Indonesia Spending more or spending better: Improving education financing in Indonesia Extended Executive Summary 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. 73358-ID Spending more or spending better: Improving education financing in Indonesia Extended Executive Summary Spending more or spending better: Improving education financing in Indonesia (Extended Executive Summary) U Table of Contents 1. The education system and the"20 percent rule" 6 Who manages and what and who pays for what in Indonesia's decentralized education system? 7 Drawbacks of the 20 percent rule: budget planning and management 9 Where did resources go? 9 2. The effect of increased spending on education outcomes 15 Explaining the limited effect of increased spending on basic education outcomes 18 3. Improving the quality of spending on basic education 25 Sources 35 B Figure 1: Public Expenditure on Education, 2001-2010 4 Figure 2: Share of enrollment by provider type and level, 2010 6 Figure 3: Public expenditure on education as a share of total government expenditure and as a share of GDP 10 and GDP per capita, selected countries in 2009 10 Figure 4: Distribution of the budget by level of education, 2008 - 2009 12 Figure 5: Change in resources by economic classification, within level of education, 2008-2009 12 Figure 6: Share of budget by major programs, 2006-2009 13 Figure 7: District spending by economic classification, 2001-2009 13 Figure 8: Central government budget for BSM (Scholarships for the Poor Programs), 2008-2010 14 Figure 9: Distribution of education funds by direct recipient, 2009 14 Figure 10: Gross and Net Enrollment Rates by Level 2001-2009, and International comparisons, 2009 15 Figure 11: Share of children enrolled in school by age and quintile, 2006 and 2010 16 Figure 12: PISA results in reading and mathematics, 2009 17 Figure 13: Indonesia PISA scores in mathematics, reading by socioeconomic decile, 2003-2009 18 Figure 14: Growth in number of students and number of teachers by level and student-teacher ratio, 2004-2010 19 Figure 15: Evolution of the number of civil servant (PNS) and non-civil servant teachers by level, 2006-2010 19 Figure 16: Student-teacher ratio, per student spending and Bahasa Indonesia test scores in public primary school 20 Figure 17: Change in public education spending and change in education outcomes at the district level 21 Figure 18: Percentage of total education budget going to salaries 21 Figure 19: Distribution of real wages for teachers, 2010 22 Figure 20: Causal impact of teacher certification on a range of outcome variables 23 Figure 21: The budgetary impact of the certification program 24 Figure 22: Significant teacher redistribution is required to implement the joint decree 28 Figure 23: Per student amount by sources of income at the school level, 2010 30 Figure 24: Students in primary schools with BOSDA perform better (even after controlling for other factors) 31 Figure 25: Per student school income and share of expenditure by category for BOSDA and non-BOSDA schools, 2010 31 Illustration 1: Decentralized education management by level of education 8 Table 1: Teacher characteristics by location, 2010 26 Table 2: Recent teacher staffing regulation 27 Box 1: The BOSDA improvement pilot program 32 B Spending more or spending better: Improving education financing in Indonesia (Extended Executive Summary) 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 (a minimum of) 20 percent of government spending 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 impetus for this report. The large increase in expenditures addressed one of the main concerns of the last Education Public Expenditure Review (PER)' - the lack of resources for education - but the problems with the distribution of resources persist. As a consequence, improving the quality of spending in the education sector is now critical. 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, 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 -- and focuses specifically on the effects of the current financing and governance systems on these factors. Figure 1: Public Expenditure on Education, 2001-2010 Rp Tr Percent 350 - 25 21.9 21.1 300 - 20 250 - 15.9 16.3 15.0 14.4 14.5 ..*** *., .15 200 - .......