Indonesia Spending more or spending better: Improving education financing in Indonesia East Asia and Pacific Region AVA *TF King UU KEMENTERRAN PSC ED _ K"U= -NEUROPEAN UNION THEWORLDBANK THE WORLD BANK OFFICE JAKARTA Indonesia Stock Exchange Building,Tower Il/1 2-13th Fl. JI. Jend. Sudirman Kay. 52-53 Jakarta 12910 Tel: (6221) 5299-3000 Fax: (6221) 5299-3111 March 2013 Spending More or Spending Better: Improving education financing in Indonesia is a product of staff of the World Bank. The findings, interpretation and conclusion expressed herein do not necessarily reflect the views of the Board of Executive Directors of the World Bank or the government they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denomination and other information shown on any map in this work do not imply any judgment on the part of the World Bank concerning the legal status of any territory or the endorsement of acceptance of such boundaries. Photo credit left to right: World Bank, Ratna Kesuma, World Bank. Report No. 73050-ID Spending more or spending better: Improving education financing in Indonesia East Asia and Pacific Region Spending more or spending better: Improving education financing in Indonesia 0 Table of Contents Acknowledgements vii List of Abbreviations viii Executive Summary 11 Chapter 1: How is the Indonesian Education System Financed and Managed? 21 A description of the National Education System in Indonesia 22 Who manages what in Indonesia's decentralized education system? 24 Who pays for what in Indonesia's education system? 25 Chapter 2: The"20 percent rule": Where did resources go? 29 Implications of the 20 percent rule 30 Is Indonesia spending too much on education after the 20 percent rule? 34 Where did the 20 percent go? 36 A simple costing of government objectives using current spending patterns 52 Chapter 3: How Have Outcomes Changed? The Unfinished Agenda for Equity and Quality of Education 55 Access 55 Equity 58 Who are out-of-school children? 60 Learning Outcomes 64 Linking resources to outcomes: Is it only a matter of time before resources improve quality of education? 68 Chapter 4: Improving the Quality of Spending on Basic Education 75 Improving and expanding existing programs: Scholarships for the poor 76 Improving Teacher Management 80 Improving the allocation of teachers: Policy Options 85 Improving district management and school discretion: effects from BOSDA pilots 92 Improving district performance using performance-based transfers to support BOSDA 99 Chapter 5: A Framework for Action: Policy options for improving the quality of spending 101 Objectives and main policy areas 102 Strategic Reallocation of Resources 103 Effective use of transfer mechanisms 104 Improving Teacher Management 105 Improving School and District Management 107 Improving Budget Planning, Transparency and Accountability. 108 References 111 U List of Figures Figure 1: Share of enrollment by provider type and level, 2010 22 Figure 2: National Education System in Indonesia 23 Figure 3: An overview of the complexity of transfers and fund flows, 2012 26 Figure 4: Public Expenditure on Education, 2001-2010 30 Figure 5: Windfalls Allocation by program, 2010 33 Figure 6: Public expenditure on education as a share of total government expenditure and as a share of GDP, selected countries in 2009 35 Figure 7: Per pupil spending on primary education as a share of per capita GDP, selected countries 35 Figure 8: Total education spending by level of education and as a share of the total education budget, 2008 and 2009 36 Figure 9: Share of education spending by level, selected East Asia countries, 2009 37 Figure 10: Spending on ECD, Central and Local Governments, 2008-2009 37 Figure 11: Total public and household spending and share of public and private spending by level of education 38 Figure 12: Spending by level of government, 2001-2009 38 Figure 13: Non-salary education expenditure by programs and level of government, 2009 39 Figure 14: Public spending by level of education, 2008-2009 39 Figure 15: Public spending by level of education, 2008-2009 40 Figure 16: Scholarship programs by level of education, 2008-2010 41 Figure 17: District spending by economic classification, 2001-2009 41 Figure 18: Real capital and recurrent spending by level of government, 2006-2009 42 Figure 19: Real spending by economic classification (total education budget), 2006-2009 43 Figure 20: Spending increase distribution after the "20 percent rule,"2009 43 Figure 21: Distribution of education funds by direct recipient, 2009 44 Figure 22: Share of total education budget spent on salaries, selected countries 44 Figure 23: Growth in number of students and number of teachers by level and student-teacher ratio, 2004-2010 45 Figure 24: Student-teacher ratio by level, 2010 46 Figure 25: Evolution of the number of civil servant (PNS) and non-civil servant teachers by level, 2006-2010 46 Figure 26: Teacher salary scale and distribution of real wages for teachers, 2010 47 Figure 27: Evolution of the percent of non-civil servant teachers by level, 2006-2010 47 Figure 28: Share of teachers with a university degree, by type of contract and level of education, 2006-2010 48 Figure 29: Percentage of non-PNS Teachers, 2010 49 Figure 30: Share of non-PNS teachers by province, 2010 49 U Spending more or spending better: Improving education financing in Indonesia Figure 31: Share of budget spent on contract teachers vs school size and STR, 2010 50 Figure 32: Estimated cost of converting all contract teachers to PNS, and of certifying all teachers 51 Figure 33: Estimated cost of achieving government goals 53 Figure 34: Gross and Net Enrollment Rates by Level, 2001-2009 56 Figure 35: International comparison of Gross Enrollment Rates, 2009 56 Figure 36: Share of children enrolled in appropriate level, out of age enrollment and out of school by age, 2010 57 Figure 37: Attainment profile of 27 year olds, 2010 57 Figure 38: Gross Enrollment Rates by Level of Education, 2010 58 Figure 39: Share of children enrolled in school by age and quintile, 2006 and 2010 59 Figure 40: Share of children enrolled by age in urban and rural areas in 2010 59 Figure 41: NER of females relative to males by level, 2006-2010 59 Figure 42: Possible supply constraints for expanding secondary education: distance and students per 9 60 Figure 43: Share of out-of-school children by urban-rural and quintile, 2010 61 Figure 44: Total household expenditure on education and as a share of household per capita consumption, by level and income quintile 61 Figure 45: District ECD enrollment rates, 2009 62 Figure 46: District SMP enrollment rates, 2009 62 Figure 47: District disparities in junior secondary enrollment rates, 2001-2009 63 Figure 48: Enrollment rates for 18-20 year olds by district poverty quintile 63 Figure 49: Indonesian 4th-Grader performance in 2006 PIRLS 64 Figure 50: Indonesian 8th-Grader math performance in 2007TIMSS 65 Figure 51: 2009 PISA results in reading and mathematics 65 Figure 52: PISA math score vs. GDP per capita 66 Figure 53: Distribution of math test scores Indonesia, Thailand and OECD, 2009 66 Figure 54: Indonesia PISA scores in mathematics, reading and science, 2000-2009 67 Figure 55: Indonesia PISA scores in science and math by socioeconomic deciles, 2000-2009 67 Figure 56: Summary of progress, challenges and highlighted Gol objectives in education, 2010 68 Figure 57: Change in public education spending and change in education outcomes at the district level 69 Figure 58: Student-teacher ratio, per student spending and Bahasa Indonesia test scores in public primary schools 71 Figure 59: Average test scores for teachers in primary and junior secondary 72 Figure 60: Student learning gains by teacher competency in subject knowledge 74 Figure 61: Education costs vs BSM amount, 2009 77 Figure 62: Targeting of BSM 77 U Figure 63: Education cost profile vs BSM benefits 78 Fig ure 64: Estimated cost of expanding coverage of BSM to all poor, for 100 percent of education costs, 2012-2015 79 Fig ure 65: Estimated cost of expanding coverage of BSM to all poor, for 100 percent of education costs, 2012-2015 79 Figure 66: Share of teachers with senior secondary or less as their highest education by province, 2010 81 Figure 67: Per student spending and student-teacher ratios at the school level, 2010 81 Figure 68: Estimates of government spending on basic education teachers for different student- teacher ratios, 2009 82 Figure 69: The estimated teacher requirements to fulfill the standards outlined in the joint decree, 2010 85 Figure 70: Percentage of primary school teachers that would need to be transferred to comply with joint decree 87 Figure 71: Population density, proportion of small schools and number of teachers in each province, 2010 89 Figure 72: Percentage of districts and provinces that report having BOSDA in primary and junior secondary schools, 2009 93 Figure 73: BOS per-student grant and median regional per-student grants in primary and junior secondary schools, 2010/11 93 Figure 74: Per student amount by sources of income at the school level, 2010 94 Figure 75: Income per student by source and school size, 2010 94 Figure 76: Per student expenditure by type at the school level, 2010 95 Figure 77: Per student school income and share of expenditure by category for BOSDA and non-BOSDA schools, 2010 95 Figure 78: Average test scores for Grade 6 students in Mathematics and Indonesian, 2010 96 Figure 79: Change due to BOSDA funding formula allocations to small and remote schools 98 Figure 80: Incentives by pilot district to improve school performance 98 Figure 81: Consequences of the different transfer mechanisms on education 100 Figure 82: Options for reform and main result for the education sector 100 Figure 83: Objectives and critical policy areas for Indonesia's education system 103 Figure 84: Strategic reallocation of expenditures: who and what? 104 Figure 85: Effective use of transfer mechanisms: who and what? 105 Figure 86: Improving Teacher Management 107 Figure 87: School and District Management 108 Figure 88: Budget planning, transparency and accountability 109 B Spending more or spending better: Improving education financing in Indonesia List of Tables Table 1: Decentralized education management by level of education 24 Table 2: Size and distribution of budget windfalls from the 20 percent rule 32 Table 3. Budget realization, 2010-2011 32 Table 4: Target Gross Enrollment Rate by level of education 2010-2014 52 Table 5: Local average treatment effect of teacher certification - selected variables 73 Table 6: Causal effects of certification on student learning outcomes 73 Table 7: Teacher characteristics by location 80 Table 8: Recent teacher staffing regulations 83 Table 9: Primary and Junior Secondary School Size, 2010 88 List of Boxes Box 1: An overview of current transfer mechanisms in the Indonesian education system from the central to sub-national governments 28 Box 2: The joint decree for teacher management 84 Box 3: The BOSDA improvement pilot program 97 U * Acknowledgements The team is grateful to officials and staff in the Ministry of Education and Culture and Bappenas for their overall support of this study. Special thanks go to lbu Nina Sardjunani (Deputy for Human Resource and Culture, Bappenas), Pak Taufik Hanafi (Expert Staff for Social Economy, Ministry of Education and Culture) and Pak Fasli Jalal (former Vice-Minister of Ministry of National Education when the report was started) for their overall support and extremely helpful conversations and comments throughout the process. This report also benefited greatly from inputs and comments by the participants at consultation workshops as part of the Thematic Education Dialogue supported by the BEC-TF and including officials from the Ministry of Education and Culture, National Planning Agency (Bappenas), Ministry of Religious Affairs, Ministry of Finance, Ministry of Home Affairs, as well as other development partners. The core team responsible for the content of the report included Pedro Cerdan-Infantes (Task Team Leader, Education Economist, EASHE); Yulia Makarova (co-Task Team Leader, Public Finance Economist, EASPR), Samer Al-Samarrai (Senior Economist, EASHE), Dandan Chen (Senior Economist, EASHE), Yus Medina, Research Analyst (EASPR), Imam Setiawan (Research Analyst, EASHE) and Steisianasari Mileiva (Research Analyst, EASHE); district budget data and significant portions of the district analysis were contributed by Cut Dian Agustina (Economist, EASPR) and Dhanie Nugroho (Research Analyst, EASPR). Hassan Noura (former co-TTL Public Finance Specialist, EASPR) was fundamental in developing the concept and the early stages of the study and contributed with extremely useful discussions about the challenges of education financing in Indonesia. Vivi Alatas (EASPR) and Jon Jellema (EASID) kindly contributed sections of their social assistance PER to be included in this report. Finally, the team is grateful for very useful comments from Bill Wallace (Senior Advisor), Enrique Blanco Armas (Senior Economist, EASPR), Anna Gueorguieva (Economist, EASPR), Erita Nurhalim (Senior Operations Officer, EASHE), Blane Lewis (Senior Consultant, EASPR), Gregorius Pattinasarany (Senior Economist, EASPR) and Ashley Taylor (Economist, EASPR). Dyah Kelasworo Negraheni and Irene E. Bouwmeester provided the most efficient team support. The report is a joint effort between HD and PREM units at the World Bank and as such it was prepared under the overall guidance of Luis Benveniste (Sector Manager, EASHE) and Mae Chu Chang (Lead Education Specialist and HD Sector Coordinator for Indonesia) from HD, and Shubham Chaudhuri (Indonesia Lead Economist, EASPR) from PREM. EduardoVelez Bustillo (former Sector Manager, EASHE) oversaw the concept development and initial stages of the report. Peer reviewers were Hanna Birdsey (AusAid, Indonesia), Juan Diego Alonso (Education Economist, LCSHE), Soren Nellemann (Country Sector Coordinator, ECSH2), Blane Lewis (Senior Consultant, EASPR) and Ashley Taylor (Economist, EASPR). The production of this report was made possible through the generous support of the Basic Education Capacity Trust Fund. U Spending more or spending better: Improving education financing in Indonesia a List of Abbreviations APBD Anggaran Pendapatan dan Belanja Daerah Provincial or District Government Budget APBN Anggaran Pendapatan dan Belanja Negara National Budget ASEAN Perhimpunan Bangsa-Bangsa Asia Tenggara Association of Southeast Asian Nations AusAid Badan Kerjasama Pembangunan Australian Agency for International Internasional Australia Development BAPPENAS Badan Perencanaan Pembangunan Nasional National Development Planning Agency BEC -TF Dana Program Pengembangan Kapasitas Basic Education CapacityTrust Fund (World Pendidikan Dasar (Bank Dunia) Bank) BERMUTU Pendidikan Lebih Baik Melalui Reformasi Better Education through Reformed Managemen dan Peningkatan Menyeluruh Management and Universal Teacher Guru Upgrading BKD Badan Kepegawaian Daerah Regional Civil Service Board BKN Badan Kepegawaian Negara National Civil Service Board BLU Badan Layanan Umum General Services Agency BOMM Bantuan Operasional Manajemen Mutu Operational Assistance for Quality Management BOS Bantuan Operasional Sekolah School Operation Fund (Provided by the Central Government) BOSDA Bantuan Operasional Sekolah Daerah Local BOS/Local School Grants BOS-KITA Peningkatan Pengetahuan untukTransparansi BOS Knowledge Improvement for dan Akuntabilitas BOS Transparency and Accountability BPS Biro Pusat Statistik Central Bureau of Statistics BSM Bantuan Siswa Miskin Scholarship for Poor Students BSNP Badan Standar Nasional Pendidikan National Education Standards Agency D1,2,3,4 Diploma Tingkat (1 -tahun), (2-tahun), (3- Post Secondary Diploma Level (I-year), (2- year), (4-tahun) year), (3-year), (4-year) DAK Dana Alokasi Khusus Special Allocation Fund DAU Dana Alokasi Umum General Allocation Fund (Central Transfer to Regional Budgets in the form of an Unrestricted Block Grant) Dekon Dekonsentrasi De-concentration (budget and decision- making transfer from the central to the provincial government) ECD Pendidikan Anak Usia Dini Early Childhood Education ESWG Kelompok Kerja Sektor Pendidikan Education SectorWorking Group EU Uni Eropa European Union GDP Produk Domestik Bruto Gross Domestic Product GER Angka Partisipasi Kasar Gross Enrollment Rate UainlBde Gol Pemerintah Republik Indonesia Government of Indonesia IDR Rupiah Indonesian Rupiah JSS Sekolah Menengah Pertama (SMP) Junior Secondary School MA Madrasah Aliyah Islamic Senior Secondary School MBS Manajemen Berbasis Sekolah School Based Management MDG Tujuan Pembangunan Milenium Millennium Development Goals MenPAN Kementerian Pendayagunaan Aparatur Ministry of State Personnel and Bureaucracy Negara dan Reformasi Birokrasi Reform MGMP Musyawarah Guru Mata Pelajaran Secondary School SubjectTeacher's Working Group MI Madrasah lbtidaiyah Islamic Primary School MoF Kementerian Keuangan Ministry of Finance MoHA Kementerian Dalam Negeri Ministry of Home Affairs MoEC Kementerian Pendidikan dan Kebudayaan Ministry of Education and Culture MoRA Kementerian Agama Ministry of Religious Affairs MoU Nota Kesepahaman Memorandum of Understanding MTs Madrasah Tsanawiyah Islamic Junior Secondary School NER Angka Partisipasi Murni Net Enrollment Rate NUPTK Nomor Unik Pendidik dan Tenaga Unique Identification Number forTeachers and Kependidikan Teaching Personnel OECD Organisasi untuk Kerjasama Ekonomi dan Organization for Economic Cooperation and Pembangunan Development PAD Pendapatan Asli Daerah Regional Own Source Revenues PAUD Pendidikan Anak Usia Dini Early Childhood Education PER Kajian Belanja Publik Public Expenditure Review PIRLS Studi Internasional tentang Literasi Membaca Progress in International Reading Literacy Study PISA Program Penilaian Siswa Internasional Program for International Student Assessment PNS Pegawai Negeri Sipil Civil Servant PP Peraturan Pemerintah Government Regulation PT Perguruan Tinggi Higher Education Institution PTAI PerguruanTinggi Agama Islam Islamic Higher Education Institution PTK PerguruanTinggi Kedinasan Higher Education Institution held by other Technical Ministries besides MoEC dan MoRA Puspendik Pusat Penilaian Pendidikan, Kementrian Center for Education Assesment, Ministry of Pendidikan dan Kebudayaan Education and Culture RA Raudhatul Atfal Islamic Kindergarten/Early Childhood Center RENSTRA Rencana Strategis S-year Strategic Plan RKP Rencana Keuangan Pendidikan Education Financial Plan Indoesin Rpia Spending more or spending better: Improving education financing in Indonesia RPJMN Rencana Pembangunan Jangka Menengah Medium-term National Government Nasional Development Plan Si, S2, 53 Sarjana 1, Sarjana 2, Sarjana 3 Degree equivalent to Bachelor's Degree, Master's Degree and PhD respectively SBM Manajemen Berbasis Sekolah (MBS) School Based Management SD Sekolah Dasar Primary School SIKD Sistem Informasi Keuangan Daerah Regional Financial Information System Sisdiknas Sistem Pendidikan Nasional National Education System SK Surat Keputusan Decree SKPD Satuan Kerja Perangkat Daerah Local GovernmentTechnical Department (or Regional Government's Working Unit) SMA Sekolah Menengah Atas Senior Secondary School SMK Sekolah Menengah Kejuruan Vocational Secondary School SMP Sekolah Menengah Pertama Junior Secondary School SNP Standar Nasional Pendidikan National Education Standards SPM Standar Pelayanan Minimal Minimum Service Standards (MSS) STR Rasio Siswa Guru Student-Teacher Ratio SUSENAS Survei Sosial Ekonomi Nasional National Socio Economic Survey TIMSS Studi Internasional tentang Matematika dan Trends in International Mathematics and lImu Pengetahuan Alam Science Study (TIMSS) TP Tugas Pembantuan Co-administered Task TK Taman Kanak-Kanak Kindergarten (PreSchool) UKP4 Unit Kerja Presiden bidang Pengawasan dan Presidential Working Unit of Supervision and Pengendalian Pembangunan Control of Development UKMP3 Unit Kegiatan Menteri Pengawasan dan Ministerial Working Unit of Supervision and Pengendalian Pembangunan Control of Development UNESCO Organisasi Pendidikan, lImu Pengetahuan, United Nations Educational, Scientific, and dan Kebudayaan PBB Cultural Organization UT Universitas Terbuka Open University UU Undang Undang National Law UUD Undang-Undang Dasar Basic Law WB Bank Dunia World Bank Dinas A Provincial, District, Sub-District Office with sectorol responsibilities Kabupaten District (Regency) Kota Autonomous City Komite Sekolah School Committee Madrasah Islamic School Re in lFn n ilIfr ainSse I Executive Summary Indonesia has made a clear commitment to education - passing a constitutional mandate to allocate at least 20 percent of the total government budget to education (the "20 percent rule"). This has led to a large increase in resources, more than doubling education spending in real terms since the passage of the constitutional amendment in 2002. The rule's mandate was fully met for the first time in 2009, when the government allocated more than 20 percent of the state budget to education. The goal of this report is to understand how these additional resources were spent and the extent to which they have translated into educational outcomes.The report provides recommendations to improve the quality of spending by improving the education financing system. The"20 percent rule" has resulted in a rapid increase in the education budget, but has also complicated budget management. One of the biggest drawbacks of the 20 percent rule is that earmarking funds reduces the incentives to optimize their use.The link between policy planning and resources is broken; large exogenous increases to the budget (regardless of results and independent of planning) can create inefficiencies in spending. Earmarking funds also creates rigidities in the budget and limits the capacity of the government to optimize the allocation of resources across sectors. Lastly, because the rule applies to both the planned and the revised budgets, it makes the education budget unpredictable and creates large budget "windfalls": with the budget highly dependent on the price of oil due to fuel subsidies, every adjustment to oil prices results in large fluctuations in total government spending. These windfalls have proven difficult to manage, despite Government of Indonesia (GOI)'s creation of the National Education Development Fund. The biggest payoff for this increase in spending has been in terms of access and equity, but access to senior secondary and tertiary education still remains extremely low for the poor. There has been rapid progress in access and equity, with children from poor families enrolling earlier and staying in school longer. The share of 15 year olds from the poorest consumption quintile who are enrolled in school increased from 60 to 80 percent between 2006 and 2010. However, beyond the age of 15, enrollment from this quintile drops dramatically, and by higher education, falls to less than 2 percent. Learning outcomes are still poor and show some worrisome trends in math and science. Meanwhile, the share of top performers is extremely low. Indonesia generally scores on the bottom on international tests (TIMMS, PIRLS and PISA), including compared with other countries in the region. On the PISA, the test with the most recent result and the longest time trend, the scores are mixed: while reading scores have shown steady improvement since 2000, math and science scores have not. In math, a promising increase in scores between 2000 and 2006 was set back in 2009, when scores declined for all socioeconomic deciles. In science, the trend has been flat since 2000. Equally alarming is the small number of top performers on both the TIMSS and the PIRLS, while on PISA, no student in the sample performed at level 6 (the highest) in math or science in 2009. Given recent evidence linking PISA scores and the share of top performers with GDP growth, it is evident that increasing quality of education must be a priority. U Spending more or spending better: Improving education financing in Indonesia Part of these trends is explained by patterns in spending: most of the additional resources went to basic education, for teacher salaries and teacher certification. Two-thirds of the additional resources went for teacher salaries and certification, yet as we shall see, studies have shown that adding teachers does not correlate with improved educational results. Higher education also received a significant increase in resources, as did scholarship programs after 2009. Capital spending also increased in the last decade, especially at the district level (presumably on schools). The Gol's focus on expanding access to 9 years of basic education was reflected in the budget, but there are signs the government is