CAMBODIA EDUCATION SECTOR PUBLIC EXPENDITURE TRACKING AND QUALITY OF SERVICE DELIVERY SURVEY European Union © 2018 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. CAMBODIA EDUCATION SECTOR PUBLIC EXPENDITURE TRACKING AND QUALITY OF SERVICE DELIVERY SURVEY European Union Acknowledgements This report was prepared by a World Bank team co-led by Sokbunthoeun So (Senior Public Sector Specialist) and Tsuyoshi Fukao (Senior Education Specialist). Other core team members were Russell Craig and Samsen Neak, consultants of the World Bank. The design of the study instruments was contributed by Seo Yeong Hong (Consultant) and Takiko Igarashi (Education Specialist, World Bank). Data collection was undertaken by BN Consult. The team would like to thank H.E. Academician Hang Chuon Naron, Minister of MoEYS; H.E. Pit Chamnan, Secretary of State of MoEYS; and H.E. Thoung Boran, Under-secretary of State of MoEYS for their support, valuable guidance, and comments. The team also benefited greatly from comments and interactions with the PETS Committee at various stages of the study, from the concept note and survey instruments meetings to presentation of findings. Throughout the study, the team also received inputs and support from Mr. Tep Phyorith, Director of Finance Department, and his colleague Mr. Soun Sokhom, who acted as a focal point between the World Bank team and PETS Committee in coordinating meetings and discussions. The team received valuable guidance from Fily Sissoko (Practice Manager), Robert Taliercio (Practice Manager), Harry Anthony Patrinos (Practice Manager), Miguel Eduardo Sanchez Martin (Senior Country Economist), Shabih Ali Mohib (Program Leader), and Simeth Beng (Senior Education Officer). The team would like to thank peer reviewers Sachiko Kataoka (Senior Economist); GED02 Review Team; and Bill Dorotinsky (Adviser) for all the useful comments provided. The team particularly thanks Magnus Saemundsson for his support and intellectual inputs to the study. Da Lin, Linna Ky, Lyden Kong, and Pengseng Tan provided excellent logistical and administrative support, which is gratefully acknowledged. This report was funded by Externally Funded Output (EFO) contributed by Sweden and Public Financial Management Multi-Donor Trust Fund contributed by the European Union, Australia, and Sweden. Table of Contents ACKNOWLEDGEMENTS............................................................................. 1 TABLE OF CONTENTS................................................................................. 3 LIST OF TABLES........................................................................................... 5 LIST OF FIGURES......................................................................................... 7 ABBREVIATIONS AND ACRONYMS.......................................................... 9 EXECUTIVE SUMMARY............................................................................. 11 Education funding........................................................................................ 11 Sampling and data collection........................................................................ 12 The surveyed schools and personnel.............................................................. 12 School fund flows......................................................................................... 12 Quality of service delivery in education........................................................ 13 Summary of findings on fund flows.............................................................. 15 Summary of findings on service delivery quality........................................... 16 Policy recommendations............................................................................... 16 1. BACKGROUND AND CONTEXT........................................................ 18 Objectives of the Cambodia Education PETS-QSDS................................... 18 Cambodia’s education system........................................................................ 19 Education financing in Cambodia................................................................ 19 Previous PETS studies in Cambodia............................................................. 23 2. METHOD............................................................................................... 24 Sampling...................................................................................................... 24 Survey instruments ...................................................................................... 24 Data collection and cleaning......................................................................... 25 4 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey 3. CHARACTERISTICS OF THE SAMPLE SCHOOLS........................... 27 School types and enrollments....................................................................... 27 School Directors........................................................................................... 32 School support committees........................................................................... 33 Teachers........................................................................................................ 34 Secondary students....................................................................................... 36 4. EDUCATION REVENUES AND FUNDING........................................ 38 Flow of funds to schools............................................................................... 38 Flow of SOB funds....................................................................................... 39 Flow of SIG funds........................................................................................ 45 5. QUALITY OF SERVICE DELIVERY IN EDUCATION........................ 47 School spending........................................................................................... 47 Development of quality indices.................................................................... 50 School quality index............................................................................ 51 Financial quality index........................................................................ 53 Classroom quality index...................................................................... 55 Environmental quality index............................................................... 56 6. QUALITY ASPECTS............................................................................... 59 School quality aspects................................................................................... 59 Financial quality aspects................................................................................ 62 Classroom quality aspects............................................................................. 65 Environmental quality aspects...................................................................... 67 Summary quality aspects............................................................................... 70 7. MATHEMATICS AND PHYSICS ACHIEVEMENT............................. 71 Mathematics and physics tests...................................................................... 71 Mathematics results...................................................................................... 72 Physics results............................................................................................... 74 8. SUMMARY OF FINDINGS AND POLICY IMPLICATIONS.............. 77 Summary of findings on fund flows.............................................................. 77 Summary of findings on service delivery quality........................................... 78 Policy recommendations............................................................................... 79 9. ANNEXES............................................................................................... 81 Correlational aspects of the indices and school characteristics....................... 81 Primary correlations............................................................................ 82 Secondary correlations......................................................................... 83 Correlations of Mathematics and Physics test results .......................... 84 Bibliography................................................................................................. 85 List of Tables Table 1: SOB Allocation Formula.................................................................... 21 Table 2: SIG Allocation Formula ..................................................................... 21 Table 3: Survey Instruments............................................................................. 25 Table 4: Sampling Distribution of the Survey and Actual Implementation....... 26 Table 5: Sample and National Schools Compared............................................ 28 Table 6: Sample Schools’ Average Enrollment by Province and Type................ 29 Table 7: Characteristics of School Directors by Gender and Level.................... 33 Table 8: Interview Status of SSC Members by Level and Location of School.... 34 Table 9: SSC Membership by Location and Level............................................ 34 Table 10: Teacher Characteristics....................................................................... 35 Table 11: Teaching Career.................................................................................. 35 Table 12: Secondary Student Characteristics...................................................... 36 Table 13: Secondary Students’ School Experience .............................................. 37 Table 14: Flow of SOB Funds from Provincial Treasury (PT) in FY 2015 (Khmer Riels)..................................................................................... 40 Table 15: Flow of SOB funds from Provincial Treasury (PT) in FY 2016 (Khmer Riels)..................................................................................... 40 Table 16: Average SIG Amounts per School Reported by POE and School 2014/15.............................................................................................. 45 Table 17: Average SIG Amounts per School Reported by POE and School 2015/16.............................................................................................. 46 Table 18: SIG Receipt Date 2014-15 and 2015-16............................................ 46 6 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Table 19: Average Total Operational Funds (SOB+SIG) Received in 2015-16, by Quintile and Level......................................................................... 48 Table 20: Items in the School Quality Index...................................................... 