***** 11.4 150 -10 100 - 50 - 0 0 O 2001 2002 2003 2004 2006 2007 2009 2010 Im Household Public - %Government - Spending % GDP (Public)% - GDP (Total) Source: World Bank staff estimates based on MoF, APBD and BPS data. Notes: Realized expenditure data are unavailable after 2009 as sub-national 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), 1 Indonesia Economic Quarterly, World Bank (2012a) 2 World Bank (2007) Investing in Indonesia's Education, Allocation, Equity and Efficiencyof Public Expenditures B This report is organized as follows. The first three sections show that the vast increase in education resources has been accompanied by a decline in learning outcomes, and thus highlight the urgent need to improve the quality of spending. Section 1 starts by briefly describing the complex financing and governance systems and 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. Section 2 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. In Section 3 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 report 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, the last section concludes and summarizes these recommendations. B Spending more or spending better: Improving education financing in Indonesia (Extended Executive Summary) 1. The education system and the "20 percent rule" 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 includes the management of the system's most important assets: 59 million students, 330,000 schools and close to 3 million teachers.4 While many ministries register spending on education in their budgets,' 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, with public support in the form of civil service teachers (at all levels) and direct school grants (in basic education). While the first 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 due to more 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 2). Figure 2: Share of enrollment by provider type and level, 2010 Tertiary 62 Public under MoNE Senior Secondary 3P u U Public under MoRA Senior Secondary 1 N Other Public Primary U Private 0% 50% 100% Source: SUSENAS 2010. 3 For a more detailed description of the system, see Cerdan-Infantes, P., and Y Makarova, 2013. Spending More or Spending Better: Improv- ing Education Financing in Indonesia. Jakarta: World Bank. 4 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 special needs schools, civil servant and non-civil servant, permanent and contract teachers (MoEC, 2010 and NUPTK, 2011). 5 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. B The education system and the"20 percent rule" 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 different budget sources: some come directly from the central government, some from local governments6 (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. The "20 percent rule.'a 2002 constitutional amendment which stipulates that at least 20 percent of the total state budget must be spent on education, further complicates financing and management. Both central and local government budgets are subject to the rule, which includes budget revisions. This rigidity creates significant distortions in decision-making, as elaborated below. Who manages and what and who pays for 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. Schools have considerable autonomy when it comes to operational, budgetary and programmatic decisions. The central government formulates policy, issues regulations/guidelines and standards at the national level, and still directly controls higher education. 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: Ministry of Education and Culture (MoEC), Ministry of Religios Affairs (MoRA), Ministry of Home Affairs (MoHA), 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. In managing schools and teachers, MoRA's management structure is different than that of MoEC, 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. 6 Throughout this report, both "Local Government" and "Sub-national Government" are terms used to cover both "Provincial Govern- ment"and "District Government" As of November 2012, Indonesia consists of 34 Provincial Governments. A Province consists of several Districts. B Spending more or spending better: Improving education financing in Indonesia (Extended Executive Summary) Illustration 1: Decentralized education management by level of education policisRegCuriculu Provision/ standards andov develpmen service MI at schoolc level delivery NOON U CENTRAL Early Childhood DISTRICT soon I CENTRAL Basic ONE DISTRICT NOON e I CENTRAL Secondary P I moo DISTRICT Mgag Ngag Mgag MgOg CENTRAL Tertiary Source: e aboration 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, scho arship funding, and until 2011 it also administered School Operation Funds (BOS). Provincial governments also provide partial financial assistance in senior secondary education and to specia needs schools. The funding system for the education sector is also complex, involving multiple sources and transfers across various levels of government. Expenditures for education are paid for by central government funds, transfers to local governments, sub-national governments' own-source revenues, and central government spending at the sub-national level that is not recorded in sub-national budgets. Central government transfers are the main source of revenue for district government budgets (APBD). Central government transfers account for 88 percent of district budgets and 44 percent of provincial budgets in 2009. The main transfer to sub-national governments is the Basic Allocation Fund (DAU), which mostly 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 2009.7 The DAU, which is calculated largely based on the salary bill for civil servants in the district or province, implicitly incentivizes civil service hiring An earmarked transfer, the Specific Allocation Fund (DAK) is also allocated yearly through no specific formula, and covers a significant portion of school and classroom reconstruction. 