putting more emphasis on expanding access and quality to senior secondary and beyond. The increase in spending on teacher salaries was driven both by increases in the total number of teachers and by "regularizing" contract teachers to civil servant status. The patterns were different across levels. Conversions occurred mostly in early childhood education (ECD), and at the junior and senior secondary levels, while the increase in teachers in primary education was due mainly to the addition of contract teachers. Whereas 60 percent of the additional teachers were contract teachers, the trends in regularization and the Gol's implicit policy to have all teachers converted to civil servant status and certified are worrisome. What keeps the salary bill low in Indonesia despite the extremely low student-teacher ratio (STR) is the high prevalence of contract teachers. Estimates of the cost of conversion and certification suggest it would lead to an unsustainable doubling of the salary bill. It is unlikely that the observed patterns of spending will translate into increased quality of education. At the current student-teacher ratios, there is no correlation between increasing the number of teachers and learning outcomes. At 17-1 in primary and 13-1 in secondary schools, STRs in Indonesia are low by international standards; given these ratios, adding more teachers is unlikely to lead to improved outcomes. Moreover, the teacher certification program, which absorbed a large chunkof additional funds, is not (yet) showing the expected impact. An early evaluation of the teacher certification program shows that while the program improved the livelihoods of teachers, it did not lead to improvements in learning outcomes for students. Indonesia needs to alter its education spending patterns to ensure that resources produce improved outcomes. Judging by international comparisons, ECD and senior secondary education are underfunded. Scholarships received only 1 percent of the total budget in 2009. While the planned budget for scholarships increased significantly between 2009 and 2011, there is room for expansion and improvement of the program. However beyond these broad indications, improving the quality of spending is not straightforward. Within each level of education, it implies different adjustments. Given the importance of basic education in the budget and as the basis for future learning, this report goes into detail on improving the quality of spending in this area. Some on-going studies have looked at senior secondary education in detail. Future work should examine other levels. What does "spending better" mean in basic education? Spending better means ensuring resources are allocated more effectively to produce better outcomes with regard to access and learning. Given the complexity of the governance of the system in Indonesia, this is a difficult task, and one that should focus on the districts'capacity to manage the system. Districts own schools and manage teachers, the two main inputs in the learning process. But since districts rely on central government transfers for the majority of their budgets, spending is influenced by the incentives embedded in the transfer formulas. A look at these incentive structures goes a long way to explaining why districts spend mostly on one of these inputs (teachers) while their support for schools is still very limited. Improving spending in basic education will mean achieving more balanced spending at the district level and ensuring that incentives are in line with the objective of improving the quality of education. U Executive Summary This report highlights three broad areas for improvement in education financing, focusing on increasing efficiency, equity and performance: In terms of efficiency in the use of resources, while the overall availability of resources seems largely adequate (relative to the state budget), the breakdown by level is still biased towards basic education when compared to other countries in the region. While ECD, senior secondary and higher education are likely to receive a larger share of resources in the future, liberating the resources for this increased spending will necessarily mean improving efficiency in current spending, especially when it comes to basic education, and especially on teachers. If Indonesia wants to be able to afford increased spending on non-basic education, and meet its goal for all teachers to be certified civil servants, it is imperative that it improve the efficiency of spending. In terms of equity, while the focus on basic education is more progressive than spending on higher levels of education (when children from poor backgrounds frequently are no longer enrolled), an uneven distribution of teachers results in vast differences in spending per student across schools and across districts. In addition, uneven distribution in terms of teacher qualifications (measured by education level) compounds the problem. Rural and remote areas have fewer and less qualified teachers. In terms of incentives for performance, the financing system has very few. With the focus of financing still on inputs (teachers, school infrastructure), districts face strong incentives to manage the inputs mechanically, adding inputs proportionately to the budget increases, and not necessarily to optimize their performance. That said, at the school level, administrators face pressure from communities to improve the quality of education. Ultimately, it is at the school level that most of the potential to optimize spending lies, but the central government can provide the right incentives to districts and schools to achieve these goals. OBJECTIVES Expasion Learning Outcomes Strategic Effciencyude realocaion eacer Shooand laning U.otase aaeet itit tasaec U,,,,ue mcaissMaaeen n accunabiit Spending more or spending better: Improving education financing in Indonesia The key to improving quality of spending in basic education is to improve district performance in managing budgets - however, the central government's role is fundamental. As noted, districts manage two of the most important inputs in basic education, schools and teachers, but they receive most of their funds from central government transfers and are subject to central government regulations. As a consequence, the first tool the government has at its disposal to improve the quality of spending is a strategic reallocation of spending towards underfunded programs and levels of education, including scholarship programs and early childhood as well as post-basic education. However, given that the funds are managed locally, improving the quality of spending has to occur at the district and, ultimately, the school level, and districts must be incentivized to make good decisions. We recommend the following areas for central government action to improve district performance: first, setting up the right incentives through a more effective use of transfers mechanisms; providing tools for districts to improve teacher management, and increasing the role of districts in managing and funding schools. Improving transparency and accountability in budget management can support all these reforms. Strategic reallocation of resources Indonesia can reallocate resources strategically towards programs that are underfunded, such as expanding access to ECD or senior secondary. However, increasing resources is not sufficient. It is important to keep in mind that additional resources do not always result in improvements in outcomes, so drastically expanding access requires careful planning to ensure sustainability and efficiency. While this report does not go into these sub-sector issues because they are beyond its scope, estimates of the cost of expanding access to senior secondary education show that the model chosen for expansion can have dramatic effects on the cost. Small changes in per student spending have big effects on the budget. Different financing mechanisms have different effects on equity and access. ECD and senior secondary education are clear candidates for more resources. The evidence supporting ECD internationally is overwhelming. Given the current low level of spending on ECD, increasing resources in this area seems desirable. Senior secondary education, which still relies largely on student fees, should also receive increased funding; the Gol's plans to expand compulsory education to this level demonstrates its commitment to expanded access. Whether the expansion is carried out by providing free senior secondary or through targeted scholarships will have serious budget implications. Expanding and improving the Scholarship for Poor Students program (Bantuan Siswa Miskin) is critical for supporting expansion and equity: this will require increasing funding and coverage, improving the targeting of the program, aligning the disbursement of the benefit with the timing of expenditures and possibly providing a transition bonus. The ideal scenario would cover 100 percent of the cost for all poor students from SD to SMU, providing a transition bonus for those moving from SD to SMP and from SMP to SMU. This would ensure that cost is not a reason for dropping out, and provide incentives for students who are deciding whether to continue their education. Improving district spending A key element in improving the quality of spending on education is improving teacher management. Districts face incentives to overstaff from two sources: i) the sub-national General Allocation Fund transfer formula (Dana Alokasi Umum, DAU) and ii) the guidelines for teacher distribution. The Basic Allocation component of the DAU formula (the main transfer from central to sub-national governments) is based partly on the number of civil servants in a district, which provides an incentive to hire teachers rather than redistribute them when needed. In addition, the guidelines for teacher entitlement formulas, which establish the minimum number of teachers a school should have depending on its characteristics, tend to support high teacher numbers, especially in small schools, as they are based on teaching groups rather than on student- Executive Summary teacher ratios. The combination of these two factors results in schools requesting additional teachers based on entitlement formulas, and districts preferring to hire before redistributing teachers from overstaffed to understaffed schools. If there is one message to take away from this report, it is that tackling these two issues is crucial to meeting Indonesia's goals. a) Effective use of transfer mechanisms Breaking the link between transfers and hiring.The consequences of the DAU have been studied extensively, and there seems to be a consensus in this body of research that the revisions of Laws no. 32 and 33 governing decentralization ought to eliminate incentives for overstaffing by removing the Basic Allocation element from the DAU formula. This would go a long way towards making future teacher hiring more efficient by reducing incentives to overstaff, but this change would not rationalize teacher management instantly. Teachers already hired need to be managed more efficiently. Most importantly, while it would eliminate one of the clearest incentives for districts to overstaff, changing the transfer formula alone may not guarantee the elimination of the problem of overstaffing completely. Faced with the obligation to spend on education and the lack of alternative ways to allocate resources, districts may still choose to hire additional teachers, even if the DAU formula is changed. Increase performance-based transfers. In order to further incentivize districts to focus on performance, the performance component of central government transfers, such as the incentive grant to local and provinicial governments earmarked for education (Dana Insentif Daerah, DID), should be expanded. These performance- based incentives can be linked to either specific policies (for example, to the provision of local school grants (Bantuan Operasional Sekolah Daerah, BOSDA) by the district orto enrollments and/or other outcome indicators. Both of these changes are largely out of the control of Ministry of Education and Culture (MoEC) - lying instead under the responsibility of the Ministry of Finance and Ministry of Home Affairs. These changes by themselves will not solve the problems with the quality of spending on education, but they have important implications for education. Regardless of whether they occur, the problems with teacher management still need to be addressed. b) Improving Teacher Management Teacher management is critical to improving the efficiency and equity of spending in education. The current low STR would be fiscally unaffordable if it were not for the high prevalence of contract teachers. Continuing the trend of converting contract teachers to civil servants in secondary education will drive up the salary bill up ifthe allocation ofteachers is not improved. Finding a sustainable way to deal with contract teachers will be crucial in the future. On equity, despite the low STR, there are massive differences in the availability and qualifications of teachers across schools and regions, with a clear disadvantage for rural and remote areas, but large variability even within districts. The estimated efficiency gains from reallocating teachers are substantial. Small improvements in the efficiency of teacher allocation can have large budget implications: increasing STRs by five students can result in a one-third reduction of in per student spending. Increasing STRs can be the key to liberating resources to achieve the desired results for the future. However, existing efforts to increase STRs have had limited success. Currently, the guidelines are based on learning groups (rombels) and not on student-teacher ratios. Thus if schools break down students into smaller learning groups, it can lead to an overestimation of the number of teachers needed. The magnitude of the reallocation needed to improve efficiency and equity in the allocation of teachers is massive. Using the latest government guidelines for teacher allocation, which came out of a joint decree for teacher management and establish the minimum number of students in a school, about 340,000 teachers or 17 percent of the teaching force would need to be reallocated to ensure that all schools have at least the minimum U Spending more or spending better: Improving education financing in Indonesia number of teachers. While most of this reallocation can occur within districts, which is logistically simpler, there is still a need to reallocate teachers across districts and even across provinces. The sheer size of the reallocation, and the lack of clear mechanisms for these transfers, make realizing these efficiency and equity gains difficult to achieve in the short period of the decree's validity. A large constraint in Indonesia's education system is the high number of small schools. The high prevalence of small schools increases the need for teachers. Revisiting school planning should be considered as a medium-term strategy, but more immediate measures (some already underway) could be expanded. Multi- grade teaching can be an effective way to deal with small schools, but there need to be provisions in place to support districts in its implementation. In senior secondary, cluster teaching (whereby teachers are allowed to teach more than one subject) can go a long a long way to improving efficiency. Lastly, incentives for teaching in remote areas should be elaborated and expanded. In order to adapt more quickly, contract teachers could be used more strategically, but their working conditions and career prospects need to be improved. Ideally, teachers would be fungible and flexible, available for assignment where they are most needed. However in the absence of this mobility across subjects, levels of education and schools, contract teachers can be used to deal with shortages. In many countries contract teachers have proven to be effective at supporting expansion, and they are significantly easier to move across schools. However, without effective regulations on contract teachers, defined career prospects and improved working conditions, there is a risk that high turnover and the use of unqualified teachers may harm the quality of education. The teacher certification program needs to be reassessed, since it is not delivering the expected results. Studies have shown that while certified teachers are less likely to hold second jobs, students of certified teachers do not perform significantly better on tests. To a large extent, the creation of a simplified procedure for certification (based on submitting a "portfolio" of teaching experience, training courses or workshops) seems to have weakened the effects of the program, but the introduction of recertification requirements may improve the outlook. A promising step in 2012 is the inclusion of a competency test to determine eligibility for the certification process. c) Improving School and District Management While districts are responsible for both teachers and schools, 80 percent of their spending goes to salaries. School budgets, therefore, are left with only a small portion, if any, of district funds. Almost half of schools report receiving no funds from district or provincial governments. Still many districts and provinces do support schools directly by providing (small) discretionary funds through the BOSDA program. More than half of local governments receive BOSDA, and a majority of them provide top-up funds using the same per student formula as the central government's School Assistance Program (BOS). A few are experimenting with alternative formulas (per school, per teacher, equity-based formulas and even some performance-based formulas). Additional discretionary funds from BOSDA are associated with better learning outcomes. Using detailed budget data, the report shows that schools that get BOSDA do not shift their big categories of spending in major ways.The additional funds, however, have a positive effect on learning outcomes both in math and Bahasa Indonesian. Expanding the BOSDA program offers a promising format for increased district support to schools. By experimenting with different formulas, BOSDA has the potential to play a stronger role in improving district management by allocating resources in a more equitable way and by introducing a more active district involvement in school management. U Executive Summary Improving Budget Planning,Transparency and Accountability. In addition to the specific measures above, ensuring accountability and transparency in the process of planning and allocating resources is crucial to improving the quality of spending. The 20 percent rule makes it more challenging to hold spending agencies accountable because budgetary decisions are divorced from educational results (ie. the funding for education is guaranteed regardless of results). The 20 percent funding should not be taken for granted, however, but rather viewed as an opportunity to improve educational outcomes and access. Every effort should be made to ensure that funds are spent in a transparent manner, based on long-term development objectives and linked to specific program needs. Improving budget transparency requires improved education budget reporting at the sub-national levels and improved reporting on budget priorities for mid-year windfalls at the national level. This report mentions several problematic areas in the current education data reporting, particularly when it comes to sub-national data, which lags two years behind, covers only 70 percent of districts and does not provide basic data breakdowns. Encouraging more up-to-date and detailed sub-national data reporting and providing reporting guidelines to districts and provinces to ensure consistency in methodology and calculations is an important element in improving education budget transparency. Also, given the recent large windfalls resulting from the 20 percent rule, it is important to renew and expand the education line ministries' 2010 practice of providing detailed budget plans describing windfall allocations to specific programs and activities. Ensuring the link between program planning and budgeting, and linking these to performance is likely to result in better quality spending. MoEC has led the way by measuring outcomes and evaluating performance and using this information to inform its budgeting. Since 2010, MoEC has been collaborating with the Vice President's Office to report on improvements in the government's priority programs and ensure consistency between program planning and implementation. Expanding and improving the current Monitoring and Evaluation (M&E) effort and moving towards performance-informed budgeting should allow MoEC to reap further rewards in terms of improved quality of spending and improved outcomes. Moreover, given the constraints and challenges associated with the 20 percent rule, it will be important to develop better contingency planning strategies in education budget management. As previously mentioned, the 20 percent rule makes the education budget unpredictable, especially as it also applies to the revised budget. High volatility in the education budget should be mitigated by better use of the National Education Development Fund; this will require the development of clear guidelines and a mid-term expenditure framework on windfall allocations to the Fund, as well as instructions on how and under what scenarios the Fund's resources can be used. Going forward... Implementing all of these reforms will be a major challenge: the political economy of these reforms is complex. Reforms that involve one stakeholder can be relatively easy to implement given the will and the proper resources. Some of these reforms, such as expanding BSM or funding for higher education, are relatively straightforward. However reforms that involve multiple stakeholders, each with their own interests, can prove impossible if the interests are not aligned. Reforming the transfer system is outside of the control of MoEC, and has consequences for many sectors, not only education. Balancing the various interests and incentives will be complex, requiring better coordination among stakeholders. The sequencing of reforms is important. Some issues are more pressing than others. One of the most important is teacher management, which must be dealt with urgently in order to liberate resources and stop the trend of excessive hiring. Improving budget planning and transparency are also key. These two reforms would set the stage for a more sustainable expansion in the future. Other reforms require careful reflection. For example, U Spending more or spending better: Improving education financing in Indonesia while increasing spending in ECD seems desirable, current funding sources need to be evaluated and a more sustainable financing scheme developed. Reforming transfer mechanisms is notoriously difficult-but even in the absence of such reforms, MoEC can still improve the system significantly. Expanding and improving strategic programs such as BSM, and incentivizing and supporting districts, would be major steps in the right direction. Providing districts with clear guidelines for teacher management and extending multi-grade teaching in small schools would improve the efficiency and equity of the system, even if the current DAU formula remains. Such reforms would liberate resources to fund key programs, including to improve teacher quality. Modifying the teacher certification program to ensure that it results in improved outcomes is also a priority. Exploiting the BOSDA model to expand district support and accountability can also have a major effect on the quality of education, as can the expansion of national programs for accountability and transparency. It is time to translate the strong commitment to education into higher quality education and better outcomes. Indonesia is ready for that qualitative leap. It will require the collaboration and coordination of many actors, but the potential gains are plentiful. As Indonesia consolidates its status as a middle-income country, the skills of its labor force will be crucial. Improving the quality of education by improving the quality of spending would be a major step toward ensuring the labor force is prepared. U Executive Summary * Introduction: Why another public expenditure review? Few countries in the world have increased public expenditure on education by over 60 percent in real terms over a five-year period, as Indonesia did between 2005 and 2009. During that time, a constitutionally mandated allocation of 20 percent of government expenditure to education (hereafter"the 20 percent rule") was defined and implemented. The rule led to a massive increase in resources for education, making