BANGLADESH EDUCATION SECTOR PUBLIC EXPENDITURE REVIEW Saurav Dev Bhatta (Sr. Economist, GED06) Maria Eugenia Genoni (Sr. Economist, GPV06) Uttam Sharma (Consultant, GED06) Buyant Erdene Khaltarkhuu (Statistician, DECDG) Laura Maratou-Kolias (Consultant, GPV06) T.M. Asaduzzaman (Operations Analyst, GED06) JANUARY 2019 Acknowledgement The report was prepared under guidance of Jaime Saavedra (Senior Director, GEDRR, World Bank), Qimiao Fan (Country Director for Bangladesh, Nepal, and Bhutan, SACBN, World Bank), Keiko Miwa (Director, GEDR2, World Bank), Mario Cristian Aedo Inostroza (Practice Manager, GED06, World Bank), Rajashree Paralkar (Operations Manager, SACBN, World Bank Bangladesh Office) and Tekabe Ayalew Belay (Program Leader, SACBN, World Bank). The peer reviewers for this report were Sangeeta Goyal (Senior Economist, World Bank) and Yoko Nagashima (Senior Education Specialist, World Bank). The authors would like to thank the peer reviewers as well as Tazeen Fasih (Lead Economist, World Bank), Zahid Hussain (Lead Economist, GMTSA, World Bank), Johannes G. Hoogeveen (Lead Economist, GPV06, World Bank), Shwetlena Sabarwal (Senior Economist, GED06, World Bank), Syed Rashed Al-Zayed (Senior Economist, GED06, World Bank), Mokhlesur Rahman (Senior Operations Officer, GED06, World Bank), Shiro Nakata (Senior Economist, GED06, World Bank), Tashmina Rahman (Research Analyst, World Bank), Ruth Hill (Lead Economist, GPV07, World Bank) , Urmila Chatterjee (Senior Economist, GPV06, World Bank), Alan Fuchs Tarlovsky (Senior Economist, GPV06, World Bank), Maria Teresa Balestrini (Consultant, LCRCE, World Bank), and Kelly Yelitza and for their very helpful feedback and suggestions. We would also like to thank Naibur Rahman (Consultant, GED06, World Bank) for research support, and Nazma Sultana (Program Assistant, SACBD, World Bank), Mahtab Alam (Program Assistant, SACBD, World Bank), and Shourov Kumar Sharma (Team Assistant, SACBD, World Bank) for providing necessary logistical support during the course of this study. i Table of Contents Acronyms ...................................................................................................................................................... v Executive Summary...................................................................................................................................... vi 1. Country Context and Motivation .............................................................................................................. 1 2. Overview of the Education Sector in Bangladesh ..................................................................................... 3 3. Recent Trends in Education Outcomes ..................................................................................................... 6 3.1 Overall trends...................................................................................................................................... 6 3.2 Disparities in education outcomes ................................................................................................... 11 4. The Education Budget: Process, Allocations and Utilization .................................................................. 27 4.1. The budget process .......................................................................................................................... 27 4.2 Government expenditures on education: size and trends ............................................................... 28 4.3 Utilization of earmarked resources in education ............................................................................. 31 4.4 Education sector funding sources and shares .................................................................................. 35 5. Equity in Spending on Education ............................................................................................................ 37 5.1 Equity in public spending .................................................................................................................. 37 5.2 Household expenditure on education .............................................................................................. 43 5.3 The role of public spending in total spending................................................................................... 48 6. Relationship between Spending and Outcomes ..................................................................................... 50 7. Conclusions and Recommendations ....................................................................................................... 55 References [to be completed] .................................................................................................................... 60 Annex A: Supplemental Tables and Graphs ................................................................................................ 62 Table A2.1: Student enrollment for different levels of education, 2017................................................ 62 Table A3.1: Linear probability model for being out of school ................................................................ 65 Table A3.2: Reasons for not attending to school by group in 2016........................................................ 66 Table A3.3: Percentage of students who are on or below grade proficiency by urban/rural area 201767 Figure A3.1: Household education expenditures, 2000-2016 ................................................................ 68 Table A3.4: Gender-wise percentage of students in different performance......................................... 68 Figure A3.2: Gender disparity in band ≥ 6 .............................................................................................. 69 Figure A3.3: Gender disparity in Band≤ 2 .............................................................................................. 69 Table A3.5: Percentage of students in different performance bands by urban/rural area.................... 70 ii Figure A3.4: Percentage of Students on Band Level ≥ 6 ......................................................................... 70 Table A3.6: Median Monthly Expenditures by type, 2016 (in Takas) ..................................................... 71 Table A6.1: OLS Regression between primary level outcomes and spending primary level .................. 72 Table A6.2: OLS Regression between net attendance rate in secondary and spending ........................ 73 Table A6.3: OLS regression between primary level learning outcomes and spending........................... 74 ANNEX B: School Stipend Programs in Bangladesh .................................................................................... 75 List of Tables Table 2. 1 : Number of students and institutions by level of education Level ............................................. 3 Table 3.1 : Internal efficiency indicators (2005-17) ...................................................................................... 7 Table 3.2 : Share of students in different performance bands in NSA 2011, 2013, 2015, and 2017 ........... 9 Table 3.3: Percentage of students in different performance bands by grade in LASI 2015 ....................... 10 Table 3.4 :Average annual wage by education level................................................................................... 10 Table 3.5: Rate of return to additional year of education at different levels ............................................. 11 Table 3.6: Characteristics of children out of school.................................................................................... 21 Table 4. 1: Percentage of total allocated budget for education that is spent ............................................ 32 Table 4. 2: ADP expenditure in different months (2011-2016) .................................................................. 32 Table 4. 3: MoPME budget and MTBF 2010/11 – 2015/16 ........................................................................ 33 Table 4. 4: Original and revised budgets in MoE and MoPME, 2012/13 - 2016/17 (in million Taka) ........ 34 Table 4. 5: MoPME Budget Execution Rates for 2011/12 – 2015/16 ......................................................... 35 Table 4. 6: Education sector financing—sources and shares...................................................................... 36 Table 4. 7: Contribution of domestic resources towards ADP financing (in ‘0000000 Taka) ..................... 36 Table 5. 1: Share of public education expenditure by level, 2015-16 ........................................................ 37 Table 5. 2: Per student public spending on education by level .................................................................. 38 Table 5. 3: Government expenditure per student (% of GDP per capita)................................................... 38 Table 5. 4: Incidence of Public Education Expenditure ............................................................................... 41 Table 5. 5: Characteristics of students receiving stipends in 2016 ............................................................. 41 Table 5. 6: Characteristics of students receiving tuition waivers ............................................................... 42 Table 5. 7: Distribution of household education expenditures education level, 2016............................... 46 Table 5. 8 : Total education expenditures per student, median takas per month ..................................... 49 Table 6. 1: Division level learning outcomes in primary education grouped by per-student expenditure and grade-level proficiency......................................................................................................................... 52 List of Figures Figure 3. 1: Gross and net attendance rates by education level (2000-2016).............................................. 6 Figure 3. 2: Completion of primary and secondary schooling across age groups (2016) ............................. 8 Figure 3. 3: Percentage of grade 3 and grade 5 students at/above grade level proficiency in different subjects (2011-17)......................................................................................................................................... 9 Figure 3. 4: Net attendance rates by gender .............................................................................................. 12 iii Figure 3. 5: Female-male difference in school completion by level (2000-16)........................................... 12 Figure 3. 6: Primary and secondary school completion across age groups by gender (2016) ................... 13 Figure 3. 7: Net attendance rates by consumption quintile, 2000-2016 .................................................... 14 Figure 3. 8: School completion across age groups by poverty status (2016) ............................................. 15 Figure 3. 9 : Net attendance rates by division, 2000-2016 ......................................................................... 16 Figure 3. 10: Share of 15-25 year olds who have completed different levels of schooling ........................ 17 Figure 3. 11: Variations across districts in access and internal efficiency indicators at the primary level (2016) .......................................................................................................................................................... 18 Figure 4. 1: Education’s share in total government budget and GDP in South Asian countries, 2000-2015 .................................................................................................................................................................... 28 Figure 4. 2: Budget allocation, development and non-development budget share of the education sector .................................................................................................................................................................... 29 Figure 4. 3: Budget allocation and development and non-development budget share, MoPME .............. 30 Figure 4. 4: Budget allocation and development and non-development budget share, MoE ................... 31 Figure 5. 1: Public spending per student by division .................................................................................. 39 Figure 5. 2: Spending per student across districts ...................................................................................... 39 Figure 5. 3: Relationship between spending per student and poverty at the district level ....................... 40 Figure 5. 4: Household education expenditures, 2000-2016...................................................................... 43 Figure 5. 5: Median expenditures on education per month (in 2016 Takas) ............................................. 44 Figure 5. 6 : Education expenditures by quintile, 2016 .............................................................................. 45 Figure 5. 7: Gini coefficient for household education expenditures, by year............................................. 45 Figure 5. 8: Distribution of expenditures by type across consumption quintiles, 2016 ............................. 47 Figure 5. 9 : Share of public expenditure in total education expenditure, by quintile ............................... 49 Figure 6. 1: Relationship between access and spending per student, 2014............................................... 50 Figure 6. 2: Relationship between internal efficiency and spending per student at the primary level, 2014 ............................................................................................................................................................ 51 Figure 6. 3: Relationship between student learning outcomes and spending per student at the primary level ............................................................................................................................................................. 53 Figure 6. 4: Student to teacher ratio........................................................................................................... 54 Figure 6. 5: Student-teacher ratios at the district level .............................................................................. 54 iv Acronyms ABES Annual Bangladesh Education Statistics ADB Asian Development Bank ADP Annual Development Plan AOP Annual Operation Plan APSC Annual Primary School Census BANBEIS Bangladesh Bureau of Education Information Statistics CGA Controller General of Accounts DME Directorate of Madrasah Education DPE Directorate of Primary Education DSHSE Directorate of Secondary and Higher Secondary Education DTE Directorate of Technical Education FY Financial Year GAR Gross Attendance Rate GDP Gross Domestic Product GoB Government of Bangladesh HIES Household Income and Expenditure Surveys HOI Human Opportunity Index HSC Higher Secondary Certificate IBAS++ Integrated Budget and Accounting System LASI Learning Assessment of Secondary Schools MTBF Medium Term Budget Framework MEW Monitoring and Evaluation Wing MoE Ministry of Education MoF Ministry of Finance MoPME Ministry of Primary and Mass Education MPO Monthly Pay Orders NAR Net Attendance Rate NGO Non-Governmental Organization NU National University OECD Organisation for Economic Co-operation and Development OLS Ordinary Least Squares PC Planning Commission PER Public Expenditure Review PESP Primary Education Stipends Program SEQAEP Secondary Education Quality and Access Enhancement Project SESIP Secondary Education Sector Investment Program SSC Secondary School Certificate TVET Technical and Vocational Education and Training UGC University Grants Commission UNESCO United Nations Educational, Scientific and Cultural Organization WDI World Development Indicators v Executive Summary Background Adequate investment in human capital development is critical for enabling Bangladesh to reach its goal of becoming an upper middle-income country. Bangladesh, currently a lower-middle country with an annual per capita gross national income (GNI) of USD 1,470 (WDI 2019), aims to attain upper- middle income status by 2021 and eliminate poverty by 2030. Recognizing the importance of investing in education for building its human capital base, the government of Bangladesh (GoB) has been allocating a large portion of the national budget to the education sector each year during the past two decades. Effective utilization and equitable distribution of allocated public spending is important for ensuring adequate progress in education outcomes. This report analyzes major spending and outcomes trends in the overall education sector in recent years, with a focus on primary and secondary education. Responding to the recommendation of the 2015 Bangladesh Public Expenditure Review Update for more analytical work on public spending in different sectors, including education, the current study analyzes the trends in major education expenditures, access to education, quality of education, and disparities in education outcomes in the past two decades. It also looks at the composition of education expenditure, consistency between budget allocations and actual expenditures, equity in education spending, and potential links between spending and key educational outcomes. Because of data limitations, this report focuses mainly on primary and secondary education. It is expected that this analysis will add to the literature on investments in the Bangladesh education sector, and inform discussions on identifying policy priorities and making resource allocation decisions in the sector. The analysis presented in this report utilizes data from household income and expenditure surveys, public expenditure datasets, and student learning assessments. The data used in this report come mainly from the following sources: Bangladesh Household Income and Expenditure Surveys (HIES 2000, 2005, 2010, and 2016), Bangladesh public expenditure BOOST dataset (2014), National Student Assessment (NSA 2011, 2013, 2015, and 2017), and Learning Assessment of Secondary Institutions (LASI 2015). Other data sources include the Annual Primary School Census (APSC), Annual Bangladesh Education Statistics (ABES), and Annual Budgets. Findings and recommendations Trends in education outcomes During the past two decades, Bangladesh has made impressive progress in expanding access to education at all levels and increasing the internal efficiency of the school education system. The HIES data show that the net attendance rates (NAR) at all levels of education increased substantially between 2000 and 2016 (Figure E1). At the same time, the drop-out and repetition rates decreased and the survival rates increased at the primary and secondary levels, reflecting an overall increase in the internal efficiency of the school system. Individuals from the younger generations are now significantly more likely to have completed both primary and secondary schooling compared to individuals from older age groups (Figure E2). As a result, the education profile of the overall adult population changed substantially during this period: the percentage of adults who had completed at least primary schooling increased from 30% to 43%, and the percentage of adults who had completed secondary schooling increased from 8% to 13%. The key remaining challenges in access are twofold: at the primary level, bringing the remaining out of school children into regular schooling; and increasing overall access to vi higher levels of education, especially at the tertiary level. There is also a need to address the problem of high student drop-out, especially at the secondary level. Figure E1: NAR by education level Figure E2: Schooling completion across (2000-2016) age groups (2016) 120 93 100% 100 72 78 83 72 80% 80 64 60% 60 50 54 40% 40 20% 16 0% 20 7 12 0 Age 15-19 Age 20-25 Age 26-30 Age 31-40 Age 41-50 0 Primary Secondary Tertiary Primary Junior Secondary School Secondary School Linear (Primary) 2000 2005 2010 2016 Linear (Secondary School) Source: Authors' calculations using HIES 2000, 2005, 2010 and 2016/17. Near-gender parity in access has been achieved at the primary and secondary levels, with males lagging slightly behind females; on the other hand, the tertiary NAR is significantly lower for females and the male-female gap is growing. As indicated in Figure E3, there is a continuing need to focus interventions on increasing enrollment and retention of males at the primary and secondary levels. However, the bigger challenge is addressing the growing male-female gap in access at the tertiary level. This will require a concerted effort at the policy level to gain a better understanding of the constraints faced by women in accessing tertiary education, and implementing interventions targeted towards them. Figure E3: Female-male difference in Figure E4: Tertiary NAR by NAR (2000-2016) consumption quintiles (2000-2016) 0.20 0.16 0.12 0.13 50% 0.09 0.10 0.03 0.05 0.02 0.02 34% -0.01 -0.03 0.00 -0.05 25% -0.11 -0.10 5% 0% -0.20 2000 2005 2010 2016 2000 2005 2010 2016 Poorest quintile 2nd quintile 3rd quintile Primary Secondary Tertiary 4th quintile Richest quintile Source: Authors' calculations using HIES 2000, 2005, 2010 and 2016/17. Disparities in access across income groups and geographical regions have declined at the school level, but not at the tertiary level. School attendance rates across different consumption quintiles and across divisions have converged over the years, especially at the primary level. However, the attendance rates at the tertiary level across quintiles have diverged since 2000 and show no signs of convergence in recent years (Figure E4). The disparity in tertiary attendance rates across divisions has also increased over the years. These findings indicate that more targeted support to the poor and to lagging regions is needed to help equalize opportunities for accessing tertiary education. Analysis using the human opportunity vii index (HOI) for attendance rate finds that the main circumstances leading to disparities in school attendance (especially at the secondary level) are household economic status and the education levels of adults in the household. Hence targeted support to poor households would help enhance equitable access to primary and secondary education as well. While access to education has improved over time, the quality of education, as reflected in student learning outcomes, remains low and has be declining. The NSA results for 2017 show that in grades 3 and 5, significant percentages of students are performing below their grade levels in Bangla and Mathematics (Figure E5). The results are especially poor for grade 5, with only 17% of the students achieving grade-level proficiency in Math and 12% achieving grade-level proficiency in Bangla. The LASI 2015 findings for grade 6 and 8 students also show significant percentages of children performing below grade level in language1 and Math. Equally alarming is the fact that, except in grade 3 Bangla, the performance of students in both grades 3 and 5 declined between 2011 and 2017. As for disparities in learning outcomes, the NSA and LASI data indicate that while gender differences in outcomes are small, there are significant disparities in learning outcomes across both divisions and districts, especially in Math. Improving student learning outcomes, and reducing disparities in these outcomes, are now the biggest challenges facing the Bangladesh education sector. Figure E5: Percentage of grade 3 and grade 5 students at/above grade level proficiency in different subjects (2011-17) 80% 75% 74% 70% 68% 68% 60% 57% 50% 50% 40% 41% 41% 30% 32% 29% 25% 25% 20% 17% 10% 10% 12% 0% 2011 2013 2015 2017 Year Grade 3 Bangla Total Grade 3 Math Total Grade 5 Bangla Total Grade 5 Math Total Public spending on education Though Bangladesh has been making substantial investments in the education sector, the education budget share has remained low by international standards and shows a declining trend. The education expenditure in current prices increased by 36.3% per year on average between FY 2002/03 and FY 2016/1. However, in 2015, the education budget as a share of GDP and total government budget was only 2.2% and 11.7%, respectively. Not only are these figures lower than the corresponding shares for all other countries in the region except Sri Lanka, but they are also substantially smaller than the budget shares (15%-20%) recommended by the 2015 Incheon Declaration to ensure adequate investments in the sector for improving the quality and quantity of education services. 