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. 7 Total sub-national revenues include Deconcenctration (Dekon)/Co-administered Task (TP) transfers spentat the sub-national level but recorded in the centra government budget. 8 Until 2008, DAU covered 100 percent of civil servant salaries (at both the provincial and district evels). It was reduced to 89 and 87 percent coverage for provinces and districts respectively, and further reduced in 2009 to 80 and 72 percent. B The education system and the"20 percent rule" Drawbacks of the 20 percent rule: budget planning and management Managing such a large and complex system is clearly a challenge, and a vast increase in resources provides a good opportunity to test the incentives embedded in the way the system is managed and finance. To further complicate the management, the earmarking introduced by the 20 percent rule introduces some perverse incentives in the system. Earmarking undermines the ability of the government to allocate resources efficiently across sectors. It also 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 Gol'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 of the fiscal year that need to be spent on short notice.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. 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. The allocation of these windfalls was clearly detailed in 2010. However since 2011, information on the revised budgets and windfalls has been very limited. 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). The Financial Notes of the 2011 and 2012 Draft Revised Budget documents, however, no longer provide allocation information by program, but offer instead a broad categorization of where additional resources from the budget revisions were directed to9. Perhaps the biggest drawback of the rule and its implicit 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. When resources increase dramatically and in a manner exogenous to the demands of the sector, education planners, pressured to spend the increased 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 civil servant teachers, for instance, it can be difficult to make adjustments in the future. With this complex system in mind, let us look at where resources from the 20 percent rule went. Where did resources go? Before analizing the breakdown of resources it is important to note that despite the fast increase in public spending, Indonesia's government still spends a smaller share of its GDP on education than other middle-income countries. At 3.7 percent of GDP, Indonesia spends less than Thailand, Vietnam or Malaysia in 9 (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. 1 of 2010 Concerning Acceleration of the National Development Priorities, and (4) expand scholarship programs for poor students,The technical criteria for allocating additional funds in 2011 and 2012 are also quite broad indicating that funds can cover 1) areas on teachers' professional allowance payments; 2) programs due to completion at the end of the fiscal year; 3) programs demonstrating outcome improvement; and 4) existing pro- grams B Spending more or spending better: Improving education financing in Indonesia (Extended Executive Summary) the region, and spends half as much as high-income, high-performing countries such as Norway. Indonesia fares well when compared to lower income countries in the region, spending a higher share of GDP on education than Lao, Cambodia and the Philippines (Figure 3a). Relative to its wealth, Indonesia's 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. As seen in Figure 3b, 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. Figure 3: Public expenditure on education as a share of total government expenditure and as a share of GDP and GDP per capita, selected countries in 2009 Figure 3a Thailand Norway Sweden Indonesia Argentina Viet Nam Netherlands Malaysia ____France Malaysia Brazil Brazil Lower-Middle Income Viet Nam Mexico Norway Republic of Korea Philippines Colombia Argentina Germany Chile Sweden Japan Lao Thailand Indonesia Netherlands Lao France Singapore Germany India Philippines Singapore C=_______*_*_* ______ Cambodia 0 5 10 15 20 25 0 2 4 6 8 Share of total Government Expenditure Percentage of GDP Figure 3b 30 - Primary Secondary ~5 15 - 25 0 Source: Jnesco Institute for Statistics, 2008 or latest year (not earlier than 20061; except Indonesia, own calculations using Ministry of Finance