH cash benefit is then transferred directly to Budget (IDR Trillion) Rastra PKSA mothers only. Starting in 2016, following the di- Beneficiary households (Millions) 6.0 JKN/PBI JSPACA rection of the National Financial Inclusion Strat- PIP ASLUT egy, MoSA has begun to shift the PKH payment BLT/BLSM/KKS&KSKS PKH from previously cash-based model to an elec- tronic cashless model supported by a group of state IDR TRILLION owned banks (MoSA, 2016d) with the aim to render YEAR 0 25 50 75 100 all payments via this system by 2017. 2004 Verifying the household achievements in 8.5 2005 health and education, which trigger contin- 3.5 2006 ued PKH transfers, is done jointly with service providers. PKH facilitators at the local level 2007 visit nearby schools, health centers, and hospi- 2.8 2008 tals to confirm that mothers and children from 6.1 2.4 PKH households have presented themselves 2009 and are acquiring or attending the services re- 5.2 2010 quired. At some facilities and in some regions, PKH program facilitators visit the service 1.5 2011 providers and meet with staff to jointly verify 3.6 1.1 2012 attendance. Verification forms are then most 0.8 often manually submitted to the Management 0.7 0.7 2013 Information System (MIS) in the national PKH 1.9 0.4 database (MoSA, 2016e). 1.6 2014 1.12 0.9 0.9 0.6 2015 101 As of 2017, PKH families are found in 504 districts out of 2016 514. In the third quarter of 2016 due to lower than projected 2007 2016 government revenues, MoSA as well as other ministries faced budget cuts; PKH had to reduce its budget to IDR 9 trillion 102 As of 2016, the previous implementation unit, UPPKH, is Source: MoSA (2014/2015/2016) and Ministry of Finance now merged within the institutional structure and is referred (2008-13) Source: Ministry of Finance; Bappenas; and World to as the Sub-Directorate of Family Social Insurance (Jaminan Note: 2011-13 data are realized budget; 2014, 2015 and 2016 Bank staff calculations. Sosial Keluarga, or JSK) under the Directorate General of data are realised budget data. Note: Data for 2011-2016 are realized budget. Social Protection and Insurance in MoSA (DG Linjamsos). 65 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 FIGURE 2.34 PKH payment cycle (top panel) and operational cycle (bottom panel) 4 O RGANIZATIO N O F 5 3 AS S E MBLIE S TO VA LID ATE & UP D ATE HO USEHOLD R E GISTRATI ON SE L E CT I O N O F BAS IC INFO HO U S E H O LD S P R E PARATI ON FRO M U D B 2 6 S UP P LY S I D E 1 ST PAYME NT AS S ES M EN T 1 8 7 2 ND & V ERI FI CATI ON OF SEL E CT I O N O F COMPLI ANCE WI TH PROV I N C E S / S UCE E D ING PAYME NTS CONDI TI ONS DIST R I CTS / S UB – DIST R I CTS TARG ETI N G 1 Aggregate targeting data From the UDB District proposals and provincial EL IGIBL E No END Yes recommendations D E T ERM INE LOCAT IONS DG Decree RECRUIT & SEL ECT Online & offline recruitment systems 2 FAC I L ITATORS & OPERATORS Administrative selection, Psychological tests & FGD / Interviews 3 Fascilitators & Operators T RA INING PKH flowchart & mechanism Validation P KH IM PL EM ENTAT ION 4 Disbursement T RAN SITIO N Verification Update ( 6 Y EA RS) Yes RECERT IF ICATIO N: T RAN SF ORMAT ION P OVERT Y STATU S & ELIGIBILIR No GRADUATIO N 66 Chapter 2 – PKH Local governments provide support through tended by other PKH mothers) organized by derly (70 years and older) within PKH families service-sector coordination and manage- program facilitators. At these meetings, they uncovered by other social assistance programs ment, but do not otherwise co-execute the receive guidance in fulfilling PKH conditional- (such as the old age assistance program, ASLUT) PKH program. Local implementation is com- ities and advice for remaining in good standing and the severely disabled. The conditions and pleted almost entirely by the regionally based vis-à-vis the PKH program. Any family deter- roll-out for PKH recipients targeted to receive MoSA-PKH program units. As a result, PKH mined to be poor after 6 years of PKH can be these components are still under development accounts for more than three-quarters of the provided with and an additional 3 years of PKH and may be implemented more widely in 2017 budget of the Ministry of Social Affairs’ Family transfers accompanied by additional livelihood (MoSA, 2016c). Welfare unit. and income support from programs, such as KUBE-PKH. The government has also been PKH benefit adequacy was increased in 2013 PKH eligibility depends on both household aiming to increase the integration of other so- and 2015 to better help poor households to im- level of consumption and demographic com- cial assistance programs such as PBI/JKN, PIP prove human development outcomes. Before position. To be eligible in the current coverage and Rastra with PKH throughout the program’s 2013, PKH benefits represented an approximate target level, households must be considered implementation cycle to raise the effectiveness 10-percentage-point share of beneficiaries’ av- “poor”, or in the bottom 14 percent of house- of social assistance and make it more likely erage expenditures, with a slightly higher share holds, as defined by the UDB. For the health- that families are sustainably better off due to in 2007 and a slightly lower share in 2013 (Fer- and education-related conditions, households having participated in the program. The initial nandez and Hadiwidjaja, 2012). PKH benefit lev- must meet at least one of the following condi- step towards this sort of integration was the in- els were raised in early 2015 and again in 2016 tions: a household member is pregnant or lac- troduction of the UDB by TNP2K in 2012 as a (Table 2.8) with the maximum (minimum) tating; the household has one or more children single source of beneficiary data for all social annual transfer per household at IDR 3.7 mil- below 5 years of age; the household has children assistance programs. lion (IDR 800,000), or about US$284 (US$61).104 from 6 to 15 years of age attending primary or In 2017, the benefit structure was changed to a middle school; or the household has children In 2016, several policy changes were made to single unified benefit of IDR 1,890,000 (US$ aged from age 16 to 18 that have not yet com- the PKH program. In the face of expansion 140) per family per year. At these transfer mag- pleted basic education.103 Disbursement of PKH beyond 6 million families, the target group nitudes, a PKH household receives transfers cash transfers is completed quarterly as house- was revised to become the “poor” (it used to worth about 13 percent of their regular expen- holds are verified to have achieved the relevant be the “very poor”) to allow for the inclusion ditures on average. Figure 2.35 shows an aver- conditions listed in Table 2.7 below. of more families. Via the use of e-Warong and age PKH transfer measured as a proportion of a network of agents under a collection of state- the out-of-pocket costs of a regular outpatient Participating families receive PKH transfers owned banks, 1.2 million families have re- visit, or an average year of schooling. While for 6 years, if they comply with conditionali- ceived PKH payments made electronically via multiple health visits can be financed from ties and remain eligible. In addition to regular bank accounts. MoSA aims to roll out digital PKH transfers, only one year of education can education and healthcare service attendance, payments to all PKH beneficiaries by 2018. In be purchased with a PKH transfer. PKH mothers attend monthly meetings (at- addition, MoSA has also begun including new components to extend PKH transfers to the el- Ibid. Disabled children who attend Sekolah Luar Biasa, a school for disabled children, will also become eligible to receive PKH. 103 A household’s total transfer is based on demographic composition. For example, a household with one child in elementary school and one in junior high school would receive at least IDR 1.7 million 104 yearly. Households receive a fixed amount of IDR 500,000 (US$38) even if no conditions are fulfilled. PKH transfers are disbursed quarterly. TABLE 2.7 TABLE 2.8 Core PKH eligibility and corresponding conditions PKH transfer size by component in 2016 H OU S E H O L D S W I TH … . . . M U ST AC C O M P L I S H AT L E AST T HESE CON DIT ION S TO CON T IN UE P OOR HOUSEHOLDS …R EC EIVE YEAR LY RECEIVING PKH WIT H… Pregnant or lactating women 1. Complete four antenatal care visits and take iron tablets during pregnancy. Children aged < 6 IDR 1,200,000 (US$92) 2. Be assisted by a trained professional during the birth. or mothers who are 3. Lactating mothers must complete two post-natal care visits before the new born becomes pregnant or lactating one month old. Children attending IDR 4500,000 (US$39) Children aged 0-6 years 4. Ensure that the children have complete childhood immunization, take Vitamin A capsules elementary school (SD/ twice a year and take children for monthly growth monitoring check-ups MI/Paket A) Children aged 6 - 21 years 5. Enroll their children in the relevant levels of school and ensure attendance at least 85% of Children attending junior IDR 750,000 (US$58) school days. high school (SMP/MT/ Paket B) Elderly people aged 70 years 6. Complete health check ups at health facilities or receive these at the household via home or older care and follow day care or social activities if available Children attending senior IDR 1,000,000 (US$77) high school (SMA/MA/ People suffering from heavy 7. Complete health check ups as needed at health facilities or receive these at the household Paket C) disabilities via home care and follow day care or social activities if available. Source: MoSA, 2016c; Exchange rate: US$1 = IDR 12,900 as Source: MoSA 2016c of October 2016. 67 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 FIGURE 2.35 TABLE 2.9 Average cost of C OV E RAG E , Characteristics within health visits & Indonesian populations education as a TA RG E T I N G , & share of average PKH benefit I M PACTS % of all % poor % of PKH (percent) PKH Indonesians population recipients who: who: who PKH total coverage as of the end 2016—at Do not have access to 18 31 27 6 million families—means it could provide bottled, tap or well water ATTENDING benefits to almost all of Indonesia’s poorest SCHOOL families, but there is some leakage of PKH to Do not have access to 27 53 49 private sanitation non-poor families (Figure 2.36). Generally, based on 2014 data, coverage is much higher Live in rural areas 50 62 63 in the poorest decile of families, but there are a significant number of families with PKH in Live with more than 5 25 44 43 the second, third, and fourth deciles. There are household members also some families in the richest 60 percent that receive PKH transfers. The coverage head- Have not completed primary 10 13 12 count shown in Figure 2.36 is taken over all education families. If instead coverage is taken only over Are illiterate 8 14 11 demographically-eligible families—for exam- ple, about 67 percent of families in the poorest Work in the agriculture 34 58 49 decile are demographically eligible105—then 11 sector percent of demographically eligible families in Source: Susenas 2014. the poorest decile are covered. On average, PKH is allocated to families exhib- iting correlates of income poverty. While there are some non-poor PKH families, an average FIGURE 2.36 PKH household “looks” very much like an aver- PKH coverage, by expenditure decile 2014106 age poor household along all the non-income dimensions of poverty listed in Table 2.9. Urban Rural PKH’s targeting is the most progressive of the 76 transfer programs covered in this report se- ries. 107 In 2014, just over two-thirds of PKH ben- 8% eficiaries were from families in the lowest three % 7% deciles. Since PKH’s goal is to find the poorest of the demographically eligible households, any leakage to non-poor families would be notice- 5% able (Figure 2.37). However, relative to the oth- er social assistance programs discussed in this 4% 3% 3% report series, PKH’s leakage to non-targeted 2% 2% populations is minimal. There are likely sever- 2% 2% HEALTH 1% al factors that together lead to better targeting 1% VISIT 1% 1% 1% results in PKH108 including PKH’s early adop- 0% 0% 0% 0% 0% tion of the UDB-based beneficiary selection and a verification system that includes two-way 1 2 3 4 5 6 7 8 9 10 dynamic updating of program participants and eligibility status.109 DECILE 19 Source: Susenas 2014. % 105 In estimating coverage and incidence, an adjustment is made to better simulate targeting and incidence outcomes. PKH benefits change a household’s per capita consumption. To fairly reflect targeting of pre-program household consumption, consumption has been adjusted. The adjustment assumes all of the monthly value of PKH benefits is captured in Susenas when families are surveyed. 106 The latest available survey data tracking PKH are for September 2014. Susenas data for 2015 and 2016 do not contain the PKH variable used to track program participation; this variable is set to return in the 2017 March Susenas. 107 By “progressive” we mean that shares of PKH benefits decrease as income shares increase. Note that we compare across programs with different target and coverage levels. 108 For example, having well-defined, central institutional control over the PKH program may make it less likely that local variation in preferences or administrative skills affects program outcomes. Or, non-targeted households may feel that pursuing limited benefits from a program meant for the “worst off” in their communities is too costly socially. 109 PKH has been using the UDB since its initial compilation in late 2011. Other social assistance programs covered in this report series may generate overall Source: Susenas 2014. program quotas (via queries to the UDB) or may generate suggested lists of beneficiaries (via queries to the UDB) without fully integrating the UDB-based beneficiary selection, verification, and updating system. 