1 The subjects included in LASI 2015 were Bangla, English and Math. viii Public spending per student has also been increasing but remains relatively low as a share of GDP per capita. In nominal terms, spending per student has increased for all levels of education in recent years (Table E1). Despite this increase, however, spending per student as a percentage of GDP per capita remains low compared to the OECD average and figures for neighboring countries (Table E2). Hence, ensuring adequate financing remains an unfinished agenda in the Bangladesh context, especially as the nation strives to become an upper-middle income country. There is clearly a need to increase the education budget share and spending per student to international standards. Table E1: Annual per student public Table E2. Public spending per student (% of spending on education by level (Takas) GDP per capita) Level of Education 2010-11 2015-16 Country Primary Secondary Tertiary Primary (Gr. 1-5) 4,728 7,213 Bangladesh 9 10 25 Bhutan 14 32 55 Junior Secondary 4,788 6,497 India 10 17 49 (Gr. 6-8) Maldives 15 .. 29 Secondary (Gr. 9-10) 8,578 9,598 Nepal 13 11 25 Pakistan 10 11 27 Higher Secondary Sri Lanka 11 11 30 17,100 20,872 (Gr. 11-12) OECD members 20 23 26 Source: Authors' calculations using BANBEIS data; figures in nominal Takas. Source: WDI circa 2016. Most of the government budget for education is allocated to revenue or non-development expenditure each year, leaving limited resources for development spending. While both non- development and development budgets include recurrent as well as investment expenditures, development budget primarily includes investment expenditures while non-development budget consists mainly of recurrent expenditures. Development spending is, therefore, particularly important from the perspective of implementing focused education interventions, including those aimed at quality improvements. Typically, over 70% of the education budget is allocated to non-development expenditure each year (Figure E6). Furthermore, because of their inability to fully utilize the allocated development budget, both MoE and MoPME have tended to reallocate part of their original development funds to the non- development component during annual midyear budget revisions each year. This points to a need for a more effective planning and budgeting process, that adequately takes into account the capacity constraints in the sector. There is a bunching of development spending in the last trimester of the fiscal year and in June as in other sectors. The overall budget utilization rate in the education sector is high, and typically over 90% of the annual budget allocated to the education sector is spent each year. However, both MoE and MoPME generally spend over 50% of their development budgets in the last quarter (March-June) of the fiscal year, with disproportionately high expenditures taking place in the very last month (June). The bunching of expenditures in the last trimester is likely a result of different factors including the slow release of the budget and inefficiencies in program/project implementation. This practice is not conducive to effectively implementing interventions involving activities that need to be undertaken earlier in the fiscal year. Figure E6: Budget allocation and development and non-development budget share, education sector (2002/3-2016/17) ix Source: Authors’ calculations based on Ministry of Finance data Equity in public expenditure on education Findings from a benefit incidence analysis indicate that that pubic spending at the primary level is progressive. In 2016, while 30.7% of the primary age children were classified as poor, they received 35% of the public expenditure on education. Furthermore, there is evidence that public spending contributes substantially to reducing the gap in household per student spending on primary education between the rich and poor households. Education spending per student—including private spending—at the household level has been growing in the past few years. In the absence of public spending, the richest quintile would have spent about 7.5 times more per student than the poorest quintile; public spending reduces that gap to 2 times. A key contributor to the progressivity of public spending at the primary level is the Primary Education Stipend Program (PESP). This is evidenced by the fact that the monthly PESP benefit represents about 70% of the monthly private spending [on education] of households in the poorest quintile. Furthermore, there is a significant positive relationship between stipend receipt and district poverty. However, although PESP is now supposed to be cover all children, about 2 in 5 children not receiving stipends are in the poorest quintile. Public spending at the secondary level, on the other hand, is marginally regressive, and there is no statistically significant relationship between stipend receipt and district poverty. This indicates the need for poverty targeting of secondary stipend programs. Relationship between public expenditure on education and education outcomes There does not appear to be any statistically significant correlation between district level public spending per student and the various educational outcomes analyzed in this report. None of the access and efficiency indicators at the primary level is significantly associated with district level public spending per student. While the bivariate relationship between spending and access is significant at the secondary level, the correlation disappears when other explanatory variables are taken into account. Similarly, while there appears to be a positive association between public spending and primary level x learning outcomes at the division level, analysis using district level data finds no evidence of a relationship. It should, however, be noted that this absence of a significant correlation between public spending and the various educational outcomes could be due to the highly aggregate nature of district level information and the limited degrees of freedom available when using district level data. An analysis using school-level data across time would be needed to get a better picture of the links between spending and education outcomes. Despite the increasing investments in education in absolute terms, learning outcomes at the primary and secondary have not improved since 2011, highlighting the need for public spending to be more quality focused. Analysis of the reasons behind why out-of-school children are not attending school also indicates that the quality of the learning environment plays a key role in student retention. The key areas where investments could have a direct impact on learning outcomes, especially at the primary level, include the following: (i) ensuring that teachers are recruited in sufficient numbers and deployed rationally across schools; (ii) expanding access to quality early childhood development programs to improve school readiness at primary school entry; (iii) enhancing early grade reading and mathematics skills; (iv) strengthening examinations and assessments; (iv) improving teacher training; and (iv) strengthening teacher accountability. xi Table E3: Summary of key issues and recommended measures Area Key issues Recommended measures Access and • A large number of primary age children • Ensure effective implementation of ongoing government programs that are wholly internal remain out-of-school focused on or have components focused on out of school children (e.g., the WB financed efficiency • Student drop-out rate is high at the Reaching Out of School Children Project; Quality Learning for All Program) secondary level • Strengthen measures to enhance the quality of the school learning environment to • Access to tertiary education is still limited, enhance retention and reduce drop-outs especially for women, and for students • Provide poverty-targeted scholarships to economically disadvantaged students at the from poor households secondary and higher levels • Further expand financial assistance for women to access tertiary education Public • Inadequate financing: the budget as a • Increase budget share of education to at least 3% of GDP as proposed in the 7th Five expenditure share of GDP, and as a share of overall Year Plan in the short run; increase it to international standards (4%-6%) in the long run on national budget is relatively small; • Widen the tax base, enhance tax compliance, and tighten loopholes to increase tax education spending per student is low revenue; strengthened coordination among DPs, MoE, MoPME, and MoF to mobilize • The share of development budget is low, more external resources and there is a tendency to reallocate • Strengthen the planning and implementation of capacity of MoE and MoPME to ensure development budget to non-development that the budget planning is done more effectively and the planned development budget is budget in mid-year fully utilized • There is a bunching of development • Fully roll out iBAS++ in both MoE and MoPME as soon as possible spending in the last trimester and in June • Strengthen the delivery of the primary education stipends program to ensure that it • Some poor primary level students are not reaches all poor students in practice receiving stipends • Provide poverty-targeted financial support/scholarships to economically disadvantaged • Public spending at the secondary level is students at the secondary and higher levels marginally regressive Quality • Quality of education at the primary and Focus public investments on the following: (i) providing school-based remedial support to secondary levels, as reflected in student low performing students; (ii) ensuring that teachers are recruited in sufficient numbers and outcomes, remains low and uneven deployed rationally across schools; (iii) expanding access to quality early childhood • Public spending does not seem to be development programs to improve school readiness at primary school entry; (iv) correlated with quality improvements implementing focused interventions to enhance early grade reading and mathematics skills, and develop the reading habit; (v) strengthening examinations and assessments; (v) improving teacher training, and teacher support systems; and (vi) strengthening teacher accountability. xii 1. Country Context and Motivation Bangladesh has made remarkable progress during the past two decades in economic growth, and has recently joined the ranks of lower-middle-income countries. The country has achieved steady economic growth of about 6 percent annually since the late nineties. It now has a per capita gross national income (GNI) of US$1,470 (up from US$420 in 2000), which places it among the group of lower-middle- income countries of the world. During this period, the bottom 40 percent of the population has experienced a higher growth in per capita income than the rest of the population, and the poverty rate has declined substantially. The percentage of people living below the national poverty line dropped from 48.9 percent in 2000 to 24.3 percent in 2016 (WDI 2019). At the same time, Bangladesh has made impressive gains in key areas of human development, including health and primary education. Child and maternal mortality, as well as fertility rates, have decreased substantially since 2000, immunization coverage has improved, and the incidence of communicable diseases has decreased. There has also been some progress in reducing child malnutrition, though a significant percentage (around 36%) of children under 5 years of age remain stunted. Similarly, equitable access to primary education has improved drastically: around 98% of primary-age children are enrolled in school, and gender parity in enrolment has been achieved at both primary and secondary levels. While these achievements provide a solid foundation on which Bangladesh can build to reach its goal of becoming an upper- middle-income country by 2031, the country is at a crossroads. In the years ahead, Bangladesh is expected to benefit from a demographic dividend resulting from a higher share of working-age population and a declining dependency ratio. However, with 88.5 percent of the labor force in informal employment, and 41 percent of workers with no education at all, the economy is at risk of falling into a low-productivity/low-wage trap. The majority of the youth are stuck in low-wage, labor intensive and insecure informal work. This is especially the case for poor youth, who often do not have the required education and skills needed for accessing remunerative, formal wage employment. There is thus a pressing need for the country to systematically upgrade its human capital. However, given that resources allocated to the social sectors such as education are likely to remain scarce in the face of competing development priorities and trade-offs, it is critical to develop education programs that are properly targeted to address both short- term vulnerabilities and long-term investment requirements. Accordingly, it is essential to take stock of the progress made in recent years to inform the design and implementation of programs that will equip the youth of tomorrow with the necessary skills to steer the country to the next stage of economic and social development. This Public Expenditure Review (PER) of the education sector in Bangladesh analyzes major spending and outcomes trends in the overall education sector in recent years, with a focus on primary and secondary education2. It reviews trends in major education expenditures, access to education, quality of education, and disparities in education outcomes in the past two decades. It also looks at the composition of education expenditure, consistency between budget allocations and actual 2 Note: This PER is part of a larger set of studies being undertaken under a World Bank Programmatic Advisory Services and Analytics activity entitled “Expanding Quality Basic Education for All in Bangladesh”. Oth er key studies in this set cover the following topics : profiles of out-of-school children and reasons why they are out of school; gender parity in basic education; education service delivery in urban slums; citizen engagement mechanisms in the education sector; and trends in determinants of student learning outcomes. 1 expenditures, equity in education spending, and potential links between spending and key educational outcomes. Thus this PER seeks to shed light on the extent to which resources are used effectively and equitably in the education sector. Since it also outlines some of the key challenges confronting the education sector, it is hoped that the analysis will inform discussions on the way forward among various stakeholders and assist the Ministry of Primary and Mass Education (MoPME) and the Ministry of Education (MoE) in formulating policy priorities and making decisions related to resource allocation and utilization. This PER takes advantage of various data sources to gain a deeper and better understanding of current issues in education. The 2015 Bangladesh Public Expenditure Review Update, which provides an update of the fiscal trends based on the 2010 Bangladesh Public Expenditure and Institutional Review, points to the need for more analytical work to gain a deeper and better understanding of current issues in different sectors, including education. The current study attempts to respond to this need. It uses data from four rounds of the nationally representative Household Income and Expenditure Surveys (HIES 2000, 2005, 2010, and 2016), the Bangladesh public expenditure BOOST dataset that includes disaggregated public spending (budget) data for 20143, and nationally representative assessments of student learning at the primary and secondary levels4 to analyze different educational outcomes over time and link these outcomes to public spending. Information from the Annual Primary School Census (APSC) and Annual Bangladesh Education Statistics (ABES) from the Bangladesh Bureau of Education Information Statistics (BANBEIS) are also used to gauge the progress Bangladesh has made in various education indicators. Data from Annual Budgets are used to identify major expenditure trends in education, and better understand resource allocations within the education sector. The rest of the report is organized as follows: Section 2 provides a short overview of the education sector in Bangladesh. The recent trends in education outcomes are explored in Section 3, while the education budget process and issues related to education financing are discussed in the following section. The government’s spending patterns in education across different population groups and geographical areas are analyzed in Section 5. Section 6 briefly explores the relationship between public spending on education and education outcomes. Section 7 concludes by summarizing the findings, and presenting some policy recommendations. 3 BOOST Public Expenditure Database, World Bank Group. 4 The two assessments are as follows: 1) National Student Assessment (NSA 2011, 2013, 2015, and 2017); and 2) Learning Assessment of Secondary Institutions (LASI 2015). 2 2. Overview of the Education Sector in Bangladesh The education system in Bangladesh is large and complex. It caters to approximately 17.3 million primary level students (grade1-5), 13.9 million secondary level students (grades 6-12), 4.5 million tertiary level students. These students are served by 133,904 primary level institutions, 34,036 secondary level institutions, and 5,983 tertiary institutions (Table 2.1). Table 2.1 : Number of students and institutions by education level Education Level No. of students (%) No. of institutions (%) Primary (Gr. 1-5) 17,251,568 (49.8 %) 133,904 (77.0% Secondary (Gr. 6-12) 13,878,242 (37.2%) 34,036 (19.6%) Tertiary 4,513,119 (13.0%) 5,983 (3.4%) Overall 34,642,929 (100%) 173,923 (100%) Source: BANBEIS 2018 [Bangladesh Education Statistics report 2018] Note: Both secondary and tertiary levels include technical and vocational education (0.89 million students in 5,897 institutions) There are two ministries responsible for overseeing the education system in the country—the Ministry of Primary and Mass Education (MoPME) and the Ministry of Education (MoE). MoPME handles pre-primary to grade 5, as well as non-formal education, and MoE is responsible for secondary education (grades 6-10), higher secondary education (grades 11-12), technical and vocational education, Madrasah education5, and tertiary education. The Directorate of Primary Education (DPE) is the implementing arm of MoPME. Similarly, the Directorate of Secondary and Higher Secondary Education (DSHSE), Directorate of Madrasah Education (DME), and the Directorate of Technical Education (DTE) manage post-primary education under MoE. While MoE is responsible for policy formulation and allocating funding for tertiary education, the University Grants Commission (UGC) is responsible for coordinating university education, and for quality assurance of both public and private universities. Additionally, the National University (NU) is responsible for overseeing the large number of government and non-government colleges affiliated with it. Different streams of education are offered at different education levels. There are mainly two streams at the primary level: general education (one year of pre-primary and grades 1-5)6, and Madrasah education or ebtedayee, which is equivalent to primary education under the general stream. The vast majority of the students (over 89%) are enrolled in the general education stream. In addition to institutions under the general education and Madrasha streams, there are also a small number of privately managed, relatively expensive, English medium schools offering classes from kindergarten to high school7. The General and Madrasha education streams are offered at the junior secondary (grades 6-8), secondary (grades 9-10) and higher secondary (grades 11-12) levels as well. However, a third stream—the 5 Islamic religious education. 6 In addition, there are separate initiatives aimed at providing educational opportunities for out-f-school children (those who have never enrolled till age 8 or have dropped out of primary). Examples include the World Bank supported Reaching Out of School Children II (ROSC II) project of the government, and different non-government initiatives undertaken by organizations such as BRAC and Save the Children. 7 Unlike schools under the general education stream which follow the national curriculum and generally use Bangla as the medium of instruction, English medium schools follow curricula from other countries such as the Cambridge International Education (CIE) curricula. 3 vocational stream—is also available at the secondary and higher secondary levels. After grade 10, students in the vocational stream have the option of enrolling in either higher secondary vocational education or in four-year diploma programs offered by polytechnic institutes. Of the more than 13.8 million students enrolled at the secondary level, 83.5% are enrolled in the general stream, 13.8% in Madrashas and 2.6% in vocational schools (see Annex A, Table A2.1 for details). Students who have completed higher secondary education in the general stream can pursue advanced degrees in universities or colleges. Similarly, the Madrasha stream offers tertiary level education (bachelor’s and master’s equivalent) for students who have completed higher secondary Madrasha education. Students in the different streams take different completion exams at the end of the secondary and higher secondary levels. At the end of grade 10, students in the general and vocational streams take the Secondary School Certificate (SSC) and SSC-Voc exams, respectively, while those in the Madrasha stream take the Dakhil exam. Similarly, students in the general, vocational, and Madrasha streams take the Higher Secondary Certificate (HSC), HSC-Voc, and Alim exams, respectively, to complete their higher secondary education. Different models of financing and service delivery are used at the primary and secondary levels, with most schools in the primary sector under government management. In 2017, around 56.7% of the 133,904 primary institutions in the country were government primary schools, fully financed and managed by MoPME through its Directorate of Primary Education (DPE 2017b). These schools cater to 77.7% of the primary level students. The remaining primary institutions are mostly non-government funded and privately managed, and are under the purview of other ministries and government authorities, such as the Ministry of Commerce, Ministry of Social Welfare, and the NGO Bureau. In contrast, the majority of the 19,848 schools at the secondary level (grades 6-10) are publicly subsidized and privately managed. In 2017, for example, 98% of the secondary institutions were under private management, and 82% of these non-government secondary schools received Monthly Pay Orders (MPOs) from the government for the payment of teacher salaries (BANBEIS 2018). Furthermore, development partner assisted government projects such as the World Bank financed Secondary Education Quality and Access Enhancement Project and Transforming Secondary Education for Results Operation, and the ADB financed Secondary Education Sector Investment Program and Teaching Quality Improvement in Secondary Education Project have also provided different types of support8 to many of these non-government schools. Most tertiary and TVET institutions are privately managed, but many of them receive government subsidies. Public sector TVET institutions, which enroll around 22% of the TVET student population, are fully financed by the government. However, a large number of private TVET institutions also receive subsidies from the government, mainly in the form of MPOs for teacher salary payments and through grants from donor supported government projects. At the tertiary level, public universities, which enroll around 25.5% of the total student population, are fully supported by government funds received through UGC. Government colleges affiliated to NU enrolling 37.9% of total tertiary level students also receive full funding from MoE. Nongovernment colleges, on the other hand, are largely privately funded and generate around 80% of their income from student fees. But they also have access to some public funds in the form of MPOs for teacher salaries and through donor funded government projects. Students are charged nominal tuition and examination fees in government higher secondary schools, government 8 Examples of such support include infrastructure development (classrooms and WASH facilities), teacher training, and assistance for developing the reading habits of children. 