education the largest government expenditure after energy subsidies.' This rapid and significant increase in resources, and the need to evaluate where resources went and whether they translated into better educational outcomes, are the motivation for this report. The large increase in expenditures addressed one of the main concerns of the last Education Public Expenditure Review (PER)2 - the lackof resources for education - but the problems with the distribution of resources persist. That PER, carried out in 2007 using 2006 data, highlighted that Indonesia's spending on education was lower than would be expected for the country's level of development. As this report shows, the massive increase in spending has brought Indonesia closer to other middle income countries in terms of spending as a share of the total government budget. The second challenge identified in the last PER has not yet been addressed however: the distribution of resources. The structural inconsistencies between central and district spending and responsibilities are still present. Districts control the largest share of the budget, but they spend mainly on salaries. Problems with teacher management are more serious now, as the number of teachers has continued to increase since 2006.The high cost of certification anticipated in the previous PER and in subsequent World Bank reports on teacher management' has materialized and promises to become more severe if the problems with teacher oversupply and distribution are not addressed. As a consequence, improving the quality of spending in the education sector is now critical. Education is crucial for Indonesia's plans to consolidate as a middle-income country and to accelerate its transformation into a high-income country. Human resource development is a pillar of the recently approved Masterplan for Economic Transformation4, and as the country integrates with ASEAN and other trading partners, improving the quality of human resources must be central to its development strategy. It is also key to avoiding the middle- income trap, whereby countries fail to transition from middle-income to higher-income status because they remain specialized in lower value-added activities. However, after the rapid increase in resources in recent years, total spending on education is unlikely to increase at a faster pace than the overall government budget. In this context, it is all the more important to examine how resources are currently spent and to ensure they are used efficiently and effectively to meet the remaining challenges in the sector. This report focuses on how changes to the financing and governance systems can improve equity in access to education and the quality of education, and provides recommendations to improve government spending. Analyzing the consequences of the 20 percent rule for education budget management, this report describes where the additional resources went and highlights the main challenges in the sector - continuing to expand access, especially for the poor, and improving quality. With those two problems being too complex to tackle comprehensively in a Public Expenditure Review, the report focuses on the effects of the current financing and governance systems on access and quality. When it comes to teachers and teacher management, 1 Indonesia Economic Quarterly, World Bank (2012a) 2 World Bank (2007) Investing in Indonesia's Education, Allocation, Equity and Efficiency of Public Expenditures 3 See Transforming Indonesia's Teaching Force, World Bank (201 Ob) 4 Masterplan for the Acceleration and Expansion of Indonesia Economic Development 2011-2025, retrieved from http://www.ekon. go.id/media/filemanager/2011/05/27/p/d/pdf_mp3e.pdf U Spending more or spending better: Improving education financing in Indonesia for example, the report examines how transfer mechanisms and differences in district capacity lead to a poor distribution of teachers, which eventually has a negative effect on learning. This ignores, for example, problems with pre-service teacher training, or ineffective teacher career management.This compromise is necessary given the nature of this report. Other work from the World Bank does cover these issues.' While the analysis of the trends in spending span over all levels of education and includes the various ministries involved in education, the specific analysis on quality of spending is focused mainly on basic education. This, for two reasons: basic education absorbs most of the budget and, being managed by district governments and with school autonomy, basic education has the most complex financing and governance structure in the sector. In addition, a high-quality basic education is critical both for economic growth and employability. Ensuring good quality basic education is a necessary condition for the development of the rest of the education system. The report is divided into two parts. The first three chapters show that the vast increase in education resources has not been accompanied by a similar increase in learning outcomes, and thus highlight the urgent need to improve the quality of spending. Chapter 1 starts by describing the complex financing and governance systems. With this policy framework in mind, Chapter 2 focuses on the 20 percent rule, analyzing its consequences for budget planning and management and examining where resources went after this vast increase in spending. The chapter also provides some estimates of the cost of meeting future government objectives, in the context of discussions on expanding compulsory education, and broadening teacher certification. Chapter 3 looks at education outcomes, reviewing the significant improvements in access and equity, as well as the worrisome trends in learning outcomes. It points to improving the quality of education and expanding access to secondary and above (especially for the poor) as the main challenges in the sector. Given the cost of achieving these goals, and the fact that current expenditure patterns are unlikely to lead to improvements in learning outcomes, the chapter concludes that improving the quality of spending in education is now critical. The second part of the report focuses on how to improve the quality of spending in order to continue expansion and improve learning outcomes. In chapter 4, the quality of spending issue is divided into two areas: i) reassigning or improving programs at the central level, and ii) improving management at the district and school levels. At the central level, this chapter analyzes the Scholarships for the Poor program (BSM) and recommends that it be expanded and improved. At the sub-national level, the report explores how a combination of more efficient teacher management and stronger support for schools can improve efficiency and learning outcomes, and what the central government's role should be in ensuring that these changes occur. Finally, chapter 5 summarizes these recommendations, providing a framework for improving the quality of spending and, ultimately, ensuring that this spending leads to an improvement in the quality of the Indonesian education system. 5 A comprehensive report on teacher policies is currently under preparation. U I Chapter 1: How is the Indonesian Education System Financed and Managed? The education system in Indonesia is a very large, highly decentralized system, with over 500 district governments playing a strong role in its management. This role includes the managing the system's most important assets: 59 million students, 330,000 schools and close to 3 million teachers.6 While many ministries register spending on education in their budgets,7 the Ministry of Education and Culture (MoEC) and the Ministry of Religious Affairs (MoRA) are responsible for setting policies and managing the system. Under both ministries, public and private provision co-exist and receive public support in the form of civil service teachers (at all levels) and direct school grants (in basic education). While the nine years of basic education (primary and junior secondary) are compulsory and heavily subsidized, household contributions are high in senior secondary education and very high in higher education. This is partly related to the greater reliance on private provision and limited public support for these levels. While public MoEC schools cover a majority of enrollment in basic education, private provision is more prevalent in senior secondary and higher education (Figure 1). The system's institutional complexity is matched by complexity in financing mechanisms. Despite efforts to simplify school budgets by providing per student block grants to all schools in Indonesia (BOS), schools still receive funds from seven different budget sources: some come directly from the central government, some from local governments (mainly districts).This complicates school planning. District and provincial governments also receive funds from different transfer mechanisms, each with specific associated incentives. Meanwhile, the higher education system is centralized and support is limited to public universities, except in the case of civil service lecturers placed in private universities. To further complicate financing and management, a constitutional amendment passed in 2002 establishes that at least 20 percent of the total state budget has to be spent on education. Both central and local government budgets are subject to the rule, which includes budget revisions. This rigidity creates significant distortions in decision-making. 6 The number of students and schools mentioned includes all levels of education (from early childhood/ECD to higher education), pub- lic and private. The number of teachers is from ECD up to senior secondary, including inclusive education civil servant and non-civil servant, permanent and contract teachers (MoEC, 2010 and NUPTK, 2011). 7 The Ministries of Finance, Agriculture, Industry, Energy and Mineral Resources, Transportation, Health, Forestry, Marine Affairs and Fish- eries, Tourism and Creative Economy, Youth and Sport, Defense, and Manpower and Transmigration, the National Land Agency, the Meteorological-Climatological and Geophysical Agency, the National Nuclear Energy Agency, the Ministry ofYouth and Sport, Ministry of Defense, Ministry of Manpower and Transmigration, National Library, and Ministry of Cooperatives and Small & Medium Enterprises. U Spending more or spending better: Improving education financing in Indonesia Figure 1: Share of enrollment by provider type and level, 2010 Tertiary 62 Public under MoNE Senior Secondary 3 N c U Public under MoRA Senior Secondary 1 1 Other Public Primary Private 0% 50% 100% Source: SUSENAS 2010. This chapter provides an overview of the financing and governance of the hig hly decentralized education system. It provides a picture of the system and its different tracks, explains the roles that central and regional governments play, and provides an overview of the transfer mechanisms that support the system. A description of the National Education System in Indonesia The national education system in Indonesia' accommodates three tracks: formal, non-formal and informal education (Figure 2). Formal education is structured consecutively, starting with pre-school or early childhood education, running through basic education (primary and junior secondary education) and senior secondary education, and concluding with higher education.9 Primary and junior secondary are compulsory (nine years). Senior secondary education has two tracks: general and vocational. Both last three years but the curricula vary. The higher education system also provides both general and technical options, offering diplomas from polytechnics and professional schools, and undergraduate degrees (Si) from four-year universities, as well as Masters (S2) and PhD (S3) degrees. A national examination is held to assess student competency in the last year of primary, junior secondary, and senior secondary, and to determine whether a student is eligible to continue. Non-formal education enables lifelong learning by allowing people to pursue an educational path at any point of one's life, substituting for and/or complementing formal education; such learning takes place in training institutes or community learning centers. Informal education may take the form of home schooling or other independent learning activities. 8 Law 20/2003 serves as the legal basis for Indonesia's National Education System. This Law is the latest revision of Law 2/1989 on National Education System decentralization. This revision under ines the shift in the education system's management in the context of reformation and decentralization, eliminating discrimination between government-run (pub ic) and community-run (private) in- stitutions, and between religious and general education. The system introduced in this law takes into account the implementation of regional autonomy according to Law 22/1999 on Regional Government and Law 25/1999 on Fisca Balance between Centra and Local Government. 9 More discussion on the goal of each level of formal education; see for example World Bank's "Preparing Indonesian Youth forTransi- tion,"World Bank 2012. B How is the Indonesian Education System Financed and Managed? Figure 2: National Education System in Indonesia Seonary EducatoiD (3 years: 16-18 years old) SM /AM /M sM /A or 7 yeInformalj Formal Education Non Formal Education Education Source: Elaboration based on Law 20/2003 and The Southeast Asian Ministers of Education Organization/SEAMEO (2006). Two ministries administer the national education system in Indonesia-the Ministry of Education and Culture (MoEC) and the Ministry of Religious Affairs (MoRA). The education levels of MoRA-administered schools mirror those of MoEC schools (from pre-school to higher education), enabling movement from one to the other. Both deliver the same compulsory curriculum, with a distinct emphasis on religion in MoRA schools. But while MoRA's governance is centralized, with provincial and district offices reporting directly to MoRA instead of to the head of the local government MoEC's governance is decentralized, with the bulk of decision-making and service delivery devolved to local governments. The private sector plays a substantial role in the delivery of education in Indonesia, but quality varies enormously. Private schools are run by non-governmental institutions, such as foundations, religious or other grassroots organizations. These schools have largely filled the gap left by public schools in poor and rural areas, while some cater to students from wealthier families.10 Up to 91 percent of MoRA's Islamic schools are run by foundations linked to mass Islamic organizations." When it comes to MoEC schools, private providers represent a larger share of the total as the education level increases. The government supports private schools through regulations and bycdeploying civil service teachers (a relatively small number compared to the number of school- hired teachers).The BOS program disburses blockgrantsto both public and private schools at the basic education level. Thus, in addition to complying with government regulations, private primary and secondary schools are now required to report BOS fund allocations on a quarterly basis to their respective district governments. 10 World Bank (2010Ob), Transforming Indonesia's Teaching Force. 11 MoRA (2008) in Ali et al.(201 1), Quality of Education in Madrasah. - C, Spending more or spending better: Improving education financing in Indonesia Who manages what in Indonesia's decentralized education system? Since the local government autonomy law in 2001, district governments are responsible for managing the two main assets at the primary and secondary education levels: schools and teachers. Legally, primary and secondary schools are owned by district governments. In fact, when it comes to budgets, the school's legal status is similar to that of a district government department. Similarly, civil service teachers are legally district government employees, although the hiring process, like that of other civil servants, depends on a number of central government ministries, including the Ministry of Finance (MoF) and the Ministry of State Personnel and Bureaucracy Reform (MenPAN). Even contract teachers are largely district employees, although some are hired directly by schools. MoRA's management structure is different, since it maintains a centralized system for its public schools and civil service teachers, and relies more heavily on private funds. Provincial governments have very limited authority when it comes to schools, mostly coordinating districts at the basic and secondary levels of education, including with regard to staff development and the provision for education facilities. Table 1: Decentralized education management by level of education poliies,tor Curriculum Provision/ standards and deveop ent service MIS atschoolleve delivery NOON U CENTRAL Early Childhood MEN DISTRICT EUOM I CENTRAL Basic MEN DISTRICT NOON e I CENTRAL Secondary P MEN DISTRICT agag agag MgNg Sggg CENTRAL Tertiary Source: elaboration based on Law 20/2003, King et al. (2004) and PP 38/2007. Notes: Financing responsibilities in the table reflect the main responsibilities under education program assigned to specific level of government. However, partial financing in some programs is shared between different levels of governments: besides higher education, central government provides financing to other education programs, including through school rehabilitation financing, scholarship funding, and until 2011 it also administered School Operation Funds (BOS). Provincial governments also provide partial financial assistance in senior secondary education and to special needs schools. The central government formulates policy, issues regulations/guidelines and standards at the national level, and still directly controls higher education. MoEC, together with the National Education Standards Agency (BSNP), develops national education standards in eight areas: content/curriculum, processes, graduate competencies, education staff, facilities and infrastructure, management, financing and educational assessment." MoEC also issued Minimum Service Standards (MSS) for all education levels and the recent Joint Decree (2011) 12 According to PP 19/2005. How is the Indonesian Education System Financed and Managed? on redistribution of civil service teachers." When it comes to higher education, the Directorate General of Higher Education within MoEC has a key role, with authority to issue and revoke permits for institutions of higher learning and tight control over the public higher education system. Salaries for civil servant lecturers in both public and private universities are disbursed by the central government. However, few higher education institutions have some autonomy, as long as they are in compliance with the norms and quality assurance standards developed by MoEC. Both central and district governments are responsible for developing and managing the teaching force. The current teacher certification program, for instance, is led by MoEC in coordination with district governments. The ongoing scheme to redistribute civil service teachers provides another example: regulations and technical guidelines are introduced by MoEC, while the analysis of teacher needs and redistribution are led by districts (for redistribution within districts) and by provincial governments (for redistribution across districts). The central government also sets quotas for professional certification. Other central government agencies remain in charge of setting pay rates for civil servants and transferring district government budgets. MenPAN, MoF and the National Civil Service Board (BKN) play key roles in the hiring of civil service teachers and determining civil service quotas, while the selection, deployment and management of civil service teachers are handled by district governments. Schools have considerable autonomy over operational, budgetary and programmatic decisions. Since 2003, School Based Management (SBM) has applied to all stages of formal education. A degree of decision- making power and management have thereby devolved to the school level, taking account of local norms and encouraging community involvement. Evidence of SBM is seen in the joint role principals, teachers and school committees4 play in the allocation of BOS funds, and in the development of school budgets and school work plans. Who pays for what in Indonesia's education system? The funding system for the education sector is complex, involving multiple sources and transfers across various levels of government. Expenditures for education come from central government funds, transfers to sub-national governments, sub-national governments'own-source revenues, and central government spending at the sub-national level that is not recorded in sub-national budgets. Currently, schools receive funds from eight different sources and four different budgets, including the national, provincial, district and school budgets (Figure 3). Central government transfers are the main source of revenue for district government budgets (APBD). Central government transfers to sub-national governments have more than doubled in real terms since decentralization, accounting for 88 percent of district budgets and 44 percent of provincial budgets in 2009. While the majority of transfers are not earmarked - making it impossible to determine exactly what they are spent on - transfers are estimated to finance about 90 percent of sub-national spending on education, and 60 percent of the total national education budget." Sub-national governments receive many types of transfers for education spending, including the General Allocation Fund (Dana Alokasi Umum, DAU), the Specific Allocation Fund (Dana Alokasi Khusus, DAK), Special Adjustment Funds (Dana Penyesuaian), Co-administered Tasks (Tugas Pembantuan), De-concentration (Dekonsentrasi/Dekon) Funds, as well as BOS for elementary and junior secondary schools."6 13 MoEC put together technical guidelines as a reference for district and provincial governments during the mapping of teacher needs and eventually the redistribution process. 14 School committee representatives include parents, members of the community and education experts and/or practitioners, and should be chosen via a democratic and transparent election process (as regulated in the PP 17/2010 on Management and Implemen- tation of Education), 15 2009 Revised Planned Budget (APBN-P) includes planned total education budget and estimates of regional transfers allocated for education, 2009 realized regional budget data (SIKD), MoF. 16 BOS program was initially designed as a direct transfer from central budgets to schools, but was rerouted through districts in 2011 and provinces in 2012. B Spending more or spending better: Improving education financing in Indonesia - E .0O 0 r -- - - - --E E z z c Ou0 E CD ---------r----- CMQ 7DD7 So 0 w2)C uC G)J E G) G OO 2c o 'o, cI Ii ZCE .5" 8 ~Is u.. C CD 26 DC How is the Indonesian Education System Financed and Managed? The main transfer to sub-national governments is the DAU block grant, which provides funding for the salaries of district civil servants, including civil service (PNS) teachers. DAU transfers represented about 60 percent of district and 20 percent of provincial budgets in 2009.17 The DAU is allocated through a two-part formula consisting of the"Basic Allocation"and the"Fiscal Gap"(See Box 1 for details on each transfer type).The Basic Allocation, which is calculated largely based on the salary bill for civil servants in the district or province, implicitly incentivizes civil service hiring. Covering about 72 percent of the salary bill, it accounts for about 45 percent of the total DAU.11 An earmarked transfer, the Specific Allocation Fund (DAK) is also allocated yearly through no specific formula,and coversa significantportion ofschool andclassroom reconstruction and school improvement. Although relatively small - about 8 percent of district and 1 percent of provincial revenues in 2009 - the DAK is an important financing mechanism for school improvement projects. De-concentrated funds (Dekon) play a similar role, but may also cover other functions, such as social assistance and capacity-building programs. De- concentrated funds and co-administered tasks contribute about 9 percent of total sub-national revenues but are not typical transfers - although disbursed at the sub-national level and administered by provincial and district education offices (Dinas), they are not recorded in sub-national budgets (APBD). At the school level, the BOS and BOMM (Operational Assistance for Quality Management or Bantuan Operasional Manajemen Mutu) programs constitute the majority of funds at the school level and are fundamental for operational assistance and quality management. Managing such a large and complex system is clearly a challenge. The decentralized system means the central government has limited influence over district decisions. In this context, regulations may prove difficult to enforce. Incentives, on the other hand, can have great influence over district decisions, as evidenced in this report by the issue of teacher hiring. Thus with the right incentives, transfer mechanisms can be powerful tools to guide district spending. However, this may not be sufficient. Additional support from the central government may be needed for low-capacity districts or schools. As we will see in this report, neither the right incentives nor the additional support are currently in place, so given the vast increase in resources following implementation of the "20 percent rule," defining roles and providing the right incentives for each actor is crucial to ensure that resources are spent effectively. This will be a key subject of this report. 