51 Table 21: School Quality Items, by School Type................................................. 52 Table 22: School Quality Items, by School Location.......................................... 52 Table 23: Items in the Financial Quality Index................................................... 53 Table 24: Financial Quality Items, by School Type............................................. 54 Table 25: Financial Quality Items, by School Location....................................... 54 Table 26: Items in the Classroom Quality Index................................................. 55 Table 27: Classroom Quality Items, by School Type........................................... 55 Table 28: Classroom Quality Items, by School Location.................................... 56 Table 29: Items in the Environmental Quality Index.......................................... 57 Table 30: Environmental Quality Items, by School Type ................................... 57 Table 31: Environmental Quality Items, by School Location.............................. 58 Table 32: Quintile Divisions of Total Operational Funds Received by Schools in 2016-17............................................................................. 60 Table 33: Operational Fund Receipt Quintiles, by School Level......................... 60 Table 34: Quintile Divisions of SIG Quality Spending Percentages.................... 62 Table 35: Summary Quality, Funding, and School Characteristics, by Level and Location....................................................................................... 70 Table A1: Primary correlation matrix ........................................................................... 82 Table A2: Secondary correlation matrix.............................................................. 83 Table A3: Test results correlation matrix............................................................. 84 List of Figures Figure 1: Comparison of SOB and SIG Processes............................................ 22 Figure 2: School Types by Province................................................................. 29 Figure 3: Average Class Size by Province and School Level.............................. 30 Figure 4: Student-Classroom Ratios by Province and Level, with and without shift adjustment................................................................................ 31 Figure 5: Student-Teacher Ratio by Province and Level, with and without shift adjustment....................................................................................... 32 Figure 6: Grade 9 Students’ Age by Gender..................................................... 37 Figure 7: Knowledge of SOB Amounts by Person Responsible for Accounts... 38 Figure 8: SOB Funds Sent and Reported, Fiscal Year 2015.............................. 39 Figure 9: SOB Funds Sent and Reported in Fiscal Year 2016........................... 41 Figure 10: SOB Amounts Sent/Received by Level and Location of School........ 42 Figure 11: Average Months of Delay from Start of Quarter in POE Request for SOB Transfers.................................................................................. 43 Figure 12: SOB Fund Processing Times for POE to PT and PT to POE........... 44 Figure 13: Turnaround Time from POE back to POE....................................... 44 Figure 14: Average Operational Funds Received by Type and Level of School... 49 Figure 15: Percent of SIG Funds Spent on Quality Aspects by School Type and Location........................................................................................... 49 Figure 16: Average SIG Amount Spent on Quality Aspects, by Quintile and School Type...................................................................................... 50 Figure 17: School Quality Index Scores by Province ......................................... 59 Figure 18: Average School Quality Index Scores, by School Type and Location.60 8 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Figure 19: School Quality Spending by Operational Spending Quintile............ 61 Figure 20: Average School Quality Index, by Quality Spending Quintile.......... 62 Figure 21: Average Financial Quality Index Score, by Province.......................... 63 Figure 22: Average Financial Quality Index Score, by School Type and Location........................................................................................... 63 Figure 23: Mean Financial Quality Score, by SIG Quality Percent Quintiles..... 64 Figure 24: Mean financial quality index by operational funding quintile........... 64 Figure 25: Mean Classroom Quality Index, by School Type and Location......... 65 Figure 26: Mean Classroom Quality Index, by Province.................................... 66 Figure 27: Mean Classroom Quality Index Score, by SIG Quality Spending Quintile........................................................................................... 66 Figure 28: Mean Classroom Quality Index Score, by Operational Funding Quintile........................................................................................... 67 Figure 29: Mean Environmental Quality Index Scores, by Province.................. 68 Figure 30: Mean Environmental Quality Index, by School Type and Location....68 Figure 31: Mean Environmental Quality Score by Operational Funding Quintile........................................................................................... 69 Figure 32: Mean Environmental Score, by SIG Quality Spending Quintile....... 69 Figure 33: Math and Physics Test Percentage Distribution................................ 72 Figure 34: Mean Mathematics Test Scores, by Type and Location of School...... 73 Figure 35: Mean Mathematics Test Scores, by Province..................................... 73 Figure 36: Mean School Quality Score, by Math Test Quintile.......................... 74 Figure 37: Mean Physics Test Results, by Province............................................ 75 Figure 38: Mean Physics Test Results, by Location............................................ 75 Figure 39: Mean School Quality Scores, by Physics Test Quintile...................... 76 Figure 41: Mean Classroom Quality Score, by Physics Test Quintile................. 76 Figure 40: Mean Financial Quality Scores, by Physics Test Quintile.................. 76 Figure 42: Mean Environmental Quality Score, by Physics Test Quintile.......... 76 Abbreviations and Acronyms ADB Asian Development Bank BMC Budget Management Center BSEC Budget Strategy and Enforcement Center CCLS Cambodia Child Labor Survey CDRI Cambodia Development Research Institute COM Council of Ministers CSES Cambodia Socio-Economic Survey DBFA Department of Budget and Financial Affairs DC Domestic Canvassing DCO Direct Contracting DOE District Office of Education DHS Demographic and Health Survey DP Direct Purchase DPP Department of Public Procurement ECE Early Childhood Education EFA Education for All EMIS Education Management Information System ESP Education Strategic Plan IFAPER Integrated Fiduciary Assessment and Public Expenditure Review MEF Ministry of Economy and Finance MOEYS Ministry of Education, Youth and Sport MOH Ministry of Health MRD Ministry of Rural Development NA National Assembly NAA National Audit Authority NBC National Bank of Cambodia 10 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey NT National Treasury O&M Operation and Maintenance OPM Office of Prime Minister PAP Priority Action Program PB Program Budget PEAC Pre-qualification, Evaluation and Award Committee PEFM Public Expenditure and Financial Management PFMRP Public Financial Management Reform Program POE Provincial Office of Education PETS Public Expenditure Tracking Survey PNBC Provincial Branch of the National Bank of Cambodia POEF Provincial Office of Economy and Finance PT Provincial Treasury PU Procurement Unit QSDS Quality Service Delivery Survey RGC Royal Government of Cambodia Sida Swedish International Development Cooperation Agency SIF School Improvement Fund SIG School Improvement Grant SNEC Supreme National Economic Council SOB School Operational Budget SSC School Support Committee WB World Bank Executive Summary The Ministry of Education, Youth and Sport (MoEYS) and the Ministry of Economy and Finance (MEF), together with key Development Partners, conducted a Public Expenditure Tracking Survey (PETS) and Quality of Service Delivery Survey (QSDS) for the education sector in Cambodia. PETS identify resource use and leakages by examining flows of funds and materials from the central government to local service providers via regional and local governments. QSDS are multi-purpose surveys that examine the efficiency of frontline service delivery and the dissipation of resources by collecting information on service providers and various agents in the system. Several PETS were conducted in Cambodia in the 2005-2013 period. They generally found leakage to be low but the timeliness and consistency of funds delivery to service providers to be poor. This PETS-QSDS will assist the Royal Government of Cambodia in its aim to improve the efficiency and effectiveness of public service delivery through several key reforms. The educational structure of schooling in Cambodia is six years of primary, followed by three years of lower secondary then three years of upper secondary. Prior to primary, there is at least one year for pre-school education. Cambodia also has technical and vocational education, which commences after completion of grade 9. Higher education commences after grade 12. Education funding Over the past decade, Cambodia’s experience with public funding of education has been mixed. Public recurrent funding of education has been increasing in Cambodia, with government commitment to further increases in both teacher wages and school operational funding. However, education budgets per student are still very low in Cambodia compared to neighboring countries at all levels of schooling. Furthermore, actual expenditure has not always reflected even the limited growth in budgets. Execution was only 86 percent of budget in 2014/15, although this improved to 90 percent in 2015/16 and 94 percent in 2016/17. 