68 Chapter 2 – PKH FIGURE 2.37 FIGURE 2.38 PKH coverage & PKH incidence by incidence expenditure decile Poor The rest Urban Next 30 Total Rural As it has expanded, PKH has improved stages. The first verification stage early COVERAGE (% Decile its ability to deliver benefits to poor on in the year has rather low compliance 7 OF HOUSEHOLDS RECEIVING families. In order to facilitate compari- rates, while the other three are converse- 1 37% BENEFITS) sons between PKH and other social assis- ly very high.112 In terms of actual im- 6 33% tance transfers (which have slightly dif- pacts, the original PKH pilot (launched ferent target groups), Figure 2.37 shows in 2007) was designed to accommodate 4 3 2 17% coverage and incidence for the “Poor”, a randomized, controlled trial (RCT)- 3 the “Next 30” percent, and “The rest”. based impact evaluation, which involves 2 2 20% 2 2 This figure demonstrates that while the experimentally comparing two groups 1 1 poverty headcount rate fell by about 2 of families that differ only in whether 1 3 17% percentage points between 2010 and they received a “treatment”—in this case, 2012 2013 2014 12% 2014, the share of PKH benefits account- the PKH program—or not (World Bank, ed for by the “Poor” group rose by about 2011b). Both a mid-line and end-line sta- 8 percentage points. The “Next 30” tistical evaluation have been conducted; INCIDENCE 4 10% group, which contains the same propor- the former re-visited families after about (% OF TOTAL 10% tion of near-poor families in every year, 3 years of experience with the program 45 43 BENEFICIARIES has a roughly constant PKH share over and the latter re-visited families after 40 BY GROUP) the same time period, while “The rest” more than 6 years of experience. Re- 35 33 5 32 4% has grown larger (by the same 2 percent- sults from these evaluations, which indi- 27 10% age points that the poverty headcount cate that the PKH program was directly 23 22 fell), this group’s share has fallen. This responsible for greater investments in trajectory indicates that as the micro-lev- education and healthy behaviors while 6 7% el poverty situation changes—many providing consumption budget support, 6% families exit poverty year to year, while are summarized below. fewer enter—PKH has continued to add significant numbers of the poor families PKH improves welfare and can bring 2012 2013 2014 7 4% that remain.110 families above poverty-line expenditure 5% levels. The mid-line evaluation demon- Source: Susenas 2014. PKH benefits are concentrated in poor strated that PKH families experienced families (Figure 2.38). In 2014, the poor- a statistically significant 10 percent in- 8 2% est 10 percent of families received over crease in average monthly expenditures. 2% one-third of the benefits available. The The increase was used mainly to buy bottom 20 percent received over half of high-protein foods and to cover health 9 2% the benefits available, while the bottom costs. The end-line evaluation showed 1% 30 percent received over two-thirds of that beneficiary expenditure increased the benefits available. This puts PKH on by 3.3 percentage points, while beneficia- a par with similar programs such as Bra- ry families’ expenditure on food was 3.4 10 1% zils’ Bolsa Familia and the Philippines’ percentage points higher than non-ben- 0% Pantawid Pamilya Pilipino Program eficiary families. For protein consump- (4Ps), which registered CCT benefits ac- tion, the impact was considerably lower, cruing to 57 and 52 percent of the poorest at 1 percentage point (TNP2K, 2015a). Source: Susenas 2014 20 percent of families, respectively.111 and World Bank staff 8 calculations. PKH motivated healthy behaviors gen- Overall, conditionality compliance ap- erally, and maternal and neo-natal prac- pears high and PKH’s initial positive tices improved noticeably. The mid-line impacts have continued as the program evaluation demonstrated that PKH was + PP and the families in it have matured. Compliance for both the education and health components is around 80 percent responsible for statistically significant increases in pre-natal care. The likeli- hood of attending at least four prenatal for both, averaged across four verification visits increased by 9 percentage points, “…while the poverty headcount rate fell by about 2 percentage points be- tween 2010 and 2014, the share of Coverages of small programs like as PKH (3.5 million families in 2015 and mid-2016) are likely underestimated PKH benefits accounted for by the 110 in the Susenas household survey; actual PKH coverage is slightly more than double that displayed in Figure 2.37. “Poor” group rose by about 8 per- 111 World Bank Aspire Database. centage points” 112 2016 PKH MIS data. 69 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 while newborn delivery at a facility or attend- ing (height for age) decreased by 3 percentage ed by a professional increased by 5 percentage points. PKH improved neo-natal visits by 7.1 points. Post-natal care improved by almost 10 percentage points but it had no significant im- percentage points, while immunizations and pact on outpatient visits or increased intake of growth monitoring check-ups increased by 3 iron tablets. Contrary to the mid-line results, and 22 percentage points, respectively. PKH there appeared to be no significant impact of had some impact on severe stunting as well, PKH on post-natal visit to health facilities (TN- up to 3 percentage points. Unconditioned P2K, 2015a). The end-line report noted possible health behaviors also increased, indicating that explanations as a prevailing belief among mothers PKH was responsible for increases in general that if their delivery went well, there was no need health-seeking behavior in beneficiary fami- for post-natal check-ups and that some women not- lies. Visits by any household member to either ed the difficulty in arranging appointments with private or public health facilities increased healthcare professionals (TNP2K, 2015). more in PKH families than in eligible fami- lies in non- PKH areas, albeit at a more mod- PKH’s positive impacts on education appeared est rate of 0.5 of a percentage point. The study later. The mid-line evaluation indicated that also showed that PKH impacts were more pro- children from PKH families spent more time in nounced in urban areas in Java, due to the high- school (if they were already attending), but the er availability, better quality, and proximity of estimated impact was small in magnitude, with health facilities (IPC, 2013). attendance increasing by just 0.7 of an hour per week. In general, the mid-line evaluation These positive impacts were less pronounced indicated that though education-related behav- in the end-line evaluation results. Impacts on iors were improving over time everywhere— healthcare-professional-assisted deliveries or participation, enrolment, and transition rates delivery at health facilities were not significant. all rose—PKH families did not experience a Significant impacts were registered in the like- greater improvement than non-PKH families lihood of children receiving immunization. (World Bank, 2012i; 2011b).113 By the time of PKH families saw an increase of 7 percentage the end-line evaluation, however, PKH families points in immunization, while severe stunt- were demonstrating positive, if small, changes in these practices. For example, according to end-line results there were statistically signifi- 7.1 cant increases of 2 percentage points in the gross participation rate for elementary school and almost 10 percentage points in the junior high PP "PKH improved neo-natal visits school gross participation rate (TNP2K, 2015a). by 7.1 percentage points" While the probability of a PKH child continuing to secondary school increased by 8.8 percentage points, there was no significant impact on the probability of dropping out of secondary school. Child labor continues to decrease in Indone- sia but there is no statistically significant im- pact attributable to PKH, according to either the mid- or end-line evaluations. Child-labor indicators in the mid-line evaluation showed similar patterns as the education-related indi- cators discussed directly above: while in gener- al rates of child labor were falling in both PKH and non-PKH regions, PKH was not responsi- ble for larger reductions in beneficiary fami- lies. The end-line evaluation, meanwhile, has indicated a small (but still statistically signifi- cant) decrease in the rate of child labor in PKH families of just over 1 percentage point, which is about equal to the fall in the elementary school drop-out rate (IPC, 2013; TNP2K, 2015a). When the mid-line evaluation was completed, the 113 following reasons were offered to explain the lack of impact on conditioned education behaviors: payments did not coincide with the academic school year, so parents did not have the funds when needed while the amount received was not adequate to cover education fees that parents must usually pay. 70 Chapter 2 – PKH ACCESSI B I LI T Y By design, PKH’s M&E system provides ev- idence for implementing units to use in im- proving program efficiency and functionality. Operational monitoring is jointly completed by the central PKH implementation team with the cooperation and assistance of many region- al and local implementation teams. The PKH M&E cycle begins with data entry into the MIS. These data are collected and entered by local PKH facilitators and summarize key household and administrative indicators.114 The facilita- tors liaise with the district coordinator and the operators115 available at the sub-district level. Meanwhile, the district PKH offices liaises with these two sub-district coordinators, as well as with the five PKH-dedicated working groups at the district level: the data team, the health and education services team, the fund allocation team, the verification team and the M&E team. Finally, district and provincial PKH offices in coordination with local service providers re- and transfer disbursement schedule with ed- for the management of PKH socialization ac- main jointly responsible for helping to ensure ucation service provider billing cycles—was tivities (and those activities’ budget). However, that local basic services are available and func- made part of PKH’s standard operating proce- by establishing an “in-house” communications tioning so that PKH beneficiaries face few con- dure through regulation and MIS functional- team at the central level and by providing firm straints when fulfilling conditions. The MIS ity was enhanced so that it could signal when direction to the Ministry of Communications system provides a conduit for program-related delays in these procedures were accumulating. and Information, PKH has been able to produce information to proceed directly from the field Ensuring this process runs smoothly is key to and disseminate widely and deeply a greater vol- to the implementing unit at the central level. the effective functioning and responsive na- ume and variety of media related to program In addition, the structure of PKH’s oversight ture of PKH. Likewise, once it was discovered benefits, eligibility criteria and accessibility. and its M&E procedures include many two-way that PKH transfers were not commensurate information flows between local, regional, and with the actual cost of schooling, the BSM/ PKH’s grievance redress system is theoret- national levels as well as two-way information PIP cash transfer for poor and at risk students ically easy to access but has functioned only flows at any level between service providers and PKH benefits were linked, while the PKH weakly. PKH participants (and community and PKH implementers or between families transfer levels were increased. Both actions in- members) can submit complaints directly to and facilitators. creased the likelihood that PKH families could the village facilitator, the PKH facilitator, and access education and remain compliant. the PKH implementing units at the district, Program monitoring has led to administrative province, and central levels, by making either revisions, making it easier for beneficiaries to PKH has increased its public information a direct, unstructured report, or by filling in a access the PKH transfers effectively. As PKH campaign efforts as the program has expand- standardized form (MoSA, 2016e). There is an commenced operation in 2007, bottlenecks in ed. PKH’s dissemination of program and pol- operational sub-manual dedicated to grievance household verification, compliance monitor- icy information, known as “socialization” in reporting and redress that describes tasks and ing, and payment delivery, as well as a weakly Indonesia, has suffered from the same incon- responsibilities from the village level upward functioning MIS system, meant that PKH trans- sistency as do most other policy and program (Oxford Policy Management, 2012). However, fers to families were not synchronized with the implementation functions in Indonesia’s thor- the grievance redress system was found to be due date for school fees. Once this constraint oughly decentralized administrative environ- mostly un-operational and not used effective- on PKH families’ access to education was iden- ment (World Bank, 2012i). This is due partly to ly to improve outcomes for PKH families: as of tified, its solution through 2015 —the harmo- institutional boundaries: The Ministry of Com- 2013, 7 percent of beneficiaries had submitted nization of the entire compliance verification munications and Information is responsible written complaints (World Bank, 2011b). 114 Key indicators include: the share of participants unable to meet PKH conditions in health and education; the type and content of complaints received through the centralized grievance system and entered into the MIS; the disbursement of funds, fund disbursement timeliness, and cause for disbursement delay (when applicable). 115 The operations coordinator monitors and assists service facilities that beneficiaries attend to remain PKH-compliant as well as oversees and assists the sub district-based administration and MIS teams. 71 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 CHA LLEN G E S, to 6 million families in late 2016, the CCT could reduce the poverty rate by around 0.8 of a per- less than 30 percent of PKH families in the poorest decile received PIP, JKN-PBI and Ras- RECOMMENDAT I O N S & centage point. Therefore, this expansion alone could bring the headcount poverty rate down tra, even though they are automatically eligible for all three programs. Through the office of a from its current 10.6 percent to 9.8 percent. 