4 TVET polytechnics, public universities and NU affiliated government colleges. Private institutions charge substantially higher fees at all levels. 5 3. Recent Trends in Education Outcomes9 3.1 Overall trends During the past two decades Bangladesh has made remarkable progress in expanding access to education. Over the period 2000-2016, the gross attendance rate (GAR)10 increased from 91 to 114% at the primary level, from 54% to 67% at the secondary level, and from 2% to 25% at the tertiary level (Figure 3.1). Also, over the same period, net attendance rates (NAR) increased by 21 percentage points for primary, 22 points for secondary and 16 points for tertiary education. These figures show that Bangladesh is rapidly approaching universal attendance at the primary level, and has made good progress at the secondary level as well11. There is, however, a continuing need to expand access to higher levels of education. In particular, despite the very rapid growth in tertiary attendance, the attendance rate at this level remains relatively low. Figure 3.1: Gross and net attendance rates by education level (2000-2016) a. Gross attendance rates (%) b. Net attendance rates (%) 120 114 120 101 100 91 92 100 93 83 78 80 80 72 72 67 64 58 54 52 60 60 50 54 40 40 25 19 16 20 10 20 12 7 2 0 0 0 Primary Secondary Tertiary Primary Secondary Tertiary 2000 2005 2010 2016 2000 2005 2010 2016 Source: Authors' calculations using HIES 2000, 2005, 2010 and 2016/17. 9 Sections 3-5 draw from a draft background note entitled “Equity in Education Outcomes and Spending” (Genoni et al. 2018). 10 The HIES only collects information on whether an individual is currently attending school, and not on the enrollment status of the person. Hence, the estimates presented in this report ( attendance rates) are lower than the official enrollment rates based on administrative data. For example, while net enrollment rate (NER) at the primary level for 2016 is 98%, the corresponding NAR presented here is 93%. The gross attendance rate for primary education is defined as the ratio between the number of students attending primary school and the number of students aged 6-10. The gross attendance rates for the secondary and tertiary levels are estimated similarly except that the age ranges used are 11-17 and 18-22, respectively. The corresponding net attendance rate for any level of education is calculated as the percentage of children in the appropriate age range attending school at that particular level. 11 To put the performance of Bangladesh in perspective, it would be useful to compare different educational statistics for the nation with those from other countries in the region. However, international databases do not necessarily have the required data for the different counties of interest at common points in time. For example, the UIS database (UIS 2019) and the WDI database (2019) do not include the net enrollment rates for Bangladesh for years following 1990, even though they do present gross enrollment rates (GER) for these years. According to the WDI database, in 2016, the primary and secondary GERs for Bangladesh were 119% and 69%, respectively. The corresponding GERs for South Asia as a whole were 113% and 71%, and those for the whole world were 104% and 76%. The secondary NER for Bangladesh that year was 63%, slightly higher than the figure for South Asia (60%). 6 The gains in access have been accompanied by improvements in the internal efficiency of the education system (Table 3.1). For instance, between 2005 and 2017, the repetition rate at the primary level declined from 10% to 5.6%, the cycle drop-out rate fell from 47.2% to 18.8%, survival rates increased by more than 50%, and the coefficient of efficiency rose by more than 34%.12 There has also been an overall improvement in these indicators in secondary education as well, though the problem of high student drop-out, in particular, continues to be a major issue at this level13. Table 3.1 : Internal efficiency indicators (2005-17) a. Primary level 2005 2010 2016 2017 Repetition Rate (%) 10 13 6 6 Cycle drop-out rate (%) 47 40 19 19 Survival rate (%) 54 67 82 83 Coefficient of efficiency 0.61 0.62 0.81 0.82 Source: DPE various years [APSC reports]; BANBEIS various years [Bangladesh Education Statistics report] b. Secondary level 2010 2016 2017 Repetition Rate (%) 4 3 3 Cycle drop-out rate (%) 57 37 37 Survival rate (%) 63 65 65 Coefficient of efficiency 0.50 0.73 0.71 Source: BANBEIS various years [Bangladesh Education Statistics report] The sustained increase in access over many years is changing the education profile of the adult population. The literacy rate for adults aged 18+ improved from 43% in 2000 to 60% in 2016.14 In addition, during this period, the percentage of adults who had completed at least primary schooling increased from 30% to 43%, and the percentage of adults who had completed secondary schooling (grade 12 The repetition rate measures the rate at which pupils from a cohort repeat a grade. It is defined as the ratio between the number of repeaters in a given grade in a given school year (t+1) and the number of pupils from the same cohort enrolled in same grade in the previous school year (t). The survival rate is percentage of a cohort of pupils (or students) enrolled in the first grade of a given level or cycle of education in a given schools year expected to reach successive grades, regardless of repetition. This rate is calculated following UNESCO reconstruction cohort model. The coefficient of efficiency is an indicator of the internal efficiency of an educational system. It summarizes the consequences of repetition and dropout on the efficiency of the educational process in producing graduates. It is defined as the ideal (optimal) number of pupil years required (i.e. in the absence of repetition and dropout) to produce a number of graduates from a given school cohort expressed as a percentage of the actual number of pupil years spent to produce the same number of graduates. The coefficient of efficiency therefore ranges from a low of 0 to a high of 1. 13 Note: According to the UIS database, the repetition rate at the secondary level in Bangladesh was 2.16% in 2016, lower than the figures for Nepal and Bhutan but higher than the figures for other South Asian countries. 14 A person is considered literate if she can write a letter. 7 10) increased from 8% to 13%. Progress in schooling is even more evident when different adult age groups or cohorts are compared (Figure 2). For example, while only 36% of the population aged 41-50 had completed primary education in 2016, 83% of the population aged 15-19 had completed this level. Similarly, while 21% of the population aged 41-50 had completed junior secondary, 58% of the population aged 15-19 had achieved this level. These changes reflect the increased opportunities for schooling now available to the younger generations, and the tremendous improvements in both education access and completion made by the country during the past two decades. Figure 3.2: Completion of primary and secondary schooling across age groups (2016) 100% 80% 60% 40% 20% 0% Age 15-19 Age 20-25 Age 26-30 Age 31-40 Age 41-50 Primary Junior Secondary School Secondary School Linear (Primary) Linear (Secondary School) Source: Authors' calculations using HIES 2016/17. Junior Secondary School and Secondary School completion refer to completion of grade 8 and grade 10, respectively. While Bangladesh has performed well in enhancing access to education, the quality of primary education, as reflected in student learning outcomes, has been fluctuating over time without showing an improving trend. According to the government’s National Student Assessment (NSA)15 data, learning outcomes in Bangla language and Math for students in grades 3 and 5 have generally not improved since 2011 (Table 3.2). In the case of Bangla language, only 68% of grade 3 students were found to be at or above grade level proficiency in 2011. This figure increased to 75% in 2013, dropped to 68% in 2015, and increased again to 74% in 2017, indicating the absence of an upward trend in learning outcomes (Figure 3.3). Student performance in grade 3 Mathematics is worse than in Bangla—the proportion of students at or above grade level proficiency in 2011 was already quite low (50%) to begin with; it increased to 57% in 2013 but declined to 41% in both 2015 and 2017. The findings for grade 5 students are even more alarming: in all four years, only about a quarter of these students achieved grade level proficiency in Bangla, and the percentage of students who achieved this level in Mathematics declined from 32% in 2011 to 17% in 2017 while fluctuating between 10% and 32% over the years. 15 The NSA assesses the learning outcomes of a nationally representative sample of grade 3 and 5 students in Bangla language and Mathematics. Temporally comparable NSAs were conducted by DPE in 2011, 2013, 2015 and 2017. The NSA results from the different years can be compared to monitor student learning outcomes over time. 8 Table 3.2 : Share of students in different performance bands in NSA 2011, 2013, 2015, and 2017 2011 2013 2015 2017 Subject Below On/above Below On/above Below On/above Below On/above grade 3 grade 3 grade 3 grade 3 grade 3 grade 3 grade 3 grade 3 Bangla grade 3 32% 68% 25% 75% 32% 68% 26% 74% Mathematics 50% 50% 43% 57% 59% 41% 59% 41% grade 3 Below On/above Below On/above Below On/above Below On/above grade 5 grade 5 grade 5 grade 5 grade 5 grade 5 grade 5 grade 5 Bangla grade 5 75% 25% 75% 25% 71% 29% 88% 12% Mathematics 68% 32% 75% 25% 90% 10% 83% 17% grade 5 Source: DPE 2016a, 2018 Figure 3. 3: Percentage of grade 3 and grade 5 students at/above grade level proficiency in different subjects (2011-17) 80% 60% 40% 20% 0% 2011 2013 2015 2017 Year Grade 3 Bangla Total Grade 3 Math Total Grade 5 Bangla Total Grade 5 Math Total Source: DPE 2016a, 2018 Learning outcomes at the junior secondary level (grades 6-8) also leave much room for improvement. Findings from the government’s Learning Assessment of Secondary Institutions (LASI) 16 show that in 2015, around 23%, 29%, and 6% of grade 6 students performed at the “band 2 or below” level in Mathematics, English, and Bangla, respectively (Table 3.3). The corresponding figures for grade 8 were 5%, 9% and 1%, respectively. “Band 2 or below” is the lowest achievement level in LASI, and represents performance below the grade 6 proficiency level.17 16 LASI 2015 assessed the learning outcomes of a nationally representative sample of grade 6 and 8 students in Bangla, English, and Mathematics. It was conducted by the Monitoring and Evaluation Wing (MEW) of DSHE. Since only one round of nationally representative data are available for LASI, comparison over time is not possible for the secondary level. 17 The design of LASI does not provide cut-off scores for grade 8 level proficiency. Furthermore, it uses a single set of bands for both grades 6 and 8. 9 Table 3.3: Percentage of students in different performance bands by grade in LASI 2015 Subject Band 2 Band 3 Band 4 Band 5 Band 6 or below and above Bangla grade 6 6% 24% 36% 24% 10% English grade 6 29% 36% 20% 10% 5% Mathematics grade 6 6% 14% 24% 33% 23% Bangla grade 8 1% 12% 32% 32% 22% English grade 8 9% 42% 30% 12% 7% Mathematics grade 8 5% 38% 35% 16% 6% Source: ACER 2016 Workers with higher educational qualifications have higher wage rates, indicating that job market views individuals with more education as more capable or higher quality workers. For instance, in 2005, the average annual wage for workers with tertiary education was almost 4.5 times higher than the wage for workers with no schooling, and 1.5 times higher than the average wage for those with secondary level education (Table 3.4). This pattern of higher wages for workers with higher levels of education can be observed in 2010 and 2016 as well. The table also shows that real wage increased significantly for all levels of education during this period. Table 3.4 :Average annual wage by education level Education level 2005 2010 2016 Average annual wage in nominal Taka No schooling 22643 40979 81057 Primary level completed 31588 49471 101804 Secondary level completed 67184 114839 182782 Tertiary education completed 101482 173715 284055 Average annual wage in 2016 Taka No schooling 49815 62288 81057 Primary level completed 69494 75196 101804 Secondary level completed 147805 174555 182782 Tertiary education completed 223260 264047 284055 Source: Authors’ calculations based on HIES data (2005, 2010, 2016/7) The returns to primary and secondary education have decreased over time, but returns to tertiary education have remained consistently high18. As shown in Table 3.5, the private rates of return to each additional year of education are positive for all levels of education, and range from 4% for primary to 20.5% for tertiary. But the rates of return at the primary and secondary levels have been slightly decreasing over time. However, the rates of return to tertiary level education have remained consistently over 20 percent between 2005 and 2016. One interpretation of this finding is that that the labor market in Bangladesh is increasingly putting a greater value on higher order skills and education. Another potential reason for the significantly higher returns to tertiary education compared to the returns to primary and 18 For this analysis, secondary level covers grades 6 to 12. 10 secondary education is that better performing students continue to higher education and most of the less able students drop out of school and start working after just completing primary and secondary schooling. The finding that tertiary education has the highest returns is also consistent with the conclusions derived by Montenegro and Patrinos (2014) in their study comparing returns to schooling globally. Table 3.5: Rate of return to additional year of education at different levels Education level 2005 2010 2016 Primary (grades 1-5) 7.5% 5.5% 4.0% Secondary (grades 6-12) 6.8% 5.4% 4.6% Tertiary (bachelors and above) 20.5% 22.8% 20.5% Source: Authors’ calculations based on HIES data (200, 2010, 2016/17) 3.2 Disparities in education outcomes Along with the progress made in enhancing access and school completion, Bangladesh has seen a reduction in disparities in these outcomes at the primary and secondary levels. Disparities in access and school completion across males and females, income groups and geographical areas are analyzed below. This is supplemented by a brief analysis of the relative importance of the circumstances that explain disparities in access, using the human opportunity index (HOI). The discussion below also includes an analysis of disparities in student learning outcomes between males and females and across geographical areas19. Gender disparities in access and school completion Males lag behind females in primary and secondary school attendance, but are more likely than females to complete tertiary education. In 2016, the primary school attendance rate for girls was 2 percentage points higher than that for boys (Figure 3.4). This gap has not changed much since 2000. At the secondary level, attendance was higher for females in 2016, but the gender difference has been declining with time. In contrast, the rapid increase in tertiary attendance rates has been led by males, which is reflected in an increasing difference in attendance by gender in favor of men. In 2016, the attendance rate for women aged 17-22 was 12% compared to 23% for men. 19 Note: it is not possible to analyze distribution of learning outcomes across income groups due to data limitations. 11 Figure 3.4: Net attendance rates by gender a. By level in 2016 b. Difference between females and males 100% 92% 94% 0.20 0.16 0.13 77% 0.15 0.12 80% 68% 0.09 0.10 0.05 60% 0.03 0.02 0.02 0.05 -0.01 0.00 -0.03 40% -0.05 23% -0.05 20% 12% -0.11 -0.10 0% -0.15 Primary Secondary Tertiary 2000 2005 2010 2016 Male Female Primary Secondary Tertiary Source: Authors' calculations using HIES 2000, 2005, 2010 and 2016/17. Note: The lines in (b) show NER for women minus NER for men. Secondary education refers to grades 6-12. Even though adult females are overall less educated than males, the new generations are reversing this disadvantage. As shown in Figure 3.5, which presents the difference between females and males in the percentage of adults who have completed various levels of school education, gender disparities in school completion have been declining with time. In 2000, 67% of adult women had no schooling compared to 60% of men (7 percentage points difference); by 2016, this difference had narrowed to 3 percentage points. The reduction in gender gap in terms of completing at least primary education has been particularly large—the gap changed from a disadvantage of 7 percentage points for women in 2000 to only 1 percentage point in 2016. Figure 3.4: Female-male difference in school completion by level (2000-16) 0.10 0.08 0.06 0.04 0.02 0.00 -0.02 -0.04 -0.06 -0.08 -0.10 No schooling At least primary At least JSC complete At least SSC complete complete 2000 2005 2010 2016 Source: Authors' calculations using HIES 2000, 2005, 2010 and 2016/17. The bars present the difference between the share of women and share of men achieving an education level. 12 As a result of the faster progress in increasing women’s attendance, young men are now less likely than women to complete primary and secondary schooling, though they are still disproportionately represented at the tertiary level. For the population aged 41-50, the percentage of females who have completed primary or secondary education is significantly lower than the percentage of males. In contrast, for the group aged 15-19, females surpass males in terms of both primary and secondary school completion (Figure 3.6). But the advantage experienced by women progressively declines among the older generations--the percentage of women who have completed primary is lower than that of men starting with the 26-30 year age group, while in the case of secondary education the reversal in advantage starts earlier with the 20-25 year age group. Figure 3.6: Primary and secondary school completion across age groups by gender (2016) a. Primary school completion b. Secondary school completion 100% 87% 100% 80% 80% 79% 60% 60% 40% 43% 40% 29% 29% 31% 20% 20% 17% 7% 0% 0% Age 15- Age 20- Age 26- Age 31- Age 41- Age 15- Age 20- Age 26- Age 31- Age 41- 19 25 30 40 50 19 25 30 40 50 Male Female Male Female Source: Authors' calculations using HIES 2016/17. Note: Secondary completion refers to completion of grade 10. Disparities in access and school completion across income groups The improvement in access to primary and secondary schooling has been broad based, leading to a convergence in attendance rates across different income groups. Children from economically well-off households are more likely to attend school than children from poorer households. In 2000, for example, only 66% of the children aged 6-10 in the poorest quintile were attending primary school, compared to 92% of the children from this age group in the richest quintile (Figure 3.7). However, this gap of 26 percentage points has progressively narrowed over the years. In 2016, there was only an 8 percentage point difference in attendance rates between the two quintiles. As secondary schooling is more expensive than primary, the disparities in secondary school attendance across quintiles are much larger than in primary. Nevertheless, there has been a significant convergence in attendance rates across quintiles at the secondary level as well. At the tertiary level, however, attendance rates across quintiles show no signs of convergence, indicating that the increase in attendance rate since 2000 continues to be largely driven by the top quintiles. 13 Figure 3.5: Net attendance rates by consumption quintile, 2000-2016 a. Primary b. Secondary 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% 2000 2005 2010 2016 2000 2005 2010 2016 c. Tertiary 100% 80% 60% 40% 20% 0% 2000 2005 2010 2016 Poorest quintile 2nd quintile 3rd quintile 4th quintile Richest quintile Source: Authors' calculations using HIES 2000, 2005, 2010 and 2016/17. Note: Quintiles are defined based on household per capita consumption deflated across space to account for differences in the cost of living across 16 different regions. There are also substantial differences in primary school completion between the poor and the non- poor, but they are smaller for the younger generations, suggesting that the gaps have been shrinking over time. For instance, in the 41-50 year age group, there was a 23 percentage point gap (41%-18%) between the non-poor and poor population in the percentage of people who had completed primary schooling (Figure 3.8). On the other hand, the corresponding gap among the 15-19 year olds was only 15 percentage points, pointing to a reduction in disparity across income groups over time. However, the figures for secondary education are less encouraging—the gap between the non-poor and poor in school completion is larger for the 15-19 year age group than for the 41-50 year age group. 14 Figure 3.6: School completion across age groups by poverty status (2016) a. Primary school completion b. Secondary school completion 100% 100% 86% 80% 80% 60% 60% 71% 40% 41% 40% 33% 20% 18% 20% 17% 14% 0% 3% 0% Age Age Age Age Age Age Age Age Age Age 15-19 20-25 26-30 31-40 41-50 15-19 20-25 26-30 31-40 41-50 Non poor Poor Non poor Poor Source: Authors' calculations using HIES 2016/17. Secondary completion refers to completion of grade 10. Disparities in access and school completion across geographical areas There has been a convergence in primary school attendance rates across divisions. In 2000, Dhaka division had the lowest attendance rates (67%) followed by Sylhet division (69%) (see Figure 3.9). These two divisions were far behind Khulna and Barisal, which had 82% and 80% attendance rates, respectively. By 2016, all divisions had substantially higher net attendance rates ranging from 90% in Chittagong to 97% in Khulna. For Sylhet, Rangpur, Rajshahi, Khulna, and Dhaka, gains accelerated after 2010. However, in Barisal, progress slowed down between 2010 and 2016. There has also been a substantial increase in secondary level attendance rates in all divisions, accompanied by a small reduction in disparities across divisions. Between 2000 and 2016, secondary attendance rates increased by more than 1% per year on average across all divisions. The increase in attendance was in general faster for lagging divisions. Nevertheless, in 2016, there were still large differences in attendance rates across divisions, ranging from 64% in Sylhet to 81% in Khulna. Similarly, at the tertiary level, attendance rates have increased across the board starting from less than 1% in 2000. Between 2010 and 2016, Barisal and Sylhet divisions posted the highest increases in attendance rates (over 50%), followed by Chittagong (44%) and Rangpur (40%). In contrast, Dhaka division showed the smallest increase over the 2010-16 period (13%). In sharp contrast to the primary education sector, in particular, these increases in attendance rates have been accompanied by an increased disparity in attendance rates across divisions. One reason for the increase in disparity is that opportunities for tertiary education were quite limited in all divisions in 2000, leading to relatively low but more uniform attendance rates across the board. 