12 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Several changes have been made to the process of funding the operational aspects of schools in the past few years. Funding of School Operational Budgets (SOB) is now all based on the Program Budget and accounted for in a uniform manner. All schools are required to have bank accounts and receive their transfers of funds from the government directly into those accounts. Since 2014, all schools have received School Improvement Grants (SIG) from the Swedish International Development Cooperation Agency (Sida). For SOB funds, the Cambodian government has a clear formula for allocation, with a fixed amount per school that differs according to the location and size of the school. In addition, the SOB’s formula has a per capita student allocation. The SIG fund also has a clear formula, albeit simpler than that of the SOB for the fixed amount per school. The budgeting processes of SOB and SIG funds also differ. Sampling and data collection The study selected a sample of 400 schools (300 primary and 100 lower secondary) which provides reliable national estimates for primary and secondary schools and for provincial primary schools. The sample was drawn using two- stage stratified random sampling. To capture information, the study employed 13 survey instruments to gather information from key actors involved in the SOB and SIG fund processes. Data collection was undertaken by a contracted firm and conducted from December 2016 to March 2017. The surveyed schools and personnel Some key aspects of the surveyed schools: • Of the 300 primary schools, 127 (or 42 percent of primary schools) were simply standalone primary schools, while 173 (58 percent) had an attached early childhood center. Three in five of the sampled secondary schools were lower secondary, and the remainder were full secondary schools covering grades 7-12. • Primary school directors typically went to post-secondary college for their training, while secondary school directors attended a higher education institution and received a Bachelor’s or Master’s degree. • On average, the school support committee—the joint school-community committee that is responsible for school planning and budgeting, overseeing spending of the operational funds, and keeping records—had about six members, with around 20 percent of the committee being female. Secondary rural schools had the largest committees, while secondary urban schools had the smallest. The latter also had the smallest proportion of women members, while primary urban schools had the largest. School fund flows The analysis of fund flows yielded three key findings regarding leakage, timeliness, and recordkeeping. SOB funds flow from the MEF and through the Provincial Treasury to school accounts. SIG funds come from Sida and flow through MoEYS to separate school bank accounts.  | 13 No leakage in fund transfers to schools No leakages of funds took place in the transfers to schools. Almost all school accountants knew the amount of SOB to which their schools were entitled, which implies that schools would know if there were shortfalls in the amount of funds transferred to their bank accounts. All schools in the study’s sample received their full amount of the SOB fund. Late or slow disbursement of funds The main challenge is the delay in disbursement at the beginning of the fiscal year, which could result in inefficiency of funds use. Schools expected to receive their first wave of funds in January but received it several months later, for example in April/May 2015 and in February/March 2016. The significant delay in funds transfer to schools at the beginning of each fiscal year is related to the budget request and approval processes. Like SOB, disbursement of SIG also faced some delays, albeit shorter. The timeliness of fund receipts improved to some degree in the 2015-2016 school year compared to 2014-2015, but nonetheless, delays continued. Rigidity of SOB subaccounts was identified in previous PETS and remains an ongoing challenge. There are now 12 subaccounts in two SOB chapters (60 and 61) in which government funds flow to schools and in which the schools have to account to their respective POE. Most schools receive their funds allocated to the subaccounts even if these amounts do not reflect the needs of specific schools. (e.g. they may receive funds for paying for electricity even though the school is not connected). Moving funds between these lines is difficult and time consuming, taking many weeks or even months. Poor recordkeeping at the school level Over-reporting and underreporting of fund receipt is related to poor recordkeeping for both SOB and SIG funds. No single pattern of poor financial recording can be found according to type or location of school from 2015 to 2016. There are no indications that secondary schools do better than primary schools nor that schools in urban areas do better than schools in rural and remote areas in proper recording of their finances. Quality of service delivery in education Wide variation in funding between schools Fund availability for schools’ operations, which can affect school quality, varies considerably. Total operational funds available to schools (SOB and SIG combined) ranged from USD 248 to USD 25,833 per year. A school’s level and location were strongly linked to the average total operational funds available to the school. 14 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey The proportion of SIG funds spent on quality aspects1 is also important in assessing management and community intentions to improve the school. Total SIG funds reported to be spent on quality aspects ranged from USD 17 to USD 7,359 in 2015-16. It seems clear that even with a high level of management skills, small schools find it hard to make an impact on educational quality when total spending is at an average of USD 118. Four indices measuring quality High-quality schools facilitate student achievement and personal growth. Individual school quality has a number of dimensions and is a consequence of ‘internal’ aspects of school operation and ‘external’ aspects related to the socio- economic status of the parental community from which students are drawn. The indices here focus on operational aspects of the school, financial practices, and classroom and wider school environments. They provide information on things that can be changed through policy, training, and resource provision. School quality aspects. Spending more on quality is strongly associated with a higher school quality index score.2 Schools that spent about half or less of their SIG funds on quality scored less than half of the possible school quality index score. The one-fifth of schools in the sample that received the lowest amount of operational funds received between USD 248 and USD 1,092, while schools in the top quintile each received more than USD 3,103. Phnom Penh and Kampong Cham provinces were clearly ahead of the remaining provinces on this index, while Preah Vihear, a remote province in northwestern Cambodia, was well behind the other provinces in terms of overall school quality measures. Financial management quality aspects. The total amount of operational funding was not significantly associated with financial quality in the sampled schools. Only those schools that spent 85 percent or more of their SIG on quality aspects had a higher mean for their financial quality score. More than 20 percent of all schools scored a maximum of 12 points in financial management, showing that a substantial minority of schools could cover all the financial management 1 Schools can allocate their SIG quality spending to the following areas: (a) increase teaching and learning materials in the classroom; (b) introduce experimental and corner subject materials in the school; (c) increase library materials and rooms; (d) improve the school environment and playground to be clean and green; (e) Initiate life skill practices, including agriculture; (f ) support children with disabilities and vulnerable groups; (g) support slow learners and dropout prevention; (h) provide staff training and workshops, and (i) strengthen the quality and efficiency of school management. 2 Four quality indices were developed to measure general school quality, financial management quality, classroom quality and environmental quality. The school quality index has 16 items and was developed from the Director’s form, which covered much of the school functioning. It covers several aspects of the school, including recordkeeping, human resources, physical aspects, educational resources, and library resources. The financial quality index has 12 items which are part of the standard Ministry financial procedures and activities, and responsible school staff should have received training and manuals to cover them. The classroom quality index has ten items, and the data was gathered during a class observation session in either a Grade 3, Grade 5, or Grade 8 class. The first three items reflect the presence of a prepared teacher with a class, while the remaining seven reflect the presence of essential items for quality education such as textbooks, learning material, and suitable and adequate furniture for teacher and students. There are nine items in the environmental quality index which refer to both safety and utility. They include items related to safety, amenity, and utility. These can be improved with the quality funds available through SIG, and most can also be addressed through community involvement.  | 15 requirements. All provinces had fairly high average scores on financial management. Remote and rural schools of all types were more likely to have lower financial management scores. Either rural schools have had less training, or their school directors and financial officers have less capacity, than those in urban areas. Classroom quality aspects. The 20 percent of schools that spent the lowest proportion of their SIG funds on quality had the lowest mean classroom quality scores. However, the relationship between spending on quality and classroom quality then reverses and falls as schools spend a higher proportion of their SIG funds on quality, which is unexpected. Not a lot of differences were found among the sampled provinces on this measure, suggesting that addressing this issue will be a task for the respective Provincial Departments of Education and District Departments of Education since poor schools on this measure are scattered rather than located in a few provinces. Lower and full secondary schools had relatively low average scores on classroom quality. Environmental quality aspects. There is a strong positive relationship between the amount received by schools as operational funding (SOB plus SIG) and the environmental quality score. Schools that received the lowest amounts of operational funding had the lowest average scores, and schools that spent the lowest proportion of their SIG funds on quality outputs had the lowest mean environmental quality scores. Remote schools do very poorly in the assessment of school environmental quality. Addressing most of these environmental aspects can be costly—e.g., fencing the school grounds, providing sufficient safe toilets, establishing a bore for safe water in rural and remote regions. It is difficult to see how schools in the lowest quintile will be able to address such issues with their limited operating funds. Mathematics and physics test outcomes Levels of student achievement on mathematics and physics tests are low. Only eleven schools (five urban and six rural) have an average score of 50 percent or more on the mathematics test. Eight schools (two urban and six rural) averaged 50 percent or more on the physics test. Only five schools of the 100 in the sample averaged 50 percent or more on both tests, only one of which was urban. Summary of findings on fund flows Fund flows are expected to continue to be satisfactory, even with the anticipated change from two separate operational funding sources (SOB and SIG) to School Improvement Fund (SIF). The anticipated change is that the two amounts (SOB and SIG) will be combined at the national level to flow through the provincial treasury then into school accounts. Even if funds are received in full, the efficiency of school operations and hence educational quality are affected if fund delivery is sufficiently delayed. Discrepancies between what the Provincial Office of Education (POE) recorded as sent and what schools recorded as received suggest that recordkeeping at the school level is not sufficiently accurate. This situation is found in both SOB and SIG funds. 