117 social assistance ombudsman or deputy, local MOVI NG FO RWA R D governments could be mobilized to assist the To successfully manage an expanded PKH, the poorest families in accessing all of the avail- PKH should continue to build on its well-con- program should strengthen its own human able transfers for which they are eligible. As structed foundation as it expands to serve resources as well as its institutional capacity, espoused under the early 2016 verification/val- more families, while incorporating innova- its IT systems, as well as the capacity of the idation exercise led by MoSA and district gov- tions that increase its relevance and effec- service providers on which beneficiaries rely. ernments, further local government support tiveness for Indonesian communities. PKH Continuous increase in number and enhance- of the UDB, and associated revision and up- has twice demonstrated, under rigorous ex- ment through training of the local, regional, dating protocols will be important to support perimental protocols, that it generates positive and national teams that oversee the core pro- the programs goals and positive perception in impacts in welfare, in health-seeking behavior, gram functions is essential for efficient deliv- the short to medium term. In the longer term, and in education. It has also demonstrated that ery of benefits and effective access for families. a permanent data updating mechanism should it can be somewhat flexible with its operation- Key program functions and enhancements in- be built such as through the Integrated Refer- al protocols and varied in its approach to ser- clude: timely and complete verification of ben- ral System (SLRT) and On Demand Application vice-provider coordination and assistance in eficiaries’ status and conditionality fulfillment; (ODA) to help achieve better integration of the order that PKH families are better served. For regular MIS updating, adjustment of benefit lev- entire social assistance portfolio.119 example, in 2014, “Family Development Ses- els and timely disbursements; determination sions” were piloted to serve families exiting the of local level capacity for supporting program PKH benefit levels should remain commen- program. These benefits are available to families implementation; and suggestions for remedi- surate with regional or local price changes still considered poor after six years in the PKH ation of local supply inadequacies in health, in the cost of conditioned services. Benefit program and they provide training modules in education and program socialization, are some adequacy was improved in 2013, 2015 and only early childhood education, parenting, health and of the aspects to be improved.118 PKH’s recent, marginally in 2016, or by three times over a nutrition, household finances, small business de- and potentially further, expansion will require 9-year period. This is too infrequent to keep up velopment and entrepreneurship, while extend- additional consolidation and strengthening of with inflation and, given the increase in health ing (for up to 2 years) the receipt of cash transfers. program delivery systems and in particular the and schooling costs specifically, too infrequent In 2017, program implementers are moving to process of the verification of conditionalities. to keep the PKH transfer relevant for families ensure FDS modules are to be given at an earlier This will need to occur with careful attention that wish to comply with PKH conditions. In stage in the PKH program cycle (MoSA, 2016c). to human resources and personnel, as well as comparison to other CCT, at about 13 percent, IT systems, in the central- and regional-level PKH benefit levels are relatively low and could Now expanded to 6 million families, or about teams that manage core PKH functions. Lastly be raised: Brazils’ Bolsa Familia and Mexico’s 10 percent of the population, PKH will yield it is critical to regularly monitor the program Prospera account for about 19 percent and 22 significant impacts on poverty and some im- implementation performance and communi- percent, respectively, of household monthly pact on inequality as well, if the program is cate with stakeholders and the public at large to expenditures. 120 managed well. In early 2016, based on a request ensure transparency and confidence. from MoSA, simulations were conducted to es- timate poverty and inequality reductions that PKH should continue to pursue explicit links could be expected from an expanded CCT.116 It to complementary programs targeted to poor was predicted that an expanded PKH, from 3.5 families, especially JKN-PBI and PIP. In 2014, 116 Simulations from Susenas 2014. Eligible new PKH beneficiaries’ per capita consumption is increased and poverty and inequality are re-estimated using the current poverty line. New beneficiaries are targeted using the proxy means test approach used by the UDB. Poverty and inequality impacts exclude any effect of future economic growth, increased household incomes or higher inflation. This simulation was based on current administrative and operational costs and quality; a PKH expansion may lead to an increase in per-beneficiary implementation overheads. Actual poverty and inequality impacts will depend on all these factors. 117 Not taking into account growth and inflation, which would affect these results. The same simulation indicated that the suggested PKH expansion would be, all else remaining equal, responsible for a modest drop in inequality as well. A larger expansion estimation—to 8.4 million families —was expected to have a poverty reduction impact of 1.5 percentage points, which would put the current administration on track to meet its 2019 RPJMN target. 118 All of these PKH processes (as well as some others) were found to be not operational or only sporadically operational in a first round of implementation “spot checks” completed over 2008 and 2009; see Centre for Health Research, University of Indonesia, 2010. 119 SLRT, Sistem Layanan Rujukan Terpadu or the Integrated Referral System (for social protection programs) has been implemented in 59 districts by MoSA under the guidance of Bappenas, while ODA, a complementary an initiative for updating UDB, has been piloted in 12 districts by TNP2K. 120 Susenas 2014 and Aspire database 2015. [No reference in references section.] 72 Chapter 2 – PKH To successfully manage an expanded PKH, the program should strengthen its own human resources as well as its institutional capacity, its IT systems, as well as the capacity of the service providers on which beneficiaries rely. P.73 The “Life-Cycle” Approach to Social Assistance & Safety Nets OVERVIEW E ffective, efficient social assistance sys- (for example). An effective social assistance tems help households and their mem- system should also be stabilizing over the en- bers mitigate risks. Social assistance tire course of an individual’s or a household’s programs are concerned with helping life cycle. For example, should a newborn who households absorb, mitigate, and overcome lacks access to weight checks and immuni- risks to their well-being. Social assistance pro- zation become a student unable to afford col- grams are usually targeted to poor and vulner- lege-entrance-exam tutoring and then an older able households unable to afford access to the laborer contemplating retired life without a (publicly- and market-provided) goods and pension, the social assistance system should re- services that non-poor households regularly main accessible and effective to this individual consume, including the investments in hu- at any age. This section identifies key life-cycle man capital, such as health and education that risks faced by poor and near-poor populations directly reduce risks to well-being. They also in Indonesia in order to determine when and provide basic needs for those households that where the current social assistance programs find it difficult to afford even basic necessities are addressing salient risks, as well as whether and provide an alternative to negative coping any SA solution currently available can be rel- strategies, such as asset sales or forgoing invest- evant for all poor and vulnerable households. ments in human capital, that sacrifice future Using the rich individual- and household-level stability and productivity for a reduction in the data in the Susenas socioeconomic survey, we likelihood of falling into poverty now. are able to generate an empirical catalogue of the risks (as proxied by outcome gaps) faced at The individual at risk, the salient risk, and every stage in the life cycle of an individual or a the right time to offer SA-specific benefits or household. This catalogue then provides a log- strategies, all depend on an individual’s tra- ical benchmark for reviewing social assistance jectory, or her position in her own “life cy- programming, following the GoI’s Masterplan cle”. An effective social assistance system will for the Acceleration and Expansion of Poverty combine instruments and strategies so that it Reduction (2013 to 2025), which also uses the is as effective for poorer individuals without life-cycle approach as a framework for deter- employer-based pensions nearing retirement, mining what positive characteristics a social as for secondary education students unable to assistance system should have (Box 3.1). afford tutoring for university entrance exams CHAPTER 3 75 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 BOX 3.1 Master Plan for the Acceleration & Expansion of Poverty Reduction in Indonesia 2013—2025 Indonesia desires comprehensive social protection programs. According to the Master Plan for Acceleration and Expansion of Poverty Reduction (MP3KI) 2013-2025, social protection programs should help the poor and vulnerable cope with crisis or socio- RI SK I DE NTI FI CATI ON political, economic or environmental shocks BASE D ON THE LI FE -CY- through direct and indirect transfers. To CLE A PPROACH achieve this, the MP3KI calls for preventive, promotional social protection programs to comprehensively address risks to welfare at the individual, household, and community-level. INFANTS SCHOOL AGE PRODUCTIVE ELDERLY AGE 0 6 19 Social protection in Indonesia will use the life cycle approach to identify risks during each stage of life. The social protection framework will focus on two form of risks: 1. Risk at individual and household level, both for men and women, in each age group, which TO 5 TO 18 TO 60 >60 can occur during the life cycle. Since these MALNOURISHED DROP OUT OF UNEMPLOYED NO PENSION risks might be long term or permanent, it is SCHOOL FUND NO IMMUNIZATION CONTAGIOUS important to have social assistance and social CHILD LABOR DISEASE ABANDONED insurance programs which operate on regular ABANDONED basis, including protection from violence and DISEASE ACCIDENT AT DISCRIMINATION exploitation. WORK AT WORKPLACE 2. Risk at community-level, which occur due EARLY MARIAGE/ to external factors, such as natural disaster, TEENAGE DIVORCE HEALTH economic shock, and social conflicts. This type PREGNANCY DEGRADATION of risk should be tackled by temporary social CANNOT SEND assistance program which can be distributed ABANDONED CHILD TO SCHOOL during the disaster or crisis and targeted to CHILD VIOLENCE REPRODUCTIVE particular beneficiaries. HEALTH RELATED DISEASE The MP3KI indicates that social protection is composed of (i) social insurance; (ii) social UNSAFE assistance transfers; and (iii) voluntary, DELIVERY/LABOR individual, privately-purchased insurances. Each of these three components should be DOMESTIC implemented via: VIOLENCE • Institutional strengthening, policy integration GENDER and social protection interventions, including DISCRIMINATION the involvement of social workers and DISCRIMINATION facilitators, as well as community and social AT WORKPLACE institutions. • Identification of, and strengthening linkages POOR HE A LTH CONDI TI ON I NCLUDI NG DI SABI LI TY between, social protection providers and the sector-based facilities and resources, including infrastructure, that also promote sustainable livelihoods. NATURA L DI SASTE R, E CONOM I C A ND SOCI A L SHOCK • Consistent, sustainable, automatic fiscal Source: Bappenas, 2013 support for all social protection initiatives. Source: Bappenas, 2013 76 Chapter 3— The “Life-Cycle” Approach to Social Assistance & Safety Nets FIGURE 3.1 Indonesian population pyramids, 2014 Urban CHALLE N G E S , Rural RECOMM E N DAT I O N S & AGE 0—5 6—18 19—24 25—60 60+ –6 MOVI N G FORWA R D POOR HOUSEHOLDS –4 –2 Poor households are predominantly rural. Figure 3.1 below presents population pyramids 0 for three different groups: (i) the poor, or indi- viduals whose consumption expenditure is less than the national poverty line; (ii) the “near- 2 poor” or vulnerable, or individuals in roughly the bottom 40 percent of households ranked 4 by per-capita consumption expenditure who are not counted as poor; and (iii) the rest, or 6 all individuals who are neither poor nor near- poor.121 In 2014, the Indonesian population was split evenly into rural and urban areas,122 8 but poor households and individuals were con- AGE centrated in rural areas: just under two-thirds 0—5 6—18 19—24 25—60 60+ of all poor individuals were found in rural lo- 20 cales.123 In the chapters that follow, we evaluate VULNERABLE HOUSEHOLDS (THE NEXT 30%) SA program targeting—or the ability to locate 15 and provide benefits to poor and vulnerable individuals or households—and return to this 10 characteristic as a benchmark. 5 Poor and vulnerable households have a great- er number of younger-than-school age, school- 0 age, and retirement-age dependents. For ex- ample, children from age 0 to 5 years account 5 for about 15 percent of all poor individuals and 10 13 percent of all vulnerable individuals, but only 10 percent of all non-poor, non-vulnerable 15 individuals. Frequency rates for “School Age” individuals (from 6 to 18 years old) in poor, 20 vulnerable, and non-poor, non-vulnerable populations are 28, 25, and 21 percent, respec- 25 tively; and for “Elderly” (61 years old or older) AGE 8.4, 7.0, and 7.2 percent, respectively. In other 0—5 6—18 19—24 25—60 60+ words, the “Working Age” population (from 25 20 to 60 years old) is squeezed on both sides by THE REST more dependents in the poor and vulnerable (NEITHER POOR NOR VULNERABLE) population. In addition to creating a larger 10 cumulative burden on household income from labor, more dependents usually means greater expenditures on health and education services. 