15 Figure 3.9: Net attendance rates by division, 2000-2016 a. Primary b. Secondary 100% 95% 95% 85% 90% 75% 85% 65% 80% 55% 75% 45% 70% 65% 35% 60% 25% 2000 2005 2010 2016 2000 2005 2010 2016 c. Tertiary 35% 30% 25% 20% 15% 10% 5% 0% 2000 2005 2010 2016 Barisal Chittagong Dhaka Khulna Rajshahi Rangpur Sylhet Source: Authors' calculations using HIES 2000, 2005, 2010 and 2016/17. Note: Secondary includes grades 6-12. The percentage of adults who have completed different levels of school education has also increased in all divisions. Between 2000 and 2016, the percentage of primary school completers increased by 10 to 16 percentage points in the various divisions. The largest improvement was seen in Sylhet where the percentage of adults completing primary rose from 21% to 37%. Focusing on the younger cohorts (15-25 years old), it can be observed that there has been a reduction in disparities in primary completion across divisions between 2000 and 2016, with more rapid convergence between 2000 and 2005 (see Figure 3.10). The percentage of junior secondary completers also increased, but at more similar paces across divisions. As a result, the reduction in disparity across divisions has been limited. For example, Sylhet — the lowest performing division—continued to significantly lag behind the other divisions in 2016 with only 43% of young adults having completed grade 8. 16 Figure 3.7: Share of 15-25 year olds who have completed different levels of schooling a. Primary school completion b. Completion of grade 8 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% 2000 2005 2010 2016 2000 2005 2010 2016 Barisal Chittagong Dhaka Barisal Chittagong Dhaka Khulna Rajshahi Rangpur Khulna Rajshahi Rangpur Sylhet Sylhet Source: Authors' calculations using HIES 2000, 2005, 2010, and 2016/17. The spatial disparity in access and completion are even more evident when districts are compared. At the primary level, disparities in attendance across districts are small but there is still room for improvement in other internal efficiency indicators. The net attendance rates for 56 of the 64 districts are above 98%. However, in terms of repetition, survival rates and dropout rates, there is significant variation across districts, with some districts performing relatively well but many others still at levels comparable to the national averages from 2005 and 2010. For instance, 25% of the districts still have repetition rates above 8%. Furthermore, survival rates range from 59% to 93%, dropout rates range from 8% to 47%, and about 14% of the districts have dropout rates above 28% (see Figure 3.11). 17 Figure 3.8: Variations across districts in access and internal efficiency indicators at the primary level (2016) a. Net attendance rates b. Repetition rates c. Survival rates d. Dropout rates Source: APSC 2016. While there is little difference in access among districts at the primary level, disparity in access across districts remains an issue of concern at the secondary and tertiary levels. District level secondary NARs range from 59% to 87%, with NARs for 30% of the districts below 71% and 14% of the districts below 65%. Attendance rates at the tertiary level also vary significantly across districts and range from 7% to 31% (see Figure 3.12). 18 Figure 3.9: Net attendance rates in secondary and tertiary education across districts (2016) a. Secondary b. Tertiary 100% 100% 90% 90% 80% 80% Net enrollment rate 70% Net enrollment rate 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% District District Source: Authors' calculations using HIES 2016/17. Progress in access corrected for disparities in attendance: the HOI perspective The HOI20 for attendance rate is an adjusted measure of attendance rate that extracts a penalty for inequities in attendance observed among children living under various circumstances outside their control. When there is 100% coverage of the population (i.e., attendance rate is 100%), the HOI for attendance rate is equal to the attendance rate as there is no disparity in access. On the other hand, the greater the disparities in access across children, the bigger the divergence between the two measures. The discussion below explores the relative importance of the expansion in education access and the distribution of access across children, and also shows how different personal characteristics such as the area of residence or the gender of a child may affect her access to education. Consistent with the expansions in school attendance presented above, the HOI for attendance rate increased between 2000 and 2016 at both the primary and secondary levels. As shown in Figure 3.13, at the primary level, the HOI grew from 68% to 91%, driven both by a rise in coverage/attendance and a reduction in the inequality of coverage across children (seen by the shrinking difference between the HOI and the attendance/coverage rate). At the secondary level, the HOI rose from 40 to 66%, and the gap between the HOI and attendance rate decreased by four percentage points. 20 The Human Opportunity Index (HOI) measures how individual circumstances (i.e. characteristics such as place of residence, gender, and education of the household head) can affect a child’s access to basic opportunities such as education, electricity or water and sanitation. It is a synthetic measure of how far a society is from universal access to an essential good or service, and how equitably access is distributed across distinct groups of individuals (circumstances). The HOI is thus an economic indicator that combines coverage rates and equality in a single measure. The HOI is based on discounting a penalty for inequality of opportunity P from the overall coverage rate C so that: HOI=C-P. The penalty is chosen such that it is zero if all circumstance group specific coverage rates are equal, and it is positive and increasing when differences in coverage among circumstance groups increase. For more information about the HOI see Barros et al 2009; and Barros, Molinas Vega and Saavedra 2010. 19 Figure 3.10: Human opportunity index for attendance rates, 2000-2016 100% 93% 90% 83% 78% 91% 80% 72% 72% 70% 80% 64% 73% 60% 68% 54% 66% Percent 50% 50% 56% 40% 45% 30% 40% 20% 10% 0% 2000 2005 2010 2016 2000 2005 2010 2016 Primary Secondary HOI Coverage Source: Authors’ calculations using HIES 2000, 2005, 2010 a nd 2016/17. Note: Primary school attendance for children aged 6-10. Secondary school attendance for children aged 11-15. Overall, the increase in the HOI was mainly driven by the increase in access (attendance rate), but reduction in disparities also played a role (Figure 3.14). For the primary level, a decomposition of the changes in the HOI across time shows that the increase in attendance rates or coverage explains about 72% of the change in the HOI between 2000 and 2016, and 84% of the change between 2010 and 2016. The remaining change is explained by the reduction in disparities in access. For the secondary level, the rise in attendance explains about three quarters of the change in the HOI. Thus, reductions in disparities in access have played a particularly important role in increasing educational opportunities at the secondary level. Figure 3.14: Decomposition of the change in the HOI a. 2000-2016 b. 2010-2016 Enrollment ages, 11-15 19.73 6.12 Enrollment ages, 11-15 7.46 2.55 Enrollment, ages 6-10 16.86 6.43 Enrollment, ages 6-10 9.05 1.68 0.00 10.00 20.00 30.00 0.00 2.00 4.00 6.00 8.00 10.00 12.00 Coverage Composition Coverage Composition Source: Authors’ calculations using HIES 2000, 2005, 2010 and 2016/17. Primary school attendance for children 6-10. Secondary school attendance for children 11-15. Note: Primary school attendance is for children 6-10, and secondary school attendance is for children 11-15. The change in the HOI between two selected years can be decomposed into the composition effect (due to changes in 20 the distribution of circumstances), and the coverage effect (the contribution of changes in the coverage rates of different circumstance groups). Household resources and the educational levels of adults in the household are the main circumstances behind the disparities in school attendance (see Figure 3.15). In 2006, for primary education, in 2016, the consumption level of the household and the years of education of the household head explain 38% and 31% of the disparities in attendance, respectively. At the secondary level, these two factors together explain 74% of the disparities in attendance rates. The next important circumstance is the gender of the child, which explains 14% of the disparities in secondary attendance. Figure 3.11: Circumstances that explain disparities in attendance rates, 2016 a. Primary b. Secondary 1 Log per capita expenditure 2 2 Years of education of head 7 2 14 30 9 Total number of children (0-17 38 5 years old) 6 Urban/rural location 12 Gender Female household head 31 44 Both parents present in household Source: Authors’ calculations using HIES 2000, 2005, 2010 and 2016/17. Who are the children not attending school? Between 2010 and 2016, there was a remarkable decrease in the number of out-of-school children. In 2010 about 5.5 million children aged 6-14 years were out of school. In 2016, this number was about 3 million, a 45% reduction in 6 years. These children represented 9% of the population aged 6-14 years in 2016 (Table 3.6). A large majority of the out of school children, 6 out of 10, never attended school. Out of school children are more likely to be male and come from the poorest households (Table 3.6). Regression results also highlight that children living in households with less resources and with less educated adults are significantly more likely to be out of school (Appendix Table A3.1). They also confirm that out of school children are more likely to live in urban areas. Table 3.6: Characteristics of children out of school Children 6-14 years old In school Out of school All 91% 9% Area Rural 76% 68% Urban 24% 32% 21 Gender Female 50% 40% Male 50% 60% Quintile 1 22% 35% 2 21% 26% 3 20% 19% 4 19% 12% 5 17% 8% Note: Author's calculations using HIES 2016/17. Even though resources are an important constraint mentioned by households to explain why children are not attending school, the main reasons are lack of interest and age. For children not attending primary, 51% of households state that lack of interest in attending or thinking it is too late to go back to school are the main reasons for not attending (Figure 3.16). Other key reasons include resource constraints (27% of cases), and distance to school from home (11%). Some specific reasons are more relevant for some groups as expected: availability of schools nearby is cited more often in rural areas and cost constrains are cited more frequently by poor households. As in the case of primary education, lack of interest or being too old to go back are the most frequently cited reasons (40%) for not attending secondary school. The need to work is the next major reason (26%), particularly for males (34% of males compared to 14% of females cite this). In the case of women, family chores and marriage are also cited as key reasons for not attending secondary school (30% of women not attending secondary school cite this). For tertiary-aged people not attending school, 40% of respondents cited not wanting to go back or being too old as the reasons. Work (for males) and marriage (for females) follow as the main reasons for not attending tertiary education (Appendix Table A3.2). Overall these results also align with the HOI findings indicating that children living in households with more resources are more likely to attend primary and secondary school. They also suggest that creating a quality learning environment that the children find interesting and engaging could help reduce student drop-out rates. 22 Figure 3.12: Reasons for not attending school by level, 2016 35% 30% 25% 20% 15% 10% 5% 0% Do no want To old to go No No schools Have to Attending For to study back money/too close to work family marriage more expensive home chores Primary Secondary Tertiary Source: Authors’ calculations using HIES 2016/17. Disparities in learning outcomes The NSA data indicate that gender differences in learning outcomes at the primary level are small, and slightly in favor of females in Bangla. As shown in Figure 3.17, the percentages of girls reaching grade level proficiency are similar to those for boys for almost all tested grades/subjects, though the figures are generally slightly higher for females except in grade 3 Math. This near-gender parity in the different subjects and grades can be seen in all four NSA rounds (2011, 2013, 2015 and 2017). Figure 3.13: Female-male difference in percentage of grade 3 and grade 5 students at/above grade level proficiency in different subjects (2011-17) 5 4 Percentage difference 3 2 1 0 -1 2011 2013 2015 2017 -2 -3 Year Bangla Gr 3 GD Math Gr 3 GD Bangla Gr 3 GD Math Gr 5 GD Source: DPE 2016a, 2018 23 The LASI 2015 data show that there is near-gender parity in learning outcomes at the junior secondary level as well except in Mathematics. In both grades 6 and 8, there is little difference between males and females in the percentage of students not achieving grade level proficiency in Bangla and English (Appendix Table A3.4). On the other hand, in the case of Mathematics, a larger percentage of girls than boys are not performing at the grade level in both grades 6 and 8. There are, however, notable disparities in learning outcomes across geographical areas both at the primary and junior secondary levels. The NSA and LASI data show that there are significant disparities in learning outcomes across divisions and districts. For example, in 2017, the percentage of students not achieving grade level proficiency in grade 5 Bangla (Math) ranged from 87% (73%) in the Barisal division to 98% (92%) in the Sylhet division (Figure 3.18). As shown in Figure 3.19, there is a wide variation in outcomes across divisions at the junior secondary level as well, with Sylhet performing significantly worse than other divisions. Figure 3. 14: Percentage of students who are below grade level proficiency in different grades and subjects by division, 2017 98 100 91 92 90 92 87 87 87 89 87 90 86 84 85 82 82 80 73 71 70 64 62 57 57 Percentage 60 56 49 49 50 42 40 31 28 27 30 23 23 23 20 16 10 0 Barisal Chittagong Dhaka Khulna Mymenshing Rajshahi Rangpur Sylhet Name of Divisions Bangla Grade 3 Bangla Grade 5 Math Grade 3 Grade 5 Source: DPE 2018 24 Figure 3.15: Percentage of junior secondary students who are "below band 2" proficiency level in different grades by division, 2015 50 45 40 35 Percentage 30 25 20 15 10 5 0 Barisal Chittagong Dhaka Khulna Rajshahi Rangpur Sylhet Grade 6 Bangla Grade 8 Bangla Grade 6 English Grade 8 English Grade 6 Mathematics Grade 8 Mathematics Source: ACER 2016 The NSA 2017 data show large differences across districts in student learning outcomes at the primary level. The percentage of grade 3 students who are at or above grade level in Bangla varies between 43% and 92% across districts (Figure 3.20). The variation in outcomes across districts is even greater for grade 3 mathematics, with figures ranging from a low of 13% in Cox’s Bazaar to a high of 76% in Barisal. While district-level outcomes are on average lower for grade 5, differences in grade 5 outcomes across districts are nevertheless quite significant in both Bangla and Math. Furthermore, as in grade 3, there are larger variations in the percentage of students achieving grade level proficiency in Math than in Bangla. Figure 3.16: Disparities across districts in the share of students performing at or above grade level a. Proportion of grade 3 students at or b. Proportion of grade 3 students at or above grade level in Bangla by districts above grade level in Maths by districts 1 1 0.9 0.9 0.8 0.8 0.7 0.7 0.6 0.6 0.5 0.5 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 0 0 25 c. Proportion of grade 5 students at or d. Proportion of grade 5 students at or above grade level in Bangla by districts above grade level in Maths by districts 0.45 0.45 0.4 0.4 0.35 0.35 0.3 0.3 0.25 0.25 0.2 0.2 0.15 0.15 0.1 0.1 0.05 0.05 0 0 Source: Authors’ estimates based on NSA 2017 data Note: The horizontal axis represents districts There are also some differences in outcomes across urban and rural areas. At the junior secondary level, the percentages of children in the lowest performance bands (bands 1-3) are higher in rural areas than in urban areas in all tested subjects and grades, indicating that rural areas are at a disadvantage in terms of learning outcomes (Figure 3.21). However, no clear pattern emerges at the primary level when we look at the NSA 2017 data, as Bangla outcomes are better in urban areas while Math outcomes are better in rural areas. Figure 3.17: Rural-urban disparity in the share of students performing below grade level in different subjects and grades a. Junior secondary students b. Primary students performing performing at “band 2 or below” level, below grade level, 2017 2015 0 10 20 30 40 Grade 5 Math Grade 6 Bangla Grade 3 Math Grade 8 Bangla Grade 6 English Grade 5 Bangla Grade 8 English Grade 3 Bangla Grade 6 Math 0% 20% 40% 60% 80% 100% Grade 8 Math Percentage Percentage of Students Urban Rural Rural Urban Source: LASI 2015 Source: NSA 2017 Source: Authors’ estimates using NSA 2015 and NSA 2017 data 26 4. The Education Budget: Process, Allocations and Utilization 4.1. The budget process The government agencies in charge of budget formulation and management are the Ministry of Finance (MoF) and the Planning Commission (PC). The budget consists of non-development and development budgets21 , which are prepared by MoF and PC, respectively, with MoF responsible for guiding and coordinating the overall budget preparation process. The Controller General of Accounts (CGA), an entity under MoF, is the key institution responsible for payment processing, internal controls and accounting. The annual budget process starts with the determination of the budget envelope for both development and non-development budgets by MoF in consultation with PC and Bangladesh Bank. MoF also assigns indicative budget ceilings to individual line ministries to guide their budget preparation. The overall non-development budget for the country is based on budget proposals submitted by individual ministries to MoF by October following a budget call circular. The development budget, which is approved by the PC, is based on annual operational plans (AOPs) for each project submitted by the different ministries. The overall development budget for the country is presented as a consolidated Annual Development Plan (ADP) prepared by the PC. The inclusion of a project in the ADP requires the prior approval of a Development Project Performa for the project by the PC22. For non-approved projects, an allocation can still be made provided that it is consistent with the projected resource allocation presented in the Medium Term Budget Framework (MTBF) 23. A series of consultations take place between MoF and MoPME/MoE between October and March in the process of finalizing the budget. MoPME and MoE prepare their draft budget proposals taking into account the MTBF and the indicative budget ceiling provided by MoF, and submit them to MoF by October 31. The proposals are based on inputs received from their subordinate agencies. Discussions between MoF and MoPME/MoE on the budget proposals begin in November and continue for the next five months. The final negotiations take place in March, after which MoF proceeds with the finalization of the budget. Consultations also take between the ministries and PC during this period to determine the development budgets that will be included in the ADP. As required by the Bangladesh Constitution, the budget is presented to the Parliament by the finance minister in June for final approval. The cabinet of ministers must approve the budget before tabling it in the parliament. Once the parliament approves the budget, MoF sends the budget book to MoPME and MoE, which then send allotment orders to their directorates and other subordinate agencies authorizing them to make expenditures as approved. The directorates and other subordinate agencies, in turn, issue allotment orders to divisional offices, district offices, and educational institutions. After the budget has been approved by parliament, any increase in the total budget can only be made by the National Assembly through a supplementary appropriation bill. However, the line ministries are allowed to re-allocate their budgets across the development and non-development 21 While both budgets include recurrent as well as investment expenditures, development budget primarily includes investment expenditures while non-development budget consists mainly of recurrent expenditures. It should also be noted that non-development budget is often referred to as revenue budget in Bangladesh. 22 The DPP is basically a project document with a detailed expenditure plan following GoB rules (with segregation of expenditures according to GoB chart of accounts). 23 The MTBF provides actual budget figures for the first year, and projections for the next two years. However, MTBF forecasts only take into account ongoing projects and projects already under preparation, and often exclude required future interventions and resources for achieving sector targets as reflected in NEP and its costed strategic plan. The MTBF is updated each year by MoF. 27 categories during the year so long as the total allocation does not exceed the approved amount. Typically, budget revisions are made after the first six months of the fiscal year. 4.2 Government expenditures on education: size and trends Public spending on education is relatively low in Bangladesh. Public expenditure on education as a share of the gross domestic product (GDP) was only 2.2% in 2015. Apart from Sri Lanka, which also spent 2.2% of its GDP on education, all other countries in South Asia spent much more on education than Bangladesh in 2015 (see Figure 4.1). Similarly, Bangladesh ranked second from the bottom in the region in terms of the share of the national budget devoted to education (11.7%). The Incheon Declaration, adopted in May 2015 by many multilateral organizations, including the World Bank, and participants from 160 countries during the World Education Forum, urged countries to devote at least 4% to 6% of GDP and/or at least 15% to 20% of public expenditure to the education sector to improve educational outcomes. The evidence presented in Figure 4.1 indicates that Bangladesh has a long way to go to meet these targets.24 The low spending on education imposes severe constraints on improving both the quality and quantity of education services. Figure 4.1: Education’s share in total government budget and GDP in South Asian countries, 2000- 2015 a. Education budget: share of total budget b. Education spending: share of GDP 30 7 % of total budget 25 6 % of GDP 5 20 20.5 4 15 3 13.7 12.6 11.7 2 2.1 2.2 1.8 1.8 10 1 5 0 2000 2005 2010 2015 2000 2005 2010 2015 Bangladesh Bhutan India Bangladesh Bhutan India Maldives Nepal Pakistan Maldives Nepal Pakistan Sri Lanka Sri Lanka Source: UIS Stat (2019) and calculations based on data from Ministry of Finance, Bangladesh The share of the public education expenditure as a percentage of GDP has been has fluctuating, and the share of the total government budget allocated to education has been decreasing. As shown in Figure 4.1b, in 2000, 2.1% of GDP was devoted to education in Bangladesh. This percentage decreased to 1.8% in 2005, was maintained in 2010, and increased to 2.2% in 2015. Except for Sri Lanka, all other South Asian countries devoted a higher share of GDP to education in most years. At the same time, the share of the education budget as a percentage of total government budget has generally been decreasing. While in 2000, the education budget share for Bangladesh was 20.5% -- the highest share in South Asia— it had decreased to 11.7% by 2015. 