16 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Summary of findings on service delivery quality Quality aspects were analyzed relative to total operational funding and to percentage of SIG spending on quality using the four quality indices developed from questions and observations on several of the school-based survey forms. Phnom Penh ranked at the top on three out of four indices, and Preah Vihear ranked last on all four. Clear differences were found in most of the quality indices between the provinces, suggesting that it would be useful initially to identify and target lower-quality provinces to get rapid gains in quality rather than simply address individual schools across all provinces. Secondary schools had higher average school and environmental quality scores and lower average classroom quality scores than primary schools. Urban schools generally scored higher than rural and remote schools on financial quality. Secondary schools and non-urban schools were less likely to provide classroom storage and to have other books and reference material in the classroom. The total amount of operational funds (SOB plus SIG) received by a school was positively related to all quality indices except classroom quality. More funds meant better quality if they were dedicated to quality improvement purposes. The proportion of SIG spent on quality was also positively related to all indices except classroom quality. Plans for SIF expenditure on schools to 2021 as part of the new Sida funding proposal indicate a sharp rise in per school operational funds over the period commencing in 2018, almost doubling school operational funds in SIF over the current combined SOB and SIG funds to schools. Poor achievement on mathematics and physics tests reflect the quality of school outcomes and a need for quality enhancement, particularly for teaching and learning. With only five out of 100 schools in the sample scoring an average of 50 percent or more on both mathematics and physics tests, the challenge for quality improvement is huge. Policy recommendations Improve the timeliness of fund flows and align procedures for more efficiency in the management of school funds as SOB and SIG funds flow together as SIF. Currently, the separate provision of SIG funds through a different disbursement process means that they bridge the funding gap caused by the delay in the first quarter of SOB disbursement. Providing authorization for expenditure in quarter one to mirror that in quarter four of the previous year, with any rectification made in quarter two to balance the books, could be considered. Ease the rigidity of SOB subaccounts to enable schools to execute budget in a way that matches their needs. Addressing this will require allowing SOB expenditure based on school’s actual needs without the constraint of line items imposed in the 12 subaccounts of two SOB chapters (60 and 61). Schools can report on actual expenditures, while reporting of expenditure following line items  | 17 could still be retained by consolidating such expenditure at the provincial level. This can create efficacy in the use of SOB funds and improve development at the school level. Consider providing SOB payments to small schools in full early in the fiscal year. Receiving one-quarter of a small amount four times a year precludes small schools from easily making major expenditures, such as those commonly needed to enhance the school environment. Some measures to improve quality outcomes do not involve extra funding.3 Providing the intended number of instructional hours can be a powerful improver of outcomes. The strongest determinant of this in Cambodia is the prevalence of double-shifting in schools, which reduces overall instructional hours. This may involve school management measures such as providing the full number of instructional days and the full number of instructional hours. Place top priority for school funds on boosting the quality of teaching and learning to improve substantive student achievement outcomes. The items could include support for slow learner students from week one of the school year, coaching of less experienced teachers by more experienced teachers in the school or from neighboring schools, teaching and learning materials, and sufficient drinking water and toilets for boys and girls. Improve community knowledge and involvement in school budgets and spending. This could be achieved through (i) improving compliance checks with required budget disclosures to the school, committee, local community through public display, and anyone requesting school budget information during provincial and district staff visits to schools and (ii) regularizing disclosure of the budget and actual expenditures as part of opening and closing parental meetings at all schools. Address widespread poor financial recordkeeping at the school level. This can be achieved through regular refresher training which could be organized on a regional basis, grouping together geographically close provinces to provide enough numbers each year and to make training courses cost-effective. Refresher training helps to address skill shortages, especially as attrition removes trained staff and replacements have no training. Furthermore, it is important that soft skills be included in the regular management training in addition to hard skills such as accounting and recordkeeping. 3 There may be some costs if this involves ensuring that replacement teachers are available during regular teachers’ absences, and the provision of such replacements is much more difficult in remote and small schools than in large and urban schools. 1 Background and context The Royal Government of Cambodia aims to improve the efficiency and effectiveness of public service delivery through several reforms, including the Public Financial Management Reform Program (PFMRP), led by the Ministry of Economy and Finance (MEF). The Ministry of Education, Youth and Sports (MoEYS) is one of the key stakeholders in this reform process. To contribute to implementation of the PFMRP at the sector level and to gain knowledge about how funds allocated to the education sector could be better targeted in a timely manner toward priority areas, the MoEYS and MEF together with key Development Partners conducted a Public Expenditure Tracking Survey (PETS) and Quality Service Delivery Survey (QSDS) for the education sector. The PETS is also an important fiduciary risk management measure. The previous PETS in the education sector was undertaken in 2005, and three small-scale PETS were completed in 2012 and 2013. Objectives of the Cambodia Education PETS-QSDS PETS and QSDS seek to analyze and address the efficiency and equity of public service delivery and can inform relevant policy reforms. To improve service delivery performance, PETS allows policymakers to follow how resources move from origin (funders) to frontline service delivery facilities and examines the efficiency of public spending. QSDS examines relevant incentive structures and various elements of service delivery by frontline providers. QSDS therefore assists in understanding how incentives and accountability systems are working in practice and how they can be improved. PETS identify resource use and leakages by examining flows of funds and materials from the central government to local service providers via regional and local governments. They mainly evaluate the proportion of public resources Background and context | 19 (financial, human, and in-kind) that reaches each level, particularly frontline service providers. QSDS are multi-purpose provider surveys that examine the efficiency of frontline service delivery and the dissipation of resources by collecting information on service providers and various agents in the system. These two instruments could be applied jointly to obtain a more complete picture of the efficiency and equity of public service delivery. The main objective of this Cambodian PETS-QSDS was to trace fund flows, especially school funds, through the system from provincial treasuries to government schools, including early childhood centers up through full secondary schools. The survey also measured the timeliness of fund flows from the central level to schools and the impacts of the funds on school quality, teacher performance, and student learning outcomes. This is the second national PETS- QSDS in Cambodia. The first national Cambodian PETS-QSDS was conducted in 2004. This second PETS-QSDS study was conducted following reforms by the Cambodian government to transfer funds through banking systems rather than through physical cash deliveries from treasuries to provincial offices of education and to schools. Cambodia’s education system The formal educational structure of schooling in Cambodia is formulated as 6+3+3. Prior to primary school, there is also at least one year for pre-school education (kindergarten) for children from age 3 to below age 6. Then follows nine years for the completion of basic education, which is divided into six years of primary education (grades 1 to 6) and three years of lower secondary general education (grades 7 to 9). Upper secondary education consists of three years (grades 10 to 12). Post-school formal education includes technical and vocational education, which