0 10 121 More precisely: the “Poor Households“ are those containing individuals with per-capita consumption under the 2014 poverty line; the vulnerable are the next-poorest 30 percent (or the “Next 30 percent”) of individuals who are not counted as poor; and “The rest” is everyone else. The headcount 20 poverty rate in 2014 was 11.3 percent, so the “Poor” are 11.3 percent of individuals, the “Vulnerable” are 30 percent of individuals, the “Poor” together with the “Vulnerable” are 41.3 percent of individuals, and “The rest” are 58.7 percent of individuals not counted as “Poor” or “Vulnerable”. 122 In results not shown, we examined age-group population 30 shares within each of Indonesia’s 33 provinces and found no significant provincial differences in age-group shares. Source: Susenas 2014 and World Bank staff calculations; bars represent 123 56 percent of vulnerable individuals and 42 percent of non- millions of individuals within the age ranges listed on the x-axis. poor, non-vulnerable individuals are found in rural areas. 77 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 nearly 16 percent of primary-school-enrolled households has grown since 2004. It also illus- children from poor households did not start trates that while the rate of labor-market entry school on time and the poor-non poor gap in for young adults from non-poor and non-vul- this indicator has increased since 2004 (in oth- nerable households has finally caught up to the er words, the difference in late enrolment rates rate for young adults from poor households, between poor and non-poor, non-vulnerable those from poor households are more fre- households has increased since 2004). While quently bringing a drop-out’s credentials (that KEY RI SKS OV E R T H E we cannot observe induction into the basic is, they more frequently lack a certificate). education system in a single cross-section of L I FE CYC LE households (such as Susenas), it is not unrea- Working-age individuals in poor households sonable to suggest that children from poor have lower-quality jobs. Nearly 90 percent of households are more often unprepared—so- poor working-age individuals bring to the la- cially, emotionally, or mentally—for the formal bor market a “basic education or less” creden- schooling system and that that lack of readiness tial. While this does not prevent them from may stem from less prior time spent in struc- finding jobs—for all households labor force tured developmental programs, such as ECED activity rates are nearly 80 percent while un- and kindergarten. The 13- to 15-year-old cohort employment rates (among the “active” labor Children from poor households face disad- (those near the end of Indonesia’s 9-year basic force) are quite low—it likely limits them to vantage early. Figure 3.2 presents selected out- education mandate) from poor households job types that provide few (if any) benefits or comes and gaps by age groups and by the same show a slightly higher risk of non-enrolment extra-salary compensation. For example, while welfare levels, namely the “Poor”, the “Vulner- in the junior-secondary level and of drop-out the rate of informality among employed males able”, and “The rest”. Pre-school124 might still in, or non-continuation from, the primary level has stayed about the same in all households be viewed as a luxury in Indonesia in that less (Figure 1.2). The 13- to 15-year-olds from poor since 2004, the rate of informality among em- than half of non-poor, non-vulnerable house- households also show higher rates of late en- ployed females has increased noticeably since holds have children enrolled. But for children rolment (or grade repetition) than those from 2004, and it has increased especially for poor from poor households, it is almost a rarity: non-poor households, and this gap too has in- working women. In addition, coverage of gov- nearly 70 percent of pre-school age children creased since 2004. Compounding these risks ernment-subsidized public health insurance from poor households are not enrolled in any is a higher frequency of labor contributed by schemes among poor working-age individu- pre-school initiative. Furthermore, while pre- 13- to 15-year-olds from poor households. In als is about 50 percent, which means that only school enrolment rates have risen since 2004 other words, even enrolled children may have half of poor households with more dependents for all households, the poor-non poor gap125 has additional responsibilities outside of school (both young and old) and where the primary actually widened: 2004 pre-school enrolment that make them less productive at school. wage-earner(s) likely has a job with few ex- rates have risen faster in non-poor households tra-salary benefits can depend on help with than in poor households. As a 9-year basic education ends, the major- healthcare expenditures. 131 ity of students from poor households exit Risks for the youngest children can materi- the education system. For example, nearly For the elderly, income security is a serious alize even earlier. While rates of unattended 60 percent of 16- to 18-year-olds from poor issue as the elderly absorb the lingering im- birth in poor households have been halved households are not enrolled in senior sec- pacts of labor-market choices. As detailed in since 2004, pregnant mothers in the poorest ondary school, while nearly 50 percent of the Figure 3.3, rates of poverty and vulnerability 20 percent of households access ante-natal ser- same group have dropped out of school already. rise markedly for those older than 64. Over 40 vices from a general practice doctor, a doctor While about one-third of 16- to 18-year-olds percent of the elderly are poor or vulnerable, as specializing in obstetrics and gynecology, or a from poor households are active in the labor compared to 31 percent for the general popula- nurse less than 5 percent of the time (Riskesdas, market and could thereby be adding work expe- tion. In addition, a large share, approximately 2013).126 In contrast, pregnant mothers in the rience to their skills base, the “unemployment 35 percent, of the elderly poor and vulnerable wealthiest 20 percent of households access an- rate” among that active population is 5 percent. are either living by themselves or with one oth- te-natal care from a doctor more than one-third For 19- to 24-year-olds, Figure 3.2 illustrates er person revealing another risk inherent to of the time.127 Average rates of malnutrition for that university education is an option only for getting older in Indonesia Pensions for retire- 0 to 5 year olds at between 20 and 36 percent128 non-poor and non-vulnerable households, and ment-age individuals from poor and vulnera- are 3 to 4 times higher in Indonesia than the again the gap in university enrolment rates ble households are extremely rare (Figure 3.2). East Asia and Pacific Developing Country aver- between poor and non-poor, non-vulnerable But low coverage is only half the story: it is esti- age.129 Average height is significantly lower for both boys and girls from rural areas than those from urban areas, and this significant differ- 124 Which includes kindergarten, daycare, and ECED centers (known by their Bahasa Indonesia acronym as PAUD centers). 125 The poor-non poor gap is defined as the percentage point difference in, for example, enrolment rates, between poor ence is apparent at age 5 (if not before) and per- households and non-poor, non-vulnerable households. sists until adulthood. Since poor households 126 Instead, pregnant mothers from poor households are most often attended by midwives when they acquire antenatal services. The household ranking here referred to is based on a wealth index compiled independently from the Susenas-based measures of are more likely to be located in rural areas, it expenditure consumption. In addition, this service-provider quality difference (between poor and rich households) is conditional upon utilization. The publicly-available Riskesdas data summaries do include the average rate of completing the recommended can be inferred that the rates and severity of four ante-natal care visits among pregnant or recently-pregnant women: 70 percent in 2013, up from 61 percent in 2010. The malnutrition are higher for children from poor summaries do not publish ante-natal care rates by quintiles of the wealth index; it is expected that ante-natal care rates are lower among poor households. households than those from rich households 127 Riskesdas summaries also indicate that those mothers with at most primary education (regardless of wealth level) acquire ante-natal services from midwives approximately 95 percent of the time; Susenas indicates that nearly 80 percent of working- (Riskesdas, 2013).130 age individuals from poor households have at most a primary education (see below). The constraints to acquiring high-quality ante-natal care, therefore, are not just financial. 128 When measured by weight-for-age and height-for-age indicators, respectively These risks may have lingering impacts as 129 According to the World Development Indicators database (accessed on June 4, 2015). The East Asia and Pacific Developing Country set includes Cambodia, China, Fiji, Indonesia, the Rep. of Korea, Lao PDR, Malaysia, Mongolia, Myanmar, Papua New standardized, basic, compulsory education Guinea, the Philippines, Thailand, Timor-Leste, Vietnam and several small Pacific island nations. begins. Very few 7- to 12-year-old children do 130 We are implying that the rural-urban difference in a height-for-age measure is indicative of a higher burden of malnutrition for poor households. The publicly-available Riskesdas results do not include rates of malnutrition by wealth quintile. not enroll in primary education. However, 131 Susenas 2004 & 2014 analysis 78 Chapter 3— The “Life-Cycle” Approach to Social Assistance & Safety Nets FIGURE 3.2 mated that total benefits from the contributory pension schemes for government employees Outcomes & gaps by age group, 2014 (percent) received by the non-poor, non-vulnerable pop- ulation are 2.8 times greater (when measured on a per-capita basis) than those received by the 0—6 YEARS OLD 7—12 YEARS OLD vulnerable population (World Bank, 2016).132 In other words, pension coverage is low for all Not enrolled (PAUD) Not enrolled (SD) population groups, while non-poor, non-vul- Not enrolled (kindergarten) Late SD enrollment nerable households capture the overwhelm- ing majority of the pension benefits available. 97 This result could logically be tied to the quality 96 96 of jobs to which individuals from poor house- 75 holds are matched: informal employment does 70 not often come with any non-salary benefits 61 (World Bank, 2014a).133 The disabled are also more likely to be or be- 24 17 come poor or vulnerable as they are often lim- 9 12 ited in their opportunity to generate income. 7 8 In addition, they may face above average expen- ditures in health. To date, there is only scant Poor Vulnerable The Rest Poor Vulnerable The Rest data available on the prevalence and trends of disability; the Susenas survey does not include key variables on the subject and so there is little 13—15 YEARS OLD 16—18 YEARS OLD ground to build an analysis on at this point.134 Not enrolled (SMP) Not enrolled (SM) Poor households face overall risks associated Some labor contributed Droppwd out with the lack of access to basic services and Dropped out Joined labor force Unemployment rate (among active) poor housing conditions. Poor households of- ten use lower-quality, lower-cost materials for their residential structures and those residen- 36 tial structures are more often located in areas 9 28 36 where publicly-provided utilities do not reach 16 6 28 28 25 9 6 (see Box 3.2). For example, most poor house- 6 16 holds lack access to improved sanitation facil- 9 6 9 6 5 16 9 6 ities, and endure non-permanent housing as reflected by building materials. Such residen- Poor Vulnerable The Rest Poor Vulnerable The Rest tial location choices can reasonably be viewed 132 About 12 percent of the labor force (or 5 to 6 percent of 19—24 YEARS OLD 25—60 YEARS OLD the population) was covered by pensions in 2012 with the bulk of that coverage due to the contributory pension schemes for public employees (known then as Taspen and Asabri). Not enrolled (University) Basic education or less 133 About 60 percent of all employment is considered Joined labor force Active labor force “informal” in Indonesia (circa 2010). 134 The UDB does contain a mix of disability variables. unemployment rate (among active) unemployment rate (among active) No certificate (if education complete) Subsidized public health insurance coverage FIGURE 3.3. 