central budget and SIKD data. The education system and the"20 percent rule" When private spending is included in the total resources devoted to the education sector, the scale ofthe increose in resources is even more impressive (see Figure 1).Total education spending from public and private sources went from about 3 percent in 2001 to over 5 percent of GDP in 2010. In fact, household spending on education grew faster than public spending, resulting in a larger contribution of households to education spending despite the massive increase in public resources. Households went from directly contributing 20 percent of the total resource envelope in 2001 to contributing over 30 percent in 2010. This increase was largely due to expansion in non-basic education levels (senior secondary and higher education) which are largely reliant on fees and households instead of public spending. With this increase in resources, one would expect drastic improvements in education outcomes. However, it is well known that increases in resources do not necessarily result in better outcomes, especially quality of education (Patrinos 2012). Before examining trends in education outcomes, it is thus important to see where the resources went. Unfortunately, the complexity of the system's management is matched by the difficulty in gathering reliable spending data. District's spending autonomycoupled until recentlywith relatively rudimentary reporting mechanisms meant that detailed district spending data was unavailable before 2008. More reliable district spending data was recently made available which allows, with some caveats, a detailed analysis of spending trends from 2008. ")These data cover realized spending in 2008 and 2009 for only 70 percent of the over 500 districts in Indonesia, but they do allow us to examine spending trends by level of education, economic classification and even specific programs over the 2008-2009 period. Afirst breakdown ofspending by level of education shows a biastowards basic education (comprising compulsory primary and junior secondary education). Figure 4a shows the share of spending by level of education in 2008 compared to the share of additionol resources that went to each level of education in 2009. Senior secondary and, especially, higher education received a disproportionately larger share of additional resources. Still almost half of additional resources from the achievement of the 20 percent rule went to basic education. On the contrary, early childhood education (ECD), which received an extremely low share of resources in 2008, received an even lower share of additional spending. As a result, senior secondary and higher education received somewhat higher shares of the total budget in 2009, but ECD received a smaller share overall. Comparing Indonesia to other countries in the region it is clear that this composition will have to change in the future (Figure 4b).The share of the budget going to non-basic education will likely increase more rapidly in the near future, in part through the natural expansion of ECD and post-basic education. However, a conscious effort to increase spending in early childhood education may be needed in light of the extremely low public spending in Indonesia and the large returns ECD has been shown in the literature both internationally and in Indonesia (World Bank 2013). 10 An important caveat of the data is in the salary classification.The data do not allow distinguishing teachers from other public servants working on education (i.e. district officials). It also does not allow the breakdown of teacher salaries by level of education, which we do using the share of teachers by level from administrative data. U Spending more or spending better: Improving education financing in Indonesia (Extended Executive Summary) Figure 4: Distribution of the budget by level of education, 2008 - 2009 4a Share of additional resources by level 4b Share of the budget spent on non-basic education 100% -80%- 90% - mOther mPre-Primary mSecondary Tertiary programs 70% - 80% - 70% - 0 Universities 60%- 60% - 50% - 50% - Senior Sec. 40%- 40% . education 30% - E Basic Education 30% - 20% - 20%- 10%- 10% -H ECD 10% - 0% - Share of 2008 Share of Additiona 0 I Budget Resources in 2009 Indonesia Thailand Philippines Korea Mexico Brazil Malaysia Source: Indonesia Ministry of Finance Source: World Deve opment Indicators But where did money go within each level of education? The allocation of additional funds differed significantly by level of education. On the one hand, in higher education, the share going to salaries decreased, while the share going to capital, and goods and services increased rapidly. On the other hand, in basic education and senior secondary, most of additional spending went to salaries and teacher certification. In Figure 5, teacher certification is included in "Social Aid', a category used