commences after basic education (after completion of grade 9), as well as higher education, which commences after grade 12. In practice, actual schools may take several forms. Most early childhood education (ECE) centers are incorporated physically in primary schools. Most secondary education can be found in a stand-alone lower secondary school, a stand- alone upper secondary school, or a full secondary school with classes from grades 7 to 12. In addition, Cambodia has a small number of stand-alone ECE centers and a small number of basic schools (grades 1 to 9). Any of the above types of school may be incomplete, either because the school is new and has not yet progressed students through all its appropriate grades or, in the case of primary schools, because it is isolated and makes provisions for young children who cannot walk to the more distant full school. Education financing in Cambodia Public funding of education in Cambodia has been increasing in recent years, with government commitment to further increases in both teacher 20 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey wages and school operational funding. The budget for wages has been increasing considerably and stood at 77 percent of the current budget in 2015. The capital budget has increased significantly, as well (14.3 per cent in 2016). Nonetheless, Cambodia ranked only above Myanmar in the sub-region for education expenditure either as a percentage of GDP or as a percentage of government expenditure. Education expenditure per student in Cambodia is very low compared to neighboring countries at all levels of schooling. This largely reflects the low levels of teacher salaries in Cambodia, although there are also school operational effects. Actual expenditure has not always reflected the growth in budgets. Execution was only 86 percent of the budget in 2014-15, although this improved to 90 percent in 2015-16 and 94 percent in 2016/17. Measures to increase the efficiency of disbursement, such as payment of salaries and school operating budgets into teachers’ and schools’ bank accounts, will continue to improve expenditure-to- budget ratios. However, planning, financial management, and audit capacities also need to be increased. In the past few years, several changes have been made in the funding process for the operational aspects of schools. These include changes in the composition of funding, the funds disbursement process, and the amounts available at schools. Public funds supporting the operations of schools come from two main sources: School Operational Budgets (SOB) from the government budget and School Improvement Grants (SIG) from the Swedish International Development Agency (Sida) starting from 2014. SOBs are now all based on the Program Budget (PB) and accounted for in a uniform manner. All schools are required to have school bank accounts and receive their fund transfers from the government directly into those accounts. Since 2014, all schools have received SIGs which, although considerably less than SOB funds, have arrived in schools generally well before SOB funds and with less constraints on their use. In addition to SOB and SIG, other sources of funding include contributions from parents, communities, NGOs, and other humanitarian organizations, but these sources are not properly recorded at schools. The Cambodian government has a clear formula for allocation of SOB funds, which was developed jointly by MEF and MoEYS. The allocation formula has a fixed amount per school, but to accommodate differences in the sizes and hardship levels of schools, the fixed amount per school differs by type, location, and school size—more disadvantaged or bigger schools receive higher amounts of funding. In addition to the fixed amount per school, the SOB’s formula also has a per capita student formula, allowing flexibility to accommodate schools with more students and schools with fewer students. For example, a primary school with six or less classes would receive the fixed amount of KHR 800,000 (USD 200) per year and KHR 10,000 (USD 2.5) per student per year under this allocation formula (Table 1). Background and context | 21 Table 1: SOB Allocation Formula School Type School Operating Budget (KHR) Per School Per Student 400,000 9,000 Typical Area with 3 or fewer classes ($100.00) ($2.25) 500,000 8,000 Typical Area with 4 or more classes ($125.00) ($2.00) Pre-School 450,000 10,000 Disadvantaged Area with 3 or fewer classes ($112.50) ($2.50) 500,000 9,000 Disadvantaged Area with 4 or more classes ($125.00) ($2.25) 800,000 10,000 Typical Area with 6 or fewer classes ($200.00) ($2.50) 1,000,000 9,000 Typical Area with 7 or more classes Primary ($250.00) ($2.25) School 1,000,000 12,000 Disadvantaged Area with 6 or fewer classes ($250.00) ($3.00) 1,200,000 10,000 Disadvantaged Area with 7 or more classes ($300.00) ($2.50) 1,500,000 19,000 Typical Area with 10 or fewer classes ($375.00) ($4.75) 2,000,000 17,000 Typical Area with 10 or more classes Secondary ($500.00) ($4.25) School 2,000,000 21,000 Disadvantaged Area with 11 or fewer classes ($500.00) ($5.25) 2,500,000 19,000 Disadvantaged Area with 11 or more classes ($625.00) ($4.75) Source: Joint MEF-MOEYS Prakas No. 508 (2013). The SIG fund also has a clear formula, albeit simpler than that of the SOB for the fixed amount per school. For the SIG, the fixed amount per school is the same regardless of school size and location. The per capita student formula differs between schools located in typical areas and disadvantaged areas. Schools in disadvantaged areas receive more funds per student than those in typical areas (Table 2). Table 2: SIG Allocation Formula Per Student (US$) School Type Per School (US$) Typical Area Disadvantaged Area Pre-School 69 1.41 2.12 Primary School 94 1.41 2.12 Lower Secondary School 144 2.12 3.17 Upper Secondary School 144 1.76 2.65 Source: SIG Financial Management Manual (2010). 22 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Since SOB funds come from the Cambodian government, SOB goes through government budget procedures which include budget planning, review, negotiation, and approval processes. The budgeting processes start in June and are completed in December each calendar year. In theory, the SOB process starts with estimates of the budget needed at the school level based on the School Development Plan (SDP). Once each school completes the budget proposal, they submit to their District Office of Education (DOE) for consolidation before submitting to the Provincial Office of Education (POE) for further consolidation at the central offices of MoEYS before submitting to MEF. The SOB also goes through review and approvals from cabinet ministers and Parliament (Figure 1), since it is part of the government budget processes and envelope. Figure 1: Comparison of SOB and SIG Processes SIG Approval SOB Approval Parliament Approve the budget Dec Cabinet Ministers Submit to Parliament Budget hearing and submit Jan Steering MEF national budget to cabinet Committee ministers Adjustment SIG: Submit SIG SOB: Consolidated national Oct/Nov Sep proposals to SC for MoEYS education budget and submit approval it to MEF Adjustment Review and consolidate budgets Aug PDEF in a province and submit the Adjustment POE province-level budget Adjustment Review and consolidate SDPs SOB Only process DOE of all schools in a district SIG&SOB process Adjustment Prepare SDP (including activity SIG Only process June items for SOB, SIG, and other) School For SIG, the budgeting process starts in October or November, which is the beginning of the school year in Cambodia. In terms of funds request and approval, the process does not follow the steps of the government budget process since the funding source is a development partner (Sida). The process for SIG is simplified, although it starts with the same School Development Plan at the school level for budget request and consolidation processes at DOE and POE. It mainly involves entities within MoEYS, particularly POEs and Departments of Finance. In addition to the process within MoEYS, SIG fund requests go through a review and approval process by the Project Steering Committee, which has representatives from Sida and other government ministries including MEF. Background and context | 23 Previous PETS studies in Cambodia Several PETS were conducted in Cambodia in the 2005-2013 period. They generally found leakage to be low but the timeliness and consistency of funds delivery to service providers to be poor. The findings were taken into account in the directions recommended for this PETS-QSDS and the emphasis in the data to be collected and analyzed. The views of the Technical Working Group on the objectives, as well as national and sector financial system reforms that have taken place since the previous PETS were conducted, were also taken into account. Four education sector PETS have been conducted in Cambodia, although none had major QSDS components. All four took place in the past ten years, with three of them being in the past four years. One focused on ECE, two on primary schools, and one on senior secondary schools. Some common issues emerged across these four studies: • Reported leakage is low, apart from textbooks. • In many provinces, timeliness is a problem between the Provincial Treasury and the POE. • Most disbursement is timely within the education system, from POE to DOE to School. • Disbursements appear to be mainly twice a year rather than the quarterly program that is mandated in the system. • Use of credit or delayed purchases due to delayed disbursements to schools results in inefficiency and lack of effectiveness in spending resources. • Problems with record keeping, incorrect reports, and poor monitoring heighten fiduciary risk. • Inequitable distribution and rigidity in spending of funds further undermine effectiveness. • Some stakeholders have poor knowledge of the financial system, making them unable to function as informed users and effective monitors. Even though three of these PETS were undertaken in the past four years, a number of substantial reforms in education and its financial management system have taken place since then. One example of a major change is that over the 2014-2016 period, all government schools in Cambodia were required to open a bank account, which has transformed how schools receive both government funds and agency funds. Reforming the resource planning, delivery, and reporting system has been a high priority for the Ministry. Currently, the Ministry is discussing with an external agency support to develop a unified fund flow to schools, called a School Improvement Fund (SIF), with a view to introducing a single fund for school operations as a channel for both government and external funds with a single financial manual. 