99 95 89 79 75 80 Poverty, vulnerability & household size 76 72 80 Poor 62 64 62 55 Poor & Vulnerable 40 Share of poor & vulnerable 23 living alone or with one 5 9 7 6 5 other person 3 1 1 1 Poor Vulnerable The Rest Poor Vulnerable The Rest 44.34 38.38 38.35 33.16 31.54 31.34 20.41 ≥60 YEARS OLD 55 16.78 13.40 45 Subsidized public health insurance coverage 9.60 9.23 5.21 Pension coverage 28 6 1 2 18—54 55—64 65—74 75+ Source: Susenas 2014 and World Bank staff calculations. Poor Vulnerable The Rest Source: Susenas 2016 and World Bank staff calculations 79 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 as poor households absorbing fiscal risk, or the ter and sanitation) become greater. Figure 3.4 risk that an electricity or water-supply program summarizes the indicators mentioned above will not be fully funded. In absorbing this risk, for the poor, the vulnerable, and the non-poor, total residential operating costs (which in- non-vulnerable (“The rest”) households. clude the cost of essential utilities such as wa- BOX 3.2 Housing Policy in Indonesia Housing is a critical part of public infrastructure considered too small (in magnitude) to make a ket’s terms; the middle 40 percent cannot af- as well as an essential service for all families. significant impact. In 2013 spending on these ford formal-market housing without subsidy In Indonesia the demand for affordable hous- housing programs reached about five-hun- support; the bottom 40 percent are unable ing units exceeds the available supply: there dredths of a percent of GDP, which is much to afford even a subsidized basic starter unit are about 64 million housing units in Indonesia smaller than, for example, in the Philippines (which is valued at IDR 15 to 30 million). Low- (while the number of households is greater than (0.31 percent of GDP), Thailand (2.15 percent er-income households also tend to dedicate 65 million), but 20 percent of these units are GDP), or the United Kingdom (1.42 percent of less disposable income (proportionally) to in poor condition. Formal, private real estate GDP). Public housing expenditures in Indonesia housing, and larger (proportional) amounts and construction firms are producing about are regressive, however, as the budget favors on other essentials such as water, food and 400,000 units each year, and about 50,000 middle-, upper-middle-, or upper-class-target- transport. Lower-income households more to 100,000 of these are part of a subsidized ed. Housing programs targeting lower income frequently finance housing services or home mortgage program. A second subsidy program households account for less than half the annu- improvement projects by turning to microfi- provides 150,000 to 200,000 new units annu- al budget for housing subsidies. nance institutions which generally have worse ally by helping finance renovations (for existing terms (higher nominal rates and shorter re- housing in poor condition), rental housing, or Home Improvement and Incremental Expan- payment terms) than formal lenders. The social housing. That leaves about 200,000 new sion for Low Income Households BSPS program, managed by Ministry of Hous- households with no alternative in the formal About 70 percent of Indonesia’s total stock of ing, provides to lower-income households a market; these households—primarily lower-in- housing was at least partially self- or informal- subsidy (either IDR 7.5 million for home im- come households with high population growth ly-constructed; among low-income households provement or IDR 15 million for new construc- rates—turn to the informal housing market. this number is likely even higher. Most low-in- tion) for such incremental construction. The come households prefer building and improving BSPS grant is designed to cover only a portion Government Spending on Housing their dwellings incrementally as it is difficult for of the total cost; the remainder is to be paid by Indonesia has developed programs to improve poor household to access mortgage finance: the homeowner’s savings or other assets. housing conditions and provide purchase assis- only the top 20 percent (ranked by income) of tance to lower-income individuals, but they are households access formal housing on the mar- BSP S PROGRA M I M PLE M E NTATI ON SCHE M E 03 05 Selected 06 01 02 Consultants Money is The regional districts are 04 GoI selects Beneficiaries list help distributed government visited & several beneficiaries to several regions in is approved directly to material selects the poor households identify proper suppliers in two each province districts are randomly construction installments selected materials Source: Bappenas, 2011 Housing units with various Housing units with various Coverage substandard characteristics, 2014 substandard characteristics, 2014 While over 80 percent of families own their 80% 15 IDR MILLION homes, nearly 30 million housing units from the 60% current stock are considered substandard (due 10 40% to overcrowding, poor-quality building materi- 5 als, or lack of access to basic services). Low-in- 20% come households suffer substandard housing 0% 0 more frequently: 27 percent of first-decile 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 households make do in an overcrowded house Overcrowding Monthly HH income and 22 percent do not have access to basic No water & sanitation MOnthly HH expenditure utilities. However, BSPS covers about 140,000 All poor quality materials Funds for housing investment Overcrowded or no basic utilities or households annually, so 1 percent of Indonesian substantial material households (of any income level) with substan- Overcrowded and no basic utilities or dard housing are covered by the program. substantial material Source: Bappenas, 2015 80 Chapter 3— The “Life-Cycle” Approach to Social Assistance & Safety Nets FIGURE 3.4 Poor households are more exposed to macro- economic and fiscal risk than non-poor house- Housing & basic service gaps, 2014 Light materials on walls (bamboo, others) holds. If we think of a household as a collection Light materials on roof (palm, others) Light materials on floor (soil, others) of behaviors and preferences, we can analyze the outputs this collection produces. For exam- LOW-COST HOUSING MATERIAL USAGE ple, household budget expenditure shares for 41 food, housing, transportation, and services (and so on and so forth) are one such output: poor 36 and vulnerable households dedicate two-thirds 31 of their budgets to food alone; and non-poor, non-vulnerable households dedicate just over half of budgeted expenditures to food.135 Less diversification in household welfare sources136 implies greater risk when consumer prices are 18 volatile.137 9 7 2 5 2 S O C I AL ASSISTANCE PRO G RAM S & L IF E-CYCL E R I S KS ADDRESSED Poor Vulnerable The Rest 61 BASIC SERVICE ACCESS RATES Current social assistance programming ad- dresses many of the risks explored above. For No safe water example, the Rastra program provides month- No sanitation ly subsidized-rice packages to targeted house- No toilet holds; program implementers intend both 42 No electricity to keep the local price of rice stable, as well as provide a direct near-cash transfer to poorer 32 households. The JKN-PBI health insurance 29 or fee-waiver program and the CCT program 25 24 (Program Keluarga Harapan, or PKH) both 20 19 work to increase access to healthcare services; the former lowers the cost of access for indi- 9 11 viduals of any age, while the latter conditions the receipt of cash transfers upon healthcare 3 2 visits for pregnant mothers, their newborns, and their toddlers. The PIP and PKH programs should also work in concert in the education sector: both programs provide cash transfers to targeted households that have demonstrated Poor Vulnerable The Rest they are making investments in education for their children. Finally, the BLSM unconditional cash transfer protects general welfare in times Source: Susenas 2014 and World Bank staff calculations 135 Susenas (2014). 136 Here we take the value of household expenditures as a proxy for household welfare. 137 Imagine the expenditure-share weighted price change in Poor households are more the household consumption basket when the price of food increases by 10 percent: in poor households, 66 percent of the consumption basket has become 10 percent more expensive while in non-poor households only 50 percent exposed to macroeconomic of the consumption basket has become 10 percent more expensive. Diversification as a macroeconomic risk-reduction strategy works at any level of disaggregation. For example, and fiscal risk than non– poor households spend one quarter of their food budget on rice alone (and no other food item has a larger food budget share) while non-poor, non-vulnerable households spend one-quarter of their food budget on prepared food alone poor households. (and no other food item has a larger food budget share). But prepared food is itself composed of inputs beyond raw agricultural inputs; prepared food often requires fuel (for cooking and for transport), transport and logistics services, capital investments and infrastructure, and labor. Therefore, when the price of rice goes up by 10 percent, 25 percent of a poor household’s food consumption basket will increase by 25 percent. For a non-poor household, the food consumption budget will go up less than that amount as the price of fuel, labor, and capital (and any other inputs to “prepared food”) are unaffected. 81 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 of acute macroeconomic stress occasioned by A recent law on disability143 paves the way for subsidy and managed-price reform. In other some reform at least in terms of inclusion into words, the current array of social assistance the labor market: public and private enter- programs is a good match to the current array prises must have 2 and 1 percent of employ- of risks faced by poor households. ees hired be disabled. The law also details the duty of government to protect and rehabilitate However, fragmentation produces a system the disabled, for instance via social assistance that is less effective. Nonetheless, social assis- among other pathways, but without an imple- tance programs are not currently integrated mentation regulation on the law it falls short at any level of provision or implementation. on stipulating exactly what programs and ini- Household survey analysis shows that, of the tiatives would implement the law. An existing poorest households that are nominally eligible program, ASODKB has been in existence but for each of the Rastra, PIP, and JKN-PBI pro- has been mired in low coverage (much like the grams, only 8 percent are actually in receipt of child cash grant and old age social assistance all three programs (in 2015), while about 37 per- program). As with the program for the elderly, cent receive at most two of the three.138 There- ASLUT, ASODKB may be merged and expanded fore, for a household in need of strategies and within the flagship CCT program, PKH. tools for addressing all risks encountered when they are salient, Indonesia’s social assistance Indonesia also provides very little assistance system is less effective.139 through labor-market “activation” programs. For example, those young adults with poor cre- Meanwhile, some important risks do not yet dentials entering the labor market will more have an adequate counterpart SA solution. often than not become adults with low-qual- For example, low enrolment in structured pre- ity jobs and retirees without pensions. Pro- school activities likely reduces school-readi- viding a low-cost “workplace training” initia- ness in poor households, while international tive through employers, through schools, or evidence shows that thoughtful140 ECED ini- through community-based institutions could tiatives can lead to improved child nutrition provide a post-basic-education path to high- and health, higher enrolment rates (when basic er-quality jobs.144 Or the high rates of part-time education begins), and increased mental apti- employment among the poor—33 percent of tude, and also that the positive results are often poor, working-age individuals are estimated to greater specifically for poor households (World be “underemployed”—could be partially ame- Bank, 2012b). Gains to abilities and skill can liorated with workfare programs, including a accumulate through a child’s entire education “basic income transfer” to families for whom career, so “good starts” may go some way toward workfare labor does not provide a large enough reducing the intergenerational transmission of income boost to make the program worthwhile.145 poverty by enhancing learning ability, school- ing and future skills development among others. Increased exposure to macroeconomic also re- mains uncovered. While the 2012 Social Assis- The elderly are more likely to be poor or vul- tance Public Expenditure Review recommend- nerable as they retire without access to pen- ed that an automatic, temporary, emergency 138 PKH was still a very small program sion and to a large extent (40 percent of those income support facility be established for the covering about 5 to 6 percent of Indonesian households in 2014. The percent of over 65 years of age141) with no health insur- poorest households experiencing adverse ad- eligible households receiving all four main ance as well. Indonesia has begun to address vents as a result of macroeconomic stress, no programs—PKH, Rastra, PIP, and PBI—iswas just over 2 percent in 2014. this risk through SJSN’s BPJS labor program, such facility has yet been established. The 139 The rest of the chapters in this report explore the institutional complexities of which integrates and expands the existing suite nascent CMRS has established a robust moni- SA provision in Indonesia that lead to this of labor-related insurance programs including toring protocol hosted by TNP2K, but the coun- mismatch between an “effective on paper” system and its “in the field” results. pension and old age savings programs. Howev- tercyclical, automatic SA response, triggered by 140 Internationally, effective ECD initiatives include parental involvement in program er, the risk is likely to remain in the medium pre-defined adverse events, is still under discus- design, service provision, and scheduling term for the poor and vulnerable elderly, espe- sion. Therefore, the CMRS can help poor and and provide parent training; collaboration with local stakeholders and active NGOs; cially those with a history of informal work. To vulnerable households to anticipate upcoming and cost-sharing between governments and beneficiaries so that each side has “skin in respond to this risk, that is apparent especially shocks, but it does not (yet) give them any addi- the game”. for the poor and vulnerable elderly with a his- tional flexibility in greeting those shocks when 141 Susenas 2016. 142 Indonesia previously had a very small tory of informal labor, a non-contributory pen- they arrive. In 2013 following a significant re- non-contributory pension program covering only around 26,500 individuals. When the sion for those who reach retirement age with duction in energy subsidies and a noticeable pension initiative under the National Social no pension in place would clearly be beneficial spike in inflation, a temporary compensation Security Plan (SJSN) is fully operational, the Government intends to offer subsidized for poor and vulnerable retirees themselves, scheme (known as BLSM; see also Section 4) contributions as a benefit to targeted poor and vulnerable households. For a more as well as the households that support them.142 was negotiated and delivered. BLSM improved detailed look at ageing in Indonesia and A specific cash transfer for vulnerable elderly, on the share of resources distributed to poor beyond, see World Bank 2016c. Live Long and Prosper – Aging in East Asia and Pacific. ASLUT, exists but has had a very low coverage and near-poor households (relative to previous 143 UU 2016 no. 8 144 See, for example, World Bank 2011c for the entirety of its operation versions of the same program), but it was not “From evidence to policy. Do vouchers certain that there would be any compensation. for job training programs help?” Human Development note No. 65766 for evidence Addressing the risk of disability will require a from Kenya. 