17 Total sub-national revenues include Dekon/TP transfers spent at the sub-national level but recorded in the central government bud- get. 18 Until 2008, Basic Allocation covered 100 percent of civil servant salaries (at both the provincial and district levels). It was reduced to 89 and 87 percent coverage for provinces and districts respectively, and further reduced in 2009 to 80 and 72 percent. U Spending more or spending better: Improving education financing in Indonesia Box 1: An overview of current transfer mechanisms in the Indonesian education system from the central to sub-national governments This box provides a brief description of the objectives and means by which the various transfer mechanisms from the central government to sub-national governments within Indonesia are determined. These transfers represent the major source of financing for sub-national governments and thus, to a large extent, explain the level and composition of their spending. General Allocation Fund (Dana Alokasi Umum, DAU) The DAU, according to Law No. 33/2004 Article 1 (21), is a discretionary block grant sourced from the Central Budget (APBN) and aims to equalize the fiscal capacities of sub-national governments. It is transferred monthly and directly from central to sub-national governments. The DAU is allocated based on a national formula and is the sum of a basic allocation (a portion of the sub-national budget for public servant salaries) and the"fiscal gap" (the difference between the estimated fiscal needs and fiscal capacity) of the sub-national government. The basic allocation accounted for about 45.5 percent of the DAU in 2010. Fiscal needs are based on regional variables such as population, area, GDP per capita, and the human development index. Fiscal capacity is measured by a region's own-source revenue and a fraction of total revenue-sharing. Based on Government Regulation No.55/2005, provinces only receive 10 percent of the total DAU, while districts receive 90 percent. Specific Allocation Fund (Dana Alokasi Khusus, DAK) DAK is an earmarked grant allocated to finance specific investment expenditures that are aligned with national priorities and carried out under the jurisdiction of sub-national governments. The DAK cannot be used for research, training, administration, or official travel. In 2011, 19 economic sectors received DAK allocations including education, health, agriculture, forestry, trade and various infrastructure sectors (road, irrigation, water, sanitation, rural electricity, housing and local government and remote areas infrastructure). Education is a key priority for DAK spending, with about 40 percent of DAK transfers allocated for education and used primarily for school rehabilitation and quality improvement. The DAK allocation has a formula component that takes into account the fiscal gap and has a 10 percent matching requirement. DAK is transferred in three tranches: the first is allocated after the budget is submitted to the central government; the next two depend on the depletion of the previous tranche. Although DAK is earmarked to fund capital spending, the government allowed some routine maintenance expenditures. Revenue Sharing Fund (Dana Bagi Hasil, DBH) Unlike DAU, which is a horizontal equalization grant, DBH is a vertical equalization grant which consists of revenue sharing from natural resources and taxes. Local governments are obliged to use 0.5 percent of their receipts from the natural resources part of DBH on basic education." DBH represented approximately 20 percent of total sub-national government revenues in 2009. Special Autonomy and Adjustment Funds Special Autonomy Funds include specific grants for Papua, Papua Barat and Aceh (Dana Otsus) and Special Adjustment Funds (Dana Penyesuaian) which include additional allowances for teachers, such as professional benefits for certified teachers and for uncertified civil service teachers, a School Operational Assistance program (Bantuan Operasional Sekolah, or BOS), and local incentive grants (Dana Insentif Daerah, or DID) for education. Central government spending at the sub-national level not recorded in sub-national budgets (APBD) De-concentration (Dekon) and Co-Administered Tasks (Tugas Pembantuan, TP) Dekon and TP funds originate from the central government's budget (APBN), and are administered by the provincial Dinas. The funds cover a variety of projects and activities, including school and classroom reconstruction and school quality improvements, social assistance programs (which included BOS until 2011) and capacity building programs for civil servants. Sources: MoF and various laws 19 MoF on DBH at http://www.djpk.depkeu.go.id/information/22/. U I Chapter 2: The "20 percent rule": Where did resources go? The evolution of public spending on education over the last decade has been impressive. From 2001 to 2009 the Gol's education budget increased by over 120 percent in real terms, with most of the increase taking place in 2009, when the budget increased by 20 percent in real terms over the previous year. What drove this rapid increase in resources was, to a large extent, the constitutional amendment passed in 2002 that mandated the government to spend 20 percent of its total budget on education. The 20 percent rule was first met in 2009, after the rule was clarified, bringing education spending to over 20 percent of the total state budget. The rule, however, has had some negative implications, complicating budget management and potentially creating distortions when it comes to spending decisions. This chapter looks at the consequences of the 20 percent rule and at where the additional resources went in 2009, examining the total education budget as well as changes in the central budget in 2010-2011. In addition to breaking down expenditure by level of education, level of government and economic classification, the chapter identifies the programs and policies that absorbed the majority of the increase in resources. Due to the fact that sub-national data are available only until 2009, we look at budget trends by program only until that year, revealing the impact of 20 percent rule on the total education budget the first year after it was met. However, we also examine changes in the composition of the central budget for which more recent data are available (until 2011) to understand whether any significant changes occurred in the central education budget after 2009, both as a result of the 20 percent rule and of periodic windfalls - as we will see in the next section, the majority of the windfalls have been allocated to the central budget through the MoEC or MoRA budgets. It is important to keep in mind the limitations of the data, especially when it comes to sub-national government budgets, when looking at the results in this chapter. A simple look at the budget does not allow some of the basic breakdowns shown in this chapter. The biggest problem is with detailed sub-national budget data, which is notoriously difficult to obtain. In fact, until the end of 2011, it was not possible to obtain sub-national budget data broken down by program (equivalent to level of education). Even now, these data are only available for 2008 and 2009, and only for 413 out of 500 districts. Although the central budget data is more up-to-date and detailed by program and activity of expenditure (2010 realized expenditures and 2011 partial breakdown realized expenditures are available), it has no consistent methodology for recording salary spending across years. In both central budget and regional data, it is impossible to distinguish between teacher and non-teacher (administrative) salaries. In regional data, civil service teacher and staff salaries are reported at the aggregate level as part of"indirect spending" (or non-program), while non-civil service teacher and staff salaries are reported as part of "direct spending" (or specific program expenditure). In addition, a change in the classification of central government teachers in 2008 makes comparisons across years difficult when looking at central budget data, since non-teacher salaries (reported as"others" before 2008) are broken down by education program sub-function starting only that year. U Spending more or spending better: Improving education financing in Indonesia We are therefore forced to use the share of teachers by level of education to estimate sub-national salary spending. This has two caveats. First, it assumes that teacher salaries are uniform by level of education, which may not be accurate if teacher characteristics between basic and secondary education differ significantly. In effect, since primary school teachers tend to possess fewer qualifications than secondary school teachers, we may be overestimating primary teacher salaries and hence the share of sub-national spending that goes to primary education. A second assumption we are forced to make is that the breakdown of administrative staff by level follows that of teachers. Since we cannot distinguish teachers from adiminstrative staff in the data, we use the share of teachers by level to break down all salaries. While we lack information to check the accuracy of this assumption, its effect on the overall picture is likely very small. Even if the supervision structure of different levels is significantly different, the number of teachers relative to other district office staff is so large that this assumption is unlikely to affect the overall picture. We first look at the importance of the 20 percent rule for the education budget and budget management. Implications of the 20 percent rule The "20 percent rule" was originally introduced in a 2002 amendment to the Constitution, establishing that a minimum of 20 percent of the total state budget must be spent on the education sector. Since education stood at 15 percent of government spending at the time, the rule was not implemented immediately, and was met only in 2009, after the Constitutional Court defined and clarified it. While there was considerable debate for many years over how to interpret it, since 2009 the 20 percent rule has been interpreted such that it: (1) applies to both the planned and revised budgets (APBN and APBNP, respectively); (2) includes salaries and estimates of subnational spending on education funded from central transfers; and (3) is calculated as a ratio of total state expenditures including subsidies, interest payments and transfers to the regions. The rule applies to both central and sub-national governments.20 At the central government level this requirement has been met, and while in the majority of districts, spending on education exceeds the mandated 20 percent, there is a wide range of district spending on education as a share of total budgets (from 11 percent to 48 percent). However well-intentioned the rule was, its effects have been mixed. On the one hand, the 20 percent rule has resulted in a massive increase in resources - adding 6 percenttothe state budget for education between 2008 and 2009. The rule clearly ensures political commitment to investing in education and prevents opportunism by policy makers, thus insulating priority education expenditures. On the other hand, it has complicated budget management and introduced perverse incentives. Figure 4: Public Expenditure on Education, 2001-2010 DR trillion Percent 350 25 21.9 21.1 280 16.4 21.7 20.9 20 mReal spending, 15.0 159 16.3 2009 prices (LHS) 210 14.15 15.6 15 Nominal spending (LHS) 14.5 -realized education expenditure 11.4 as% total national expenditure 140 10 -planned education expenditure as% total national expenditure 70 4 5 0 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Source: World Bank staff estimates based on MoF, APBD and BPS data. Notes: Realized expenditure data are unavailable after 2009 as sub-nationa realized data after that year are unavailable. Planned education expenditures reflect calculations of the budget including central government and regional transfers in the planned revised budget laws (APBN-P). Unlike planned data, realized data also includes sub-national own-source revenues allocated for education, which explains the slightly higher realized expenditure amount compared to planned expenditure in the graph. 20 Law No.20/2003 on National Education System U The"20 percent rule": Where did resources go? Earmarking 20 percent of the budget for education undermines the ability of the government to allocate resources efficiently across sectors. In general, earmarking is problematic as: (1) rigidities in the budget prevent government from moving resources to meet changing needs; (2) earmarked allocations undermine managerial incentives and planning capacity; and (3) there is a tendency for earmarks to proliferate, which increases overall budget rigidity. Indonesia's Health Law (September 2009) requiring the Central Government to allocate 5 percent of its budget to health (exclusive of salaries) is a sign that of the tendency towards more earmarking." Earmarking makes the education budget unpredictable, especially since it also applies to the revised budget. Volatility in the state budget (especially due to highly volatile energy prices, the GOI's largest budget item) automatically leads to volatility in the education budget, which complicates and discourages long-term planning. The fact that the rule also applies to the revised budget further complicates matters, introducing windfalls in the middle ofthefiscal yearthat need to be spent on short notice.This has been especially problematic in recent years, when oil prices climbed higher than had been anticipated in the budget. The windfalls can be large, sometimes even surpassing the size of some flagship government programs, such as BOS. And since these windfalls often arise late in the budget process, they risk being poorly spent, since short planning times result in hastily implemented programs. It is also unclear what the consequences would be of a downward revision of the budget due to oil prices. To mitigate problems in education budget planning due to "windfalls;' the Indonesian Government established the "National Education Development Fund" in 2010. It consists of an Endowment Fund to ensure the sustainability of educational programs for the next generations (intergenerational equity) and an Education Reserve Fund to secure funds for projects related to rehabilitation of educational facilities damaged by natural disasters." The Endowment Fund and the Reserve Fund are not part of the MoEC budget and are currently managed by the Government Investment Center (PIP), though a General Services Agency (BLU)- a working unit at the Ministry of Finance with relatively high budgetary autonomy. The Government's establishment of the National Education Development Fund is a positive step towards contingency planning, but the share of the windfalls the Fund has absorbed has been relatively small and varied across the years. Most of the budget windfalls still go through normal budget revision procedures and are mostly assigned to the central government budget through allocations to MoEC and MoRA. State budget revisions in the years 2010-2012 increased the education budget by about 7 percent over the initial planned budget each year (equivalent to USD 1.8 billion).The central government received the majority of these windfalls, with MoEC receiving 60 percent and MoRA 14 percent on average, while the regional governments received only about 5 percent, and the Education Fund's share varied from 6 percent in 2010 to 9 percent in 2011 and 29 percent in 2012 (Table 2).This is partly due to the lack of clear regulatory guidelines as to the percentage of"windfalls"to be transferred to the Fund. Because revisions to the state budget take place in the middle of the fiscal year windfalls have not been fully absorbed by the central government and assigned ministries. Budget revisions in Indonesia usually take place in July-August and need to be spent by December, which leaves little time for planning and executing the additional budget. Table 3 shows that MoEC and MoRA have been unable to fully absorb the windfalls in recent years. The unrealized amount as a share of the received windfalls in 2010 and 2011 represented about 42 and 46 percent respectively, contributing to low realization rates of the total revised central government education budget in those years. 21 Law No. 36/2009 on Health 22 The Minister of finance Regulation 238/PMK.05/2010 on Procedures for Procurement, Disbursement, Management, and Accountability of the Endowment Fund and the Education Reserve Fund, U Spending more or spending better: Improving education financing in Indonesia Table 2: Size and distribution of budget windfalls from the 20 percent rule APNAPBN APBN AP3- 2010 Realized 2011 Realized2 (budget) (rvsd buclget (revsedgudge (budget) (eie budget)ubudget Total expenditure 1047.7 1126.1 1229.6 1320.8 1435.4 1548.3 Oil & Gas subsidy 69 89 96 129.7 123.6 137.4 Total Education Exp. 209.5 225.2 249.0 266.9 290.0 310.8 Ministries 83.2 96.5 91 89.7 105.4 98.1 102.5 117.2 MoEC 54.7 63.0 59.2 55.6 67.3 61.2 64.4 77.2 MoRA 23.7 26.5 24.6 27.3 30.4 28.4 32.0 33.5 other ministries 4.8 7.0 7.1 6.9 7.6 8.5 6.2 6.6 Transfers 126.4 127.7 n.a 158.2 159.0 n.a 186.4 186.6 BOS 16.8 16.8 23.6 23.6 D3H (Part of shared 0.6 0.7 0.8 0.9 0.8 1.0 revenue to edu.) DAK 9.3 9.3 10.0 10.0 10.0 10.0 DAU 95.9 95.9 104.3 104.3 113.9 113.9 Teacher Professional 11.0 11.0 18.5 18.5 30.6 30.6 Allowance Additional Allowance for 5.8 5.8 3.7 3.7 2.9 2.9 PNSTeachers DID 1.4 1.4 1.4 1.4 1.4 1.4 Special Autonomy Rpein Futnfory 2.3 2.3 2.7 2.7 3.3 3.3 Region Fund for Edu. Other transfers 1.3 0.6 National Education Deveopntl und 1.0 1.0 2.6 1.0 7.0 Development Fund Source: 2010-2012 revised plan budget laws (APBN-P), MoF. Table 3. Budget realization, 2010-2011 % Unrealfized%of % f o % o Unrealized E. Amount l Amount Pln la Pan Pln la Pan as % of Pa Pln as % of Windfalls Windfalls MoEC 96.1 96.0 95.1 108.3 94.1 45.1 110.5 91.2 50.4 MoRA 92.9 91.2 103.9 92.7 67.6 104.0 93.4 64.4 Others 86.0 84.6 147.8 101.4 -4.4 122.9 110.8 -110.3 Total Ceta 97.0 95.4 93.5 94.6 109.3 94.2 41.7 109.5 93.3 45.5 Central Source: MoF Notes:The first significant windfall was received in 201 O.The windfall of 2009, when the 20 percent rule was met for the first time, contributed to just about DR 1 trillion., thus the difference between the initial and revised budget in 2009 was insignificant compared to later years. B The"20 percent rule": Where did resources go? The allocation of these windfalls was clearly detailed in 2010. In 2010, which was the first year the education sector received a significant windfall, MoEC and MoRA provided a detailed budget plan describing the windfall's allocation to specific programs and activities (Financial Note of the 2010 Draft Revised Budget). Both ministries indicated the specific amounts allocated to each level of education, to teacher quality improvement and to service management programs (Figure 5), and provided a description of activities supported by each program, including expansion of services in ECD, increasing scholarships for the poor in basic, secondary and higher education programs, improving the BOS management system, developing and rehabilitating school buildings, training of school principals and supervisors, accelerating the teacher qualifications improvement program, and providing functional and professional allowances for teachers. MoEC and MoRA made a great effort in 2010 to ensure transparency and accountability in education and windfall budget planning. However since 2011, information on the revised budgets and windfalls has been very limited. The Financial Notes of the 2011 and 2012 Draft Revised Budget documents no longer provide allocation information by program, but instead broadly indicate that additional resources from the budget revisions were directed to: (1) support the national priority program RPJMN 2010-2014; (2) strengthen educational achievement of development priorities in the RKP in 2011 and 2012, (3) support the implementation of Presidential Instruction No. I of 2010 Concerning Acceleration of the National Development Priorities, and (4) expand scholarship programs for poor students.23 Figure 5: Windfalls Allocation by program, 2010 MoEC Budget Plan MoRA Budget Plan Edu service R&D 2% EDED 3% management 6% SD/SMP 31% TeacherII quality 8% Non-Formal 1% Teacher SD/SMP 33%9 quality 8% HE 32% Non- formal 3% Source: MoF Perhaps the biggest drawback of the rule and its earmarking is that it introduces perverse incentives when it comes to managing the government budget, distorting the link between resource planning and planning for programs or policies. This is true at the overall government budget level and within the education budget. First, rigid earmarking rules may decrease what is known as allocotive efficiency, preventing government from allocating resources optimally across sectors. Since any increase in government spending in any sector raises the overall government budget, it also causes an increase in spending on education; thus the marginal cost of allocating resources to non-educational activities increases, while the marginal benefit decreases (i.e. part of the increased spending goes to education, not to the intended activity). This earmarking may thus bias government allocation of resources across sectors. Second, earmarked allocations reduce technical efficiency by undermining managerial incentives and planning capacity. When resources increase dramatically and in a manner exogenous to the demands of the sector, education planners, pressured to spend the increased 23 The technical criteria for allocating additional funds in 2011 and 2012 are also quite broad indicating that funds can cover 1) arrears on teachers' professiona allowance payments; 2) programs due to completion at the end of the fiscal year; 3) programs demonstrating outcome improvement; and 4) existing programs. U Spending more or spending better: Improving education financing in Indonesia resources, may not face the right incentives. For example, they may choose activities that easily absorb additional budget (i.e. adding staff), or short-term programs over long-term programs that may require more planning.This may have lasting consequences on the budget and on educational outcomes; if a large share of the additional budget is committed to wages, for instance, it can be difficult to make adjustments in the future. Going forward, the 20 percent rule will also limit the growth of the education budget. The rule will be interpreted as both a "floor" and a "ceiling" for education expenditure. As a consequence, expenditure on education is unlikely to grow higher than 20 percent of the government budget in the foreseeable future, regardless of need. Education is considered to be well-funded by both the government and parliament. Given pressing demands to allocate more resources to currently underfunded sectors, including infrastructure and social assistance programs, it is unlikely the education budget will continue to increase at the rate it has in recent years over the next decade. Nonetheless, it is important to recognize the contribution the rule has made to increasing the education budget. It is equally important to assess the quality and efficiency of spending to ensure that the increase in resources improves access to and the quality of education. The big jump in resources allocated to education between 2008 and 2009 offers an opportunity to evaluate the results of such a drastic increase in spending, and to examine the incentives built into the current system and how they affected that spending. We will therefore analyze what happened to the education budget during this period of rapid growth. But first we should ask: is 20 percent of the budget an appropriate amount to spend on education? The next section provides some international comparisons. Is Indonesia spending too much on education after the 20 percent rule? By all international measures, Indonesia does not spend too much on education. Indonesia is one of the top spenders on education if we consider the educational budget as a share of total government spending. Of its regional peers, onlyThailand allocates a higher share of the government budget to this sector. That said, budget allocation is a central element in education financing, and an allocation of 20 percent of the national budget is widely considered to be an indicative threshold for a strong commitment to education. Indonesia is now among the low- or middle-income countries that have met that threshold - only about one third of low-income countries have either achieved or surpassed the 20 percent threshold.24 As a share of GDP Indonesia still spends less than other middle-income countries, both in East Asia and in other regions. At 3.7 percent of GDP, Indonesia spends less than Thailand, Vietnam or Malaysia in the region, and spends half as much as high-income, high-performing countries such as Norway. Yet Indonesia fares well when compared to other countries in the region, spending a higher share of GDP on education than Lao, Cambodia, the Philippines or even Singapore. Relative to its wealth, therefore, Indonesia's per student spending is still low, especially when it comes to secondary education. As a share of per capita GDP, Indonesia spends less per student than most developed countries and its regional comparators. In both primary and secondary education, Indonesia is at the lower end of the distribution.That said, when it comes to primary education, it is positioned above the Philippines or Chile, and only slightly below Mexico or Malaysia, yet it lags behind not only high-income countries but also some other middle-income countries such as Vietnam orThailand. In secondary education, the differences are greater. Only the Philippines and Thailand fall below Indonesia on the graph. 