24 According to the figures from the Bangladesh Ministry of Finance (MoF), education budget as a share of the total budget has been lower than 15% since 2008 and has not exceed 2.5% as a share of GDP since 2000. The 7th Five Year Plan of GoB (2016-2020) envisions increasing the allocation to education to 3% of GDP; but even this vision falls significantly short of the recommendation made by the Incheon Declaration. 28 The education budget has been steadily increasing in absolute terms in recent years. As shown in Figure 4.2, between FY 2002/03 and FY 2016/17, education expenditure in current prices increased from 65 bil Taka to 395 bil Taka, representing an average increase of around 36.3% each year. In real terms, the average annual increase in education expenditure during this period was around 9%. As pointed out above, however, despite the increase in the education budget in absolute terms, education spending as a percentage of GDP and the overall budget has remained low by international standards. Figure 4.2: Budget allocation and development and non-development budget share, education sector (2002/3-2016/17) Source: Authors’ calculations based on Ministry of Finance data During the past 15 years, in general, at least 70% of the education budget has been allocated to non-development expenditure each year (see Figure 4.2). As non-development expenditures are mostly related to recurrent costs, this expenditure pattern indicates that investment expenditures in the education sector are relatively small. Furthermore, the non-development budget has been increasing more rapidly than the development budget. This has been the case for both MoE and MoPME25. While the share of non-development budget is larger than the share of development budget in both ministries, the development budget share has historically been higher for MoPME than for MoE (see Figures 4.3 and 4.4). In 2016-17, for example, the development budget share in MoPME was 35% while it was 25% for MoE. Over the last 15 years, these figures have fluctuated widely, especially for MoPME—it was as high as 50% in 2002-03 for MoPME and 29.8% in 2003-04 for MoE, and as low as 25 MoE has been receiving between 51% and 62.5% of the annual budget allocated to education in the last 15 years. The highest education budget share that has gone to MoE during the past fifteen years is 62.5% (in 2004/5). 29 31% in 2015-16 for MoPME and 15% for MoE in 2008-09. The relatively larger share of the development budget in MoPME could be partly attributed to the fact that, compared to MoE, the primary education sector receives more external funding from development partners. It should also be noted that the total budget for MoPME shows more fluctuations than the budget for MoE mainly due to its higher share of the relatively more volatile development budget. Figure 4.3: Budget allocation and development and non-development budget share, MoPME 100% 200,000 50% 40% 32% 44% 36% 36% 38% 41% 38% 32% 41% 38% 35% 31% 35% 90% 180,000 Percentage of Development and Non Development Budget 80% 160,000 Allocated Budget (Million Taka) 70% 140,000 68% 68% 69% 60% 64% 64% 62% 65% 65% 120,000 60% 62% 62% 59% 59% 50% 56% 100,000 50% 40% 80,000 30% 60,000 20% 40,000 10% 20,000 0% 0 Year Development Budget Share in total Allocation Non Development Budget share in total allocation Total Non Development Allocation Total Development Allocation Total Budget Allocation Source: Authors’ calculations based on Ministry of Finance data 30 Figure 4.4: Budget allocation and development and non-development budget share, MoE 100% 250,000 30% 30% 27% 20% 18% 16% 15% 16% 17% 19% 20% 22% 26% 21% 25% 90% Percentage of Development and Non Development Budget 80% 85% 84% 200,000 82% 84% 83% 81% 80% Allocated Budget ( Million Taka) 80% 78% 79% 70% 73% 74% 75% 70% 70% 60% 150,000 50% 40% 100,000 30% 20% 50,000 10% 0% 0 Year Development Budget Share in total Allocation Non Development Budget share in total allocation Total Non Development Allocation Total Development Allocation Total Budget Allocation Source: UIS Stat (2019) and calculations based on data from Ministry of Finance, Bangladesh 26 4.3 Utilization of earmarked resources in education The overall budget utilization rate in the education sector is high but variable. Full and proper utilization of earmarked resources is essential for the resource to have maximum impact on education outcomes of interest. As shown in Table 4.1, typically over 90% of the annual budget allocated to the education sector is spent each year. However, the budget utilization rate has fluctuated substantially over the years, and has been as high as 110.5% in 2010 and as low as 88.0% in 2008. This points to a need for a more effective planning and budgeting process, that adequately takes into account the capacity constraints in the sector. 26 Rahman et al. (2016) used Ministry of Finance data to calculate the share of GoB budget devoted to education for 2015. 31 Table 4.1: Percentage of total allocated budget for education that is spent Year % of allocated amount for Year % of allocated amount for education that is spent education that is spent 2002 94.9 2009 92.4 2003 92.7 2010 110.5 2004 94.5 2011 102.3 2005 88.6 2012 94.0 2006 90.9 2013 96.1 2007 90.5 2014 102.1 2008 88.0 2015 89.5 Source: Rahman et al (2016) A disproportionately high percentage of the development budget is spent in the last trimester of each fiscal year (Table 4.2). The budget utilization rates for the annual development budget for both MoPME and MoE are consistently higher than the corresponding rates for the overall GoB annual budget. Development budget expenditure as a share of the ADP budget for MoE and MoPME has ranged between 92% and 99% over the last five years, while the development budget utilization rate for all the ministries taken together has ranged between 83% and 96%. However, as in other ministries, the bunching up of expenditures in the last trimester is an issue of concern for both MoE and MoPME. For example, between 2011-2016, MoE never spent more than 50% of the total allocated amount in the first eight months (July to February) of the fiscal year. Furthermore, a significant percentage of the expenditure was made only in June, the last month of the fiscal year (e.g., 45% in 2014). The situation for MoPME is also not very different. Table 4.2: ADP expenditure in different months (2011-2016) Year Ministry 2011 2012 2013 2014 2015 2016 Cumulative ADP Expenditure share (%) spent between July and February MoPME 53 38 54 47 42 41 MoE 49 41 50 37 45 37 All GoB Ministries 37 38 44 38 38 34 Cumulative ADP Expenditure share (%) spent between July and June MoPME 97 98 96 99 97 94 MoE 95 96 99 96 99 92 All GoB Ministries 92 92 96 95 91 83 ADP Expenditure share (%) spent in June MoPME 15 15 25 22 18 30 MoE 25 29 22 45 40 32 All GoB Ministries 20 23 26 25 20 22 Source: Ministry of Finance, ADP Utilization (various years) (https://mof.portal.gov.bd/site/page/e207b75e-9968-420a-ad14-f95c0c356d4f) 32 One of the main reasons for the disproportionably high expenditure towards the end of the fiscal year has been the slow release of the budget. This slow release is a result of many factors such as delays in the ADP approval process, coordination issues, and delays in the approval of the Annual Operational Plans (AOPs) of the line ministries. The practice of releasing funds to the cost centers through manual allotment letters, particularly in MoPME, has also contributed to the slow funds release in the past27. Furthermore, the current quarterly fund release system of equal tranches also contributes to the bunching of expenditures towards the latter part of the fiscal year. In particular, because funds are released in equal quarterly tranches across all economic heads, activities that require higher expenditures upfront sometimes face funding crises in the initial months of the fiscal year. In such cases, projects and agencies have to wait until the next quarter to receive the funds required for implementing activities under these economic heads. Such delays in receiving funds also lead to disproportionately large expenditures in the last quarter of the fiscal year. The expenditure of funds under the ADP spikes in June, the last month of the fiscal year (Table 4.2). Since budgeted amounts for a fiscal year cannot be rolled over to the next fiscal year, there is immense pressure on all ministries, including MoE and MoPME, to spend the remaining amount before the end of the fiscal year. This proportion of the development budget spent in June is particularly high for MoE. Between 2011 and 2016, at least 22% and as much as 45% of MoE’s annual expenditure was made in June. The relatively high June expenditure for MoE compared to that for MoPME may be partly attributed to the higher number of development projects implemented by the former. While the total annual budget in education is well aligned with the MBTF, inconsistencies between the annual budget and MTBF are observed when the non-development and development components are scrutinized separately. For example, as shown in Table 4.3, the MoPME annual total budget and MTBF figures align well in most years. However, the non-development budget for MoPME has exceeded the MTBF projections in most years, and the development budget has frequently been less than the MTBF projection. The MTBF provides actual budget figures for the first year, and projections for the next two years so that activities can be planned and executed effectively. Hence the misalignment between the MTBF and the allocated development budget presents challenges in operational planning and in achieving annual targets and results (DPE, 2017a). Table 4.3: MoPME budget and MTBF 2010/11 – 2015/16 Year 2010-11 2011-12 2012-13 2013-14 2014-15 2015-16 MoPME Budget MTBF Projection (crore taka) 7,558 8,960 9,899 11,057 13,673 14,502 Actual Budget (crore taka) 8,074 8,964 9,825 11,935 13,676 14,504 % difference (actual-MTBF) 6.83% 0.04% -0.75% 7.94% 0.02% 0.01% Non-Development MTBF Projection (crore taka) 3,823 5,087 5,525 5,809 6,040 8,960 Actual Budget (crore taka) 4,867 5,450 4,382 6,657 7,898 8,963 % difference (actual-MTBF) 27.31% 7.14% -20.69% 14.60% 30.76% 0.03% Development Budget MTBF Projection (crore taka) 3,735 3,873 4,374 5,249 6,942 5,542 27 However, this situation is expected to improve once the new Integrated Budget and Accounting System (iBAS++) is rolled out to all cost centers in the education sector. 33 Actual Budget (crore taka) 3,207 3,514 5,443 5,278 5,778 5,541 % difference (actual-MTBF) -14.14% -9.30% 24.40% 0.60% -16.77% -0.01% Source: DPE 2017a The mismatch between MTBF projections and the non-development and development annual budgets is partly a result of midyear budget revisions. Since budgets allocated to ministries are not automatically rolled over to the next fiscal year if not used in the current fiscal year, budget revisions occur midway through the fiscal year, based on the actual expenditures for the first six months. As the non-development component of the budget largely includes recurrent costs that are relatively easier to incur, ministries have an incentive to reallocate funds to the non-development component when the overall budget utilization rate is expected to be low. Budget figures from recent years suggest that, in the education sector, development budgets are indeed largely revised downwards and non-development budgets are revised upwards during the midyear adjustments (Table 4.4). While upward revisions to the non-development budget can be observed in the overall national budget as well, these increases are, on average, higher in the education sector. The non-development budget increases for both MoE and MoPME were particularly high in 2015/16 (they increased by 24% and 29%, respectively)28. Table 4.4: Original and revised budgets in MoE and MoPME, 2012/13 - 2016/17 (in million Taka) Financial Budget (all Change Education Change MoE Change MOPME Change Year sectors) (O-R) sector budget (O-R) budget (O-R) budget (O-R) Non-development (revenue) budget 2011-12 (O) 878,510 141,490 87,070 54,420 (R) 918,230 5% 139,240 -2% 86,570 -1% 52,670 -3% 2012-13 (O) 994,960 144,720 90,290 54,430 (R) 1,028,920 3% 148,270 2% 92,900 3% 55,370 2% 2013-14 (O) 1,134,710 167,150 100,630 66,520 (R) 1,159,980 2% 186,500 12% 112,150 11% 74,350 12% 2014-15 (O) 1,282,310 197,880 118,930 78,950 (R) 1,273,710 -1% 201,390 2% 120,550 1% 80,840 2% 2015-16 (O) 1,645,710 218,660 129,060 89,600 (R) 1,503,790 -9% 276,010 26% 160,020 24% 115,990 29% Development budget 2011-12 (O) 460,000 56,570 21,430 35,140 (R) 410,800 -11% 44,360 -22% 19,760 -8% 24,600 -30% 2012-13 (O) 550,000 69,360 25,540 43,820 (R) 523,660 -5% 61,690 -11% 22,530 -12% 39,160 -11% 2013-14 (O) 658,700 83,780 31,000 52,780 (R) 600,000 -9% 76,770 -8% 31,480 2% 45,290 -14% 2014-15 (O) 803,150 94,250 36,470 57,780 (R) 750,000 -7% 84,750 -10% 41,420 14% 43,330 -25% 28 The sharp increase in revenue budgets for both ministries can be partly attributed to the implementation of the recommendation for remuneration increase made by the pay scale commission. In the case of MoPME, the increase can also be partly explained by the nationalization of non-government schools that year. 34 2015-16 (O) 970,000 97,390 41,970 55,420 (R) 910,000 -6% 95,040 -2% 42,570 1% 52,470 -5% Source: Ministry of Finance data, various years; Note: “O” and “R” refer to original and revised budgets, respectively. The revisions to the budget mid-way in the fiscal year make execution rates for the development budget look excessively impressive. For instance, although MoPME’s development budget execution rate in 2011/12 is relatively low (69%) when compared with the original allocated amount, it is a remarkable 98% when the revised budget is used as the reference (Table 4.5). Similar patterns can be seen for the development budget in other years as well. These figures suggest that there is plenty of room for improving the planning and execution of the development budget in MoPME. In the case of the non- development budget, there is little difference in execution rates between the original budget and the revised budget. Table 4.5: MoPME Budget execution rates for 2011/12 – 2015/16 2011-12 2012-13 2013-14 2014-15 2015-16 Budget Origin Revis Origin Revis Origin Revis Origin Revis Origin Revis al ed al ed al ed al ed al ed Developm 69% 98% 86% 96% 84% 99% 81% 92% n/a 98% ent budget Non- developme 106% 109% 104% 102% 98% 100% 88% 88% n/a 95% nt Total 91% 106% 96% 99% 95% 99% 87% 89% n/a 98% budget Source: DPE 2017a 4.4 Education sector funding sources and shares The government has been using its own sources to finance more than 80% of the expenditure incurred in education in most years since 2002 (Table 4.6). Foreign aid as a source of total education financing has been fairly uneven, ranging from a low of 7% of the total financing in 2004 to a high of 23% in 2007. Since the proportion of foreign aid in education expenditure has been declining since 2007, domestic resource mobilization—mainly tax revenue—has become increasingly important in meeting education financing needs. However, the tax to GDP ratio for Bangladesh is relatively low and has not increased much over time. According to the World Development Indicators (WDI) data, among the 119 countries with tax to GDP ratio figures available for 2015, Bangladesh had the 111th lowest ratio (8.5%). This is significantly lower than the average ratio of 17% for the 119 countries in the dataset. It is imperative that government explore ways to enhance tax compliance, expand the tax base, and tighten loopholes to increase tax revenue. Furthermore, as external funding from funding from DPs is still significant (though its share has been declining), strengthened coordination among DPs, MoE, MoPME, and MoF continues to be important. 35 Table 4.6: Education sector financing—sources and shares Percent of total financing Percent of total financing Year Domestic Foreign Aid Year Domestic Foreign Aid 2002 89 11 2009 88 12 2003 91 9 2010 88 12 2004 93 7 2011 91 9 2005 86 14 2012 89 11 2006 78 22 2013 85 15 2007 77 23 2014 89 11 2008 88 12 2015 91 9 Source: Rahman et al (2016) The share of domestic financing in the development budget is lower than in the overall national budget, but it has been increasing over the years. As shown in Table 4.7, the contribution of domestic resources towards financing the ADP has increased from 44% in 2008-09 to 65% in 2016-17, reflecting a decline in the share of foreign assistance. The main source of domestic financing for education is tax revenue. Both MoPME and MoE have had limited capacity for independently mobilizing funding from domestic resources, especially at the primary and secondary levels. They generate some non-tax revenues (mainly through different types of fees such as tuition fees, examinations fees and admission fee); but these revenues account for only around 0.5 % and 1.5% of the budgets for MoPME and MoE, respectively (Rahman et al, 2016). Table 4.7: Contribution of domestic resources towards ADP financing (in ‘0000000 Taka) 2008- 2009-10 2010-11 2011-12 2012-13 2013-14 2014-15 2015-16 2016-17 09 Total ADP 23000 28500 35588 41080 52366 60000 75000 91000 123345 financing Domestic 10011 12000 20850 26080 33866 36399 48915 61840 80345 financing Domestic as 43.53 42.10 58.59 63.48 64.67 60.67 65.22 67.96 65.14 % of ADP Source: Programming Division, Planning Commission (referred to in MoF 2017b) 36 5. Equity in Spending on Education 5.1 Equity in public spending This section discusses GoB spending patterns across population groups and geographic areas utilizing budget data from BOOST for fiscal year 2013-14 (FY14) and the HEIS data from 2016-17, the latest years available. It focuses on analyzing spending at the primary and secondary levels, since allocation of public spending for districts can be estimated for these two levels only with a few assumptions.29 It is important to interpret these results with care as GoB spending on education has been increasing since FY14. The relative comparisons across areas and groups are, nonetheless, valid under the assumption that the distribution of spending has not changed significantly since FY14. The primary education subsector receives most of the public expenditure on education, followed by the secondary education subsector. Data from FY 2015-16 show that around 44% of the total education spending goes to primary education, 38% goes to secondary education, and the rest goes to post- secondary education (Table 5.1). In terms of per-student expenditure, however, tertiary education receives the highest level public funding—around three times more than primary (Table 5.2). As students enrolled in higher educational levels are more likely be economically more advantaged, this spending pattern is likely to have a negative impact on equity. Table 5.1: Share of public education expenditure by level, 2015-16 Expenditure (%) Education level 2010- 2011- 2012- 2014- 2015-16 11 12 13 15 Primary (Gr. 1-5) 47.35 44.6 45.86 46.8 43.5 Junior Secondary (Gr. 6-8) 18.96 18.67 18.32 16.26 17.26 Secondary (Gr. 9-10) 22.61 21.42 20.75 19.42 20.52 Higher Secondary (Gr.11-12) 1.12 1.05 1.04 1.89 2 Tertiary 9.93 14.26 14.03 15.64 16.71 Overall 100 100 100 100 100 Total expenditure on education (bil Taka) 179.6 199.2 214.2 292.3 316.2 Source: BANBEIS 2012, 2013, 2014, 2016 and 2017; Note: 2013-14 data not available GoB education spending per student has been increasing in nominal terms during the past few years, but it is still lower than that in other countries in the region. As discussed in Section 4, the annual budget allocated to the education sector has continuously increased in nominal terms in recent years. This expansion has been accompanied by an increase in spending per student in nominal terms for all levels (Table 5.2a). In real terms, however, except for a substantial increase at the tertiary level, spending per student has either declined or increased only marginally (Table 5.2b). Furthermore, at the 29 The GoB expenditures included in the analysis are those reported by MoPME and MOE. Only categories related to primary and secondary education were included. For the expenditure items that could not be disaggregated at the district level, several types of expenditures were allocated uniformly based on either the share of students, teachers or institutions in each district. For instance, for MoPME about 22% of the expenditures did not have a district code, and 70% of them correspond to stipends. In this case, the amount was distributed based on the share of students receiving stipends in each district. For MOE most of the unassigned expenditures correspond to teachers’ salaries, which were allocated based of the share of teachers in each district. 37 primary and secondary levels, spending per student as a percentage of GDP per capita is much smaller than the average for OECD countries, and lower than the figures for other countries in the region (Table 5.3). Table 5.2: Per student public spending on education by level a. Annual in nominal terms (Takas) Level of Education 2010-11 2012-13 2013-14 2014-15 2015-16 Primary (Gr. 1-5) 4,728 4,676 5,017 7,173 7,213 Junior Secondary 4,788 5,358 4,781 5,761 6,497 (Gr. 6-8) Secondary (Gr. 9-10) 8,578 8,134 7,794 9,155 9,598 Higher Secondary 17,100 9,826 15,383 16,603 20,872 (Gr. 11-12) Tertiary 11,066 13,272 15,186 16,035 20,924 b. Annual in 2016 (Takas) Level of Education 2010-11 2012-13 2013-14 2014-15 2015-16 Primary (Gr. 1-5) 6468 5604 5616 7572 7,213 Junior Secondary 6552 6420 5352 6072 6,497 (Gr. 6-8) Secondary (Gr. 9-10) 11736 9756 8736 9660 9,598 Higher Secondary 23,400 11784 17,232 17,520 20,872 (Gr. 11-12) Tertiary 15,144 15,912 17,016 16,920 20,924 Source: Authors' calculations using BANBEIS for education per student costs and WDI for inflation adjustment Table 5.3: Government expenditure per student (% of GDP per capita) Country Primary Secondary Tertiary Bangladesh 9 10 25 Bhutan 14 32 55 India 10 17 49 Maldives 15 .. 29 Nepal 13 11 25 Pakistan 10 11 27 Sri Lanka 11 11 30 OECD members 20 23 26 Source: WDI circa 2016. At the same time, there are large variations in the levels of government spending per student across divisions and districts. For the primary level in FY14, Barisal division had the highest level of public spending per student (approximately Tk. 9237 in 2016 prices), while Dhaka had the lowest (Tk. 6014 in 2016 prices). For secondary, the divisions of Barisal and Rangpur had the highest spending per student 38 (about Tk. 15,000), while Sylhet and Chittagong divisions had the lowest (about Tk. 9,900) (Figure 5.1). Variations across districts are even larger. At the primary level, Dhaka had the lowest spending per student (Tk. 2114), while districts like Jhalokati in the Barisal division and Joypurhat in the Rajshahi division had spending per student of more that Tk. 13,000. A similar pattern is found at the secondary level, with spending per student ranging from about Tk. 7,000 to Tk. 23,000 (Figure 5.2). Figure 5.1: Public spending per student by division 18000 16000 14000 12000 10000 8000 Primary 6000 Secondary 4000 2000 0 Note: Authors calculations from BOOST for FY14. Annual amounts in 2016 takas. Figure 5.2: Spending per student across districts a. Primary level b. Secondary level 18000 25000 16000 14000 20000 Spending per student (Takas) Spending per student (Takas) 12000 15000 10000 8000 10000 6000 4000 5000 2000 0 0 Districts Districts Note: Authors’ calculations using BOOST FY14. Annual amounts in 2016 takas. 