145 Piloting both workfare programs more systematic and programmatic approach, and a “basic needs” transfer were also the initial steps of which were taken recently. recommendations from the 2012 Social Assistance PER. 82 Chapter Chapter 3— The “Life-Cycle” Approach 3 Assistance & Safety Nets to Social Providing a low-cost “workplace training” initiative through employers, through schools, or through community-based institutions could provide a post-basic- education path to higher-quality jobs. P.83 CHAPTER 4 Integrated Social Assistance: Possibilities & Benefits i Indonesia needs to spend percentage points reduction in more on social assistance the national poverty headcount and more importantly use poverty rate per year between 2007 available resources more and 2010, to 0.5 of a percentage effectively for continued contri- point between 2011 and 2017, the bution in poverty and inequality demand for improved implemen- reduction. While the household tation performance and the col- targeted social assistance spend- lective impact of major social as- ing have grown since 2005, as a sistance programs is greater than percentage of total public spend- ever before. ing the overall social assistance ex- penditure has stayed roughly con- Because there were about 6 million 146 fewer poor individuals in 2015 (than there stant in the decade between 2005 were in 2005), “per poor capita” real social and 2016.146 As the pace of poverty assistance expenditure as a percentage of GNI per capita had increased from 3.3% in reduction has slowed: from 1.2 2005 to 4.6% in 2015. 85 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 The existing social assistance pro- for health, education, as well as in from sharing common processes. grams could bring bigger collec- some cases riskier but more pro- The current practices is that each tive impact in poverty reduction ductive income-generating activi- agency delivering social assistance if better integrated. This report ties that reduce inequality of oppor- manage all related business prac- documents that while each in- tunity for children and adults alike. tices—data management, com- dividual program is relevant for pliance verification, updates and poverty reduction by creating a An integrated system could also payments—which require struc- transparent pathway for beneficia- tailor protection strategies to all tures and skills not often internal- ry households to mitigate a clear- vulnerable population. In Indo- ly well-developed. International ly-defined risk, none of these in- nesia, each program naturally fo- experience suggests that delegat- dividual initiatives is by design to cuses on the needs of the specific ing these program processes or help all targeted households fully population for which it is respon- sub-processes allows each agency mitigate or absorb all risks. These sible. As a consequence, individu- to focus on their own compara- programs either do not provide a als with some unique characteris- tive advantage. Likewise, programs transfer large enough, or cover a tics might fall into cracks between that operate independently of a large enough number of poor and the existing programs. For exam- common policy framework do not vulnerable households, to have a ple, there are no significant, na- naturally achieve coordination on significant impact on poverty147. tional level programs or dedicated sector-wide performance targets. Therefore, it is essential to coordi- agencies for female-headed house- nate and joint these programs in holds, informal-sector workers order to increase their collective impact. An example of a transfer who live alone or for orphans.148 If there were, the additional program POT E N T I AL package comprising of the four major social assistance transfers) would very likely be implemented in a “self-contained” manner and I MPACTS F RO M would boost eligible households’ income significantly larger than with little coordination with other relevant or even complementary I N T E G RAT I O N what have been actually seen programs. Under an integrated re- among the poor population. gime for policymaking, planning, While social assistance in Indo- and implementation, every agency nesia covers more individuals A truly integrated social assis- would have incentive to determine than ever before, few poor and tance system would help a house- individually and jointly how best to vulnerable households yet receive hold respond effectively to any reach these disadvantaged groups. a complete benefit package. The risk encountered. For example, social assistance programs covered an economy-wide shock such as Integration would help Indonesia in this series have broader or nar- a poor harvest may put consump- achieve better fiscal performance. rower upper income (or “means”) tion expenditures of many people To reduce poverty efficiently given cut-offs that delimit their target at risk. A household may respond a limited pool of resources, inclu- populations, but they all target at by pulling children from school sion errors or benefits should be least all poor households. Since or forgoing health care. A social received by as few non-targeted the publication of the 2012 Social assistance system that at least households as is politically, social- Assistance Public Expenditure partially restores consumption ly, and culturally feasible. However, Review, JKN-PBI targeted an addi- levels will save individuals and while Indonesian social assistance tional 15 million beneficiaries; PIP households from short-term pov- programs currently have the same targeted an additional 10 million erty. However, if the transfer is target population in principle, in students; and PKH targeted near- not large enough or if there is not reality they reach different popu- ly 4.5 million new families. Rastra a separate incentive for keeping lations due to uncoordinated im- coverage stayed roughly the same children in school, there may still plementation. For example, large in terms of the absolute number be a long-term negative impact on numbers of non-poor households of beneficiaries targeted. Howev- welfare from reduced schooling. receive subsidized Rastra rice and er, the proportion of poor house- Therefore, this integrated social utilize JKN-PBI. These inclusion holds receiving all four “main” assistance system would provide errors increase the cost (to the programs—Rastra, JKN-PBI, PIP, likely multiple but coordinated Government) of reducing pover- and PKH—was just over 1 percent interventions for both immediate ty through those programs. An- in 2013 and just over 2 percent in relief and long-lasting incentives other potential efficiency gain is 2014.149 147 To reduce the incidence of poverty or poverty correlates (achievement gaps in education, in healthy behaviors and health outcomes, or in wages and productivity, for example), benefits provided to targeted households should be large enough to erase their income gap or, when transfers are conditional upon behaviors, large enough to allow households to comply fully without further impoverishment. In the poorest 10 percent of households ranked by consumption expenditure, Rastra covers the most households (68 percent of this decile buys Rastra rice) but Rastra has the lowest transfer value at 4 percent of average total consumption expenditure in this decile. In contrast, PKH has the lowest coverage at 9 percent and the highest transfer value at around 13 percent. 148 Programs exist for elderly, or for disabled individuals but have been kept at very low coverage. 149 Observers suggest the percentage-point increase in complete SA-program-portfolio coverage can be traced to coordination efforts (led by TNP2K) that led to the adoption, by all main social assistance programs, of the UDB standard for determining eligibility for new beneficiaries. 86 Chapter 4— Integrated Social Assistance: Possibilities & Benefits A lack of enforceable common standards for still controlled by schools and their local and Sizeable, one-time reductions in poverty could local social implementation exacerbates neg- regional stakeholders; and the distribution of be achieved through integration at the house- ative effects of a fragmented social assistance previously Jamkesmas now Kartu Indonesiat hold level. To simulate an “integrated” trans- architecture. Low program coverage is ob- Sehat (KIS that proves the recipient is part of fer, a single benefit comprised of all the existing served across the array of social assistance, and JKN-PBI) cards has not previously been mon- cash transfers (PKH and PIP) and a monetized is partly due to leakage of benefits to non-poor, itored or evaluated. That such practices are value of the Rastra in-kind transfer, was intro- non-vulnerable households. Qualitative and tolerated indicates a low level of coordination duced into a static expenditure model. Smaller quantitative evidence from Indonesia suggests between central-government-level policy plan- and larger groups were “targeted”: the small- that idiosyncratic and varied eligibility deter- ners and funding authorities and the region- er (larger) group is defined as the bottom 10 mination procedures, targeting frameworks, al- and local-level administrations who have (40) percent of Susenas households ranked by and benefit ownership and control rights ac- been delegated crucial portions of important pre-transfer expenditure. Table 4.1 illustrates count all have negative impacts. For example, social assistance processes.150 Furthermore, that a single, integrated benefit could provide a the allocation of “rights” to purchase subsi- these entrenched and idiosyncratic operating significant boost to consumption expenditure dized Rastra rice is done by village heads and principles suggest that increasing any single of 14 to 21 percent of an average targeted house- sub-village administrations; a portion of PIP program’s coverage will bring only a small per- hold’s budget.151 benefits (and the right to distribute them) are centage of targeted households a “full” benefit. FIGURE 4.1 SA convergence in the poorest 10 TABLE 4.1 percent of households Integration scenario share out of household expenditure by target group Target group Average Average % of PKH monthly HH monthly recipients expenditure who (IDR) Poorest 10% 1,306,137 251,406 19% RAST RA Poorest 40% 1,814,236 234,972 13% 7 8% Source: Susenas 2016. 43% PBI 53% A lack of enforceable common standards for local social implementation exacerbates negative effects of a fragmented P IP social assistance 16% architecture. PKH 6.5% 150 Unlike most other public social expenditures which are in large part executed by regional governments (primarily district-level governments), 85 percent (on average) of social assistance programming expenditures are centrally executed. See the Expenditure Summary report in this series ALL FOUR 2.2% PIP,PKH 2.6% or World Bank (2012j) for more detail on the history and contemporaneous particulars of this arrangement. RASTRA, PBI, PIP 11% RASTRA,PIP 14% 151 Larger transfers are expected to have both macroeconomic RASTRA,PIP,PKH 2.5% RASKIN, PKH 6% and micro-behavioral impacts. For example, inflation PBI,PIP,PKH 2.3% (including higher charges by service providers with knowledge of an individual’s beneficiary status), family planning decisions, and consumption patterns would all be expected to change if Source: Susenas 2014 and World Bank staff calculations. SA benefits were integrated at the household level. The static Note: PKH’s overall coverage is low and is (technically) restricted to the extreme poor; due to PKH’s small size, Susenas tends to expenditure model explore here cannot account for these under estimate the actual size program, leading to lower program overlap among the poor or vulnerable populations. general equilibrium effects. 87 Toward a Comprehensive, Integrated, & Effective Social Assistance System in Indonesia – SAPER 2017 A single benefit also generates larger reduc- TABLE 4.2 efficiency and ease of access to social assis- tions in poverty, vulnerability and inequality tance transfers, the government plans to inte- than the current fragmented benefit pack- Estimated impact of integration on grate all social assistance payments under the age (Table 4.2). For example, the “overnight” poverty, vulnerability and inequality KKS card. More recent developments toward reduction in the headcount poverty rate that digitally rendered social assistance transfers would result from benefit integration is ex- Headcount Gini coefficient are being spearheaded by MoSA, with 1.4 mil- pected to be 2 to 4 percentage points. For ref- poverty (%) lion previous Rastra recipients now receiv- erence, if actual headcount poverty continues 2016 September 10.7 .397 ing an e-voucher benefit via the KKS card and to fall at the rate experienced between 2013 and E-Warong delivery system. In addition, MoSA 2016 (about 0.1 to 0.5 of a percentage point per aims to render PKH transfers to all 6 million year), it would take about 10 years to achieve the Integration: 10% 8.65 .394 beneficiaries via a collection of state owned “overnight” reduction that the least expensive level banks (HIMBARA) and also using the KKS card integration scenario achieves immediately. 