to classify most specific programs (the bigger ones being, BOS, teacher certification and scholarships). Figure 5: Change in resources by economic classification, within level of education, 2008-2009 60% - 49% 50% - Social Aid 40% - MCapital 299' 30% - 20% - IGoods& 14% 10% Services 10% - Saary 0% - -Total -10%- ECD Basic Senior Higher Ed Other Secondary Source: (World Bank 2012) Taking the overall budget, about 60 percent of additional resources went to teacher salaries and the teacher certification program (Figure 6). Meeting the 20 percent rule in 2009 implied an increase of 5.5 percentage points of governmend spending, of which more than 3 went to salaries and teacher certificationThe teacher certification program almost doubled the pay for certified teachers through an allowance equivalent to a teacher's basic pay. The certification allowance absorbed more than 2 percent of the state budget in 2009 even though only 30 percent of teachers were certified.The increase in spending on teacher salaries was driven mostly by increases in the total number of teachers, but the "regularization"of contract teachers to civil servant status also played a part. U The education system and the"20 percent rule" Figure 6: Share of budget by major programs, 2006-2009 25 - 20 - E 15 - E NMBOS > Universities 1 Teacher certification MTeacher salaries 0 Rest of programs 5 0~ 2006-08 2009 Source: (World Bank 2012) In the context of decentralization, where districts are responsible for teacher management (including paying salaries) these patterns of expenditure mean that districts spend on salaries at the expense of other things. This is obvious when looking at district budgets. The share of spending on education that went to staff remuneration increased from about half in 2001 to almost 80 percent in 2009. While the share destined to capital increased in the 2006-2008 period, it decreased in 2009 (Figure 7). Salary spending also crowds out direct support to schools, which are also owned and managed by districts. In 2010, about half of public schools in basic education reported not receiving any financial support from districts. In light of the positive effects that these districts funds have on student performance, this is problematic (World Bank 2012). Figure 7: District spending by economic classification, 2001-2009 120 100 0 II m 80 o m Personnel 60 m Goods and Services 0 m Capital E40 m Others 20 2001 2002 2003 2004 2005 2006 2007 2008 2009 Source: SIKD A large share of the increase in social assistance spending went to scholarship programs, including the Government Assistance Program for Poor Students (BSM), which almost doubled between 2008 and 2010. Yet scholarships represent a 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 U Spending more or spending better: Improving education financing in Indonesia (Extended Executive Summary) 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. Figure 8: Central government budget for BSM (Scholarships for the Poor Programs), 2008-2010 4000 3500 3000 2500 2000 Q 1500 1000 500 0 2008 2009 2010 * Basic Edu. (for poor) 0 Senior Sec. (for poor) Higher Edu (for poor) I Higher Edu. (others) Who were the direct recipients of increased education funding in 2009, after the 20 percent rule was met for the first time? Figure 9 describes the allocation of education spending among teachers, students, schools and education administrators (staff). 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). 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 9 shows, unspecified expenditures, most of which are sub- national expenditures, make up a significant share of total education spending. Figure 9: Distribution of education funds by direct recipient, 2009 Teachers & staff g e (salary & (nus2yc Allowances) 56% ment, 3% Others, 0,2% Students 2% Teachers & staff (training & management) 4% Source: MoF The effect of increased spending on education outcomes 2. The effect of increased spending on education outcomes The biggest payoff from this increased spending has been an expansion in access to education, especially for the poor. The progress in enrollment rates in the last decade has been impressive. Indonesia has achieved universal primary education and has made great advances in secondary and higher education, with an increase of 10 and 7 percentage points in the gross enrollment rates (GER) respectively. Perhaps the most impressive achievement has been in early childhood education (ECD), which now reaches half of 4 to 6 year olds (up from 25 percent a decade earlier). The improvements in GER have brought Indonesia closer to its neighbors, resulting in a higher than expected senior secondary enrollment rate for its level of GDP per capita. By 2009, Indonesia's enrollment rates profile mirrored that of China, with higherthan expected secondary education enrollment rates for its level of income but still