2 Method Sampling The study selected a sample of 400 schools (300 primary and 100 lower secondary). To provide a 95 percent confidence level with meaningful statistical results for primary and secondary schools, a sample of nearly 400 schools was necessary. The study used the stratified random sampling method and was done in two stages. Multi-stage random sampling has been widely used in social surveys over many years and does not compromise the sampling statistically. In the first stage, schools were stratified by province and district and by primary/lower secondary level, using Education Management Information System (EMIS) data from 2015- 2016. In the second stage, for each province, 40 schools (30 primary and 10 lower secondary) in five districts were selected randomly. From this two-stage sampling strategy, a total of 400 schools (300 primary and 100 lower secondary) in ten provinces were selected for conducting data collection. Survey instruments The study employed 13 study instruments to capture information about fund flows and impacts of the funds on service delivery at the school level and from key actors involved in the PB and SIG fund processes. This included survey modules for interviews with the Provincial Treasury, POE, DOE, school director, school accountant, teachers, students, parents, and School Support Committee members, as well as classroom observation. In addition to the survey instruments, the study used a national assessment test to conduct mathematics and physics tests for grade 8 students to assess their learning outcomes. The 13 instruments were applied in each lower secondary school selected for this study, while only 9 study instruments were applied for each primary school selected due to budget constraints (Table 3). The four study instruments not applied in primary schools were classroom observation, student module, parent module, and mathematics and physics tests. Method | 25 Table 3: Survey Instruments Lower Survey Modules PETS or QSDS Primary Secondary 1 Provincial Treasury PETS √ √ 2 Provincial Office of Education PETS √ √ 3 District Office of Education PETS √ √ 4 School Director QSDS √ √ 5 School Accountant (PB) PETS √ √ 6 School Accountant (SIG) PETS √ √ 7 School Support Committee PETS/QSDS √ √ 8 Teacher Roster QSDS √ √ 9 Teachers QSDS √ √ 10 Quick Classroom Observation QSDS - √ 11 Students QSDS - √ 12 Parents QSDS - √ 13 Mathematics and Physics Tests (Grade 8) QSDS - √ Data collection and cleaning Data collection was undertaken by BN Consult and conducted from December 2016-March 2017. The BN Consult team visited all 400 schools but could collect information on SOB and SIG funds from only 391 schools due to failure to access school directors or school accounts or unavailability of financial records at schools. Hence, the data on SOB and SIG is only available for 391 schools, not 400 schools as anticipated (Table 4).4 Other data such as QSDS data was available for all schools. 4 The nine schools that the data collection team could not meet or for which the team could not access financial information or for which financial information was unavailable were primary schools. Specifically, the schools were (i) Bangkan in Rovieng district, Preah Vihear province; (ii) Chunhchaing in Choam Ksan district, Preah Vihear province; (iii) Damrei Slab in Kampong Svay district, Kampong Thom province; (iv) La Ang Bo Rae in Veal Veng district, Pursat province; (v) Peak Kantel in Ek Phnom district, Battambang province; (vi) Phtas Rung in Phnom Kravanh district, Pursat province; (vii) Robang Romeas in Bakan district, Pursat province; (viii) Sena Pramouk in Koulen district, Preah Vihear province; and (ix) Thmey in Mesang district, Prey Veng province. 26 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Table 4: Sampling Distribution of the Survey and Actual Implementation Province and District Primary Lower Secondary   Target Achieved Target Achieved Battambang (Ek Phnom, Kamrieng, Koah Kralar, Maung Russey, Sangker) 30 29 10 10 Kampong Cham (Chamkar Leu, Kampong Cham, Kampong Siem, Kang Meas, Srey Santhor) 30 30 10 10 Kampong Chhnang (Chul kiri, Kampong Leng, Rolea Pa-ir, Samaki Meanchey, Teuk Phos) 30 30 10 10 Kampong Speu (Chbar Morn, Oudong, Phnom Sruoch, Samrong Torng, Thporng) 30 30 10 10 Kampong Thom (Baray, Kampong Svay, Prasat Sambo, Sandann, Staung) 30 29 10 10 Phnom Penh 30 30 10 10 (Chamkar Morn, Daun Penh, Po Senchey, Prek Pneuv, Tuol Kauk) Preah Vihear (Choam Ksan, Koulen, Preah Vihear, Rovieng, Tbeng Meanchey) 30 27 10 10 Prey Veng (Kamchay Mear, Mesang, Peam Chor, Prey Veng, Svay Antor) 30 29 10 10 Pursat (Bakan, Krakor, Phnom Kravanh, Pursat, Veal Veng) 30 27 10 10 Tbaung Khmum (Dambe, Memot, O Raing Euv, Suong, Tbaung Khmum) 30 30 10 10 Total 300 291 100 100 During data collection, the survey team met people who were in charge of work directly related to SOB and school operations. At the Provincial Treasury, the survey team met with the head and/or deputy head of the Provincial Treasury and officer in charge of SOB. At the POE and DOE levels, the survey team met the director and/or deputy director and accounting officer and/or officer in charge of primary and secondary schools. At schools, the team met with school directors and/ or school accountants. For the QSDS data, the team met with the school director or deputy director, selected teachers, sampled students and their parents, and members of the School Support Committee. Checking and cleaning the data took considerable time and effort. Not all schools kept good records, requiring significant effort to revisit and check documents. In addition, several ongoing reforms and changes in the 2014-15 school year meant that accurate records were difficult to retrieve during this time, even at the provincial level. Comparison of provincial disbursement and school-level records showed some differences, with schools recording both lower and higher amounts. 3 Characteristics of the sample schools This section describes the main characteristics of the schools and the people who make up the school community. Most of the remainder of the report will analyze aspects of either the disbursement or management of school finances (the PETS aspect) or the quality aspects of schooling that are the presumed outputs of that funding (the QSDS aspects). This section looks at the directors, teachers, students, parents, and committee members. It focuses more on the secondary schools as more data on teachers, students and their families, and school committees was collected for secondary schools than for primary schools, mainly due to cost—each extra set of data, especially for students and their families, was quite costly in terms of interviewer time. School types and enrollments Overall, the sample schools were slightly over-represented in the full secondary category and under-represented in the stand-alone primary schools compared to the total population of government schools in Cambodia (Table 5). Three-quarters of the sample were primary schools by choice, while four-fifths of all government schools (excluding ECE standalone centers) were primary schools in 2015/16. The sample schools were more likely to be in urban areas and thus larger. Thus, the sample represented 4.5 percent of government schools and 6.5 percent of government school students. Most of the differences arise from the sampling choice to select 300 primary schools and 100 secondary schools to ensure sufficient numbers of secondary schools within the limited budget available for data collection. Representative proportions would have meant 320 primary schools but only 80 secondary schools and an unacceptable margin of error. 28 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Table 5: Sample and National Schools Compared Sample National     Number Percent Number Percent Schools 400 100% 8,799 100% Standalone Primary 127 32% 3467 39% Primary with ECE 173 43% 3618 41% Lower Secondary 63 16% 1,251 14% Full Secondary 37 9% 463 5% Location 400 100% 8,799 100% Urban 90 23% 1,387 16% Rural 238 60% 7,412 84% Remote 72 18% Average student enrollment 455 322 Standalone Primary 217 {284 Primary with ECE 490 Lower Secondary 353 250 Full Secondary 1,282 1,106 Total student enrollment 181,916 100% 2,835,743 100% Standalone Primary 27,392 62% 2,010,673 71% Primary with ECE 84,855 Lower Secondary 22,240 12% 312,991 11%   Full Secondary 47,429 26% 512,079 18% The number of schools, both primary and secondary, was constant by province. However, within each province, the random sampling meant that the number of primary schools with and without attached ECEs varied, as did the number of lower and full secondary schools. Of the 400 schools in the sample, 127 (or 42 percent of primary schools) were standalone primary schools, while 173 (58 percent) had an attached ECE. Of the 100 secondary schools, 63 were lower secondary only, and the remaining 37 were full secondary schools (Figure 2). Characteristics of the sample schools | 29 Figure 2: School Types by Province 45 40 NUMBER OF SCHOOLS 35 30 25 20 15 10 5 0 Battambang Kampong Kampong Kampong Kampong Phnom Preah Prey Pursat Tbaung Cham Chhnang Speu Thom Penh Vihear Veng Khmum Secondary School 2 3 2 5 5 6 2 4 3 5 Lower Secondary School 8 7 8 5 5 4 8 6 7 5 Primary school with ECE 14 20 16 14 16 25 15 22 20 11 Primary School 16 10 14 16 14 5 15 8 10 19 Student enrollment in the sample schools varied by province, level, and school type (Table 6). In every province, standalone primary schools were on average half the size of those with attached ECEs, and junior secondary schools were about two-thirds the size of full secondary schools. Average size varied considerably among provinces. Average primary school size varied from 200 in Preah Vihear to 1,026 in Phnom Penh, a fivefold difference. The differences were nearly as great among secondary schools, from 404 in Preah Vihear to 1,728 in Phnom Penh. Preah Vihear, Battambang, Kampong Chhnang, and Pursat had small schools at both the primary and secondary levels. Table 6: Sample Schools’ Average Enrollment by Province and Type Primary Lower Full All Primary All primary Province school with secondary secondary secondary school schools ECE school school schools Battambang 186 326 252 295 980 432 Kampong Cham 256 478 404 322 1,161 574 Kampong Chhnang 154 344 255 241 1,193 431 Kampong Speu 256 480 361 514 1,018 766 Kampong Thom 186 332 267 288 1,012 650 Phnom Penh 471 1,137 1,026 883 2,291 1,728 Preah Vihear 100 300 200 233 1,089 404 Prey Veng 333 332 332 400 1,257 743 Pursat 135 293 240 395 601 457 Tbaung Khmum 279 645 413 227 1,340 783 Total 217 490 375 353 1,282 697 30 