152 as the unified payment platform. Integration: 40% 6.72 .388 level Latin America’s experience can provide Indo- nesia with possible strategies for sector-wide INTEGRATIO N N E E D S Source Susenas 2015/16. Note the following targeting accuracy assumptions are based on survey data SA incidence levels extrapolated to an integration and for integrating all levels of government to better deliver on shared social TO ACCEL E RAT E & expanded overall coverage level of 40 percent: assistance tasks. International experience sug- 10% level 55% of Decile 1 (D1), 20% of D2, 15% of D3, 10% of D4 gests that there is more than one way to effec- LE ARN L ESSON S FRO M 40% level 90% of D1, 75% of D2, 65% of D3, 50% of D4, 40% of D5, 30% of D6, 20% of D7, 15% of D8, 10% tively and efficiently distribute essential social assistance delivery processes between agencies, OTHER COU N T R I E S of D9, 5% of D10. and between central and local administrative authorities. For example, Brazil consolidated at the central administrative level its diverse landscape of social assistance initiatives into Indonesia has made the initial steps toward a single program uniting education-, health-, greater integration by launching the SLRT nutrition-, and basic income-focused transfers and ODA initiatives as well as the movement for poor and vulnerable households. Chile toward digitizing social assistance payments. and Colombia proceeded in a slightly different The Integrated Referral System (SLRT) is imple- fashion: coordinating on delivery standards mented by MoSA. The program has the purpose for national and local initiatives alike, as well as to help identify and refer poor and vulnerable strengthening the provision of complementary households to social protection programs via public services to enhance the local coverage of the facilitators and ‘single-window’ offices poor and vulnerable households. at the district level. The program facilitators use a tablet-based application that contains The establishment and refinement of a com- relevant data from the UDB, key information mon, authoritative targeting and beneficiary about national and local programs as well as an selection procedure has been an integrative assessment instrument to define a household’s catalyst. The approaches to integration de- welfare status. After an initial piloting phase during 2015-2016 the program was deemed to have a high potential and was scaled up to be implemented in 50 districts in 2016 and up to 150 districts by 2019. Up until 2016, 59 local governments have helped refer 146,000 house- holds for further assistance. The On Demand Application (ODA), managed by TNP2K, was designed to work in tandem with the SLRT and has the purpose to update household informa- tion that is used in determining eligibility and relative welfare status used in targeting. ODA has been piloted and is operating in 12 districts. In line with the governments push to achieve greater financial inclusion, the government has mandated to digitize all social assistance payments. 153With the purpose of increasing 152 In results not presented, the integration simulation is run with parameters describing known SA implementation weaknesses like local benefit deduction or local re-allocation of benefit pools to larger populations (and resulting benefit dilution). With these parameters included, poverty rate impacts are reduced by 0.12 to 0.18 of a percentage point. 153 The PIP program under MoEC and MoRA has had payments rendered with the help of the BRI and BNI banks already since 2013. 88 Chapter 4— Integrated Social Assistance: Possibilities & Benefits tailed below have in common the foundation istry of Social Development, but many sub-op- maintaining and updating the registry through (and further development) of a unified target- erations are delegated to other government applications and grievances received. 157 ing system that identifies and selects beneficia- actors at local, regional, and national levels. ries from common populations. In Colombia, For example, the state-owned Caixa Economica Colombia has deployed outreach and facilita- for example, nationally- and locally-executed Federal bank gathers and reports compliance tion to integrate a diverse program mix at the social assistance programs converged opera- data generated by health, education, and local beneficiary level. Colombia also uses the local- tionally around a common target population government providers; collects and reports up- ly-operated SISBEN system to identify families established by an authoritative household dates (made by municipalities) to the Unified for the “Together” program, which provides registry. Indonesian programs have a com- Beneficiary Registry (Cadastro Unico); and gen- to families a social worker or facilitator who mon targeted population in principle but in erates payment instructions, and then makes liaises with locally-available social programs practice each program makes a unique deter- payments, for all active beneficiaries. The and service providers, for example child care mination of eligibility and beneficiary selec- municipalities themselves register families, or and youth training providers, micro-credit fa- tion. If instead for each program an authori- update the information of those already reg- cilities, conditional cash transfer programs or tative registry of households was consulted to istered, in the Cadastro Unico; coordinate and scholarships, and housing subsidies. So while select beneficiaries according to common and monitor compliance verification reporting by integration “at the top” is achieved via adher- authoritative procedures, poor and vulnerable health, education, and local government ser- ence to a common standard (embodied by the households would have an integrated benefit vice providers; are the first point of contact to SISBEN system), integration at the household package regardless of the extent of institutional receive complaints and grievances about either level is achieved by linking benefits to house- or central-local integration.154 Bolsa Familia itself or the Cadastro Unico; and holds through facilitation and active outreach. provide links for Bolsa Familia households to Brazil has led the way in creating “single complementary services and benefits in health, Chile’s long-standing158 and authoritative window” service for social assistance benefi- education, and livelihood initiatives (includ- household registry has evolved to capture ciaries. In 2003, Brazil began to merge food, ing microcredit and professional counseling) . deprivation, vulnerability, and the relevant gas subsidy, and direct cash transfer programs risks to well-being in all their forms. By now, into a single benefit called Bolsa Familia or the Colombia has used a single targeting frame- about two-thirds of the population (3.7 million (“Family Grant”).155 Bolsa Familia remains to- work to integrate programs institutionally households containing about 11 million indi- day a conditional cash transfer in that benefit and at the household level. In 1994, Colombia viduals in 2011) are registered. There are 14 receipt depends on health care, and primary established a national targeting system—the ministries, 24 social services and 200 programs and secondary education utilization. Bolsa Fa- Colombian System for Selecting Beneficiaries that use the registry to select beneficiaries,159 milia’s target population includes any family for Social Programs (SISBEN)—to distribute which suggests that common standards, when living below the national poverty line (about all social assistance expenditures; SISBEN authoritative, can integrate operations even 13 million families in 2015), but the size of the has been updated regularly and remains the when the programmatic landscape is diverse benefit depends on household composition authoritative system for identifying and se- and the implementing agencies are numerous. and characteristics. lecting beneficiaries for an evolving suite of Also similar to Colombia, day-to-day opera- financial and social assistance programs.156 A tions such as maintenance, updates, and com- Brazil’s single transfer is executed by a single central government agency formally adminis- plaint-handling are delegated to municipali- ministry, while sub-processes are delegated ters and executes the system—for instance el- ties, while financing, planning, administering, to government agencies with comparative igibility cut-offs and eligibility determination quality control, evaluation and user guidelines advantages. Bolsa Familia is planned, adminis- are completed centrally—while municipalities are completed by the central government. Cru- tered, implemented, and evaluated by the Min- are tasked with day-to-day operations such as cially, household ranking and determination of eligibilities are also completed at the central level and communicated to municipalities. Chile is also using outreach, service provider linkages, and coordinated social assistance delivery to ameliorate poverty holistical- ly. Households found to be living in extreme poverty are provided two years of professional social counseling services from a locally-based social worker. This social worker formulates a poverty-exit strategy with the household and links them to complementary (and locally available) services to enhance the household’s 154 Indonesia has already made great strides in this direction with the establishment of the collective human capital, the value and pro- UDB for Social Protection. However, the UDB is not yet authoritative in the selection of beneficiaries, which limits its usefulness as a common standard around which program ductivity of their dwelling, and the household’s implementers can converge ability to generate income. Agencies providing 155 Created in October, 2003, Law No. 10.836, 2003. 156 The first introduction of a targeting instrument to allocate subsidies of social programs was services and programs are encouraged to coor- made in the Law 60 of 1993 (Arts. 2 and 3). A task force within National Planning Department (DNP) was in charge of designing and implementing SISBEN; the same team provided dinate on service schedules specifically for ex- assistance to local governments (departments and municipalities) in SISBEN implementation tremely poor households. Meanwhile, house- in their areas. Beneficiary eligibility is determined by a proxy means test (PMT) over a set of socioeconomic and demographic variables. holds receive direct cash transfers that can 157 Municipalities also contribute a small portion to SISBEN’s operating budget. 158 Chile’s main targeting system, the Ficha Comites de Acción Social (Social Action Committees continue for up to 3 years after the 2-year fa- Registry), was established in 1981. cilitated introduction (Galasso, 2011),160 while 159 While single initiatives use different eligibility definitions—some use the poorest 5 percent, some the first, second, third, or even fourth poorest quintile, and some use a means cutoff facilitators regularly monitor a household’s with additional characteristics like disability or advanced age—they all implicitly submit to an eligibility standard that is determined by the same household registry. progress with respect to its own plan. 160 The initiative program also now includes a social protection objective: households facing uninsured risk are protected from further impoverishment. CHAPTER 5 Summary & Main Recommendations i ndonesian administrations have pursued logical, The current collection of programs should be enhanced progressive, and empirically founded revisions to to cover more of the salient life cycle risks. Both the pre- P.89 SA operations and institutional structure, but over- vious Social Assistance Public Expenditure Review and this all the pace of reform has been tentative. For exam- report have noted that there are noticeable gaps at crucial ple, though public expenditures on household-based social junctures in an individual’s life cycle (in addition to gaps assistance programming have spiked to nearly 1 percent of in coverage for social assistance transfers that do address GDP in years in which fuel subsidies were reduced or elimi- salient risks, see below). In particular, for younger-than- nated, those increases were temporary; as a share of GDP the school age children and their parents, there is no national, level of spending on permanent household-based social as- programmatic ECED initiative. For those at the end of their sistance programs has remained about constant from 2005 schooling career entering the labor market for the first time, to 2016. Likewise, major coverage increases in most of the or for those whose lack of schooling have left them in low- main permanent programs have been too small to fully cov- skilled, low-paying jobs, there are no national labor-market er the targeted populations. At these spending and coverage activation, skills training, second-chance education, or em- levels, Indonesia’s social assistance programs eliminate less ployment services initiatives. For those nearing retirement, than 20 percent of the total poverty gap. Operational re- there is no social pension system through which welfare lev- forms have been haphazard: some programs have pursued els can be maintained even after labor-market productivity operational reforms, while ignoring their own inefficient falls. The nascent CMRS has established a robust monitor- institutional arrangements; some have been radically re- ing protocol, but the countercyclical, automatic SA response, formed institutionally, while ignoring ineffective opera- triggered by pre-defined adverse events, is yet to be insti- tional practices; and some have not made significant re- tutionalized. The current incarnation of the CMRS could forms in either area. help poor and vulnerable households anticipate upcoming shocks, but it does not give them any additional flexibility in greeting those shocks when they arrive. Independently implemented programs should ciary households to keep up with the increasing oughly each program incorporated procedures be integrated under a “One-System” type of costs of and education (in which both PKH and into standard operating procedures. For exam- approach to cover all individuals in the tar- PIP require investment). While the Rastra pro- ple, Rastra uses the UDB to generate regional geted population and provide the government gram—which provides a very small benefit to a rice quotas; the determination of the identity a more efficient and effective SA delivery sys- population with a significant proportion of un- of those who have the right to purchase Ras- tem. Three of four of the permanent social as- targeted, ineligible beneficiaries—remains the tra rice is made locally. PKH’s commitment to sistance programs dramatically increased cov- second-largest SA initiative (in terms of public UDB procedures, meanwhile, is more thorough erage between 2012 and 2016 (while the fourth expenditures), the PKH program—which pro- and even includes bi-directional updating161 of had roughly constant coverage), and yet in 2014 vides a significant benefit to a small population household status, which keeps both PKH eligi- less than 5 percent of eligible households in most of whom are actually targeted by the pro- bility lists, as well as the UDB, current. An