lagging in higher education and ECD. Figure 10: Gross and Net Enrollment Rates by Level 2001-2009, and International comparisons, 2009 120% 100% Korea, Rep. 80% Philippines Indonesia 60% China 40% Thailand Malaysia 20%Brazil 0% . Mexico ECED Primary Jun enior Tertiary Secondary Secondary 0.0 20.0 40.0 60,0 80.0 100,0 120,0 m 2001 m2002 02004 w2005 w2006 m2007 m 2009 m2010 Gross enrollment rate Gross Enrollment Rate m Net Enrollment Rate tertiary mSecondary ECID Source: Susenas various years, Source: World Bank edstats, last year available, Note: Data for 2009 except Malaysia, Philippines, Argentina and Brazil (2008) The progress in access and equity has been especially rapid over the last 5 years, with children from poor families enrolling earlier and staying in school longer. The share of 15 year olds from the poorest consumption quintile who enrolled in school increased from 60 to 80 percent between 2006 and 2010 (Figure 11). However, the equity agenda is still incomplete: beyond the age of 15, the share of children enrolled from the poorest quintile drops dramatically, and by higher education, falls to less than 2 percent. This drop is even larger for the richest quintile. Only 40 percent of 19 year olds from the richest quintile were enrolled in school in 2010, a share that, worryingly, has not changed since 2006. U Spending more or spending better: Improving education financing in Indonesia (Extended Executive Summary) Figure 11: Share of children enrolled in school by age and quintile, 2006 and 2010 100% - 100% 90% -2006 9Do 80% -2010 80%- 70% 709/ 60% 0 60o. S 50% 50/o 40% 40% 30% - 30% 20% I 20%/. I I RichestQuintile I 10% PoorestQuintile 0%'S 1 Rcet intil 0 - 0%tI 6 7 8 9 10 11 12 13 14 15 16 17 18 19 2072T 22 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 Age Age Source: Own ca culations using Susenas, 2006 and 2010 The effect of the increased spending on the quality of education has been less dramatic. Indonesia's performance in international tests has been disappointing, generally scoring at the bottom of international assessments of learning achievement (TIMSS, PIRLS and PISA). In the most recent round ofTIMSS (2011), Indonesian students in 8th grade performed significantly worse in Mathematics than the other participating countries in the region - Thailand, Malaysia and, of course, top scoring South Korea and Singapore. In fact, only five of the 45 participating countries scored lower than Indonesia (Mullis, Martin et al. 2012). The results are almost identical in the science assessments included in TIMSS and reading and literacy tests in PIRLS (Mullis, Martin et al. 2012).12 Not only are average scores low, but the share of students achieving the highest levels of performance is very small. In TIMSS 2011, only the top 20 percent performers in Indonesia achieved at least an intermediote score, only 3 percent achieved at the high level and no students scored at the advanced level according to the international benchmarks used (Mullis, Martin et al. 2012).13 In contrast, almost 50 percent of students inThailand and Malaysia (which had similar average scores) achieved at least the intermediate level, 10 percent achieved the high benchmark and one percent were assessed as being advanced. The results are similar in PISA: in the 2009 assessment no student performed at the highest level in math or science in 2009 (OECD 2010). Given recent evidence linking PISA scores and the percentage of top performers with GDP growth, this lack of top performers is a serious concern (Hanushek and Woessmann 2007; Pritchett 2009; Pritchett and Viarengo 2009). Indonesia is below its East Asia Pacific neighbors in PISA, both in Math and Reading. A majority of students did not achieve beyond a rudimentary level of proficiency in reading, math, and science. In 2009, 94 percent of Indonesian students reached only level 2 or below (out of 6 levels), and 44 percent fell below level 1. In science, around 93 percent of Indonesian students fell below level 2, with about a quarter falling below level 1. In reading, 88 percent of students reached only level 2 or below. 11 TIMSS: Trends in International Mathematics and Science Study; P RLS: Progress in Internationa Reading Literacy Study; PISA: Program for International Student Assessment, 12 In reading, only 2 countries perform statistically significantly lower than Indonesia. 