24 UNESCO(2011) The"20 percent rule": Where did resources go? Figure 6: Public expenditure on education as a share of total government expenditure and as a share of GDP, selected countries in 2009 Thailand Norway Sweden Indonesia Argentina Viet Nam Netherlands Malaysia France Ma aysia Brazil Brazi Lower-Middle Income Viet Nam Mexico Norway Republic of Korea Philippines Colombia Argentina Germany Chile Sweden Japan Lao Thailand Indonesia Nether ands Lao France Singapore Germany Idia Philippines Singapore ________________Cambodia_______________ 0 5 10 15 20 25 0 2 4 6 8 Share of total Government Expenditure Percentage of GDP Source: Unesco Institute for Statistics, 2009 or latest year; for Indonesia, own calculations using Ministry of Pinance central budget and regional (SIKD) data. Sample countries for both indicators are different because of lack of data for certain variables. Figure 7: Per pupil spending on primary education as a share of per capita GDP, selected countries Primary Secondary 25 0100 15 2 50 25 Source: Unesco Institute for Statistics, 2008 or latest year (notearlierthan 2006); except Indonesia, own caculations usng Ministry of Finance central budget and SIKD data. Judging by international comparisons, after the 20 percent rule, Indonesia's spending on education is largely adequate - certainly not too high. Relative to its level of income, Indonesia spends slightly less than other middle-income countries, especially on secondary education. Relative to its overall governmert spending, however, Indonesia is at the higher end of the distribution. Few countries spend more than 20 percent of their budgets on education, so Indonesia's effort to ensure adequate resources for education is laudable. Making sure that resources are spent well is key. The next section looks at the evolution of the budget composition during this period of growth in resources, breaking it down by level of education, economic classification and level of government, and highlighting some potential issues with the allocation of resources, especially at the sub-national level. B Spending more or spending better: Improving education financing in Indonesia Where did the 20 percent go? Between 2008 and 2009, spending on education increased by 17 percent in real terms as a result of the 20 percent rule - the equivalent to USD 3 billion. Where did the funding go? Although basic education received most of the additional budget, we see that the composition of spending by level of education changed, with senior secondary and higher education receiving a moderately larger share. Basic education received about 2 percent less of the budget in 2009 than in 2008, but still received over half of overall additional resources. Senior secondary and higher education received about 1 and 3 percent more respectively, but were starting from a smaller base. The uncategorized component of the budget ("other" in the figure) continued to be a very large component of the budget, but declined 2 percentage points over this period. Figure 8: Total education spending by level of education and as a share of the total education budget, 2008 and 2009 m2008 n2009 ECD m Basic Education mother program 140 m Senior Sec. education a Universities 120 200 a1100 90.0 80.0 C! 80 70.0 EC0asc SnirSe.Uivriie8thr20820 60 50.0 _ 40 0.40.0 30.0 20 20.0 10.0 - 0.0 ECD Basic Senior sec. Universities other 2008 2009 Education Education program Source: World Bank staff estimates based on state budget data and Regional Financial information system data (Sistem Informasi Keuangan Daerah, SIKD), Ministry of Finance Compared to other countries, the share of the education budget spent on secondary education is low despite the increase in 2009. Given the Gol's goal of expanding enrollment at these levels, financing is insufficient. The more economically advanced countries in the region allocate a smaller share of their overall education budget to primary education, making secondary and higher education a priority. As Indonesia advances, it will have to increase the availability of its higher-skill labor force, which will require improving access to secondary and higher education and allocating more resources to these programs. On higher education, Indonesia's share of spending comes closer to that of its neighbors.The increase in spending on higher education in 2009 brought Indonesia's level of public spending on higher education as a percentage of the total education budget to equal that of South Korea. As a percentage of GDP, it is still very low. If we take into account public and private funding, Indonesia spends slightly above 1 percent of GDP on tertiary education. This is far below what most developed countries spend, and even countries in the region such as Thailand and Malaysia outspend Indonesia. Moreover, only about 25 percent of Indonesia's spending on tertiary education is financed by public sources, while in Malaysia the funding is 100 percent public, in Thailand it's about 70 percent public, and in OECD countries it is close to 80 percent public. B The"20 percent rule": Where did resources go? Figure 9: Share of education spending by level, selected East Asia countries, 2009 100 > 90 - > 80 - 70 - Other 60 r oTertiary 2 50 4 Secondary 40 , 30 m Primary 0 a,10 0 Indonesia Thailand Malaysia Korea Singapore Source: UNESCO Institute for Statistics; ndonesia data own elaboration using MoF, SIKD Figure 10: Spending on ECD, Central and Local Governments, 2008-2009 mMoEC mMoRA wSub-national 4,000 o 3,500 3,000 2,500 .0 2,000 1,500 - 1,000 500 2008 2009 Source: MoF, SIKD On early childhood development (ECD), the trend is mixed. Local governments reduced their spending on ECD between 2008 and 2009 by about 20 percent in real terms, while the central government increased spending by bout the same amount, resulting in a flat budget between 2008 and 2009. When compared to other levels of education, spending on ECD is too small to be appreciated (Figure 8). However, the growing commitment of Indonesia towards ECD is clear. The central government's budget for ECD has tripled since 2009, but it still represents less than 1 percent of the budget. Spending on senior secondary and higher education has increased between 2008 and 2009 by 26 and 53 percent respectively, but households seem to have absorbed a large portion of the cost of expansion, and total household spending on education in general has increased significantly over the decade. Senior secondary education and higher education are far more reliant on fees than basic education. This makes public spending on education is more progressive since most of the spending happens at the basic education level, which has almost universal coverage, the benefits flow more to the poor. This will likely have to change if equitable expansion of senior secondary education is a priority. As noted in a recent World Bank report," there are strong indications that senior secondary education is supply constrained at the moment. In addition, the high reliance on fees will be a barrier to expansion to the poor. The cost implications of this expansion will be explored at the end of this chapter. 25 "Preparing ndonesian Youth for Transition," 2012. U Spending more or spending better: Improving education financing in Indonesia Figure 11:Total public and household spending and share of public and private spending by level of education 350 - 100% - a Household a Public 90% 300- 80% 250 70% 200 60% 50% . Household 150 40% - Public 100 30%- 20%- 50 100% 0 0% 2001 2002 2003 2004 2006 2007 2009 Basic Senior Higher Secondary Education Source: Budget data from MoF, Household Spending from Susenas, core and education modules The composition of spending by level of government has not changed significantly between 2001 and 2009 -- districts have doubled their expenditures on education, but their share of the spending has remained relatively constant. An increase in the share of central government spending in 2006 and 2007 was due to the introduction of the school operational assistance program (BOS) and the professional allowance for certified teachers. Both programs were initially part of the central government budget, but have been decentralized (certification in 2010 and BOS in 2011), which is likely to have shifted more spending back to districts, however sub-national data are not available after 2009 to confirm this shift. In 2009, districts accounted for 50 percent of total national education expenditures, while provincial governments accounted for only 8 percent. Salaries are a major component of district spending; when excluding salaries, education spending is still largely centralized. In 2009, the central government spent about 67 percent of the non-salary budget, 10 percentage points more than in 2008.The central government controls the majority of the non-salary budget at all levels of education, from the 70 percent in ECD to 99 percent at the university level. Even in basic education, almost 90 percent of non-salary spending still occurs at the central level. What brings down the overall average to 67 percent is unclassified spending at the district and provincial levels. However, it is important to note the contribution of BOS and teacher certification to this central spending (BOS accounts for 44 percent of non-salary spending in basic education), programs which, as mentioned above, have been decentralized since 2009. Figure 12: Spending by level of government, 2001-2009 250 90 200 80 70 70 48 55 5 150 60 65 66 59 65 3 9 5 50 100 40 ~.30 50 20 1 40 4/ 4 34 41 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2001 2002 2003 2004 2005 2006 2007 2008 2009 N Central w District Province a Central oDistrict nProvince Source: World Bank staff estimates based on state budget data and Regional Financial information system data (Sistem Informasi Keuangan Daerah, SIKD), Ministry of Finance U The"20 percent rule": Where did resources go? Figure 13: Non-salary education expenditure by programs and level of government, 2009 m Central m Central-BOS n District m Province * Central m District Province 100 100 90 80 80 70 60 60 50 r 40 40 30 20 20 10- Early Basic Secondary Universities Otners 0 Childhood Education Education 2008 2009 2008 2009 Education Total Non-salary Source: MoF Between 2008 and 2009, the share of the central government's budget increased dramatically in ECD, basic and senior secondary education. Central government spending increased by 7 percentage points, from 34 to 41 percent of the total education budget. This central government spending push was partially due to the need to meet the 20 percent rule, but the fact that growth occurred in central government spending even at the basic education level requires some explanation, as basic education is normally the responsibility of districts while the central government is responsible for financing and managing higher education. Nonetheless, the districts'share of the total basic education budget decreased between those two years by 8 percentage points from 68 to 60 percent. The big increase in central government spending in basic education and senior secondary was driven mainly by central government"social assistance" programs. Salaries, goods and services and capital follow traditional economic classifications, while social assistance is reserved for programs such as scholarships, teacher programs (such as certification) or BOS transfers (among others). Virtually all social assistance spending occurs at the central level, but some of these programs benefit schools directly (BOS). The fact that social assistance represented a large share of the increase in spending is desirable in principle, but it also depends on which programs drove the spending. We will analyze the spending breakdown later in this section. Figure 14: Public spending by level of education, 2008-2009 100 - 80 - m Central 60 m District i Province 0.40 - - 20 C 0 2008 2009 2008 2009 2008 2009 2008 2009 2008 2009 ECD Basic Ed ucation Senior Sec. Universities Other Program S Education Source: MoF U Spending more or spending better: Improving education financing in Indonesia Figure 15: Public spending by level of education, 2008-2009 Early Childhood Education Basic Education m2008 m2009 m 2008 m2009 3,500 - 80 o o - 3,000 70 II II 2,500 60 2,000 50 40 .= 1,500 . 3 -- 30 1,000 -2 ~ _ 20 500 -10 - Salary Goods & Capital Social Aid Salary Goods & Capital Social Aid Services Services Senior Secondary Education Higher Education m2008 n2009 m2008 z2009 14 8 o o a 12 7 10 6 Sa ary Goods & Capital Social Aid Salary Goods & Capital Social Aid Other Services Services Source: MoF A large share of the increase in social assistance spending went to scholarship programs, including the Scholarships for the Poor (BSM) program, which almost doubled between 2008 and 201 0.Yet scholarships represent a very small percentage of the education budget. The budget for scholarship programs increased by 62 percent in real terms between 2008 and 2009, while the Scholarships for the Poor program increased by 95 percent. 2010 data show a continuous albeit smaller increase in spending on total scholarship programs. However the increase in scholarships for higher education in 2010 was quite dramatic and unprecedented (over 300 percent increase in real terms), and clearly indicates an effort by the government to improve access to higher education for poor students. Yet despite this increase in scholarships and some benefits to poor students resulting from the 20 percent rule, scholarships still represent only a fraction of the total central government education budget: 4 percent in 2010, compared to 3.6 percent in 2009 and 3.1 percent in 2008. In higher education, there were widespread increases in all categories, except salaries. From 2008 to 2009, higher education saw an increase in spending of 49 percent in real terms. Spending on both capital and goods and services more than doubled in real terms. Social assistance increased slightly less, but still grew by 80 percent. The salary bill was reduced, though a new category emerged in "Other," which includes research grants for universities. As a result, in 2009, about 74 percent of the total university program budget was spent on non-salary items, particularly on new capital investment projects, such as new research laboratories and a new polytechnic university, as well as on training and capacity building, R&D and scholarship programs to university lecturers and students.These trends reflect a major push to expand higher education and an attempt to improve the quality of education by improving the quality of inputs. 5 The"20 percent rule": Where did resources go? Figure 16: Scholarship programs by level of education, 2008-2010 4000 3500 3000- a 2500 a Higher Edu. (others) 2000 Higher Edu (for poor) 1 Senior Sec. (for poor) S1500 m Basic Edu. (for poor) 1000 500 0 2008 2009 2010 Source: MoF Salary spending increased in ECD, basic education and senior secondary education. The increases were significant: between 2008 and 2009, salary spending in basic and senior secondary education increased by 16 and 18 percent in real terms, respectively. In basic education alone, this meant an increase of DR 10 trillion (about USD 1 billion) to the salary bill, while in senior secondary education, salaries increased by IDR I trillion (about USD 100 million). Given the nature of salary spending, composed mainly of civil servant teacher salaries, this represents a permanent burden on the budget, not a one shot increase. This increase in salary spending is evident at the district level: between 2008 and 2009, district salary spending increased but capital spending remained constant. Over a longer period, however, district capital spending has been growing rapidly. Spending on basic education accounts for 67 percent of district spending26. Although between 2008 and 2009, the share of the district education budget allocated to capital spending remained constant, between 2001 and 2009 it went from 5 to 18 percent, reflecting efforts to rehabilitate school infrastructure. Although spending on salaries has also been increasing rapidly, the share of salaries in district education budgets actually decreased between 2005 and 2009, from 85 to 76 percent. Meanwhile, districts spent a greater proportion of their education budgets on capital spending than the central government. Figure 17: District spending by economic classification, 2001-2009 120 0 100 C 80 o m Personnel ' 60 m Goods and Services o a Capital 40 n Others 20 2001 2002 2003 2004 2005 2006 2007 2008 2009 Source: SIKD 26 If we use the share of teachers by evel of education to approximate non-program salary spending. As explained in the beginning of the chapter, teacher salaries at the district level are not classified by level of education. Spending more or spending better: Improving education financing in Indonesia Nonetheless, districts still allocate the vast majority of their budgets to recurrent expenditures, mostly personnel. In 2009 about 82 percent of district education spending was allocated for recurrent spending, but 84 percent of that recurrent spending went to salaries for civil service teachers and administrative staff in primary and junior secondary. Figure 18: Real capital and recurrent spending by level of government, 2006-2009 90 80 70 60 0 CL 50 m2006 40 - 2007 30 m2008 -- 20 i 2009 10 Capital Recurrent Capital Recurrent Capital Recurrent District Province Central Source: SIKD, Ministry of Finance Aggregating across levels of education, salaries absorbed almost 50 percent of additional resources between 2006 and 2009, followed by capital spending (28 percent) and social assistance (15 percent). Given its low base, capital spending grew at the fastest rate (159 percent), reflecting district efforts to rehabilitate school infrastructure, as well as the central government's effort to make capital improvements in the area of higher education. Social assistance, which absorbed the majority of the increase between 2008 and 2009 and which includes large programs such as BOS and teacher certification, grew by 25 percent over this period. When looking at major expenditure items, teacher salaries and teacher certification absorbed almost two-thirds of the additional resources coming from the 20 percent rule. Figure 20 breaks down the education budget by items or programs that absorb significant shares of the budget (teacher salaries and certification, BOS, higher education and other programs).Comparing the breakdown before and after the 20 percent rule was implemented, we can see which of these expenditures absorbed a majority of the additional resources. Overall, in 2009, the education sector in Indonesia gained an additional 5.5 percentage points of the total state budget, relative to the 2006-2008 average of 16 percent. Of this 5.5 percent increase, more than 3 percent was allocated to teachers'salaries and certification subsidies, leaving only a 1 percent increase for higher education programs and slightly less than a 1 percent increase in resources for all other programs. Who were the direct recipients of increased education funding in 2009, after the 20 percent rule was met for the first time? Figure 21 describes the allocation of education spending among teachers, students, schools and civil servants. Expenditure on teachers and staff covers salaries and all subsidies (including professional, functional subsidies and special subsidies for teachers in remote areas). The amount assigned to students reflects spending on student scholarship (BSM government assistance program for poor students). All expenditures associated with school operational costs (BOS program), capacity building, management and administration, as well as school infrastructure and rehabilitation expenditures, are attributed to direct spending on schools. Expenditures defined as "training programs" for non-teachers and staff, as well as "unspecified" or "other" expenditures, have been assigned to separate categories. As Figure 21 shows, unspecified expenditures, most of which are sub-national expenditures, make up a significant share of total education spending. U The"20 percent rule": Where did resources go? Figure 19: Real spending by economic classification (total education budget), 2006-2009 250 200 0 150 35 m Salary Sm Goods & Services 100 a Capital 10Social Assistance 50 m Others 2006 2007 2008 2009 Source: MoF Figure 20: Spending increase distribution after the "20 percent rule," 2009 25 m Other programs and W20 - unspecified expenditures a BOS e 15 a Universities S10 a Teacher certification # - 0 5 Teacher and staff salaries 0 d 2006-08 2009 Source: World Bank calculations using MoF data The total salary bill was about 60 percent of the total education budget. About half of the education budget in 2009 (99 trillion rupiah) was spent on teacher salaries. If teacher allowances, including functional, professional allowances and allowances for teachers in remote regions, are taken into account, the share of the budget spent on teachers adds up to 56 percent. If the share of BOS going to contract teachers is included (an average of 20 percent) the total spending on salaries and allowances would constitute about 60 percent of the total education budget. This share, however, differs significantly by level of education. In basic education, for example, the share is closer to 70 percent. Indonesia's spending on salaries as a share of the total education budget is in line with other countries in the world. Indonesia spends about the same share as the Republic of Korea on salaries, and a significantly smaller proportion of its budget on teachers than many high performing systems (Norway, Netherlands or the USA). It also spends much less than other worse performing systems like Argentina and slightly less than the Philippines. Spending more or spending better: Improving education financing in Indonesia Figure 21: Distribution of education funds by direct recipient, 2009 Teachers & staff ( m (salary & Allowances) 56 ment, 3 LOthers, 0.2 Students 2 Teachers & staff (training & management) 4 Source: MoF Figure 22: Share of total education budget spent on salaries, selected countries Colombia Argentina Chile 1 Malaysia France USA Philippines Netherlands Norway Brazil Indonesia* Republic of Korea Australia - 0 20 40 60 80 100 Percentage of total education going to salaries Source: UNESCO I nstitute for Statistics, 2010 or latest year - except Indonesia, own calculations based on MoF data. Note: Indonesias salary expenditure includes salary subsidies from certification, and contract teachers under BOS, in addition to civil servant salaries. What drove the increases in salary spending? Most of the additional resources went to teacher salaries largely because of the continued increase in teacher hiring at all levels.The number of teachers is growing at a faster pace than that of students at all levels of education, but especially in primary education. Since 2004, the number of teachers in primary education grew by 30 percent, while the number of students remained largely constant. In secondary education, both junior and senior, the disparity in the growth in the number of teachers and students is smaller. In junior secondary, the growth rates have been very similar after teacher growth stalled in 2006. Still, in senior secondary education the growth in the number of teachers was twice that of students. The"20 percent rule": Where did resources go? As a result, the student-teacher ratio (STR) continued to decline in primary and senior secondary. By 2010 Indonesia had one of the lowest STR in the world; while the global average STIR at the primary school evel is 31:1, in Indonesia it is 20:1. At the secondary school level", comparisons are even more striking, with the average Indonesian STR currently at 12:1 - the lowest ratio in the East Asia region. In part, the low STR in Indonesia can be explained by the large number of small schools due to the low population density of many of its islands. However, school size is not the only explanation, as low STIR is not only a small school phenomenon." Figure 23: Growth in number of students and number of teachers by level and student-teacher ratio, 2004-2010 a) Growth in number of students and teachers b) Student-Teacher Ratio 20 - Number of teachers 24 1 8 - Number of student 22 1 6 20 14 -18 1.2-4 1 1214 1 .0 - 1 2 12 0.8-- - - - - - 1 000 000 C C D0 000 000 000D1 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Junior Senior Vocational Kindergarten Secondary Secondary School -+-Primary ---Junior Secondary -)-Senior Secondary Source: Own calculations using MoEC, various years In addition to the overall increase in teacher numbers, the conversion of contract teachers to civil servants contributed to the increase in the salary bill between 2006 and 2010, especially at the non- primary levels. From 2006 to 2010, the teaching body increased by about 377,000 teachers, a majority of which (60 percent) were hired as contract teachers (non-PNS). However, conversion of contract teachers to civil service teachers contributed to the increase in salaries, especially in some levels. The conversion of contract teachers to civil service positions was significant in ECD and secondary education, but not at the primary level.The number of PNS teachers in junior and senior secondary, for instance, increased by 13 and 25 percent respectively, while the number of contract teachers in junior secondary increased only slightly, and decreased in senior secondary. Meanwhile at the primary level, the increase in teachers was due mainly to contract teachers. 