inte- the poorest expenditure decile received all four gram—remains the smallest (in terms of total grated benefit package of all four of the current programs. There are limits, in other words, to public expenditures absorbed by the PKH pro- social assistance transfers, delivered reliably to comprehensive SA coverage through program gram). In other words, less effective social assis- all eligible beneficiaries, will only occur once expansion alone. Both this report and the tance transfers are still receiving large budget all programs thoroughly adopt a unified target- previous Social Assistance Public Expenditure shares, while more effective social assistance ing procedure. Review recommend achieving comprehensive transfers receive small budget shares. The first coverage of households and risks through co- solution to Rastra’s benefit dilution problem is Finally, an important SA reform that will fos- ordination and integration in the social assis- to select its beneficiaries strictly from the na- ter greater program convergence is already tance sector instead. The greater the degree of tional registry. The second solution is to deploy underway. The Government has recognized integration, however it is achieved, the less like- a standard delivery platform that is transparent the importance updating the national registry ly any eligible household will fail to receive any and accountable. The e-Warong initiative has regularly, transforming the existing static UDB particular transfer or service, or initiative that the potential to be such a platform by lever- into a dynamic and two-way registry of poor the social assistance sector provides, and the aging an e-voucher mechanism for the Rastra and vulnerable households, as exists in other more likely that poverty—which is a multi-di- benefit. If managed carefully, this program is countries such as Chile, Turkey, Brazil, and Aus- mensional problem—can be affectively amelio- likely to increase the allocation of Rastra rice to tralia. A more dynamic social registry of poor rated by a collection of one-dimensional pro- targeted beneficiaries and so decrease the pro- and vulnerable households will support SA pro- grams addressing in concert the multiple needs gram’s inclusion errors. gram integration, faster program response to of poor and vulnerable families. changes in the needs of families, and also allow The poor socialization and varied targeting poor families excluded from social assistance Simultaneously, the currently operating social that undermined performance in the past programs to potentially become included. The assistance programs should aim to deliver the have been only partially remedied with the SLRT and ODA initiatives that are essentially right benefits to the right people at the right SA-wide adoption of a unified targeting sys- integrated referral systems will require thor- time. For instance, Rastra promises beneficia- tem; additional reform is necessary. Leakage to ough M&E for them to effectively update the ries 15 kilograms of rice per month, but delivers non-targeted populations still ranges from low UDB, so that it becomes a truly dynamic and in- only a fraction of that amount. PIP revises its (for example, in PKH) to high (for example, in clusive registry of poor and vulnerable households. benefit magnitudes too infrequently for benefi- Rastra). This is partially traceable to how thor- As the Government has indicated that the current schedule of nationwide updates to the UDB—the UDB was established in 2011 and updated 2015—will cease. Therefore this bi-directional 161 updating and integrated referral system—where all initiatives making use of the UDB can provide updates on beneficiaries (or potential beneficiaries) in their own program—is likely to become a critical feature of the new national registry interface. P.91 References Bappenas (Kementerian Perencanaan Pembangunan Nasional Republik Indonesia). 2013. Masterplan Percepatan dan Perluasan Pengurangan Kemiskinan Indonesia Tahun 2013-2025. Jakarta: Government of Indonesia. Bappenas (Kementerian Perencanaan Pembangunan Nasional Republik Indonesia). 2014. Social Assistance Table. Internal documentation. Jakarta: Government of Indonesia. Bappenas (Kementerian Perencanaan Pembangunan Nasional Republik Indonesia). 2015. Indonesia Roadmap for Housing Policy Reform. Jakarta: Government of Indonesia. Bappenas (Kementerian Perencanaan Pembangunan Nasional Republik Indonesia). 2017. Strategi Pengembangan Penghidupan Berkelanjutan Bagu Rumah Tangga Miskin. Infographic. Jakarta, Indonesia. Castaneda, Tarsicio, and Kathy Lindert, with Benedicte de la Briere, Luisa Fernandez, Celia Hubert, Osvaldo Larranaga, Monica Orozco and Roxana Viquez. 2005. Designing and Implementing Household Targeting Systems: Lessons from Latin American and the United States. Social Protection Discussion Paper 0526. Washington, DC: World Bank. Centre for Health Research, University of Indonesia. 2010. PKH Spot Check: Quantitative and Qualitative Assessments to Monitor Household CCT Operations. Unpublished note. Centre for Social Protection (CSP). 2011. Are Conditions Pro-Women? A Case Study of a Conditional Cash Transfer in Indonesia. Brighton: CSP Research Report 03. Fernandez, L., Hadiwidjaja G. 2012. Analysis of the Size of the Cash Transfer of PKH (CCT) Program in Indonesia. World Bank technical note. Jakarta: World Bank. Galasso, Emanuela. 2011. Alleviating Extreme Poverty in Chile: The Short-term Effects of Chile Solidario. Estudios de Economía. 38 (1): 101-127. Junio 2011. Grosh, Margaret, Carlo del Ninno, Emil Tesliuc, Azedine Ouerghi. 2008. For Protection and Promotion: The Design and Implementation of Effective Safety Nets. Washington, DC: World Bank. IPC (International Policy Centre for Inclusive Growth). 2013. Program Keluarga Harapan (PKH): Indonesian Conditional Cash Transfer Program. Research Brief. Jakarta. IPC (International Policy Centre for Inclusive Growth). 2014. Indonesian Cash Transfer Program for Poor Students. Research Brief. Jakarta. MoF (Ministry of Finance). Financial Note Series, multiple years. Jakarta: Government of Indonesia. MoH (Ministry of Health). 2016. “Introduction to Constructive Dialog for JKN Improvement.” Presentation April 30, 2016. MoSA (Ministry of Social Affairs). 2014. Structural Organization of PKH. Internal documentation. Jakarta. MoSA (Ministry of Social Affairs). 2016a. Daftar Isian Pelaksanaan Anggaran Petikan PKH Tahun Anggaran 2016. Budget note. Jakarta: Government of Indonesia. MoSA (Ministry of Social Affairs). 2016b. “Perluasan Program Keluarga Harapan (PKH) / CCT untuk pengurangan kemiskinan dan ketimpangan.” Presentation at a TNP2K plenary meeting, February 4, 2016. Jakarta. MoSA (Ministry of Social Affairs). 2016c. “Kebijakan Program PKH 2016.” Presentation. Jakarta. MoSA (Ministry of Social Affairs). 2016d. “Social Assistance Cashless Payments through E-Warong KUBE PKH.” Presentation. Jakarta. MoSA (Ministry of Social Affairs). 2016e. Pedoman Umum Program Keluarga Harapan. Printed edition. Jakarta: Government of Indonesia. Oxford Policy Management. 2012. Review of, and Recommendations for, Grievance Mechanisms for Social Protection Programs. Final Report Summary. Oxford. ODI (Overseas Development Institute). 2010. Gender and Social Protection in Indonesia: Raskin the food subsidy program. Project briefing. Jakarta. October, 2010. Perdana, A. Prima, R.A. Satriawan, E. 2015. Lessons Learnt from Trialing the Newly Designed Raskin Programme for 2012: Findings from TNP2K’s Monitoring. Forthcoming TNP2K. Jakarta: Government of Indonesia. Peraturan President. 2016. Strategi Nasional Keuangan Inklusif. Jakarta, Indonesia. Riskesdas. 2013. Penyajian Pokok-Pokok Hasil Riset Kesehatan Dasar 2013. Ministry of Health. Jakarta. Smeru. 2008. The Effectiveness of the Raskin Program. Jakarta: Smeru Research Institute. Smeru. 2011. Qualitative Impact Study for PNPM Generasi and PKH on the Provision and Utilization of Maternal and Child Health Services and Basic Education Services in the Provinces of West Java and East Nusa Tenggara. Jakarta: Smeru Research Institute. Susenas (multiple years). “Survei Sosial Ekonomi Nasional”. National household survey dataset. Jakarta: Statistics Indonesia (BPS): Timmer, P. et al. 2017. Evolution and implementation of the Raskin Program in Indonesia. Forthcoming book chapter. TNP2K (Tim Nasional Percepatan Penanggulangan Kemiskinan). 2013. “Indonesian Experience toward Targeted Social Assistance Reform.” Presentation files. Jakarta: Government of Indonesia. Presentation by Dr. Bambang Widianto at Poverty Alleviation Academic Conference, June 12, 2013. TNP2K (Tim Nasional Percepatan Penanggulangan Kemiskinan). 2014a. “Perbaikan Pelaksanaan Program Bantuan Siswa Miskin (BSM).” TNP2K Infographics. Jakarta: Government of Indonesia. TNP2K (Tim Nasional Percepatan Penanggulangan Kemiskinan). 2014b. “Strengthening Raskin”. TNP2K Infographics. Jakarta: Government of Indonesia. TNP2K (Tim Nasional Percepatan Penanggulangan Kemiskinan). 2014c. Strengthening National Health Insurance. TNP2K Infographics. Jakarta: Government of Indonesia. TNP2K (Tim Nasional Percepatan Penanggulangan Kemiskinan). 2014d. “Peran Pemerintah Daerah dalam Implementasi Jaminan Kesehatan Nasional (JKN).” Jakarta, May 13, 2014. TNP2K Presentation files. Jakarta: Government of Indonesia. TNP2K (Tim Nasional Percepatan Penanggulangan Kemiskinan). 2014e. “Upaya Meningkatkan Ketepatan Sasaran dan Jumlah Beras yang Diterima RTS-PM dalam Rangka Percepatan Penanggulangan Kemiskinan.” TNP2K Presentation files. Jakarta: Government of Indonesia. Presentation by Sri Kusumastuti Rahayu in ISEI, Jakarta, January 2, 2014. TNP2K (Tim Nasional Percepatan Penanggulangan Kemiskinan). 2014f. “Reaching Indonesia’s Poor and Vulnerable and Reducing Inequality: Improving Programme Targeting, Design, and Process.” TNP2K (Tim Nasional Percepatan Penanggulangan Kemiskinan). 2015a. “Evaluating Long Term Impact of Indonesia’s CCT Program: Evidence from a Randomized Control Trial.” Forthcoming. TNP2K (Tim Nasional Percepatan Penanggulangan Kemiskinan). 2015b. “Tantangan Meningkatkan Efektivitas Program Raskin.” TNP2K (Tim Nasional Percepatan Penanggulangan Kemiskinan). 2015c. “Perjalanan Menuju Jaminan Kesehatan Nasional.” Jakarta. TNP2K (Tim Nasional Percepatan Penanggulangan Kemiskinan). 2016. PIP Information Page. Jakarta, Indonesia. http://www.tnp2k.go.id/id/tanya- jawab/klaster-i/program-indonesia-pintar-melalui-kartu-indonesia-pintar-kip/. Accessed December 2016. TNP2K (Tim Nasional Percepatan Penanggulangan Kemiskinan). 2017. Re port of 2016 pilots on the distribution of non-cash in-kind social assistance. Jakarta, Indonesia. World Bank. 2009. Social Assistance Public Expenditure Review. Jakarta: World Bank. World Bank. 2011a. “Current Challenges, Future Potential.” Indonesia Economic Quarterly. June 2011. Jakarta: World Bank. World Bank. 2011b. “Program Keluarga Harapan: Main findings from the Impact Evaluation of Indonesia’s Pilot Household Conditional Cash Transfer Program.” Jakarta: World Bank. World Bank.2011c. “From Evidence to Policy: Do Vouchers for Job Training Programs Help”. Washington DC: World Bank. World Bank. 2012a. “Protecting Poor and Vulnerable Households in Indonesia.” Jakarta: World Bank. World Bank. 2012b. “Early Childhood Education and Development in Indonesia, Strong Foundations, Later Success.” Jakarta: World Bank. World Bank. 2012c. “BLT Temporary Unconditional Cash Transfer: Social Assistance Program and Public Expenditure Review 2.” Jakarta: World Bank. World Bank. 2012d. “Poverty Targeting in an Urban Setting: Indonesia’s Evolution.” Presentation at Forum Kajian Pembangunan Seminar. Jakarta: World Bank. World Bank. 2012e. “Raskin: Subsidized Rice Delivery. Social Assistance Expenditure Review 3.” Jakarta: World Bank. World Bank. 2012f. “Protecting Poor and Vulnerable Households in Indonesia.” Jakarta: World Bank. World Bank 2012g. “Bantuan Siswa Miskin – Cash transfers for poor students.” Social Assistance Program and Public Expenditure Review. Jakarta: World Bank. World Bank.2012h. “PKH Conditional Cash Transfer.” Social Assistance Program and Public Expenditure Review 6. Jakarta: World Bank. World Bank. 2012i. “Protecting Poor and Vulnerable Households in Indonesia.” Social Assistance Program and Public Expenditure Review 2012. Jakarta: World Bank. World Bank. 2013a. “Slower growth, high risks.” Indonesia Economic Quarterly. December 2013. Jakarta: World Bank. World Bank. 2013b. “Adjusting to Pressures.” Indonesia Economic Quarterly, July 2013. Jakarta: World Bank. World Bank. 2013c. “The Double Burden of Malnutrition in Indonesia.” Jakarta: World Bank. World Bank. 2013d. “The Nuts & Bolts of Jamkesmas, Indonesia’s Government-Financed Health Coverage Program for the Poor and Near-Poor.” UNICO Studies Series 8. Jakarta: World Bank. World Bank. 2013e. “Indonesia: Urban Poverty and Program Review.” Jakarta: World Bank. World Bank. 2014a. “East Asia Pacific at Work: Employment, Enterprise, and Well-being.” Washington DC: World Bank. World Bank. 2014b. “The State of Social Safety Nets 2014.” Washington DC: World Bank. World Bank. 2014c. “Social Assistance Strategy for the RPJM (2015-2019) Budget Estimations.” Unpublished note. Jakarta: World Bank. World Bank. 2014d. “Urban Safety Net Case Study: Conditional Cash Transfer (PKH) Program in Urban Indonesia.” Forthcoming. Jakarta: World Bank. World Bank. 2015a. “The Social Impacts of BLSM.” Unpublished report. Jakarta: World Bank. World Bank. 2015b. “High expectations.” Indonesia Economic Quarterly. March 2015. Jakarta: World Bank. World Bank. 2015c. “Raskin Business Process Review”. Unpublished presentation. Jakarta: World Bank and Deloitte. World Bank. 2016a. “The Distributional Impact of Fiscal Policy in Indonesia.” Unpublished report. Jakarta: World Bank. World Bank. 2016b. “Indonesia Health Financing Systems Assessment.” Jakarta: World Bank. World Bank. 2017a. “Indonesia Consolidated Fiscal Dataset”. Unpublished. Jakarta: World Bank World Bank. 2017b. “Food Based Safety Net Initiatives - Chapter 7 Evolution and Implementation of the Raskin Program in Indonesia” by Peter Timmer, Hastuti, and Sudarno Sumarto. Unpublished. Washington DC: World Bank. World Food Programme. 2015. Food Security and Vulnerability Atlas of Indonesia. Dewan Ketahanan Pangan, Kementerian Pertanian and World Food Programme (WFP). Jakarta