39 Spending per student across districts increases with the level of district poverty. There is a statistically significant correlation between GoB spending per student and poverty rates at the district level. For primary and secondary education, the correlations are 22% and 30%, respectively, suggesting that, on average, GoB spending at both levels may be pro-poor. However, as differences in spending across districts also result from other factors, there are many districts with similar levels of spending per student but very different poverty rates (Figure 5.3). Figure 5.3: Relationship between spending per student and poverty at the district level a. Primary level b. Secondary level Notes: Authors’ calculation using HIES 2016 and BOOST. District poverty measured using official upper poverty line. Spending in 2016 takas. A benefit incidence analysis indicates that the distribution of GoB spending at the primary level is pro-poor. Table 5.4 shows that while 30.7% of primary school age children are poor, they receive 35% of public primary education expenditures. In contrast, according to an earlier study half of the primary school age population in 2005 were classified as poor but they received only 47% of the public primary recurrent expenditures (World Bank, 2010). A similar analysis for 2000 showed that 59% of primary school age children were living in poverty but only 56% of primary education expenditures accrued to this group (Glinskaya, 2005). Thus, GoB spending has become more progressive with time. This is likely a result of the expansion in primary school attendance coupled with the implementation of specific cash programs targeted to the poor (i.e. stipends and tuition waivers). However, the distribution of GoB spending for secondary education lags in terms of reaching the poor as they are less likely to attend this level (Table 5.4). More specifically, while about 24% of the secondary-age children are considered poor, they receive only about 22% of public spending in secondary education. These patterns are related to the fact that children from poorer families are less likely to attend secondary school, and therefore do not benefit from GoB spending. Conditional on attending secondary school, however, the shares of public spending across consumption quintiles are very similar to the corresponding shares of students. 40 Table 5.4: Incidence of public education expenditure Primary level Secondary level Share of Share of Share of public Share of public Group children expenditure children expenditure Non-poor 69.3% 65.1% 76.0% 77.7% Poor 30.7% 34.9% 24.0% 22.3% Consumption quintile 1 (poorest) 25.6% 29.1% 19.5% 17.5% 2 22.2% 24.1% 19.6% 20.0% 3 19.9% 20.3% 20.3% 20.8% 4 17.7% 15.5% 20.2% 21.7% 5 (richest) 14.7% 11.0% 19.6% 20.0% Source: Authors' calculations using HIES 2016 and BOOST 2014. Stipend programs improve the progressivity of public spending at the primary level. The Primary Education Stipends Program (PESP) supports all primary schools by providing stipends to students in grades 1-5. To qualify for stipend support, a student must maintain 85% monthly attendance, take all the school examinations and attain a minimum exam score of 33% in each subject in her grade with certain exceptions (MoF 2017). The program used to target children from poor families, but has become universal since 2015-1630. Even though the primary stipend program is now supposed to be cover all children, its reach is still not universal. In 2016-17, around 11.1 million primary students were covered by PESP, while there were 18.6 million students enrolled in grades 1-5. There is nevertheless a positive and statistically significant relationship between stipend receipt and district poverty, indicating that poorer districts tend to have a larger share of primary students receiving stipends. However, according to HIES 2016, even though children in the poorest quintiles are more likely to receive stipends (30% of them are among the poorest 20%), about 2 in 5 children not receiving stipends are in the poorest quintile (Table 5.5). Primary stipends are an important source of funding for households, though the size of the transfer has been fixed since 2002. The amount of the PESP stipend has been fixed at Tk.100 per month since the beginning of the program. Nevertheless, this stipend amount represents about a third of the private monthly education expenditure of a median household in 2016, and about 70% of the private spending of households in the poorest quintile. Table 5.5: Characteristics of students receiving stipends in 2016 Primary level Secondary level Receive Does Receive Does stipend not stipend not 30 See Annex B for an overview of school stipend programs in Bangladesh. 41 receive receive stipend stipend Area Rural 94% 71% 85% 72% Urban 6% 29% 15% 28% Gender Female 52% 47% 71% 47% Male 48% 53% 29% 53% Quintile 1 30% 22% 21% 13% 2 25% 21% 21% 17% 3 20% 20% 20% 20% 4 15% 19% 23% 23% 5 9% 17% 16% 27% Note: Authors’ calculations from HIES 2016/17. At the secondary level, there is a weak relationship between poverty and stipends distribution. Secondary stipends are more likely to benefit females (71% of recipients are female) and its recipients are more likely to live in rural areas (Table 5.6). Recipients come from different consumption quintiles, with only 20% belonging to the poorest quintile. The higher likelihood of women to receive stipends reflects earlier emphasis of the program on incentivizing female secondary school attendance, though currently the program has a pro-poor targeting strategy aiming to reach both females and males. Tuition waivers are distributed in a comparable way to stipends. According to HIES 2016/17, about 10% of students received tuition waivers, more than half of primary school students and a quarter of secondary students (Table 5.6. About 83% of recipients attending primary school live in rural areas and 52% of them are female. As in the case of stipends, while tuition waivers tend to benefit the poorest children in primary, around 1 in 4 beneficiaries still belong to the richest 40% of the consumption distribution. At the secondary level, tuition waiver recipients come from all consumption quintiles, and are significantly more likely to be female. Table 5.6: Characteristics of students receiving tuition waivers Primary level Secondary level Does Does not not Receive receive Receive receive tuition tuition tuition tuition waiver waiver waiver waiver Area Rural 83% 69% 85% 71% Urban 17% 31% 15% 29% 42 Gender Female 52% 48% 66% 48% Male 48% 52% 34% 52% Quintile 1 28% 21% 21% 13% 2 25% 19% 23% 17% 3 21% 19% 20% 20% 4 16% 21% 21% 23% 5 10% 20% 16% 27% Source: HIES 2016/17 5.2 Household expenditure on education The growth in incomes and consumption in the last two decades in Bangladesh has been accompanied by an increase in the share of education expenditures in household budgets. In 2016, about 6 in 10 households reported incurring education expenditures (Figure 5.4). The percentage of households with education expenditures has remained around 60% since 2000, except in 2010 when it was slightly higher (64%). However, among households with positive education expenditures, the share of education expenditures in total household consumption has been rising—it increased from 4.3% in 2000 to 6% in 2010 and 7.7% in 2016. Figure 5.4: Household education expenditures, 2000-2016 a. Percentage of households with b. Share of education expenditures in total education expenditures consumption 100% 7.7% 80% 6.0% 62% 61% 64% 5.3% 59% 60% 4.3% 4.5% 3.8% 3.3% 40% 2.7% 20% 0% 2000 2005 2010 2016 2000 2005 2010 2016 All households Households with education expenditures Note: Authors’ calculation using HIES 2000-2016/17 The increasing household budget share allocated to education reflects higher amounts spent on education. In 2016, the median household spent Tk. 802 per month on education, or about Tk. 516 per child (Figure 5.5a). This represents an increase of more than 148% in real terms from the figure observed in 2000 and an increase of about 47% from 2010. The average real annual growth in households’ education spending was 9% over the 2000-2016 period. 43 The rise in median expenditure per household can be seen for all levels of education. Between 2000 and 2016, the largest increase in expenditure amount was for tertiary education, with median monthly expenditure per household rising by 971 Takas, from 367 Takas in 2000 to 1338 Takas (Figure 5.5b). However, the largest growth in percentage terms is seen for primary. Between 2000 and 2016, median expenditures for primary education increased from Tk. 24 to Tk. 300 per month (1200% in real terms). The graphs also show that household education expenditures increased more rapidly between 2010 and 2016 (compared to earlier years) for the primary and secondary levels. Figure 5.5: Median expenditures on education per month (in 2016 Takas) a. All education levels by year b. Median values by education level 900 1600 800 1400 700 Takas per month Takas per month 1200 600 1000 500 800 400 300 600 200 400 100 200 - 0 2000 2005 2010 2016 2000 2005 2010 2016 Median Median per student Primary Secondary Tertiary Note: Authors’ calculation using HIES 2000, 2005, 2010 and 2016/17. The figures are calculated for households that report positive education expenditures. Figures in 2016 prices deflated spatially to account for differences in cost of living across the country. Poor households still have substantially lower private spending on education than richer households. Comparing across consumption quintiles in 2016, it is seen that the median household in the poorest quintile spent about Tk. 315 per month on education (Tk. 202 per student) compared to Tk.1933 (Tk.1310 per student) for the median household in the richest quintile (Figure 5.6). The lower spending of the poor also translates into a lower education budget share. While households from the poorest quintile allocated 5% of their total consumption to education, the richest allocated 10% of their total consumption.31 31 This is conditional on having positive education expenditures. 44 Figure 5.6 : Education expenditures by quintile, 2016 a. Median value per month in 2016 Takas by b. Share of education expenditures in quintile total consumption by quintile 2,500 10% Takas per month 2,000 9% 7% 1,500 6% 6% 5% 5% 1,000 4% 4% 3% 500 - 1 2 3 4 5 Quintile 1 2 3 4 5 All households Median Median per student Households with education expenditures Note: Authors’ calculation using HIES 2000-2016/17 However, the spending gap between the poorer and richer households has been decreasing over time. While in 2000, the top quintile spent 22 times more on education per student than the poorest quintile, in 2016 the richest quintile spent only 6 times more. Therefore, the inequality in the distribution of private education expenditures has decreased substantially. This is also confirmed by changes in the Gini coefficient for education expenditures, which declined from 0.81 to 0.74 between 2000 and 201632. Considering only households with positive education spending, the decline in inequality measured by the Gini was 12 points, with more than half of the change observed between 2010 and 2016 (Figure 5.7). Figure 5.7: Gini coefficient for household education expenditures, by year Gini coefficient for household education expenditures, by year 0.85 Total education 0.80 expenditures 0.75 Total education 0.70 expenditures (exc. Zeros) 0.65 Total education expenditures per student 0.60 0.55 Total education expenditures per student 0.50 (exc. Zeros) 2000 2005 2010 2016 Note: Authors’ calculations using HIES 2000, 2005, 2010 and 2016/17. 32 The Gini coefficient is a measure of inequality with 0 denoting perfect equality and 1 denoting the other extreme. 45 Household private spending on education is mainly used to cover school fees, books, and tutoring. On average in 2016, households spent about 20% on fees, 23% on books, 26% on private tutoring, 4% on transportation and the remaining 27% on miscellaneous items (including uniforms, internet, tiffin costs, accommodation, etc.) (Table 5.7). Across levels, fees are increasingly important for higher levels of education (16% for primary, 19% for secondary, and 31% in tertiary), and spending on tutoring is particularly high in secondary (about 31% of total spending). In addition, transport costs are most important at the tertiary level (10% of total spending). Table 5.7: Distribution of household education expenditures education level, 2016 Expenditure category Level Fees Books Tutoring Transport Others Total All 20% 23% 26% 4% 27% 100% Primary 16% 25% 22% 1% 35% 100% Secondary 19% 24% 31% 3% 23% 100% Tertiary 31% 21% 19% 10% 18% 100% Source: Authors calculations using HIES 2016/17. Note: Fees include expenditures on admissions, annual sessions, registration, examinations and tuition. Books include text, note and exercise books and stationary. Tutoring includes private tutoring and coaching. Other expenditures include uniforms, footwear, hostel, tiffin, internet/e-mail, schooling donation and others. Across quintiles, the share allocated to fees, tutoring and transportation increases for households with more resources, while the share allocated to books decreases for richer households (Figure 5.8). For the poorest quintile, expenditures on books comprise about a third of the budget spent on education at all levels. On the other hand, for the richest quintile, expenditures on books comprise a much smaller share of the total expenses; rather, tutoring expenses constitute around a third of the budget at the primary and secondary levels. At the tertiary level, the spending patterns across quintiles become more similar; this is expected since the expenditure share estimates are conditional on attending higher levels of education. The median amounts (as opposed to shares) spent on fees, books, tutoring and other items are understandably lower for the poor. For instance, in 2016, the bottom 20% spent about Tk.21 per month on fees, which is about half of what is spent by the next quintile and about 6% of the median household education expenditure of the richest quintile. These gaps are observed across all three levels of education (Appendix Table A3.6). It is likely that the substantially larger private expenditures of rich households on education in general, and on private tutoring and fees in particular, have a positive impact on the learning outcomes of their children. 46 Figure 5.8: Distribution of expenditures by type across consumption quintiles, 2016 a. All levels b. Primary 100% 100% 29% 28% 28% 26% 24% 35% 31% 80% 80% 36% 37% 36% 1% 2% 3% 4% 6% 3% 60% 1% 1% 1% 2% 60% 20% 25% 15% 27% 28% 30% 20% 23% 26% 29% 40% 40% 31% 33% 28% 25% 18% 26% 23% 21% 17% 21% 20% 20% 23% 16% 14% 15% 16% 19% 18% 18% 19% 20% 0% 0% 1 2 3 4 5 1 2 3 4 5 Fees Books Tutoring Transport Others Fees Books Tutoring Transport Others c. Secondary d. Tertiary 100% 100% 24% 24% 24% 23% 21% 15% 15% 17% 17% 21% 80% 80% 5% 7% 10% 1% 3% 3% 4% 5% 12% 11% 21% 18% 60% 24% 28% 31% 60% 19% 19% 33% 37% 19% 40% 28% 26% 32% 40% 24% 20% 17% 26% 24% 21% 17% 20% 20% 20% 19% 19% 19% 20% 31% 34% 30% 32% 31% 0% 0% 1 2 3 4 5 1 2 3 4 5 Fees Books Tutoring Transport Others Fees Books Tutoring Transport Others Note: Authors calculations using HIES 2016/17. Part of the difference in private spending patterns between poorer and richer households may be attributed to the types of school attended by children, though government funded institutions are dominant at all education levels. According to HIES 2016, 84% of primary-age students attend government schools or private government-subsidized institutions (Figure 5.9). The HIES data also indicate that children from poorer households are more likely to attend government schools. For example, while only 4% of the children form the poorest consumption quintile attended private non-government-subsidized schools in 2016, this was true for 18% of the children from the richest quintile. At the secondary level (grades 6-10), where the vast majority of the schools are publicly subsidized and privately managed, approximately 9 in 10 secondary-age school children attended government or government-subsidized private schools in 2016. Given the dominance of these schools in providing secondary education, there are no major differences in the types of school attended across consumption quintiles. Only children in the richest quintile are more likely to attend private non-government subsidized secondary schools. At the tertiary level, most students attend government subsidized private institutions (55%) or government institutions (38%), and students with more resources are significantly more likely to choose private (not subsidized) intuitions. It is interesting to note that though private, non-government-subsided educational institutions still enroll only a small percentage of the student population, during the past decade, there has been a substantial increase in the share of children attending private institutions at all levels of education. 47 Figure 5.9: Type of school attended, 2005-2016 a. Primary b. Secondary 100% 100% 6% 6% 6% 7% 1% 3% 90% 5% 9% 80% 12% 8% 80% 70% 60% 60% 50% 93% 91% 40% 40% 76% 76% 30% 20% 20% 10% 0% 0% 2005 2016 2005 2016 Other (NGO, Madrasa) Government Private (Govt. subsidized) Private (Not subsidized) Private (Not subsidized) Other (NGO, Madrasa) Government/Govt. subsidized c. Tertiary 100% 4% 3% 3% 4% 90% 80% 70% Madrasa 60% 50% Private (Not subsidized) 93% 93% 40% 30% Government/Govt. 20% subsidized 10% 0% 2005 2016 Source: Authors' calculations using HIES 2005 and 2016/17. 5.3 The role of public spending in total spending Public spending constitutes a large share of total education expenditures, particularly for the primary level. Estimates combining HIES and BOOST indicate that for the median child in primary school, about 57% of total spending comes from public resources. At the secondary level, the percentage of total cost publicly funded is 43%. Consistent with the incidence analysis presented above, public spending has a larger importance for poorer children. At the primary level, about 76% of the education expenditures of the median household in the bottom 20% of the consumption distribution come from public resources, compared to 23% for the richest 20% of the consumption distribution (Figure 5.10). At the secondary level, public expenditures 48 represent 60% of total expenditures for the median households in the bottom 20% compared to 29% for the richest 20%.33 Therefore, public spending helps reduce the education spending gap between poor and rich households. For instance, in 2016, the richest quintile spent about 7.5 times more per student in primary, compared to the poorest quintile (Table 5.8). When public spending is added, the richest quintile spends about 2 times more than the poorest quintile. At the secondary level, the ratio between mean expenditures per student between the richest and poorest quintiles declines from 11.4 times to 8.4 times when public spending is added. Thus public spending plays a very important role in enhancing equity. Figure 5.10: Share of public expenditure in total education expenditure, by quintile 100.0% 80.0% 60.0% 40.0% 20.0% 0.0% 1 2 3 4 5 Quintiles Primary Secondary Source: Authors' calculations using HIES 2016/17 and BOOST 2014. Note: Total education expenditure includes public and households' private expenditures. The figures presented are medians by group. Table 5.8 : Total education expenditures per student, median takas per month Primary Secondary Private and Private and Private Private Group public public All 204 616 417 1183 Consumption quintile 1 82 494 107 249 2 153 568 279 730 3 246 648 426 1078 4 355 719 669 1485 5 605 942 1225 2090 Source: Authors' calculations using HIES 2016/27 and BOOST 2014. Note: Private denotes the households’ expenditures net of stipends which are included in public spending. Private and public includes spending from GoB. Amounts in 2016 takas. 33 Caveat: these shares are slightly underestimated as the information from BOOST comes from FY14. Yet, the qualitative comparisons across groups should be adequate. 49 6. Relationship between Spending and Outcomes As discussed in the previous sections, overall spending on education has increased, access to education has improved, and inequalities in access across income groups have decreased during the past two decades. This section analyzes the cross-sectional relationship between levels of spending and outcomes. At the primary level, it looks at three categories of outcomes: access, efficiency and quality as represented by learning outcomes. At the secondary level, the analysis only focuses on access (attendance rates) and quality (learning outcomes) as efficiency indicators are not being readily available at the district level. Even though this analysis is not intended to explain the causal impact of more resources on education outcomes, it does provide a preliminary nuanced picture of their relationship. It should be noted, however, that an analysis at the school-level, and across time, would be needed to get a better picture of the links between spending and learning outcomes and identify how resources can be more effectively utilized to improve education outcomes. Unfortunately, the information available does not allow for such detailed analysis District level data indicate that public spending is positively associated with net attendance rate at the secondary level, but this correlation is insignificant at the primary level. As shown in Figure 6.1, on average, districts with higher spending per student appear to have higher net attendance rates at both levels, suggesting that public spending is playing a positive role in enhancing access. However, the relationship is not statistically significant for primary education, and remains insignificant when we control for other district characteristics using multiple regression analysis (Appendix Table A6.1).34 At the secondary level, however, the one-to-one correlation between spending per student and attendance is stronger, and is statistically significant at the 1% level. But in this case as well, the relationship loses significance once other explanatory factors are taken into account (Appendix Table A6.1 and A6.2). Figure 6.1: Relationship between access and spending per student, 2014 a. Primary NAR vs. spending/student b. Secondary NAR vs. spending/student Notes: Authors’ calculations using BOOST and HIES 2016/17 34 It is important to keep in mind that this analysis only presents a cross-sectional correlation between spending and outcomes, and is not meant to show a causal relationship between the two variables. Neither can provide evidence on the effectiveness of increased spending on outcomes across time. The weak relationship observed here could be explained by the fact that spending is going to groups and areas that are lagging and have relatively worse outcomes. The controls in the multiple regressions may not completely address this fact. 50 The analysis does not show any association between public spending per student and internal efficiency either. While the graphs in Figure 6.2 suggest that there may be a weak relationship between the internal efficiency indicators (survival rate, dropout rate, repetition rate, and efficiency ratio) and public spending at the primary level, none of these bivariate correlations are statistically significant. The absence of a correlation between spending and efficiency indicators is confirmed by multiple regression analyses as well (Appendix Table A6.1). Figure 6.2: Relationship between internal efficiency and spending per student at the primary level, 2014 a. Survival rate b. Dropout rate c. Repetition rate d. Efficiency ratio Source: Authors’ calculations using BOOST, HIES 2016/17 and DPE 2014b and DPE 2015 At the division level, higher spending appears to be associated with higher learning outcomes and vice versa. Table 6.1 groups the division level learning outcomes35 findings from NSA 2017 for the various subjects and grades into four categories: “low spending-low outcome”, “high spending-high outcome”, “low spending-high outcome” and “high spending-low outcome”. It is seen that most of 35 Learning outcome refers to percentage of students at or above grade level proficiency. The outcome is categorized as “high” if the value is above the mean for this variable. 