27 The breakdown of secondary education by junior and senior secondary levels is not available for all countries in edstats, so we use primary and secondary for international comparisons. 28 Chapter 4 covers in detail the issue of teacher distribution and teacher management. Spending more or spending better: Improving education financing in Indonesia Figure 24: Student-teacher ratio by level, 2010 Primary Secondary United States Japan Malaysia ndonesia Highincome High n o-e Thailand ed States Indonesia Malays a China ,ied Kingoom Si ngapore Singapore Japan China United Kingdom Korea Rep. Vietnam Mongola Korea, Rep. Vietnam Lower middle Income haiand Low & middle income Word World Middle ncome Midin come Myanmar ow & middle income Mongolia Lower middle icome Lao PDR Cambodia Philippines Myan ar Cambodia Pnilppies 10 20 30 40 50 60 5 10 iS 20 25 30 35 40 SourceU EdSttts Figure 25: Evolution of the number of civil servant (PNS) and non-civil servant teachers by level, 2006- 2010 1,200,000 8%m2006 m2008 is2010 1,000,000 .A 800,000 U 600,000 75% 0 400,000 13% 0V 2% -0.7% 25% S200,0000 0060/o Malysi Ln In L u-)Mongo ona 0-t at af0at f at aft at 0 0 0 0 C C C C TK SD SMP SMA Source: NUPTK Data Why is salary spending reasonable with such a low student teacher ratio? The large share of low pay non-civil servant teachers in the system keeps the salary bill relatively low. Thus the conversion of contract teachers to civil service status in primary education does not seem fiscally feasible. Figure 26 shows the average salaries of the working population with at least an Si1, compared to the official civil service teacher salary scale, assuming a normal progression along the scale. Even the salaries of non-certified civil service (PNS) teachers compare favorably to the average wage of the working population with at least an S1, making the teaching profession attractive, especially in the early years. But certification doubles teacher salaries; certified teachers, therefore, are substantially higher paid, in general, than the average population with an S1. Thus a civil service teaching position with the eventual possibility of certification is an attractive career choice - as was intended by the teacher law, which sought and still seeks to attract the best and the brightest to the profession. Yet when we look at the distribution of real wages for workers classified as teachers in SAKERNAS, the distribution is not concentrated around the salary scale for civil servant teachers (much less certified teachers); instead, a high percentage of teachers are earning well below the entry level civil service teacher salary, and these are contract teachers. Wol The"20 percent rule": Where did resources go? Figure 26: Teacher salary scale and distribution of real wages for teachers, 2010 Salary scale compared to workers with Si Actual distribution of teacher salaries Real Wage 2010 (base price 2002) 9.OOE+02 8000000 8.O0E+02 7000000 7.00E+02 6000000 6.O0E+02 5000000 8 5.OOE+02 4.OOE+02 4000000 3.00E+02 2000000 Non Teacher 2.00E+02 - Teacher Without Certiication 1.OOE+02 1000000 - Teacher With Certification (Age 40-59) 0.0E+00 ----- . * . . . 0 1,000,000 2,000,000 3,000,000 4,000,000 5,000,000 6,000,000 20-29 30-39 40-49 50-59 Monthly Wage Source: SAKERNAS 2010, and MoEC sa ary scales Source: SAKERNAS 2010 Since 2006, the reliance on contract teachers has increased in primary school (SD), but contract teacher qualifications have improved dramatically. In 2006, about 25 percent of SD teachers were contract teachers. By 2010, that share had increased to 35 percent. In junior and senior secondary, the reliance on contract teachers was significantly higher in 2006, but had declined slightly by 2010. Meanwhile,the qualifications of these contract teachers have improved.The share of teachers with at least a four-year degree increased 15 percentage points in SD.This upgrading was faster for contract than for PNS teachers, despite PNS teachers being eligible for upgrade programs under MoEC. Figure 27: Evolution of the percent of non-civil servant teachers by level, 2006-2010 60% m2006 m2008 2010 50% 40% o 30% 0 S20% S10% 0% Primary Junior Secondary Senior Secondary Source: NUPTK Data Spending more or spending better: Improving education financing in Indonesia Figure 28: Share of teachers with a university degree, by type of contract and level of education, 2006- 2010 100% - m2006 m2008 m 2010 E 90% 80% - v 70% 60% - r 50% - 40% ~30% S20% 0% non-PNS PNS non-PNS PNS non-PNS PNS non-PNS PNS Junior Senior Kindergarten Primary Secondary Secondary Source: NUPTK data Existing evidence on the effects of contract teachers is mixed. Hiring contract teachers is generally a cost- effective way to expand access, especially in remote areas." When placed strategically to cover hard-to-staff schools, or communities with different languages and/or cultures, contract teachers have been effective in supporting expanded access and improving equity in countries as varied as Cambodia, Nicaragua and India. In controlled experiments in different countries (India, Kenya, Niger"), the use of contract teachers has also had positive effects on learning. However, the way in which contract teachers are hired, managed and the purpose of their hiring are key factors in their impact. Contract teachers have been effective in remedial learning (i.e. teaching basic skills)," but in the absence of monitoring and accountability, their hiring has led to corruption and rent-seeking (i.e. hiring teachers and deducting "fees"from their salaries). In addition, the sustainability of the model is questionable unless contract teachers are used strategically (for substitutions, for example) and offered training, career development opportunities, and reasonable salaries. In the absence of these conditions, contract teachers may be subject higher turnover and even higher absenteeism rates, leading to lower quality education. In the case of Indonesia, the conditions are not optimal to make contract teachers effective. First, the use of contract teachers is not concentrated in remote or small schools - in fact, most schools have some share of contract teachers. In about a third of schools, less than 20 percent of the teaching force consists of contract teachers, while at the other end of the distribution, about one-fourth of schools have 40 percent or more non-PNS teachers on staff. The majority of schools fall in the middle, with between 20 and 40 percent contract teachers (Figure 29). The percent of contract teachers varies widely, but is not strongly correlated with poverty or remoteness. So while one would expect hard-to-staff districts to rely more on contract teachers, at the two extremes we find exactly the opposite (Figure 30): DKI Jakarta has a higher share of contract teachers than any other province, while Papua Barat has the smallest share. In the middle, both remote and densely populated areas have similar percentages of non-PNS teachers. 29 Duthilleul (2005), Lessons Learnt in the Use of Contract Teachers 30 See Duflo, Dupas and Kremer (2012), Bettinger and Long (forthcoming, 2006), among others. 31 Banerjee et al (2007), Remedying Education: Evidence from Two Randomized Experiments in India U The"20 percent rule": Where did resources go? Figure 29: Percentage of non-PNS Teachers, 2010 0.5- 0.4- C 0.3- 0 . - U 0.1- 0- 20% or less 20% - 40% 40% or more Source: World Bank - RAND School Based management Survey, 2010. Figure 30: Share of non-PNS teachers by province, 2010 exdludes outs[de values Source: NUPTK Data (2010) Schools use contract teachers partly to compensate for the lack of civil service teachers, but this is not the main driver for hiring contract teachers - nor is the size of the school. Since schools only have authority over contract teachers (i.e. they cannot hire PINS teachers out of their budget), one would expect that a negative correlation exists between the availability of PNS teachers and the share of contract teachers. That is not the case: spending on contract teachers is not correlated with the size of the school, (Figure 31), but it is partly correlated with the availability of civil service teachers. When plotting the student-teacher ratio using only civil service teachers to the share of the budget spent on non-civil service teachers, it is apparent that schools with a shortage of civil service teachers tend to spend a higher percentage of their disposable budgets on contract teachers (Figure 31). However, the relationship is far from perfect. The share of the budget spent on contract teachers varies enormously across schools, even across schools with the same student-teacher ratios. This indicates that while a shortage of teachers might be an important reason to hire contract teachers, schools must be hiring contract teachers for other reasons. Ae0C How Have Outcomes Changed? The Unfinished Agenda for Equity and Quality of Education Figure 50: Indonesian 8th-Grader math performance in 2007 TIMSS 100% - 80% - 60% -E i 40% ....i I |1 111 I 20% hff I -40%- -100% - om '- I e Hg m Avc a Low - B eo low Chin Shnga Chin Shanghai 90 0 ) Urs Q No0 Chi a ha gh i K re J pa Sin ap re in apoe orea r 0 m Intermediate aHgh * Advanced m Low m Below low Figure 51:2009 PISA results in reading and mathematics Math Scores Reading Scores Science Scores China Shangnai Cnina Shanghai China Shanghai KoreaJpn Kigaore Singapore Korea a Japan Singapore apan Australia Australia Australia OECD OECD OECD Turkey Turkey Turkey Non OECD Chile Chile Thailand Mexico Non OECD Chile Thailand Thailand Mexico Non OECD Jordan Kazakhtan Jordan Mexico Jordan Brazil Argentina Argentina Indonesia Kazakhtan Brazil Argentina Brazil Indonesia Tunisia Tunisia Tunisia Kazakhtan Indonesia Peru Peru Qatar Qatar Qatar Peru 0 200 400 600 800 0 200 400 600 c 200 400 600 800 Source: PISA, OECD Indonesian students' performance on PISA is consistent with its level of income. When mapping PISA results to GDP per capita, the correlation between the two is clear: non-high-income countries tend to have lower scores, and for those countries the score is highly correlated with per capita GDP (there is no correlation for high income countries). Indonesia is right over the trend line for non-high income economies. As mentioned earlier, Hanushek and WbGmann (2007) have provided convincing evidence that the quality of education has a causal effect on economic growth, so while the relative position of Indonesia given its GDP per capita is not alarming, the need to improve learning outcomes is nonetheless pressing. In addition to the low average score, the distribution of Indonesia's scores is particularly narrow. This means that the bulk of students score below basic levels of competency and that the share of top performers is extremely low. While inequality is clearly not a desirable goal in itself, recent evidence shows that the distribution of scores matters at least as much as the average score for economic growth.47 The share of students scoring above 600 has a separate effect from the share above the minimum achievement level, and 47 Hanushek and WObmann (2007), Pritchett (2009), Pritchett and Viarengo (2009). U Spending more or spending better: Improving education financing in Indonesia the effects are large, especially in poorer countries.48 The share of Indonesian students at that level in math and science is zero. Figure 52: PISA math score vs. GDP per capita 650 600 U Shanghal-China * Singapore 550 * Hongkong-Chia 5 Chinese Taipei Korea * Finland o # Switzerland * Japan* CaWo@therlands * Macao-China * New Zea~ri * Estonia NeeuAsai, 500 11olaor e F Norway atvi unga#y PortugaT ,# Portugl United States Luxemourg Lithuania M#gh-Incomeeconomes N Russia FeaderaoGreece *Non high-Icome economes 0 Croatia 450 Ser- Turkey * Israel Dubal(UEA) STha n aei * Trinidad and Tobago 400 KazakErMontenegro 0 Jord BAN Argentina 0ru Panama 350 g Kyrgyzstan 300 1 0 10000 20000 30000 40000 50000 60000 70000 80000 90000 Source: PISA (2009) and World Development Indicators; Note: China's GDP per capita is used for Shanghai. Figure 53: Distribution of math test scores Indonesia, Thailand and OECD, 2009 900 E 400 20 0 0 200 400 600 800 Math Scores - Indonesia -Thailand - OECD Looking at trends, the improvement in reading is good, but the trend in math and science is disquieting. Science scores have not increased since 2000 and seem to have decreased in 2009. Math results also decreased in 2009, after improving significantly in 2006. Half of students fall at level I and below defined as being "able to identify information and to carry out routine procedures according to direct instructions in explicit situations." What these students have not achieved (defined as level 2) is the capacity to"employ basic algorithms, formulae, 48 The authors use features of the education system as an instrument for cognitive achievement, so the results can beinterpreted as causal: i.e the share of high performers causes economic growth (subject to the usual discussions on the validity of the instrument). U How Have Outcomes Changed? The Unfinished Agenda for Equity and Quality of Education procedures, or conventions"or to "interpret and recognize situations in contexts that require no more than direct inference,' let alone more sophisticated reasoning methods. Since 15 is still an age at which many students leave school in Indonesia, the fact that so few students can go beyond carrying out routine procedures is worrisome.49 The differences in trends are similar across socioeconomic groups: math scores decreased for all socioeconomic deciles between 2006 and 2009. The differences across socioeconomic deciles are very sharp, with those in the lowest decile scoring 350 and the richest decile 420. However, even the richest students are significantly below the OCED average (500). These differences have not changed since 2003. A comparison between 2006 and 2009 shows declining scores for all socioeconomic deciles. Figure 54: Indonesia PISA scores in mathematics, reading and science, 2000-2009 Average scores, 2000-2009 Indonesia's score on PISA 410 400 390 380 0 2000 0370 < 02003 S360 350 02006 340 n 2009 330 Mathematics Reading Science Subjects Source: OECD, PISA Figure 55: Indonesia PISA scores in science and math by socioeconomic deciles, 2000-2009 Science Scores Math Scores 460 -- 460 - 440 - - 440 420 4 | |20 I 400 --400- 380 --380 360 360 340 340 320 i 320 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 - 2003 - 2005 - 2009 - 2003 - 2005 - 2009 Source: PISA 2009 (weighted averages by decile) In summary, this chapter has shown that while access and equity have improved significantly in the last decade, quality has remained low by international standards. High inequalities remain in access by income level, especially in higher education (Figure 57). In addition to the challenge of increasing access for poor or remote students to senior secondary and higher education, improving the quality of education is a priority. And the growth needed is two-fold: raising the performance of the bottom of the distribution and increasing the number of high-achievers, which is extremely small in Indonesia. 49 See Chen (forthcoming), SSS ESW for a deeper discussion of PISA results. See"Indonesia Skills Report"and Cerdan-Infantes et al (2010) in which results from employers'surveys highlight concerns with the level of basic skills of recent graduates. U Spending more or spending better: Improving education financing in Indonesia This disparity in the observed trends in resources, access and quality leads to the obvious question: did the increase in resources yield the expected outcomes? The next section argues that, despite improvements in access, the answer is largely not yet. The expected outcomes have not yet been achieved and will not be achieved giving current spending patterns. Getting better results from current spending is possible, however, through increased efficiency and improving the quality of spending - which is the subject of the next chapter. Figure 56: Summary of progress, challenges and highlighted Gol objectives in education, 2010 Sourceghigte OwnTR eOabjecntndivcesNTR Access: A m tero tBig increase in enroment rates in qualiy of cti FECD and secondary ncrease in euai s d one m iSmaller increase in access to higher become apparent. A we s i cat 2n 209 th ed nangfl Se ni sered Education Equity: teMost of the increase in access came tm esi ce from poorer backgrounds and rural arnt rule wen to t spening hte resources wiltaaeito outcome s it fuueon lgly dependte onfhte these endtre resorces aecrltdwithlerovequit outoffdsrc sednuscoratedwion? te outcomsharpdicesineuainsnig, one might expect tha delaynd beforeoredeeducliational pninoad districtyil eventuallyoleadtoimprovements. sia i tht 2,pt abige mphin spending 9tache salaresn eartification hare thelatest to ipove learning outcomes, one PISAulS expectai th sasme. tahis laeto speninghad eeninceasig i th yeas piorto 209,butnot t te smatexcimnng mpfuat and scienc testif g hsicreaeores wilaeaet improvementsimin outcomes?tei As e sw i chpte 2,mos oftheaddtioal esorce frmte i2nprceint shre wf(etfto techer loouratm tes oeio epatend ors r sndold e c ther a tion lookrsu at these correansnsearat o pnig.n ih epc ea eoe dctoault imrvmnsbcm paet sw a ncatr2 h igsjm nsedn curdi 09 h How Have Outcomes Changed? The Unfinished Agenda for Equity and Quality of Education Figure 57: Change in public education spending and change in education outcomes at the district level a) SMA NER, total spending (2002-2009) b) SMA NER, non-salary spending (2002-2009) 150 150 00 :*00 o. . * * . . 05 -50 -50* 100 0 100 200 300 05 change in lea ecucation expenditure% change in real Non Salary Education Expenditure% Change of SMA NER & Education Expenditure * Change of MA NER & Non Salary Education Expenditure Fitted Line - Fitted Line c) SMP NER, total spending (2002-2009) d) SMP NER, non-salary spending (2002-2009) 100- 100 *2 * .* -* * so - . * . * ** . . **.* **. ** .''** * :** * * * * ~ : * u-50 - -50- 10 0 100 200 300 0 500 1000 change in real education expenditure % change in real Non Salary Education Expenditure% * Change of SMP NER & Education Expenditure * Change of SMP NER & Non Salary Education Expenditure Fitted Line - Fitted Line e) SMP UN score, total spending(2005-2009) f) SMP UN score, non-salary (2005-2009) 60 60 40 . * 40 * 8* 40* * *. * 0 ** * * * .*. ****.. 20 4 20 .* * . I = 0 .* 2-0 100 0 100 200 300 100 0 100 200 300 change In real education expenditure % change in real education expenditure% Change of SMP UN score & Education Expenditure * Change of SMP UN score & Education Expenditure Fitted Line - Fitted Line Source: MoF S KD (budget), Susenas (NER) and MoEC (UN exams) U Spending more or spending better: Improving education financing in Indonesia District Spending and Outcomes One way to test whether resources will translate into outcomes is to analyze the correlation between increases in district spending and improvements in outcomes. However correlating spending per student with outcomes may miss an important point: districts with different characteristics may need to spend different amounts to reach the same outcomes. So a lack of correlation in one year may not mean that district spending does not matter for outcomes. A cleaner comparison is obtained by regressing changes in resources with changes in outcomes. If district spending on education matters for education outcomes, districts that increase their spending faster than other districts should experience a faster improvement in outcomes. By looking at the rate of these changes, this comparison controls for district characteristics that are constant over time. If we look at changes in resources versus changes in outcomes, there is no correlation between changes in total spending on education and enrollments or learning outcomes. Figure 58 plots the change in spending and the change in SMA and SMP net enrollment rates between 2002 and 2009, as well as the change in spending and the change in UN scores in SMP between 2005 and 2009.so The graphs show no correlation between spending and any outcomes. Total spending and non-salary spending also show no significant correlation, and the slope is actually slightly downwards in some cases. The fact that increasing spending is not correlated with improvements in outcomes means that further increases in spending will be unlikely to be associated with improved outcomes, and that outcomes may vary across districts regardless of changes in spending.This lack of correlation is worrisome, since district spending has been increasing rapidly in recent years. Since the main determinant of differences in spending across districts is the student-teacher ratio, this lack of correlation between changes in district spending and outcomes is an indication that additional teachers do not have the intended effect on education outcomes. As we saw in chapter 2, the differences in per student spending across districts are largely driven by student-teacher ratios. Districts with lower STRs spend significantly more on education, and districts that experienced a faster increase in resources also saw their STRs decline faster. In fact, the picture of STR and outcomes at the district level looks similar to the graphs in Figure 58, suggesting there is no correlation between STR and education outcomes. Student-teacher ratios and Learning Outcomes We can also test the correlation at the school level by looking at school-level budgets and test score data - these results confirm those at the district level. To test whether the massive increase in resources is likely to improve quality in the future, we look at the correlation between STRs and learning at the school level. Using a representative sample of public primary schools, collected in 2010 for a study on school-based management practices," it is possible to correlate the student teacher-ratios with learning outcomes in mathematics and Bahasa Indonesia for 6 th graders. The data set also includes school budgets with salary bills for civil servant teachers.12 There is no correlation between the student-teacher ratio and learning at the school level for math or Bahasa Indonesia. Figure 57 shows the correlation between the student-teacher ratio and scores in mathematics and Bahasa Indonesia, differentiating between schools below the regulatory maximum of 32 students per teacher, and schools above. For both mathematics and Bahasa Indonesia, the correlation is close to 0 if schools are below the maximum of 32. Including schools with very large classes (above 32), turns the 50 Since the number of districts increased dramatically, we only use the sample of districts that did not split. Consolidating splitting districts does not change the correlations (not shown). 