13 The microdata is not yet available so we rely on published percentiles in TIMMSS reports stating that the top 5 percent of students perform at less than the high level. U The effect of increased spending on education outcomes Figure 12: PISA results in reading and mathematics, 2009 Math Scores Reading Scores Science Scores n China Shanghai China Shanghai China Shangha Korea iapan Korea Singapore Korea Kora Japan Singapore Aust ia 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 0 200 400 600 800 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 WoBmann (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. But perhaps the most worrisome result is the lack of significant improvements in learning outcomes over the last decade, with the exception of reading. Figure 13 shows the scores on PISA 2003 and 2009, both averages (lines) and broken down by socioeconomic deciles (bars) in math and reading.The first result to point out is that the average scores in 2003 and 2009 are not statistically significantly different in mathematics and science (not shown). On the positive side, scores in reading did improve significantly. These results are reassuringly similar to those in TIMSS and PIRLS, which show the same trends between 2007 and 2011 (lack of significant improvement in 8th grade mathematics and science, and significant improvements in reading). Secondly, when breaking down results by socioeconomic deciles, the inequality in access to quality education becomes apparent when one sees the striking difference in performance between the richest and the poorest students in Indonesia, especially in mathematics. Poor students perform strikingly worse than richer students, and this difference has not improved since 2003. The results in reading are more encouraging, since in addition to improving the average scores, the differences across socioeconomic status narrowed between 2003 and 2009. U Spending more or spending better: Improving education financing in Indonesia (Extended Executive Summary) Figure 13: Indonesia PISA scores in mathematics, reading by socioeconomic decile, 2003-2009 460 m 2003 Math m 2009 Math 460 2003 Read m 2009 Read 440 *.**** Average 2003 ..** Average 2009 440 Average 2003 .... Average 2009 420 420 400 400 380 380 360 * .....---.---..* 360 340 340 320 320 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Socioeconomic Decile Socioeconomic Decile Source: Author's calculations using PISA 2003 and 2009 Explaining the limited effect of increased spending on basic education outcomes To what extent are these trends explained by the way resources are spent? Section 2 showed that the biggest absorbers of additional resources were teacher salaries and teacher certification. One thing is clear from the analysis of expenditure trends: teachers have collectively benefitted from recent increases in government education spending. About half of the Rp. 67 trillion (US$ 7 billion) increase in real spending (2009 constant prices) on education between 2006 and 2009 went to increasing the number of teachers and paying them more through a number of allowances (World Bank 2012). At the same time, this section has shown that these spending patterns have not been accompamied by a significant improvement in learning outcomes. Are these two patterns related? a) What drove the increase 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 (Figure 14a). 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. This rapid teacher hiring has resulted in significant declines in student teacher ratios in basic education (Figure 14b). Existing ratios are well below levels in comparator countries. For example, in 2008 the lower-middle income average student teacher ratio was 26 compared to a ratio of 18 in Indonesia. Existing student teacher ratios are below international benchmarks associated with good education quality. U The effect of increased spending on education outcomes Figure 14: 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 - Numberof teachers 28 26 1 8 - - Number ofstudent 0 24 16 1) 0 1.4- /9/ 1.0 12 1 1 995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2000 2009 2010 0,8 +Primary . Junior econdary Junio nal - Lower middle income primary average (2008) Kndergarten Prmary Secondary Secondary School -Lower middle income secondary school average (2008) Source: Own ca culations using MoEC, various years. Note: Figures do not include madrasahs Source: Ministry of Education and Culture teacher and enrolment data, UNESCO Global Monitoring Report 2011 In addition to the overall increase in teacher numbers, the conversion of contract teachers to civil servants (PNS) contributed to the increase in the salary bill between 2006 and 2010, especially at the non-primary levels. New PNS teachers tend to be recruited from the contract teachers, so a rapid increase in PNS combined with a smaller number of non-PNS teachers indicates conversions of contract teacher positions to civil servants. From 2006 to 2010, the teaching body increased by about 377,000 teachers, a majority of which (60 percent) were previously non-civil servant contract teachers (non-PNS). However, conversion rates varied by education level: 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. Figure 15: Evolution of the number of civil servant (PNS) and non-civil servant teachers by level, 2006-2010 1,200,000 8% m 2006 m 2008 2010 1,000,000 .a 800,000