51 results are either in “low spending-low outcome” (cell III) or “high spending-high outcome” (cell II) categories, indicating a positive relationship between spending and learning outcomes at the divisional level. Furthermore, while some learning outcome results are in the high category despite being in a low spending division (cell I), there are no low category results in divisions with high spending (cell IV). Sylhet, Khulna, and Chittagong have low spending as well as low outcomes in all subjects and grades implying that increased spending in these divisions may be required to help improve overall outcomes in these divisions. Table 6.1: Division level learning outcomes in primary education grouped by per-student expenditure and grade-level proficiency (I) (II) Grade-level proficiency in Bangla and Math in grades 3 and 5 Dhaka Grade 3 Bangla Barisal Grade 3 Bangla Dhaka Grade 5 Bangla Barisal Grade 5 Bangla Dhaka Grade 3 math Barisal Grade 3 math High Rajshahi Grade 3 Bangla Barisal Grade 5 math Rajshahi Grade 5 Bangla Rangpur Grade 3 Bangla Rajshahi Grade 3 math Barisal Grade 5 Bangla Rajshahi Grade 5 Rangpur Grade 3 math Barisal Grade 5 (III) (IV) Chittagong Grade 3 Bangla Chittagong Grade 5 Bangla Chittagong Grade 3 math Chittagong Grade 5 math Dhaka Grade 5 math Khulna Grade 3 Bangla Low Khulna Grade 5 Bangla Khulna Grade 3 math Khulna Grade 5 math Sylhet Grade 3 Bangla Sylhet Grade 5 Bangla Sylhet Grade 3 math Sylhet Grade 5 math Low High Total per student expenditure by division at the primary level Source: Authors’ estimates based on NSA 2017 data and BOOST 2014 data Note: The categorization of values for a variable into “High” and “Low” is based on whether the value is above or below the mean for that variable. However, looking deeper at district level data, there is no evidence of a significant relationship between spending per student and learning outcomes. The graphs in Figure 6.3, which plot the grade 3 and 5 learning outcomes against district level spending, do not show any consistent pattern to this relationship. While there appears to be a positive relationship between spending per student and grade 3 math outcomes, the opposite is true for Bangla. The correlation coefficients in all four cases are small and statistically insignificant, confirming the absence of a systematic relationship between spending per student and student learning outcomes at the district level. This result is further corroborated by multiple regression analyses findings (Table A6.3). 52 Figure 6.3: Relationship between student learning outcomes and spending per student at the primary level a. Grade 3 Bangla b. Grade 3 Math c. Grade 5 Bangla d. Grade 5 Math Source: Authors’ calculations using BOOST, HIES and NSA data. Note: The spending data are for 2014 and the education outcomes are for 2017. While multiple regression analyses confirm that differences in levels of spending do not explain variations in outcomes across districts, they show significant correlations between some of the other explanatory factors and district level outcomes. More specifically, the number of students per teacher and adult literacy rate significantly explain differences in primary survival rates, dropout rates and the efficiency ratio across districts. These two factors are also significantly correlated with secondary level net attendance rate. Moreover, adult literacy rate is correlated (albeit at the 10% significance level) with grade 5 Bangla and Math learning outcomes as well. Bivariate analyses also show that household private education spending per student is associated with lower repetition rates, a higher efficiency ratio, and lower dropout rates; however these correlations become insignificant in the multiple regression model. Given its importance for education outcomes, more efforts are needed to improve student-teacher ratios (STRs) in Bangladesh. In 2016, the average STRs at the primary and secondary levels were 34 and 36, respectively (Figure 6.4a). As shown in Figure 6.4, these STRs are high by international standards at both the primary and secondary levels. Compared to other countries in the region, the primary STR is 53 similar to that of India, lower than that of Pakistan and Bhutan, but significantly higher than that of other countries in the region and OECD members. At the secondary level, the ratio is higher than that of other countries in South Asia. There has been a reduction in the primary level STR over time, but the secondary level ratio has been increasing. There is a large variation in the district level STRs in Bangladesh (Figure 6.5): it ranges from 21 students per teacher in Rangamati to 53 in Cox’s Bazar (both in Chittagong) at the primary level, and from 27 (in Thakurgaon and Panchagarh in Rangpur division and Dhaka) to 69 (in Habiganj in Sylhet division) at the secondary level. As noted earlier, the high STR districts have, on average, lower educational outcomes. Figure 6.4: Student to teacher ratio a. Bangladesh by year b. Comparison across countries, circa 2016 50 60 Primary Secondary 45 40 50 35 40 30 25 30 20 20 15 10 10 5 0 0 2008 2009 2010 2011 2012 2013 2014 2015 2016 Primary Secondary Source: World Development Indicators. Figure 6.5: Student-teacher ratios at the district level a. Primary level b. Secondary level 60 60 Student-teacher ratio 50 50 Student-teacher ratio 40 40 30 30 20 20 10 10 0 0 Districts Districts Notes: Authors’ calculation using HIES 2016, BOOST, DPE 2014b and BANBEIS 2014. 54 7. Conclusions and Recommendations Bangladesh has made impressive progress in expanding access to education, with sustained increases in attendance rates at all levels of education. Between 2000 and 2016, the net attendance rate at the primary and secondary education levels increased from 72% to 93%, and 50% to 72%, respectively. The expansion in tertiary access has been even more rapid, from around 0% in 2000 to 16% in 2016. Overall, compared ot the older generation, the younger generations have had more opportunities for attending and completing school. As a result, the education profile of the adult population has changed, with the percentage of adults older than 18 years of age without schooling decreasing from 63% in 2000 to 44% in 2016. The key remaining challenges in access are twofold: at the primary level, bringing the remaining out of school children into regular schooling; and increasing overall access to higher levels of education, especially at the tertiary level. The progress in enhancing access to school education has been accompanied by an increase in the internal efficiency of the system. Repetition rates and cycle drop-out rates have decreased, and the survival rate and the coefficient of efficiency have increased at both levels. However, high drop-out rates continue to be an issue of concern, especially at the secondary level. There has also been a reduction in the gender gap in access at the primary and secondary levels, but the gap in tertiary level attendance has increased over time. While near-gender parity in net attendance rate was achieved as early as 2000 at the primary level, males significantly lagged behind females in secondary attendance rate at that time. Males continued to lag behind females by 9 percentage points in 2016 as well, but this represented a marked improvement from earlier. On the other hand, tertiary attendance rates are much higher for males compared to females, and this difference has increased over time. Hence, at the primary and secondary levels, the focus now needs to be on increasing enrollment of males, while at the tertiary level, greater emphasis is needed on understanding and addressing constraints faced by women in accessing tertiary education. It should be noted, however, that due to disadvantages faced by women of older generations in accessing education, the percentages of adult women who have completed different levels of school education are still lower than those for men. Disparities in access across income groups and geographical regions have also declined at the school level, further pointing to the broad-based nature of the expansion in access. Attendance rates across different consumption quintiles have converged over the years, especially at the primary level. For example, compared to a primary attendance rate gap of 26 percentage points between the richest and poorest quintiles in 2000, the corresponding difference in 2016 was only 8 percentage points. The disparities in secondary school attendance across quintiles have also declined, but remain much larger than in primary. There has been a reduction in disparities across geographical regions, with attendance rates across divisions converging over time. For instance, the attendance rate for the Sylhet division— which had the lowest divisional attendance rate in 2000—has increased more rapidly that the rates for other divisions (especially since 2010). Yet, there are visible gaps in school attendance by poverty status and region. Furthermore, attendance rates across quintiles show no signs of convergence at the tertiary. These findings indicate that more targeted support to the poor, especially at the tertiary level, and to lagging regions, is needed to help them overcome human capital disadvantages. Increases in HOI for primary and secondary attendance rates reaffirm the progress made in enhancing equitable access to primary and secondary education. A substantial part of the increase in HOI can be explained by the reduction in disparities in access at both levels, indicating that reductions in disparities have played a key role in enhancing educational opportunities at the secondary level. The HOI 55 analysis also finds that the main circumstances leading to disparities in school attendance are household economic status and the education levels of adults in the household, again indicating the need for targeted support to poor households. Though access to education has improved as discussed above, the quality of education, as reflected in student learning outcomes, is low and has be declining. The NSA results show alarmingly low percentages of children achieving grade level proficiency in Bangla and Math at the primary level. For example, in 2017, around 26% and 59% of grade 3 students failed to achieve grade level proficiency in Bangla and Math, respectively. The results are worse for grade 5. The LASI findings also show significant percentages of children performing below grade level. Furthermore, the performance of students from grades 3 and 5 declined between 2011 and 1017 in all subjects except grade 3 Bangla, indicating that the large investments made in school education in the past decade have not been effective in improving education quality. Improving student learning outcomes is the biggest challenge now facing the Bangladesh education sector. The NSA and LASI data indicate that gender differences in learning outcomes at both the primary secondary levels are generally small, but there are significant disparities in learning outcomes across geographical areas. While gender differences in outcomes are small, they are slightly in favor of females in Bangla at the primary level, and substantially in favor of males in Mathematics in grade 8. The near-gender parity in the different subjects and grades at the primary level can be seen in all four NSA rounds. On the other hand, there is a wide variation in the outcomes across divisions at both the primary and junior secondary levels, with Sylhet performing particularly poorly. There are also large variations in outcomes at the primary level across districts especially in Math. Though GoB has been making substantial investments in the education sector with the broad goals of increasing both access and quality, the overall education budget has remained low by international standards. Between FY 2002/03 and FY 2016/17, education expenditure increased, on average, by 36.3% per year in nominal terms and 9% per year in real terms. However, in 2015, the education budget was only 2.2% of the GDP, and constituted only 11.7% of the total government budget. These figures are lower than the corresponding shares for all other countries in the region except Sri Lanka. They are also substantially smaller than the budget shares recommended by the Incheon Declaration adopted in 2015 by 160 countries, including Bangladesh, to ensure adequate investments in the sector. Furthermore, the share of the total government budget allocated to education has been decreasing over the years. The low spending on education imposes severe constraints on improving both the quality and quantity of education services. To ensure an adequate level of human capital development as Bangladesh works towards becoming an upper middle income country, there is a need to increase the education budget share to international standards. Over 70% of the education budget is typically allocated to revenue or non-development expenditure each year, limiting the amount of development expenditures in education. Non-development expenditures, which are mostly related to recurrent costs, have been increasing more rapidly than the development expenditures in both MoE and MoPME. The development budget share in MoPME is higher than that in MoE. However, the share of development budget in MoPME has fluctuated widely over the years—from 50% in 2002-03 to 31% in 2015-16. While the overall budget utilization rate in the education sector is high, there is a bunching of development spending in the last trimester of the fiscal year and in June as in other sectors. Typically, over 90% of the annual budget allocated to the education sector is spent each year. However, both MoE and MoPME spent over 50% of their development budgets in the last quarter (March-June) of 56 the fiscal year, except in 2011. Furthermore, spending spikes in June, the last month of the fiscal year, due to the need to spend all remaining funds before the end of the year. The bunching of expenditures in the last trimester is partly a result of the slow release of the budget. This slow release is a result of many factors such as delays in the ADP approval process, coordination issues, delays in the approval of the AOPs of the line ministries, the practice of releasing funds to the cost centers through manual allotment letters, and the practice of releasing funds in equal quarterly tranches across all economic heads. To help improve the efficiency of the funds release process, there is a need to fully roll out iBAS++ to all field office in the education sector as quickly as possible. Furthermore, MoPME and MoE need to engage in dialogue with MoF to resolve challenges related to mismatches between resources needs for different economic heads and the current practice of quarterly funds release. Midyear budget revisions tend to reallocate part of the development budget to non-development budget, causing a reduction in much needed development expenditure. Budget figures indicate that MoE and MoPME have been reallocating development funds to the non-development component during annual midyear budget revisions, reflecting their inability to fully utilize the original development budget. This points to a need for a more effective planning and budgeting process, that adequately takes into account the capacity constraints in the sector. In addition, it would also be important for MoF to closely monitor allocations and expenditures of the budget across the two ministries to minimize duplication and enhance coordination of interventions in the education sector. Along with the increase in the education budget, public spending per student has also been increasing over time, but remains relatively low. In nominal terms, spending per student as a percentage of GDP has increased for all levels of education though the increase is limited to the primary and tertiary levels in real terms. Despite this increase, however, spending per student remains low compared to the OECD average and figures for neighboring countries. Hence, ensuring adequate financing remains an unfinished agenda in the Bangladesh context, especially as the nation strives to become an upper-middle income country. On the positive side, there is evidence that pubic spending at the primary level has become more progressive over time. For example, in 2000, 59% of primary school age children came from poor households, and they received only 56% of public primary education expenditures; on the other hand, 30.7% of the primary age children were poor in 2016, and they received 35% of the public expenditure. Analysis of household spending patterns shows that public spending helps reduce the education spending gap between poor and rich households. For example, in 2016 the richest quintile spent about 7.5 times more per student on primary education, compared to the poorest quintile. Once public spending is considered, that gap falls to 2 times. The pro-poor nature of public spending at the primary level has most likely helped to increase the participation of children from the poorer segments of society. Public spending at the secondary level is, however, pro-poor only among children who are already attending secondary school. Government stipends have played a key role in improving the progressivity of primary level public spending. As PESP is meant to be universal, children from all consumption quintiles are eligible to receive stipends. Accordingly, HIES data show that around 30% of the stipend recipients come from the poorest quintile, 55% come from the bottom two quintiles, and the rest belong to other quintiles. Furthermore, the size of the PESP benefit represents about 70% of the private spending of households in the poorest quintile. Additionally, there is a significant positive relationship between stipend receipt and district poverty. Thus there is evidence that the stipends program significantly benefits poorer households. It should be noted, however, that as the reach of the program is not yet universal in practice, there are also some poor students among those not receiving stipends. Another issue is that as the benefit amount of 57 PESP has been fixed since the beginning of the program, its importance to households has been declining in real terms. At the secondary level, there is no relationship between poverty and the receipt of stipends. Only 20% of stipend recipients come from the poorest quintile. This indicates the need for poverty targeting of secondary programs. Analysis of the bivariate relationships between public spending and education access and efficiency indicators using district level data indicates that spending is positively correlated with secondary attendance rates, but not with other outcomes. There does not appear to be a significant correlation between district level spending per student and access and efficiency indicators at the primary level. Even at the secondary level, the statistical significance of the bivariate relationship between spending and access disappears when other explanatory variables are taken into account. It is also interesting to note that between 2010 and 2016, the NAR at the primary and secondary levels increased substantially even though per student public spending did not increase in real terms36. At the same time, however, total household expenditure on education at both primary and secondary levels increased substantially, suggesting that there may indeed be a positive correlation between overall education spending and access. Similarly, while there appears to be a positive association between public spending and learning outcomes at the division level, analysis using district level data finds no evidence of a relationship. The NSA data show that divisions with high spending show high performance in both Bangla and Math in grades 3 and 5. And most of the divisions with low spending per student have low performance. But correlational and multiple regression analyses using district level data do not show a statistically significant relationship between spending per student and student learning outcomes. One possible explanation for the absence of any significant correlation between spending and the various educational outcomes discussed here is that district level data are at too high a level of aggregation to allow for a rigorous analysis of this relationship. As noted in section 5, an analysis at the school-level, and across time, would be needed to get a better picture of the links between spending and different education outcomes. While the lack of a significant association between spending and learning outcomes found in this analysis is not definitive, it does suggest that public spending needs to be more focused on interventions and activities that have a direct bearing on education quality. Despite the increasing investments in education, learning outcomes have not improved since 2011, both at the primary and secondary levels. This fact alone indicates that public spending needs to be more quality focused. The parsimonious multiple regression results already pointed to the need to invest in decreasing the STR to improve education outcomes. In other words, Bangladesh needs to adequately invest in ensuring that teachers are recruited in sufficient numbers and deployed rationally across schools. Additional examples of quality-enhancing potential interventions at the school level are the provision of remedial classes for lagging students, provision of supplementary reading materials for students (age appropriate children’s literature), and support for greater engagement of parents in the schooling process. Other areas of focused investments for directly enhancing learning outcomes, especially at the primary level, could include the following: expanding access to quality early childhood development programs to improve school readiness at primary school entry; enhancing early grade reading and mathematics skills and improving the reading habit; strengthening examinations and assessments; and strengthening teacher accountability. Ongoing government programs such as the Fourth Primary Education Development Program (2018-23) and the Secondary Education Development Program (2018-23) cover some of these areas. The impact of 36 Except at the primary level where it increased slightly. 58 these programs on education quality will depend on how effectively the relevant interventions are implemented in practice. More in-depth policy research scrutinizing public expenditures and education service delivery could be helpful in designing future programs aimed at enhancing the efficiency of public spending and improving education quality. The current study has looked at the expenditures primarily at the national level, presented an analysis of trends in education access and quality, and attempted to explore the linkages between spending and education outcomes. Building on the findings of this study, it would be valuable to do an in-depth examination of the flow of funds from the center to the school and student levels in terms of the major expenditure items, analyze the timeliness and potential leakages of funds at various points of expenditure trail, and identify possible links between expenditure and key service delivery inputs, processes, and outcomes at the school level. Hence, it is recommended that a comprehensive Public Expenditure Tracking Survey (PETS)/ Quantitative Service Delivery Survey (QSDS) based on a nationally representative sample of schools be conducted in the future to provide more specific knowledge inputs for enhancing the efficiency of public spending and improving accountability and transparency in in the education sector37. School-level data from such a survey could also be utilized to do a more rigorous analysis of the determinants of education outcomes38. 