51 World Bank -RAND 2010 SBM Survey. See"School Based Management in Indonesia,"Chen (2011). 52 Civil servant teachers are not paid out of the school budget, but they are recorded in it.The questionnaire included a section for sala- ries, but only half of schools filled it out. For the other half, we use the total number of teachers to estimate the salary bill, a method that proves adequate to predict salary bills for schools for which we have teacher salaries. U How Have Outcomes Changed? The Unfinished Agenda for Equity and Quality of Education correlation negative only for Bahasa Indonesia and only due to outliers." This is consistent with existing literature on the effects of class size on learning, which shows little or no effect unless classes are too large.54 It seems clear that increasing the number of teachers given the distribution of teachers in Indonesia is unlikely to lead to improvements in quality, unless they are targeted to severely understaffed schools with more than 32 students per teachers. Figure 58: Student-teacher ratio, per student spending and Bahasa Indonesia test scores in public primary schools a) MathematicsIndonesia 100 80 90 70 80 60 70 -t4 60 -5 0 50 - * 40 30 - 10 -10 * 0 0 10 20 30 40 50 0 10 20 30 40 50 * Less than 32 a More than 32 Lnear (Less than 32) # Less than 32 m More than 32 Linear (Less than 32) Source: School Based Management Survey, World Bank-RAND (2010) These analyses suggest that the drastic increase in resources in 2009 spent on the growing number of teachers will not necessarily lead to better learning outcomes and that it is not "only a matter of time." Indonesia needs to spend differently to improve education quality. If the large increase in spending was channeled mainly to hiring additional teachers or converting contract teachers to civil servants, and more teachers does not lead to improved outcomes, it is unlikely that time will improve outcomes. Furthermore, we have seen that something in the system biases spending toward increasing the number of teachers, so simply increasing spending may lead to yet more teachers hired with no effect on learning. Teacher Certification and Learning Outcomes5 Given the large share of the budget spent on teacher certification, ensuring that this program ultimately increases student learning is a vital way to improve the quality of spending. The national teacher law passed in 2005 mandated that all teachers should be certified by 2015. As of 2012, about 35 percent of all teachers have been certified, and the program is currently absorbing about 9 percent of the education budget. As originally planned, all teachers would have been required to have a university bachelor's degree and have gone through process to certify their competencies. Upon graduation, teachers would be eligible for the"professional certification', which would ensure the proper competencies and afford a significant pay raise equivalent to the base salary. During the implementation, however, the Gol opened another route to certification, based on teachers'"portfolio": teachers with enough experience and a portfolio of training, workshops and other learning activities could access certification without the coursework. So far, this has been the most common route to certification. 53 OLS regressions controlling for observables (school characteristics, principal and teacher education and experience, and district fixed effects) confirm these findings. 54 See Hyunkuk, Glewwe and Whitle (2010), Hoxby (2000), Jepsen and Rivkin (2009) 55 This section is based on De Bee (forthcoming)"Teacher certification in Indonesia: A doubling of salary, or a way to improve learning?" World Bank, Jakarta. U Spending more or spending better: Improving education financing in Indonesia The certification program was meant to improve student learning by upgrading teacher qualifications, increasing teacher motivation, and attracting better qualified people into the profession. First, requiring a bachelor's or Si degree and the certification training program were intended to upgrade qualifications. Second, it was noted that teachers usually held second jobs, which may have partly explained high rates of absenteeism; the increased salaries associated with certification were meant to improve teacher livelihoods so they could be dedicated to teaching exclusively in one school. In order to support this, a minimum teaching load of 24 hours was required to be eligible for certification. Finally, by increasing the salary potential, the program intended to attract better students into teaching, which will in the long run improve the quality of education. To evaluate the progress and impactof teacher certification, a rigorous impact evaluation study has been taking place since 2009. The study includes extensive data collection with the overall objective of evaluating the effects of certification on student learning: all together, the study will rely on three rounds of data collected from 240 public primary and 120 junior secondary schools, including the testing of students and teachers in mathematics and Bahasa Indonesia. The first round of data was collected in November 2009, the second round in March 2011, and a final round in April 2012. This report discusses some of the findings from baseline and midline data, since the final round was not yet available for analysis at the time of this writing. Results from the baseline and midline survey are available, but must be interpreted with caution - the final report, including additional data collection, will only be available in 2013. The first message from the evaluation is that on average teachers perform badly on subject matter tests, and that teachers who hold an S1 degree perform only marginally better than teachers with less education. On a test to measure the basic competencies of teachers (deemed necessary as a basis for effective teaching) and designed by the government institution responsible for educational assessment in MoEC (Puspendik), primary school teacher test scores averaged about 33 percent. Teachers with an S1 performed slightly better, at 35 percent. Scores improve at the SMP level, with teachers getting about half of the questions right. S1 holders scored only marginally better than those without an S1 degree (53 percent to 46 percent).Two lessons can be drawn from these results: that the competencies of teachers are low, and that teachers holding an 51 degree might not yield the expected results for quality of education. Indeed, the results suggest that an S1 is by no means a guarantee of teacher competency or knowledge of subject matter. And given that most certified teachers were certified on portfolio, it is unlikely that certification alone has led to improvements in their abilities. Figure 59: Average test scores for teachers in primary and junior secondary 100 Primary 100 Junior Secondary E 80 E 80 o a U v r 60 2 60 53 .2 .2 46 2 40 35 31 2 40 20 20 0- 0 1 Degree No Degree Degree No Degree Source: De Ree, 2012 In terms of impact, the study finds evidence that certification improved the livelihoods of teachers, drastically reducing the number of teachers that hold second jobs. Taking advantage of the randomized design of the evaluation, table 4 charts the local average treatment effects of certification, showing the control group average (non-certified teachers) and the change in the average associated with certification on a number of variables.6 Generally, the coefficients of the effect of certification are small, except in two areas: receiving a professional allowance and holding a second job. Indeed, the probability of getting a professional allowance 56 For a full methodological explanation, see De Ree (forthcoming) U How Have Outcomes Changed? The Unfinished Agenda for Equity and Quality of Education increases and the probability of reporting financial struggles is greatly reduced with certification. Meanwhile, teacher certification drastically reduces the probability of holding a second job, from an average of 33 percent among non-certified teachers to a mere 7 percent for certified teachers. Thus it seems clear that professional certification has a significant impact on teacher's livelihoods. Table 5: Local average treatment effect of teacher certification - selected variables Control group Causal effect of average certification Teacher SD test score 0,33 0.04 0.81 1663 Teacher SMP total score 0.53 -0.02 -0.34 1392 Being KKG/MGMP member 0.93 -0.04 -0.57 3117 Total pay, excluding professional allowance 2.48 -0.38 -1.13 3109 (in million Rp.) Having a second job 0.33 -0.26*** -2.70 3117 Teaching hours this school 22.76 2.48 1.13 3116 Teaching hours in other school 1.44 0.16 0.13 3117 Absent at least once last week (self- 0.13 -0.03 -0.32 3116 reported) Professional allowance (in million Rp.) 0.57 2.38 15.23 3114 Self-rated problems with supporting 0.49 -0.40 -4.39 3116 household financially Control group averages, causal effect of certification (local average treatment effect). All regressions are uncontrolled. Standard errors are clustered at the school level.*** means significant at the 1 percent level. On the other hand, certification shows no significant impact on learning outcomes. This failure of certification to improve learning outcomes may be partly due to the insignificant effect of certification on teacher test scores, as observed in table 4. When looking at scores in math and Bahasa Indonesia at both SD and SMP, the difference in scores between students of non-certified teachers and students of certified teachers is not significant on any test at any level. The effects of certification on learning, at least in the timeframe of this evaluation, do not seem commensurate with the high cost of the program. However, given the short-term span of the evaluation we cannot yet assess the longer-term effects of the program, which should attract better people into teaching. Table 6: Causal effects of certification on student learning outcomes SStandard error Math SD 0.08 0.16 35,536 IPA SD -0.02 0.16 35,470 Indo SD 0.03 0.18 35,525 total SD 0.05 0.18 35,162 math SMP 0.02 0.34 39,134 IPA SMP -0.07 0.43 39,531 Indo SMP -0.05 0.17 39,451 Engl SMP 0.19 0.46 37,869 TABLE 7. Causal effects on standardized student test scores. Local average treatment effects (controlled for baseline scores and district dummies. For classes without baseline scores, baseline scores are set at zero, and controlled for with a dummy). U Spending more or spending better: Improving education financing in Indonesia Teacher subject knowledge matters for learning: teacher certification should target subject knowledge. When teachers know the subject they teach, the gains in test scores for their students is larger." Teachers who scored in the top third in the teacher subject knowledge test in the baseline saw their students' scores increase by 0.08 standard deviations, while teachers who scored in the bottom third, decrease scores by 0.04 standard deviations. This result is consistent with previous studies in Indonesia, which have shown that both subject knowledge and pedagogical knowledge are correlated with student learning. Unfortunately, teacher certification as it was implemented, with a majority of teachers being certified through portfolio certification did not address subject or pedagogical knowledge and thus did not result in improved learning outcomes. Figure 60: Student learning gains by teacher competency in subject knowledge 0.1 - 0.08- 0.06- 0.04- 0.02- 0- -0.02- -0.04- -0.06- bottom 33% average top 33% Source: De Ree (forthcoming) While the teacher certification program was well-intentioned and well-developed at its inception, the implementation gave space for ineffective paths for certification. The introduction of"portfolio certification" allowed teachers with a certain number of years of work experience and prior training or workshops completed to receive a waiver on other certification training requirements.The evidence presented in this section suggests that this was not an effective method of certification. The pass rate for these teachers is nearly 100 percent. A process of assessment and targeted training, as well as periodic re-certification or review, might provide options to improve the program. The two biggest spending categories (adding teachers and teacher certification) do not show effects on learning. But if adding more teachers does not lead to better outcomes, what does? Next chapter looks at this issue in detail. 57 As De Ree (forthcoming) points out, these value added regressions suffer from mean-reversion bias, which can be sizeable The mag- nitude of the effects is unlikely to be driven completely by this bias. See De Ree (forthcoming) for results when attempting to control for this bias. I Chapter 4: Improving the Quality of Spending on Basic Education Improving the quality of spending means turning resources into outcomes more efficiently. For a certain level of spending, this means increasing access to schooling and improving the quality of education. International experience shows that this is easier to do for access than for quality. Indeed, this has been the case in Indonesia. As resources have increased, equity and access have improved but quality of education has not. In effect, in Indonesia, improving the quality of spending means improving learning outcomes without necessarily increasing resources. Countries have proven to be successful at keeping children in school through a variety of policies (school building, scholarships, conditional cash transfers). As we have seen, Indonesia has successfully implemented many of these policies. Access to basic education has been made virtually universal through a massive expansion in school building and the provision of block grants to schools to lower the cost of education. At the same time, the country also provides scholarships to the poor and a small conditional cash transfer (PKH). However, access to senior secondary and higher education remain a challenge. The low level of resources allocated to senior secondary education seems to indicate that access could be improved at that level. Similarly, the scholarships for the poor program (BSM), as we will see, suffers from problems that limit its effectiveness. Thus there is still room for improvement when it comes to increasing access by improving existing policies in a relatively straightforward manner. Learning outcomes, on the other hand, are more elusive. This is true not only in Indonesia, but around the world. A recent World Bank-UNESCO joint publication on "Strengthening Education Quality in East Asia" notes that "Low- and middle-income countries in East Asia have made considerable progress toward achieving universal primary education, but quality remains a challenge:" As has been the case in Indonesia, the study shows that increasing spending has not systematically resulted in improvements in learning outcomes, but that "in many countries, it has coincided with a decrease in quality." Indonesia is not alone in its elusive quest for quality. Recent evidence shows that system reforms have the strongest impact on educational quality.There is no straightforward answer on how to improve the quality of education. Countries that have made rapid advances in learning outcomes do so through systemic educational reforms. The financing of the system, the management of teachers and the promotion of school-base management, among other dimensions, all have impacts on learning outcomes. This chapter focuses on three critical systemic issues that can improve how spending translates into access and educational quality: strategic expansion of the centrally-administered Scholarships for the Poor program (BSM), improving district-level teacher management and improving district spending. Chapter 2 highlighted the need to increase spending on ECD and senior secondary education: one way to U Spending more or spending better: Improving education financing in Indonesia achieve this goal is through strategic expansion of the BSM program (Scholarships for the Poor), the first focus of this chapter. Next, as seen in chapters 1 and 2, despite being formally in charge of managing both teachers and schools, district governments spend most of their resources on salaries. Improving teacher management is therefore crucial to liberating the necessary resources for districts to support schools more intensely. Thus the second and third sections focus on improving teacher management. On teacher management, this chapter provides evidence of the magnitude of the inefficiency and inequality introduced by the current distribution of the teaching force, as well as its consequences for equity and quality of schooling. Teacher management is the direct responsibility of district governments but is to a large extent influenced by central government regulations and the incentives associated with transfers. Our focus, therefore, is on the overall financing system. While recommendations on general teacher management (from pre-service to career management) are beyond the scope of this report, this chapter provides some considerations for how to improve teacher management in terms of efficiency and distribution. With regard to district management and support of schools," this chapter looks at one of the most innovative government programs in education: the provision of funding from district governments to schools in block grants. The "BOSDA" (Bantuan Operasional Sekolah Daerah, local school grants) program is one of the main tools districts have to support schools; it is similar to BOS, but experiments with different formulas and provides more intensive support and supervision.Taking advantage of a rich dataset for a nationally representative cross section of schools, this section draws some conclusions about the effectiveness of the program and the implications for improving school-based management. The last section seeks to tie together various lessons to offer possibilities for improving district management. Improving and expanding existing programs: Scholarships for the poor59 Scholarships for the poor are critical to increasing enrollment. Poor household education expenditures are larger than expected and are particularly expensive for junior and senior secondary students, resulting in alarmingly high drop-out rates, mostly during transition years. Currently, the Bontuon Siswo Miskin (BSM) program is limited in its ability to provide real protection or incentives for continuous education to poor and vulnerable households. Problems with BSM are related to the amount of the scholarship, the timing of disbursement and its poor targeting. BSM amounts are not commensurate with the true costs of education, especially in senior secondary education (Figure 62). BSM cash benefits are typically received one year or more after a child has enrolled in and attended a grade; BSM is unavailable to most eligible students who are making the transition from primary to junior secondary or from junior to senior secondary; and BSM is one of the least-progressively targeted social assistance programs in Indonesia, meaning that non-poor and non-vulnerable households have received nearly an equal share of total BSM benefits as poor and vulnerable households. Furthermore, only a very small portion of total BSM expenditures are used for safeguarding and support operations, such as targeting, socialization, facilitation, monitoring and evaluation, and complaints and grievances, leading to under-provision of most of the activities and services that could help more households and educators use the BSM system more effectively. 58 Bruns, Filmer and Patrinos (2011), Making Schools Work: New Evidence on Accountability Reforms 59 This section is based on the executive summary of Alatas and Jellema (2012), "Cash Transfers for Poor Students,' background paper for the report on Protecting Poor and Vulnerable Households in Indonesia (World Bank 2011). The cost calculations, not included in the cited paper, were carried out in conjunction with the authors of the background paper. U Improving the Quality of Spending on Basic Education Figure 61: Education costs vs BSM amount, 2009 2,000,000 1,500,000 H Poor hh, education costs (ex-transport) H Non-poor hh, education costs (ex-trans port) 1,000,000 BSM amount 500,000 0 Primary Junior Senior Secondary Secondary Source: Alatas and Jel ema (2012) using Susenas and MoEC The BSM program is national in scope but reaches few students overall and does a poor job of identifying poor students. In 2009, program coverage (through the senior secondary level) was still small at 2.3 percent of all 6to I8 year olds in Indonesia. From basic to higher education, the program covers about 3 percent of students. However, students from the poorest 40 percent of households account for only about half of all BSM scholarships (and half of all rupiah distributed through the BSM program) while middle-class and richer households in the top 60 percent capture an equal 50 percent of all BSM scholarships. Figure 62:Targeting of BSM 20 --8 -- 2 Percent of6-18 yr olds with BSM 8 - SD-aged children 15 -U- a Share of tota BSM scholarships M6 - SMP-aged chidren Il SMU-aged children Houseo d EpendiureDcileousehld-Exendr Dciden 10-4 0 0- 111, Ul a. 0 1 , 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Household Expenditure Decile Household Expenditure Decile Source: Alatas and Jel ema (2012) using Susenas and MoEC BSM does not target those who are unfamiliar with the school system and its administrators. BSM initiatives typically identify potential scholarship recipients by soliciting nominations from schools and school committees. Students nominated must have already achieved consistent attendance and demonstrated "good behavior,"as confirmed by the principal. Recently enrolled students or prospective new entrants have little chance of being selected; likewise, those who have not made themselves known to the principal are unlikely to be selected. Households cannot nominate their own children and there is currently no formal appeals process. BSM does not effectively address difficult and costly transition periods - between elementary and junior secondary, and again between junior and senior secondary - when the overwhelming majority of drop- outs occur. Complex verification procedures plus slow rates of disbursement together mean that recipients typically receive a BSM transfer for their first schooling year only after their second schooling year has already U Spending more or spending better: Improving education financing in Indonesia started.The BSM program does not distribute scholarships forthe final grades of any schooling level (the sixth year for elementary and the last semester of the third year forjunior and senior secondary) since BSM is not integrated across primary, junior secondary, and senior secondary education providers or across the sub-directorates in MoEC and MoRA in charge of those levels of schooling.The result is that BSM transfers are not available for poor students as they transition to junior and senior secondary school and that scholarships for the first year in the post-transition, higher-level school are delayed until the following year. As education expenditures rise steeply between education levels and are always greater at the beginning of the registration period (at any school level), that BSM cannot facilitate expenditures during these periods is doubly unfortunate. Figure 63: Education cost profile vs BSM benefits 1,200,000 1,000,000 800,000 transition 600,000 trans iin 400,000 200,000 0 01 SD- 1st BSM-SD PKH-SD Q4 SD 6th 01 SMP 1st BSM-SMP PKH-5MP Q4 SMP 3rd Q1 SMU 1st BSM-SMU grade grade grade grade grade Source: Alatas and Jellema (2012) using Susenas and MoEC Reforming BSM means increasing the total budget for the program: increasing the amount, increasing coverage, improving the targeting, aligning the disbursement of the