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey A primary indicator of quality in schooling is the ratio of students to classes, classrooms, and teachers. These three ratios reflect the actual educational experience of the students. The average class size (the ratio of students to classes) determines many things about the students’ day-to-day school experience and the costs of schooling. Larger class size may reduce the amount of one-on-one teaching time with individual students, but it may also provide stimulus and opportunities for cooperation and competition not found in very small classes. Large classes reduce school costs as they distribute teachers’ salaries (the largest single cost in schooling) over more students. Primary school classes ranged from 25 in Preah Vihear to 40 in Kampong Speu and Kampong Cham. Average secondary class sizes were considerably larger (Figure 3). Figure 3: Average Class Size by Province and School Level 53 46 49 48 47 48 45 43 43 41 40 40 39 37 34 35 28 31 32 25 G AT G M G H M EU R AM EA N N AN N O U RS SP PE BA VE M TH CH H N PU H G VI AM H M EY K G G N CH O N H N G PO TT PR N PO EA PO N PH G BA AM AU AM PR N AM PO TB K K K AM K Primary Secondary When compared with student-class ratios, the student-to-classroom ratio reveals if there are serious problems with the number of classrooms available or if there are surplus classrooms. The student-classroom ratio for primary students suggests that most provinces had classroom shortages, as the student-classroom ratio was larger than the student-class ratio (compare Figures 3 and 4). The difference was not so great at the secondary level, suggesting that classroom provision at that level was not so poor. However, there is often the need for more specialized classrooms at the secondary level for subjects such as science, so in a well-furnished system, one would expect the student-classroom ratio to be lower than the student-class ratio. Six of the ten provinces did have lower student-classroom ratios than student-class ratios at the secondary level. Characteristics of the sample schools | 31 Figure 4: Student-Classroom Ratios by Province and Level, with and without shift adjustment Note: Left figure without shift adjustment; right figure with shift adjustment. 59 57 56 47 48 48 37 39 39 42 43 42 43 39 46 46 44 37 36 42 40 40 40 35 36 31 38 29 39 30 28 29 29 37 24 24 26 20 22 24 H AT R AM G G G M G AT H R G G M EU M M EU AM EA N N AN N EA U O EN N EN AN O U RS PE CH SP BA VE M RS TH SP H BA M CH H PU N H H VI G M M P V N H G PU T H EY K G G N VI O AM A CH N H H M EY K N G G G PO TT N N PR PO EA PO CH N O PH N H N G G PO BA TT AM AU PR PR AM N PO EA N AM PO N PO PH G TB BA K AM AU K K PR AM AM N AM PO TB K K K K Primary Secondary AM K Primary Secondary Adjusting the student-classroom ratio to take the number of shifts into account changes the picture considerably (Figure 4 above). Running two shifts effectively doubles the number of classrooms available for the same number of students and thus halves the student-classroom ratio. On average, primary schools in all ten provinces had between 20 to 30 students per classroom once shifts are taken into account. Similarly, at the secondary level, eight of ten provinces had an average number of 40 students or more per classroom without accounting for shifts. Once shifts are taken into account, only Kampong Thom had an average number of students per classroom of 40 or more. A similar pattern was observed in the ratio of students to government teachers (Figure 5). In terms of class preparation and classroom management, the shift-adjusted ratio shows what the teacher faces in the classroom. Shift-adjusted student-teacher ratios for primary schools were generally between 20 to 30 students, although Kampong Thom and Tbaung Khmum were above 30 even when shifts are taken into account. In terms of grading and other student-related tasks, the unadjusted ratio is the key figure, when the same teachers take both shifts as is usually the case. Interpreting the figures for secondary schools is not so simple, as secondary teachers are usually subject specialists and are more likely to face classes on the order of the student-class ratio in Figure 3. Nonetheless, when shifts are a part of the school day, the average number of students in the secondary classroom is reduced. 32 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Figure 5: Student-Teacher Ratio by Province and Level, with and without shift adjustment Note: Left figure without shift adjustment; right figure with shift adjustment. 66 51 47 36 37 38 40 41 42 42 35 29 29 30 32 26 26 29 26 26 26 27 29 23 24 24 25 22 22 19 19 23 16 16 19 18 15 15 12 14 G H R G AT G EU M M AM G R H G T G EU M M AM EA N N N AN PR RSA O U EA N N AN N O U RS SP PE BA VE M CH TH SP PE BA VE M CH TH H N H PU H N G VI PU H M H M AM NG EY K VI G G AM N H M EY K A G G CH O N H N G PO CH TT O PR N H N N G PO EA PO PO TT N N PO EA PH G PO BA N AM AU H G PR AM AM N BA AM AU PR AM P N PO TB K K K PO TB K AM K K AM K Primary Secondary K Primary Secondary The use of double shifts may compensate for a lack of classrooms or lack of sufficient teachers, but it also reduces the amount of instructional time available to students. Typically, the shifts run as morning and afternoon sessions. Although schools start early and finish late, there is insufficient time for a full day of instruction. The literature strongly indicates that instructional time is positively correlated with more and better learning.5 One of the Ministry’s policy aims is to reduce double shifting, but the practice is so widespread that it is difficult to see a significant reduction taking place in the near future, particularly in primary schools. School Directors Male and female school directors differed on several characteristics (Table 7). Female directors were likely to have smaller schools and schools in rural areas, but lower student-teacher ratios. They were less likely to manage a disadvantaged school. In terms of social characteristics, female directors were more likely to be older than their male counterparts but less likely to be married and had fewer children of their own on average. Primary directors typically went to post-secondary college for their training, while secondary directors attended a higher education institution and received a Bachelor’s or Master’s degree. The largest proportion of school directors lived a kilometer or less from their schools and typically, apart from female primary school directors, took less than 15 minutes to travel to and from school, with the great majority traveling by motorcycle (moto). 5 See for example Cattaneo, M et al. 2016 and Lavy 2014. Characteristics of the sample schools | 33 Table 7: Characteristics of School Directors by Gender and Level Primary Primary Secondary Secondary female male female male All Item director director director director Directors School characteristics Number 54 245 8 91 398 Pct disadvantaged 46% 55% 25% 40% 50% Pct rural 85% 78% 88% 75% 79% Average number of students 325 388 593 713 458 Average number of teachers 11 11 37 32 16 Average Student:Teacher ratio 43 44 19 24 39 Social characterics Mean age 43 42 49 43 43 Pct married 70% 83% 88% 91% 83% Mean number of children 2.6 3.2 2.9 2.8 3.0 Educational background To Year 12 only 13% 9% 0% 5% 8% Post secondary college 59% 62% 25% 35% 55% National Institute of Education 0% 0% 0% 8% 2% Higher Education to Bachelor’s degree 17% 22% 50% 32% 24% Master’s degree 6% 3% 13% 16% 7% Location relative to school Percent travel less than 15 min to school 59% 74% 75% 70% 71% Percent less than 1 kilometre to school 46% 45% 38% 44% 45% Pct travel to school by moto 70% 84% 100% 79% 81% School support committees The effectiveness and function of school support committees appear unclear. Some of their members even disagree on committee size. Every school has a school support committee (SSC), a joint school-community committee responsible for school planning and budgeting and for overseeing spending of the operational funds. At each school, the interviewers were requested to meet two members of the school support committee from the community. A total of 722 separate interviews were conducted.6 Initial analysis revealed that the respondents were not in agreement on the size of the SSC to which they belonged. In about one- third of the cases in which two SSC members were interviewed, they did not agree on the number of members in the committee and often on the number of women on the committee (Table 8). Primary school members were more likely to agree, and rural primary schools had the most agreement. 6 Interviews could not be arranged for 12 schools. In 50 schools, only one interview could be conducted. 34 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Table 8: Interview Status of SSC Members by Level and Location of School Primary Primary Secondary Secondary All SSC Interviews urban rural urban rural Schools No SSC interview 1 7 1 3 12 One single SSC interview 9 30 3 8 50 Two interviews with agreement on numbers 31 143 11 36 221 Two interviews with differing numbers 21 58 9 29 117 Total 62 238 24 76 400 Women appear to be underrepresented in the SSC. The average SSC had about six members, with 20 percent being female (Table 9). Secondary rural schools had the largest committees, while secondary urban schools had the smallest. The latter also had the smallest proportion of women members, while primary urban schools had the largest proportion. Table 9: SSC Membership by Location and Level Primary Primary Secondary Secondary Item urban rural urban rural school school school school All Schools Average SSC membership 5.4 5.8 4.3 7.4 5.9 Average number of women members 1.5 1.3 0.7 1.8 1.4 Percent women members 24 20 13 22 20 Note: Only responses from schools with two members in agreement or one single interview were used in this table. Teachers Most teachers had completed education at the post-secondary level. Seven out of ten teachers had been trained at a provincial or regional post-secondary college. More than one-third of the secondary teachers had a degree at the Bachelor or Master level. A minority of about one in twenty teachers had only a 12th grade education or below. Most teachers in the surveyed schools sample were “local” in terms of location and were a part of the geographical community served by their school. Seven in ten of the teachers lived within 15 minutes of their school, and a large minority took less than 15 minutes to get to school from home. Eight out of ten traveled between home and school by motorcycle (Table 10). Characteristics of the sample schools | 35 Table 10: Teacher Characteristics Primary Primary Secondary