37 No recent comprehensive PETS/QSDS has been implemented in Bangladesh. Hossain (2013) conducted a PETS/QSDS in 2013; but this was based on a small purposive sample of 50 primary schools. 38 An analysis of the determinants of outcomes can also look at the contribution of school resources versus efficiency in the use of the resources. For example, Hoogeveen et al. (2014) use stochastic frontier analysis to investigate the relationship between efficiency and school performance in Togo. 59 References (ACER) Australian Council for Educational Research 2016. “2015 Learning Assessment of Secondary Institutions--Public Report.” (BBS) Bangladesh Bureau of Statistics 2017. “Bangladesh Household Income and Expenditure Survey (HIES) 2016-2017.” Ministry of Planning, Government of the People’s Republic of Bangladesh (BBS) Bangladesh Bureau of Statistics 2010. “Bangladesh Household Income and Expenditure Survey (HIES) 2010.” Ministry of Planning, Government of the People’s Republic of Bangladesh (BBS) Bangladesh Bureau of Statistics 2005. “Bangladesh Household Income and Expenditure Survey (HIES) 2005.” Ministry of Planning, Government of the People’s Republic of Bangladesh (BBS) Bangladesh Bureau of Statistics 2000. “Bangladesh Household Income and Expenditure Survey (HIES) 2000.” Ministry of Planning, Government of the People’s Republic of Bangladesh. (BANBEIS) Bangladesh Bureau of Educational Information and Statistics 2010-2017. “Bangladesh Education Statistics Report 2010.” Ministry of Education, Government of the People’s Republic of Bangladesh. BOOST (2019). “BOOST Public Expenditure Database, World Bank Group.” Washington, DC: World Bank. Barros, R. F., Vega, F. J. M., and Saavedra, J., 2009. “Measuring Inequality of Opportunities in Latin America and the Caribbean.” Washington, DC: World Bank. Barros, R. F., Vega, F. J. M., and Saavedra, J., 2010. “Measuring Progress toward Basic Opportunities for All.” Brazilian Review of Econometrics 30 (2): 335-67. (DPE) Directorate of Primary Education 2018. “National Student Assessment 2017 for Grades 3 and 5 National Report.” Dhaka: Monitoring and Evaluation Division, Directorate of Primary Education, Government of Bangladesh. [Preliminary draft] (DPE) Directorate of Primary Education 2017a. “Bangladesh Primary Education Annual Sector Performance Report - 2017.” Dhaka: Monitoring and Evaluation Division, Directorate of Primary Education, Government of Bangladesh. (DPE) Directorate of Primary Education 2017b. “Annual Primary School Census 2017.” Dhaka: Monitoring and Evaluation Division, Directorate of Primary Education, Government of Bangladesh. . (DPE) Directorate of Primary Education 2016a. “National Student Assessment 2015 for Grades 3 and 5 National Report.” Dhaka: Monitoring and Evaluation Division, Directorate of Primary Education, Government of Bangladesh. (DPE) Directorate of Primary Education 2016b. “Annual Primary School Census 2016.” Dhaka: Monitoring and Evaluation Division, Directorate of Primary Education, Government of Bangladesh. (DPE) Directorate of Primary Education 2015. “Annual Primary School Census 2015.” Dhaka: Monitoring and Evaluation Division, Directorate of Primary Education, Government of Bangladesh. 60 (DPE) Directorate of Primary Education 2014a. “National Student Assessment 2013 for Grades 3 and 5 National Report.” Dhaka: Monitoring and Evaluation Division, Directorate of Primary Education, Government of Bangladesh. (DPE) Directorate of Primary Education 2014b. “Annual Primary School Census 2014.” Dhaka: Monitoring and Evaluation Division, Directorate of Primary Education, Government of Bangladesh. (DPE) Directorate of Primary Education 2012. “2011 National Student Assessment for Grades 3 and 5 National Report.” Dhaka: onitoring and Evaluation Division, Directorate of Primary Education, Government of Bangladesh. Glinskaya, E., 2005. “Education and Health Expenditures in Bangladesh. Benefit Incidence Analysis.” Journal of Development Societies. Vol 21, Issue 1-2. Hoogeveen, J. G., Mariacristina, R., and Sansone, D., 2014. "Drivers of performance in primary education in Togo." CeRP Working Papers 145. Turin (Italy): Center for Research on Pensions and Welfare Policies. Hossain, I., 2013. “Measuring Effectiveness of Public Expenditures in Bangladesh: A Survey Based Case Study in the Education Sector.” Ritsumeikan Asia Pacific University, Japan: RCAPS Working Paper Series 13002. (MoF) Ministry of Finance 2017a. “A Diagnostic Study on Stipend Programmes in Bangladesh with Focus on Primary Education Stipend Project (PESP).” Dhaka: Ministry of Finance, Government of Bangladesh. (MoF) Ministry of Finance 2017b. “Bangladesh Economic Review 2017.” Dhaka: Economic Adviser's Wing, Finance Division, Ministry of Finance, Government of the People's Republic of Bangladesh. Rahman, M., Khan, T. I., , and Sahbih, M. A., 2016. “Education Budget in Bangladesh: an analysis of trends, gaps and priorities.” Dhaka: Campaign for Popular Education, Bangladesh and Centre for Policy Dialogue, Bangladesh. Steer L., Rabbani F., and Parker, A., 2014. “Primary education finance for equity and quality.” Brooke Shearer Working Paper Series Nro. 3. Global Economy and Development at Brookings. (UIS) Institute for Statistics 2019. UNESCO database, http://data.uis.unesco.org, [accessed: 1/16/2019]. (WDI) World Development Indicators 2019. World Development Indicators Online (WDI) database, https://databank.worldbank.org/data/source/world-development-indicators,world-development- indicators,world-development-indicators# [accessed: 1/16/2019] World Bank 2010. “Bangladesh Public Expenditure and Institutional Review: Toward a Better Quality of Public Expenditure.” Washington D. C.: World Bank Group.. World Bank 2015. “Bangladesh Public Expenditure and Institutional Review Update.” Washington D. C.: World Bank Group. 61 Annex A: Supplemental Tables and Graphs Table A2.1: Student enrollment for different levels of education, 2017 % of Total Student % of Total Number of (by Type of Institution Enrollment (by Level ) Institutions Level) Primary Education A MoPME/DPE Managed School 13,403,989 77.70 75,991 56.75 1 Government Primary School (GPS) 8,875,278 51.45 38,879 29.03 Newly Nationalized Primary School 2 (NNPS) 3,932,235 22.79 26,159 19.54 Registered Non-Government Primary 3 School (RNGPS) 17,719 0.10 180 0.13 Non-Registered Non-Government 4 Primary School (NRNGPS) 275,909 1.60 3,001 2.24 5 Experimental 10,431 0.06 61 0.05 6 Community 13,296 0.08 112 0.08 Reaching Out of School Children (ROSC) 7 School 260,366 1.51 7,371 5.50 8 Shishu Kallayan 18,755 0.11 228 0.17 B MoE Managed School 1,856,781 10.76 12,182 9.10 9 High Madrasa Attached Ebtedayee 864,305 5.01 6,581 4.91 10 High School Attached Primary 612,513 3.55 1,734 1.29 11 Ebtedayee 379,963 2.20 3,867 2.89 C MoC Managed school 1,229,520 7.13 23,599 17.62 12 Kindergarten 1,224,683 7.10 23,544 17.58 13 Tea Garden 4,837 0.03 55 0.04 D Managed by other authorities 50,702 0.29 2,747 2.05 14 Temple Based Education Center 2,596 0.02 775 0.58 Social Welfare Based School (Minsitrty of 15 Social Welfare) 2,751 0.02 57 0.04 62 Deaf and Dumb (Minsitrty of Social 16 Welfare) 1,766 0.01 30 0.02 17 Blind School 218 0.00 3 0.00 18 Jail Attached 520 0.00 3 0.00 19 Mosque based Education Center 7,205 0.04 1,111 0.83 20 Hilly Parishad Directed Schools 2,181 0.01 46 0.03 21 Others Types 28,969 0.17 668 0.50 22 Quami Madrasa 4,496 0.03 54 0.04 E NGO Bureau Managed School 710,576 4.12 19,385 14.48 23 NGO Schools 224,971 1.30 4,793 3.58 24 Learning Centers(LC) of BRAC 406,610 2.36 12,394 9.26 25 Other NGO Learning Centers 78,995 0.46 2,198 1.64 Total Primary 17,251,568 100.00 133,904 100.00 Secondary and Higher Secondary Education including Technical and Vocational Education Secondary and Higher Secondary Education 1 Secondary Govt Institution 318,513 2.47 368 1.08 2 Secondary Non. Govt Institution 9,788,576 76.01 19,480 57.23 3 Higher Secondary Govt Institution 31,688 0.25 45 0.13 4 Higher Secondary Non. Govt Institution 618,136 4.80 2,512 7.38 5 Dakhil and Alim Madrasa 1,782,014 13.84 7,982 23.45 Technical and Vocational Education (SSC and HSC Level) 6 Govt Institution 1319 0.01 15 0.04 7 Non. Govt Institution 337996 2.62 3,634 10.68 Total Secondary 12,878,242 100.00 34,036 100.00 Tertiary Education Degree and Masters Education 1 Public University 283,866 6.29 34 0.57 Colleges and Madrasas affiliated with public 2 university 387,953 8.60 1,491 24.92 63 3 Bangladesh Open University 318,526 7.06 1 0.02 Colleges and Madrasas affiliated with 4 National University 2,528,649 56.03 2,269 37.92 Colleges and Madrasas affiliated with 5 Islamic Arab University 87,143 1.93 1,336 22.33 Public University + All affiliated Colleges 6 and Madrasas under all public university 3,606,137 79.90 3,640 60.84 7 Private University 354,333 7.85 95 1.59 8 Public University + Private University 638,199 14.14 129 2.16 Public University+Aff. With Public 9 University + Private University 1,026,152 22.74 1,586 26.51 Public+ Private+ Affiliated College and 10 Madrasa Under all University 3,960,470 87.75 3,735 62.43 Technical and Vocational Education (Above HSC level) 11 Govt Institution 194215 4.30 273 4.56 12 Non. Govt Institution 358434 7.94 1,975 33.01 Subtotal 552649 12.25 2,248 37.57 Total Tertiary 4,513,119 100.00 5,983 100.00 GRAND TOTAL 34,642,929 173,923 Source: APSC 2017, BANBEIS 2017 64 Table A3.1: Linear probability model for being out of school Out of school Never Drop out (1) (2) (3) Female -0.036** -0.026** -0.011** [0.005] [0.003] [0.002] Age 0.012** -0.003** 0.006** [0.001] [0.001] [0.000] Consumption quintile 1 (1) 0.089** 0.062** 0.016** [0.008] [0.006] [0.005] Consumption quintile 2 0.059** 0.046** 0.003 [0.010] [0.008] [0.004] Consumption quintile 3 0.036** 0.023** 0.005 [0.007] [0.006] [0.004] Consumption quintile 4 0.007 0.004 -0.002 [0.007] [0.005] [0.003] Female-headed household 0.018 0.013 -0.003 [0.014] [0.009] [0.006] Household head's education in years -0.006** -0.004** -0.001** [0.001] [0.001] [0.000] In urban area 0.053** 0.031** 0.011** [0.009] [0.007] [0.003] Both parents present in the household 0.003 0.004 -0.006 [0.011] [0.007] [0.005] Total number of children 0-17 years 0.003 0.002 0.002 [0.003] [0.002] [0.001] Constant -0.051** 0.058** -0.035** [0.016] [0.012] [0.008] Observations 37,211 37,211 37,211 R-squared 0.051 0.032 0.024 Note: Author's calculations using HIES 2016/17. Regressions include controls for divisions. Standard errors (presented in square brackets) calculated using survey's sampling design. ** p<0.01, * p<0.05, + p<0.1 (1) Omitted is consumption quintile 5 65 Table A3.2: Reasons for not attending to school by group in 2016 Do not Too No want to old to No schools Have Attending study go money/too close to to family For Primary more back expensive home work chores marriage All 25% 26% 27% 11% 6% 5% 1% Area Rural 20% 30% 26% 13% 5% 6% 1% Urban 37% 17% 28% 8% 7% 2% 1% Gender Female 23% 27% 31% 10% 3% 5% 1% Male 26% 26% 23% 12% 8% 5% 0% Quintile 1 27% 27% 29% 8% 4% 4% 1% 2 25% 20% 24% 15% 9% 7% 0% 3 22% 23% 38% 10% 2% 4% 1% 4 20% 38% 15% 19% 2% 5% 1% 5 26% 44% 4% 7% 14% 4% 1% Do not No want to Too old No schools Have Attending study to go money/too close to to family For Secondary more back expensive home work chores marriage All 32% 8% 18% 1% 26% 10% 6% Area Rural 30% 9% 17% 1% 25% 11% 7% Urban 34% 6% 19% 0% 27% 8% 5% Gender Female 27% 8% 20% 1% 14% 15% 15% Male 35% 9% 16% 0% 34% 7% 0% Quintile 1 28% 10% 22% 1% 23% 11% 6% 2 32% 9% 19% 1% 23% 11% 5% 3 33% 7% 16% 0% 29% 9% 7% 4 31% 6% 14% 0% 30% 10% 8% 5 39% 9% 11% 0% 27% 7% 7% 66 Do not No want to Too old No schools Attending study to go money/too close to Have to family Tertiary more back expensive home work chores For marriage All 27% 13% 7% 0% 20% 13% 20% Area Rural 25% 14% 6% 0% 20% 13% 22% Urban 33% 10% 8% 0% 20% 13% 17% Gender Female 25% 13% 5% 1% 6% 18% 32% Male 30% 12% 9% 0% 41% 7% 2% Quintile 1 26% 15% 6% 0% 20% 10% 22% 2 27% 12% 7% 0% 20% 13% 21% 3 24% 14% 8% 0% 20% 15% 19% 4 27% 11% 7% 0% 20% 16% 18% 5 32% 10% 4% 1% 20% 11% 22% Table A3.3: Percentage of students who are on or below grade proficiency by urban/rural area 2017 Band Bangla Math Grade 3 Grade 5 Grade 3 Grade 5 On Grade level (all) 74% 12% 41% 17% Urban 79% 17% 39% 15% Rural 73% 10% 43% 17% Below grade level (all) 26% 88% 59% 88% Urban 21% 83% 61% 85% Rural 27% 90% 57% 83% 67 Figure A3.1: Percentage of students who are on grade proficiency by urban/rural area 2017 Grade 5 Math 17% 15% Grade Subject Grade 3 Math 43% 39% Grade 5 Bangla 10% 17% Grade 3 Bangla 73% 79% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% Percentage Rural Urban Table A3.4: Gender-wise percentage of students in different performance Band Bangla English Mathematics Grade 6 Grade 8 Grade 6 Grade 8 Grade 6 Grade 8 Band 6 and above (all) 10 22 5 7 6 6 Female 9 22 4 6 4 15 Male 10 23 5 9 8 18 Band 5 (all) 24 32 10 12 14 16 Female 25 32 10 12 13 15 Male 22 32 10 13 15 18 Band 4 (all) 36 32 20 30 24 35 Female 36 33 21 31 24 33 Male 36 32 21 28 25 41 Band 3 (all) 24 12 36 42 33 38 Female 24 12 36 42 34 41 Male 25 13 36 41 32 34 Band 2 or below (all) 6 1 29 9 23 5 Female 6 1 29 9 25 7 Male 7 1 29 9 20 4 68 Figure A3.2: Gender disparity in band ≥ 6 Percentage difference Grade 6 Grade 8 Grade 6 Bangla Grade 6 English Mathematics Grade 8 Bangla Grade 8 English Mathematics 0 -1 -1 -1 -1 -2 -3 -3 -3 -4 -4 -5 Grade Subject Figure A3.3: Gender disparity in Band≤ 2 6 5 5 4 3 3 Percentagge 2 1 0 0 0 0 -1 -1 Grade 6 Bangla Grade 6 English Grade 6 Grade 8 Bangla Grade 8 English Grade 8 -2 Mathematics Mathematics Grade Subject 69 Table A3.5: Percentage of students in different performance bands by urban/rural area Band Bangla English Mathematics Grade 6 Grade 8 Grade 6 Grade 8 Grade 6 Grade 8 Band 6 and above (all) 10 22 5 7 6 6 Urban 24 41 14 20 8 9 Rural 5 16 1 2 5 5 Band 5 (all) 24 32 10 12 14 16 Urban 26 29 14 20 17 18 Rural 23 33 8 10 13 15 Band 4 (all) 36 32 20 30 24 35 Urban 30 22 21 25 23 33 Rural 38 36 20 31 25 36 Band 3 (all) 24 12 36 42 33 38 Urban 15 7 32 29 33 35 Rural 27 14 38 46 33 39 Band 2 or below (all) 6 1 29 9 23 5 Urban 4 1 18 5 19 5 Rural 7 1 32 10 25 6 Figure A3.4: Percentage of students on band level ≥ 6 Pecentage of Students 0 5 10 15 20 25 30 35 40 45 Grade 6 Bangla Grade 8 Bangla Grade Subject Grade 6 English Grade 8 English Grade 6 Math Grade 8 Math Urban Rural 70 Table A3.6: Median monthly expenditures by type, 2016 (in Takas) All levels Fees Books Tutoring Transport Others All All 123 150 193 0 158 802 Quintiles 1 21 73 0 0 73 315 2 61 119 112 0 118 548 3 110 146 202 0 157 773 4 191 192 281 0 209 1127 5 355 258 483 0 311 1933 Primary Fees Books Tutoring Transport Others All All 12 58 0 0 83 300 Quintiles 1 8 35 0 0 46 143 2 9 50 0 0 71 232 3 13 58 25 0 100 328 4 18 71 100 0 125 442 5 60 100 200 0 167 730 Secondary Fees Books Tutoring Transport Others All All 135 167 250 0 158 843 Quintiles 1 45 92 50 0 71 363 2 88 129 167 0 108 588 3 125 167 250 0 153 827 4 180 192 350 0 200 1100 5 275 250 600 0 292 1817 Tertiary Fees Books Tutoring Transport Others All All 350 250 167 75 167 1338 Quintiles 1 192 158 83 0 58 598 2 249 174 67 0 83 758 3 275 208 125 25 123 1004 4 358 233 200 83 158 1363 5 462 300 250 167 300 2117 71 Note: Authors’ calculations using HIES 2016/17. In 2016 Takas. Table A6.1: OLS Regression between primary level outcomes and spending primary level Net Gross Repetitio Surviva Efficienc Dropou attendance attendance n rate l rate y ratio t rate rate rate District variables (1) (2) (3) (4) (6) (7) Log public education spending per student -0.77 -5.91+ -1.76 0.58 -2.01 -5.01 (1.03) (3.19) (1.98) (3.02) (1.58) (4.94) Number of students per teacher 0.02 -0.38* -0.36** 0.59** -0.15+ 0.06 (0.05) (0.14) (0.10) (0.20) (0.08) (0.17) Log private education spending per student -0.55 -1.74 0.61 -0.19 -0.01 4.06** (0.56) (1.46) (0.71) (1.36) (0.66) (1.40) Log population -1.19* -0.67 1.14 -1.17 0.01 -0.49 (0.58) (2.08) (1.20) (2.05) (0.87) (2.05) Share of rural population -0.01 -0.06 -0.04 0.01 0.11* 0.23** (0.02) (0.07) (0.04) (0.06) (0.05) (0.07) Poverty rate -0.01 -0.10+ -0.01 0.08 -0.01 -0.03 (0.02) (0.06) (0.04) (0.05) (0.03) (0.06) Literacy rate for adults -0.07* 0.27* 0.20** -0.19* 0.06 -0.11 (0.03) (0.11) (0.06) (0.08) (0.06) (0.12) 159.52* Constant 39.88* * 77.53* 22.28 104.16** 119.34+ (16.27) (53.86) (33.53) (55.32) (25.40) (68.15) Observations 64 64 64 64 64 64 R-squared 0.17 0.35 0.47 0.42 0.20 0.21 Robust standard errors in parentheses ** p<0.01, * p<0.05, + p<0.1 72 Table A6.2: OLS Regression between net attendance rate in secondary and spending Net attendance rate Gross attendance rate District variables (1) (2) Log public education spending per student 4.06 2.94 (4.20) (4.93) Number of students per teacher -0.22* -0.18 (0.10) (0.12) Log private education spending per student 1.85 0.45 (1.51) (1.56) Log population -0.61 -2.91+ (1.46) (1.64) Share of rural population 0.25** 0.17* (0.07) (0.08) Poverty rate 0.05 0.07 (0.06) (0.07) Literacy rate for adults 0.29** 0.33** (0.09) (0.10) Constant -1.57 52.60 (50.81) (61.59) Observations 64 64 R-squared 0.51 0.45 Robust standard errors in parentheses ** p<0.01, * p<0.05, + p<0.1 73 Table A6.3: OLS regression between primary level learning outcomes and spending Bangla grade 3 Bangla grade 5 Math grade3 Math grade 5 District variables (1) (2) (3) (4) Log public education spending per student -0.07 -0.03 -0.03 -0.01 (0.05) (0.03) (0.07) (0.04) Number of students per teacher -0.00 -0.00 0.00 0.00 (0.00) (0.00) (0.01) (0.00) Log private education spending per student -0.04+ 0.00 -0.04 -0.03 (0.02) (0.01) (0.04) (0.03) Log population -0.01 0.00 -0.02 -0.01 (0.03) (0.02) (0.05) (0.03) Share of rural population -0.00 0.00 0.00 0.00 (0.00) (0.00) (0.00) (0.00) Poverty rate 0.00 0.00 0.00 0.00 (0.00) (0.00) (0.00) (0.00) Literacy rate for adults 0.00 0.00+ 0.00 0.00+ (0.00) (0.00) (0.00) (0.00) Constant 1.91* 0.22 0.88 0.40 (0.85) (0.52) (1.29) (0.79) Observations 63 64 63 64 R-squared 0.09 0.09 0.07 0.11 Robust standard errors in parentheses ** p<0.01, * p<0.05, + p<0.1 Source: Authors' estimates based on NSA 2017 and BOOST 2014 data 74 Annex B: School Stipend Programs in Bangladesh Stipends for students constitute an important demand side intervention to enhance education access and student retention. The government of Bangladesh currently provides stipends to both primary and secondary school students. At the primary level, the Primary Education Stipend Program (PESP) operated by MoPME is the main avenue though which students receive stipends. At the secondary level, there are a number of different stipend schemes implemented by MoE. Primary Education Stipend Program PESP is a conditional cash assistance program aimed at increasing the enrollment, attendance, and retention rates of primary school children. In 2016-17, it provided stipends to approximately 11.1 million students. While PESP was originally designed to support poor primary school students from rural areas, it became universal starting in 2015-16. However, to qualify for stipend support, a student must maintain 85% monthly attendance, take all the school examinations and attain a minimum exam score of 33% in each subject in her grade with certain exceptions (MoF 2017). Furthermore, only the following types of primary schools are eligible to participate in PESP: GPS, government recognized Independent Ebtedayee Madrassahs, high school attached primary schools, high madrassahs independent of Ebtedayee Madrassahs, primary schools run by Shishu Kollyan (SKT), class VI to VIII attached to primary schools, and only GPS under MoPME and Idependent Ebtedayee Madrassahs in city corportaions and municipalities. The stipend amount per student varies according to the number of eligible children in the student’s family. Students from families with up to two eligible children receive BDT 100 per month. On the other hand, families with three and four eligible children receive BDT 250 and BDT 300 per month, respectively. PESP also provides BDT 50 per month for each pre-primary student. The payment of stipends is made to each beneficiary’s mother or guardian thorough the state owned Rupali Bank using mobile banking services. The head teachers of participating schools help connect the beneficiaries’ mothers/guardians with Rupali Bank for the purpose of opening mobile bank accounts. Secondary education stipends At the secondary level, stipends are currently provided to eligible students in grades 6 to 10 through the following three projects/programs under MoE: Secondary Education Stipend Project 2nd phase (SESP), Secondary Education Development Program (SEDP), and the Secondary Education Sector Investment Program (SESIP). These stipend programs aim to increase enrollment, equity in access, and retention at the secondary level. As in the case of PESP, these are conditional cash transfers delivered through mobile banking. In 2017-18, around 1.7 million and 0.4 million beneficiaries were targeted by SESP and SESIP, respectively. In addition, 1.8 million students were targeted by the Secondary Education Quality and Access Enhancement Project (SEQAEP), which closed in December 201739. While the different secondary stipends programs are targeted mainly towards the poor, they differ from one another in the way beneficiaries are selected and in their geographical coverage. For example, SEQAEP used an objective proxy means testing approach to select beneficiaries and covered 250 upazillas. On the other hand, SESIP only covers 54 upazillas, and has provisions for selecting beneficiaries by school-based committees on the basis of certain pro-poor eligibility criteria. The selection of beneficiaries under SESP (which covers 183 upazillas) is also done by school-based committees using 39 The stipend scheme in SEDP, which was launched in 2018, builds on the experience from the Secondary Education Quality and Access Enhancement Project (SEQAEP), which closed in December 2017. 75 project-specific pro-poor eligibility criteria but covers 183 Upazillas. The recently launched SEDP, which plans to implement a poverty-targeted stipends program across the whole nation, intends to use a harmonized stipends guideline for beneficiary selection and stipend amounts. In all these programs, to receive stipends and ensure its continuation, students must maintain minimum attendance, pass in the annual examination and remain unmarried till SSC/ Dhakil. A recipient of any secondary level stipend program is not supposed to pay any tuition fee. In other words, the overall benefit available is a fee waiver along with the stipend. But the students receive different amounts of stipends depending on which program covers them, as all these programs have their own criteria of benefit size and stipend amounts for grades 6 to 10. Students under the coverage of SESIP and SESP start from yearly stipend rate of BDT 1380 in grade 6 which they continue to receive in grade 7. From grade 8 the amount increases with each grade in both programs, but SESIP provides higher increment in each grade making the yearly rate of grade 10 BDT 3510 while SESP provides BDT 2790 in that same grade. Students under the coverage of SEQAEP received yearly stipends of BDT 1500 starting at grade 6, which was slightly higher than the amountin the other two programs for the same grade, and kept getting yearly increments in each grade till grade 10 when they received BDT 3510 per year.40 40 SESIP yearly stipend rate: grades 6 &7 – BDT 1380, grade 8 – BDT 1680, grade 9 – BDT 2280 and grade 10 – BDT 3510; SESP yearly stipend rate: grades 6 &7 – BDT 1380, grade 8 – BDT 1620, grade 9 – BDT 2040 and grade 10 – BDT 2790; SEQAEP yearly stipend rate: grades 6 – BDT 1380, grade 7 – BDT 1800, grade 8 – BDT 2220, grade 9 – BDT 2760 and grade 10 – BDT 3510. 76