benefit with the timing of the expenditures and possibly providing a transition bonus. The ideal scenario would cover 100 percent of the cost for all poor students from SD to SMU, providing a transition bonus for those moving from SD to SMP and from SMP to SMU. This would ensure that cost is not a reason for dropping out, and provide incentives for students when they are deciding whether to continue. The estimated cost of this ideal reform scenario would triple the budget for BSM from 3 to 9 percent of the central government education budget. Figure 65 shows the estimated cost of covering all poor and 50 percent of near poor students, and of providing a 50 percent transition bonus between SD and SMP, and SMP and SMU. The coverage would increase from 4 percent of students to 22 percent, up to a total of 11 million scholarships. The impact on the budget would be substantial, adding an extra IDR 5 billion to the budget by 2015. U Improving the Quality of Spending on Basic Education Figure 64: Estimated cost of expanding coverage of BSM to all poor, for 100 percent of education costs, 2012-2015 30% - 12,000,000 25% 10,000,000 20% -8,000,000 IIIII Senior Secondary Scholarship (total) M Junior Secondary Scholarship (total) Primary Scholarship (total) 15% 6,000,000 - Coverage, all enrolled (%) 10% 4,000,000 5% 2,000,000 0% Current 2012 2013 2014 2015 Source: Own calculations using Susenas and MoEC Including subsidies for transportation would increase the amount significantly. Transportation costs represent a high share of education costs, especially for the poor. For quintile 1, in SD and SMP, transportation costs represent 200 percent of the cost of education: including transportation in the scholarship would triple the scholarship amount. For SMU, the scholarship amount would double the education cost (or triple the current amount). It is unclear, however, if this increased amount would have a significant impact on enrollment. A recent impact evaluation ofa scholarship program in Cambodia showed that changes in the amount ofthe transfer had no significant effects on enrollment.60 Figure 65: Estimated cost of expanding coverage of BSM to all poor, for 100 percent of education costs, 2012-2015 250% 0 C 0 200% 00 208 150%' - - Primary o0 - Junior Secondary y 100% . - Senior Secondary 0 0 50% 0% 1 2 3 4 5 Quintile Source: Own calculations using Susenas, 2009 60 Filmer and Schady (2009), Are There Diminishing Returns to Conditional Cash Transfers? U Spending more or spending better: Improving education financing in Indonesia Improving Teacher Management It is clear from previous chapters that improving teacher management is crucial for increasing efficiency of spending. This section will show it is also crucial for equity and quality. The previous chapters have highlighted the fact that teacher salaries have absorbed most of the additional resources from the 20 percent rule, and that lower STRs are not correlated with better outcomes, suggesting that increasing STRs may improve efficiency without lowering quality.This section will further show that improving the distribution of teachers can have important effects on equity and quality of education. By crowding out other spending at the district level, excessive teacher hiring may be having a negative effect on learning. But how much room is there for these gains in terms of efficiency and equity, and how can they be realized in the complex governance system of teacher management in Indonesia? This section provides some estimates for the size of the efficiency gains and the magnitude of misallocation of teachers, and provides some policy options to address these issues. Despite the low average STR at the national level, there are vast differences in the availability and qualification of teachers across schools. Student-teacher ratios are very unequal, as are the levels of teacher qualifications. For example, in 2010, student-teacher ratios in primary schools ranged from fewer than 10 to greater than 60 students per teacher. As shown in chapter 3, schools with STRs beyond 32 suffer from diminished results in Bahasa Indonesia tests. Shifting teachers from schools with low student-teacher ratios to those with high student-teacher ratios has the potential to raise overall levels of learning. Equally important are the differences in qualification. Wealthier urban areas have a higher concentration of more qualified and experienced teachers (Table 6). Table 7: Teacher characteristics by location Remote Rural Urban Borders Total number of teachers with at least an S1 degree Primary 10,407 214,667 119,869 2,233 347,176 Junior secondary 8,789 193,255 134,468 4,288 340,800 Total 19,196 407,922 254,337 6,521 687,976 Percentage of teachers with at least an S1 degree Primary 14 24 38 26 27 Junior secondary 69 76 78 76 77 Total 22 36 52 46 40 Total number of teachers Primary 74,757 890,290 317,623 8,476 1,291,146 Junior secondary 12,757 254,721 172,044 5,642 445,164 Total 87,514 1,145,011 489,667 14,118 1,736,310 Percentage of total teachers Primary 6 69 25 1 100 Junior secondary 3 57 39 1 100 Total 5 66 28 1 100 Source: NUPTK Data (2010) Notes: Border area - on the border with another district, Remote area - BPS classification based on six criteria (economy, human resources, infrastructure, fiscal capacity, accessibility, poverty) The regional differences in the distribution of teachers by education level are very sharp: richer districts, especially those in Java and Bali, have access to more educated teachers. The share of teachers with a senior secondary or below education is under 20 percent in all districts in Java, whereas in some districts in U Improving the Quality of Spending on Basic Education Papua or Sulawesi, it reaches 60 percent. Making the distribution of teachers more equitable by ensuring that poor and remote schools have an equal share of qualified and experienced teachers might raise overall levels of learning and narrow learning disparities. Figure 66: Share of teachers with senior secondary or less as their highest education by province, 2010 80 _ excludes outsidevalues 0 60 40 o Source:< ~ Ow acuainsuigNUT at,21 400 :it E OZ o In short, improving the distribution of teachers is a matter of efficiency, equity and quality of education. The equity and quality issues are evident from the previous graphs. On efficiency, as we saw in chapter 2, the student-teacher ratio is the main factor when it comes to district spending on education. Districts with low STRs spend significantly more per student than districts with higher STRs. At the school level the relationship is even stronger. As implied by the trend in Figure 68, an increase in the student-teacher ratio of 5 students per teacher reduces spending per student by about one-third. Figure 67: Per student spending and student-teacher ratios at the school level, 2010 50 45 - 40 .o 35 -30 5 310 < t:5 0 0 1 2 3 4 5 6 7 Budget for student in Rupiah Source: Own calculations using School Based Management. U Spending more or spending better: Improving education financing in Indonesia The potential for efficiency savings is large. The exact amount of potential efficiency savings from raising student-teacher ratios is hard to calculate. However, using information on the current levels of students and teachers and average teacher salaries, it is possible to estimate the teacher salary bill for different levels of the student-teacher ratio. In 2009, the teacher salary bill for basic education was IDR 67 trillion (USD 7.4 billion) and accounted for approximately a third of all government education spending. At this level of spending, approximately 2 million teachers were employed, resulting in a ratio of one teacher for every 18 students. Raising the student-teacher ratio reduces the number of teachers required for the same number of students. Holding the number of students and the average teacher salary constant, Figure 69 shows the effect on the total salary bill of raising the student-teacher ratio. Raising the student-teacher ratio to 23, for instance, a level that existed in the early 2000s, would reduce the overall teacher salary bill by DR 15 trillion or 22 percent. Raising the student- teacher ratio to 28 students, a level similar to other lower-middle-income countries, would reduce the overall salary bill by 31 percent, equivalent to the total amount currently being spent by MoEC on university education (IDR 22 trillion). Estimates of this kind are only illustrative: achieving savings of this magnitude in the short- or medium- term is not feasible. Strategies to raise student-teacher ratios rapidly frequently rely on hiring freezes, redundancies and redeployment. However, these are unlikely to be feasible and policies designed to raise student-teacher ratios over the medium- to longer-term need to be developed. The uneven distribution of teachers makes a hiring freeze undersirable without redistribution. For example, taking advantage of teacher retirement and making teachers more flexible through changes to pre-service training are potentially more effective strategies but take time to design and implement. The Gol has made significant efforts to improve efficiency and equity in the allocation of teachers. The Gol has issued various regulations over the last five years that have set standards for school staffing levels (Table 7). However the staffing norms associated with each regulation are different and provide different guidance on standards. This has caused some confusion and difficulty in interpreting the regulations and has complicated the monitoring of compliance at the school, district and provincial levels. As the hiring institution, districts also issue their own regulations on staffing norms and these can go beyond the minimum staffing levels implied by national standards. Figure 68: Estimates of government spending on basic education teachers for different student- teacher ratios, 2009 80,000 100 IIIIIIII o,oo80 ! . 60,000 80 70 50000 w ~40,000 50 r 40 * S30,000 30 7i 0 20,000 0 I-_ CD 0' M ,-0 r 0 ' C r1 1 N _ 1 C11 r _ C Student teacher ratio 0 Source: own calculations based on APBDN 120091 Improving the Quality of Spending on Basic Education Table 8: Recent teacher staffing regulations MStandards for basicTeh iagudln st t e (Permendiknas No. (521)(01 41/2007) Year 2007 2010 2011 Maximum of 32 students per Maximum of 28 students learning group (rombel). 20-3stdnsprlaig per learning group group (rombel). Primary education (rombel). students. One teacher rombel. Teachers must teach a Minimum of6 teachers per Minimum ofone sport and local minimu of 24hours. school in all regions 'except cotntehrprshol minimumspecial regions where minimum is4. Estimated primary student- teacher ratio if standards fully implemented. 19 19 NB: Student-teacher ratio in 2010 was 17.5 Maximum of 36 students per Maximum of 32 students per Juniorsecondry edcatilnparninrninggroup Maximum of 32 students Merinmumrofunptahe. o learning group. per learning group. Miimmecet ece fo Minimum of one teacher per Junior secondary education Teachers must teach a regionsubwheep inu speislsbjc minimum of24 hours. Teachers must teach a one teacher for each cluster of mnmmo 4hus subjects.m Estimated junior secondary student-teacher ratio if standards fully implemented - 21 19 NB: Student-teacher ratio in 2010Owas 18 We can use the latest guidelines (those under the joint decree) to identify understaffed and overstaffed schools and estimate the extent of the reallocation needed for all schools to meet the guidelines. The magnitude of this reallocation can be interpreted as a measure of the inequality in teacher distribution. The joint decree (Box 2) provides guidelines for teacher allocation and makes districts accountable for meeting redistribution requirements. The guidelines clearly define the minimum number of teachers required in a school according to its characteristics. Using the latest school-level information on teachers and student numbers, it is then possible to identify which schools have insufficient teachers and which have too many. Given that reallocation within districts is easier, we first assume that excess teachers can be reallocated to understaffed schools within the same district. If there are extra teachers after this reallocation, it is assumed they will be moved across districts within provinces, and if need be across provinces, It is important to note that we are not taking into consideration the logistics or cost of such a move. This exercise is meant to show the magnitude of the reallocation needed to make teacher distribution more equal according to government regulations. The results should be interpreted with caution: the estimates are based on 2010 levels of enrollment and do not take account of a future rise in enrollment due to population growth and the government commitment to expand education access. But while they are not projections of future teacher needs, they do provide important information on the impact of changes in staffing norms on national student-teacher ratios, the scale of redeployment required to create greater equity and ultimately the effect on public spending. regon whreminmu i Spending more or spending better: Improving education financing in Indonesia Box 2: The joint decree for teacher management In late 2011, a joint regulation was issued that provided guidelines to provinces and districts on restructuring and redistributing government-employed teachers. The decree was designed to exert some control over civil service teacher recruitment and improve the current distribution of teachers. The regulation was issued by five ministries collectively responsible for setting guidelines for the management of government teachers. The decree and accompanying guidelines provide a set of staffing standards to apply to the allocation of government teachers across schools. The decree instructs districts and provinces to collect the data necessary to identify the degree of redistribution and hiring required to achieve the standards, develop a detailed plan on how to redistribute teachers and finally to implement this plan. Where redistribution requires teacher transfer across districts or provinces, the Ministry of Education and Culture is responsible for coordination among local governments. The Ministries of Education and of Religious Affairs evaluate district and provincial plans for teacher redistribution.These evaluations are then passed on to key central ministries (e.g. Ministry of Finance, MenPAN) to help determine hiring levels for government teachers and consequently the size of intergovernmental budgetary transfers. Local governments failing to adhere to their obligations under the joint decree face a number of sanctions. The Ministry of Education and Culture can withhold education assistance (e.g. BOS) as well as make recommendations to other joint decree ministries to: * Refuse to issue 'formasi' to allow the recruitment of new government teachers (Ministry for Government Apparatus Empowerment and Bureaucratic Reforms or MenPAN) * Delay intergovernmental budget transfers (Ministry of Finance) * Issue a poor local government performance assessment (Ministry of Home Affairs) Challenges in redeploying teachers under the joint decree Local governments need to make plans to ensure that all schools meet the staffing standards outlined in the joint decree. There are two main strategies: * employ more teachers to fill posts in schools with teacher deficits. * transfer teachers to teacher deficit schools from schools that have an excess according to the standards. Clearly employing more teachers will further increase the size of the teaching force and lead to greater inefficiencies in education spending. The joint decree is able to influence local government hiring decisions for civil servant (PNS) teachers through the annual quota set by MenPAN and MoF.61 Where districts already have sufficient teachers to cover the standards outlined, the decree could be used to restrict further hiring of PNS teachers and encourage local governments to address inequalities in teacher distribution from their current teaching force. What would the implication be for teacher redistribution if local governments transfer teachers in order to comply with national standards rather than resorting to increased hiring? Under the technical guidelines of the joint decree, the number of teachers necessary for current levels of provision is smaller than the existing teaching force. There are large mismatches in the existing stock of teachers in primary and junior secondary schools. At the primary level, the number of teachers required is approximately 100,000 less than existing levels. Breaking down primary school teacher requirements by subject highlights large surpluses in class-based teachers and shortages in physical education and religion teachers. For example, fulfilling the joint decree would require only 1.1 million class-based teachers compared to the 1.3 million currently teaching in the system. On the other hand, the estimates point to a deficit of approximately 60,000 sports teachers. A similar picture emerges at the junior secondary level: based on the joint decree, there is a large surplus of Bahasa Indonesia and religion teachers and a deficit of computer, local content and civics teachers. 61 Kluyskens and Firdaus (2009), Teacher management: Recruitment, Selection and Data, Probation and Transfer. Improving the Quality of Spending on Basic Education Figure 69:The estimated teacher requirements to fulfill the standards outlined in the joint decree, 2010 1,600,000 600,000 1,500,000 550,000 S1,400,000 500,000 z 1,300,000 450,000 16 1,200,000 400,000 0, -0 1,100,000 350,000 E Z 1,000,000 300,000 900,000 250,000 800,000 ________ 200,000 Primary education Under Joint Decree Junior secondary school Under the Joint Decree teaching force teaching force 0 Classroom 0 Religion * Sports m All Teachers Note:The estimates include both PNS and non-PNS teachers Source: MoEC school data (2010) and NUPTK Data (2010) The magnitude of the reallocation needed to make the distribution of teachers more equitable is massive - 340,000 teachers or about 17 percent of the total teaching force would have to be transferred. Most of this redistribution would involve moving teachers within districts. However, approximately 74,000 teachers would need to be moved from districts with excess teachers to deficit districts in the same province (Figure 71). Given that the joint decree standards imply a surplus of teachers nationally, transfers across provinces could also improve the distribution of teachers. After transfers within and across districts, an estimated 34,000 teachers could transfer from one province to another to fill deficits. These national regulations provide a basis for ensuring that all schools have a minimum number of teachers but they do not tackle existing inequalities in the distribution of qualified (at least S1) or experienced teachers. So while the national standards may ensure that remote rural schools have the required number of teachers, they are less clear on the distribution of teachers according to qualifications and experience. A closer look at the redistribution implied by the recent joint decree highlights the challenges involved in improving the efficiency and equity of teacher distribution. In addition there are some hard constraints to implementing these reforms, given the absence of clear guidelines for dealing with the geographic distribution of schools and the characteristics of current teachers in Indonesia. All else equal, districts will try to avoid reallocating teachers because of the high cost, not only in monetary but in political terms. Moving teachers across districts is difficult, since due to the basic allocation formula of the DAU, transferring a teacher means losing funds associated with that teacher's salary. As a consequence, there is a serious risk that the joint decree will not be fully implemented. So what can be done to strengthen current reform efforts? Improving the allocation of teachers: Policy Options In addition to the ongoing efforts to improve teacher distribution, what can be done to speed up the process? As mentioned previously in this report, the system of transfers and regulations are a big reason for this misallocation. U Spending more or spending better: Improving education financing in Indonesia Breaking the link between central government transfers and teacher hiring The current system for hiring PNS teachers creates strong incentives for local governments to continue to increase the size of their teaching force and accelerate the conversion of contract teachers to PNS status. At present, intergovernmental resource transfers are partly determined by the size of a local government's payroll. Districts with larger numbers of civil servants receive more from the transfer system. Professional and functional allowances for teachers are also paid from the center and not out of local government budgets. It has been estimated that central government transfers cover approximately 75 percent of the salary of an additional civil servant or PNS teacher. This in effect subsidizes the cost of additional teachers and creates incentives for increased hiring.These incentives combine with strong political motivations at the local level to increase the size of the civil service. Controlling the demand for increased civil service employment are MenPAN and the Ministry of Finance, which set quotas for civil service hiring. Quotas are set annually by considering local government requests for additional staff and the current budget situation. MenPAN's actual process for determining quotas remains unclear.62 However, the process usually results in provincial and district governments receiving fewer posts than they requested. This creates incentives for local governments to exaggerate their actual needs, knowing that they will receive only a fraction of their initial request. In order to contain the growth of the teaching force, local governments must face the true cost of hiring teachers. The joint decree provides a reporting mechanism to contain teacher hiring excesses. Combining this reporting mechanism with a transparent and improved system of setting quotas for civil service hiring could go some way to improving teacher hiring decisions.63 However, improving transparency and evaluating local government staffing plans effectively at the central level is likely to be difficult. MoEC will need to evaluate staffing plans from 33 provinces covering over 500 districts, more than 300,000 schools and over 2 million teachers. Attempting to contain hiring in this way is also a second-best solution, in that it attempts to allocate new teaching posts more effectively from the center rather relying on district governments to address the complications of determining the size of the national teaching force. A more direct approach to eliminating incentives for over-hiring would be to break the link between intergovernmental transfers and teacher hiring. Key to this would be the elimination of the link between the size of the civil service and the size of a local government's DAU allocation. However, removing the Basic Allocation element from the DAU formula may not guarantee an automatic change in district behavior when it comes to teacher hiring decisions. During the last decade, districts have opted mainly to use additional resources to hire more teachers, motivated not only by financial consideratons (the majority of salary expenses being covered by the DAU transfer) but also by political reasons. DAU reform should not be viewed as a panacea for improving teacher hiring practices at the district level, however addressing the existing DAU issue is an important step toward holding districts accountable for their decisions and improving overall education spending efficiency. Transferring responsibility for paying professional and functional allowances from the central to local governments would be an additional step toward improving efficiency, allowing local governments to bear the real cost of teacher employment. 62 Kluyskens and Firdaus (2009), Teacher management: Recruitment, Selection and Data, Probation and Transfer. 63 For example, quotas for teachers could be based on national staffing standards and district school age populations rather than local government assessments of teacher need. U Improving the Quality of Spending on Basic Education Figure 70: Percentage of primary school teachers that would need to be transferred to comply with joint decree 30,000 - .-25,000 - 2 no. of existing teachers to move with n districts 0 S20,000 no. of exist ng teachers to move between districts in same province 15,000 10,000 0 5,000 I < L.hLI...l< <..... < < < < < < Z Z < < D < m < ~a t am D < DD