Secondary All government Item female male female male teachers teacher teacher teacher teacher Social characteristics Number 295 289 54 141 779 Mean age 35 36 32 36 35 Percent married 65% 75% 67% 74% 70% Mean number of children 1.4 2.0 1.2 1.9 1.7 Educational background To Year 12 only 9% 6% 4% 2% 6% Post-secondary college 79% 78% 60% 51% 72% Higher Education to Bachelor’s degree 8% 15% 33% 36% 18% Master’s or Doctorate degree 0% 0% 0% 4% 1% Location relative to school Percent travel less than 15 min to school 70% 72% 73% 72% 71% Percent less than 1 kilometer to school 46% 42% 44% 39% 43% Pct. travel to school by moto 74% 79% 81% 85% 79% Most teachers had received some degree of training prior to starting their teaching careers. Nearly all teachers in the sample (97 percent) had completed teacher training, with the primary teachers having generally (77 percent) completed two years of training at a provincial teacher’s college and nearly all the secondary teachers completing two years of training at a regional college (Table 11). They generally viewed their pre-service training as having been of good or very good quality (86 percent). They were experienced teachers with an average of over 12 years of teaching. Table 11: Teaching Career Primary Secondary All Primary Secondary Item female female government male teacher male teacher teacher teacher teachers Pre-service training Completed teacher training before teaching 95% 95% 100% 100% 97% Training location: Provincial teacher’s college 96% 95% 2% 11% 71% Training location: Regional teacher’s college 3% 5% 98% 87% 29% Two or more years of training 77% 77% 90% 85% 80% Years of experience 12.0 13.4 9.4 13.2 12.5 Pre-service training quality good/very good 88% 84% 96% 84% 86% In-service training Received some in-service training 57% 59% 48% 62% 58% Years since last in-service training 4.7 4.1 7.2 5.6 4.8 Percent received in-service in last three years 66% 68% 44% 49% 62% In-service training quality good/very good 94% 90% 88% 87% 91% 36 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Primary Secondary All Primary Secondary Item female female government male teacher male teacher teacher teacher teachers Training was applicable to teaching 98% 96% 96% 93% 96% Teaching experience Years at this school 11.7 11.4 9.3 11.0 11.3 This is preferred school 83% 83% 77% 75% 81% Taught at another school before this 45% 58% 40% 59% 52% Want to move in near future 34% 34% 38% 28% 33% Most surveyed teachers who had received in-service training viewed it as good or very good (91 percent) and applicable to their teaching (96 percent). However, only about three in five said they had received any in-service training. Of these, two-thirds of the primary teachers had received some in the last three years, compared to less than half of the secondary teachers. Eight in ten teachers said their current school was their preferred school, and their average time at this school was 11 years. Half had been at another school before their current school, and one-third wanted to move to another school in the near future (Table 11). Secondary students Most secondary school students live with their guardians in the same village where they were born, and half of them live within 15 minutes of travel from their schools. In each of the 100 secondary schools, 15 students were randomly selected from Grade 9 for interviews, so a total of 1,500 students were interviewed. The average age of secondary school students was 15.3 years, and 53 percent were female. The students overwhelmingly lived with their guardian (94 percent), and a majority lived in the same village where they were born (82 percent). A little less than half of the students traveled more than 15 minutes from home to school, and about half of them traveled to school by bicycle while another one-third traveled to school by motorcycle. Disturbingly, only half ate breakfast every day (Table 12). Table 12: Secondary Student Characteristics Item Girls Boys All students Number 789 711 1,500 Mean age 15.2 yrs 15.4 yrs 15.3 yrs Family size at home 5.4 5.5 5.4 Living with guardian 93% 94% 94% Born in current village 81% 82% 82% Travel 15 minutes or less to school 52% 59% 55% Travel by bicycle 52% 50% 51% Travel by moto 31% 32% 32% Eat breakfast every day 51% 55% 53% Characteristics of the sample schools | 37 A sizeable proportion of the secondary students interviewed were over-age for their grade. If the students commenced school at age 6 as they should, then they should be 15 years old in Grade 9. However, about two in five were age 16 or above, which suggests that they started school late or repeated at least one grade (Figure 6 and Table 13). Boys were more likely than girls to be over-age for Grade 9 and to have repeated at least one grade. In terms of learning resources, four out of five children shared textbooks in class (Table 13). Four out of five children shared textbooks in class with other students, but nearly all students took at least some textbooks home. Students were also likely to have other books at home for reading, with most having one to five reading books at home. Most students (74 percent) took private tutoring classes at school, and around 43 percent had private tutoring outside school. Only a small minority of about 15 percent had no tutoring classes at all. Figure 6: Grade 9 Students’ Age by Gender Number 11 years 12 years 13 years 14 years 15 years 16 years 17 years 18 years 19 years 20 years Boy 1 0 21 142 249 181 90 23 3 1 Girl 0 0 36 171 310 183 77 9 2 1 Age Boy Girl Table 13: Secondary Students’ School Experience Item Girls Boys All students Over-age for grade 35% 42% 38% Under-age for grade 26% 23% 25% Ever repeated grade 14% 18% 16% Share textbooks 80% 76% 78% Take textbooks home 98% 97% 97% 1 to 5 books for reading at home 63% 59% 61% In-school tutoring 74% 73% 74% Out-of-school tutoring 43% 42% 43% 4 Education revenues and funding Flow of funds to schools Almost all school accountants knew the amount of SOB to which their schools were entitled. This implies that schools know if there are shortfalls in the amount of funds transferred to their bank accounts. In the survey, school accountants were asked, “Do you know how much this school is entitled to receive PB/SOB?” to which 399 of the 400 sampled schools (99.7 percent of all schools) answered, “Yes, I know it” (Figure 7).7 Only one school—a primary school in Prey Veng province—answered “No, I don’t know.” Figure 7: Knowledge of SOB Amounts by Person Responsible for Accounts 99.7% 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0.3% 0% Ye, I know No, I don’’t know 7 POEs supplied details of SOB transfers by quarter for all sampled schools. Interviewers also calculated SOB entitlements using the formula in Table 1. At the school, this question on knowledge of funds transferred was followed by specific requests for the amount in riel. Thus, the answer was verified using three-way triangulation. Education revenues and funding | 39 Flow of SOB funds Schools in the study’s sample received the full amount of SOB funds they were supposed to get, which indicates no leakages as the funds were transferred. Following the implementation of SOB fund transfers via bank accounts in late 2014, funds were transferred from bank accounts of the provincial treasuries to bank accounts of the POE, then the POE made transfers to bank accounts of their respective schools. In tracing fund flows in ten provinces8 selected randomly for this study, it was found that in each quarter of fiscal years 2015 and 2016, POEs received all the funds to which they were entitled from their respective provincial treasuries. Tracing the funds from POEs to schools in the ten provinces also revealed that schools received the full amount of funds to which they were entitled (Figure 8, Tables 14, and 15). Figure 8: SOB Funds Sent and Reported, Fiscal Year 2015 Cambodia 100% Battambang 108% Preah Vihear 104% Kampong Chhnang 101% Pursat 101% Kampong Thom 99% Prey Veng 99% Kampong Cham 97% Phnom Penh 97% Kampong Speu 95% Tbaung Khmum 93% 85% 90% 95% 100% 105% 110% 8 Battambang, Kampong Cham, Kampong Chhnang, Kampong Speu, Kampong Thom, Phnom Penh, Preah Vihear, Prey Veng, Pursat, and Tbaung Khmum Table 14: Flow of SOB Funds from Provincial Treasury (PT) in FY 2015 (Khmer Riels) Amount of Amount of Amount of Amount of Percentage of fund POE received from PT SOB POE SOB SOB SOB % % % % Received POE Received POE Received POE Received Province Name 2015Q1 2015Q2 2015Q3 2015Q4 2015Q1 2015Q2 2015Q3 2015Q4 Battambang 7 85,275,000 785,275,000 785,275,000 785,275,000 100% 100% 100% 100% Kampong Cham 8 19,400,000 819,400,000 819,400,000 819,400,000 100% 100% 100% 100% Kampong Chhnang 483,551,500 508,664,400 520,997,700 480,915,400 100% 100% 100% 100% Kampong Speu 6 09,339,000 609,339,000 609,339,000 609,339,000 100% 100% 100% 100% Kampong Thom 683,837,300 683,837,300 673,837,300 673,837,300 100% 100% 100% 100% Phnom Penh 775,150,000 748,550,000 748,550,000 748,550,000 100% 100% 100% 100% Preah Vihear 248,000,000 246,318,300 247,158,900 247,160,800 100% 100% 100% 100% Prey Veng 833,935,000 833,935,000 833,935,000 833,935,000 100% 100% 100% 100% Pursat 363,450,000 363,450,000 360,325,000 343,825,000 100% 100% 100% 100% Tbaung Khmum 505,719,100 578,333,500 577,766,500 577,813,500 100% 100% 100% 100% Table 15: Flow of SOB funds from Provincial Treasury (PT) in FY 2016 (Khmer Riels) Amount of Amount of Amount of Amount of Percentage of fund POE received from PT SOB POE SOB SOB SOB % % % % Received POE Received POE Received POE Received Province Name 2016Q1 2016Q2 2016Q3 2016Q4 2016Q1 2016Q2 2016Q3 2016Q4 Battambang 1,068,406,300 1 ,619,149,500 889,659,800 819,434,800 100% 100% 100% 100% Kampong Cham 8 94,850,000 894,850,000 894,850,000 894,850,000 100% 100% 100% 100% Kampong Chhnang 480,423,600 537,086,000 480,055,900 469,001,500 100% 100% 100% 100% Kampong Speu 6 25,438,000 625,438,000 625,438,000 625,198,000 100% 100% 100% 100% 40 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Kampong Thom 719,543,500 877,143,500 708,643,500 708,643,500 100% 100% 100% 100% Phnom Penh 846,525,000 819,225,000 750,225,000 750,225,000 100% 100% 100% 100% Preah Vihear 268,309,500 268,309,500 268,309,500 268,309,500 100% 100% 100% 100% Prey Veng 846,500,000 846,500,000 846,500,000 847,500,000 100% 100% 100% 100% Pursat 388,132,000 506,287,900 387,998,300 382,079,700 100% 100% 100% 100% Tbaung Khmum 639,485,600 639,485,600 639,485,600 603,729,800 100% 100% 100% 100% Education revenues and funding | 41 However, the study found that schools in some provinces either over- reported or under-reported the amount of SOB funds they received in 2015. As Figure 8 shows, schools in some provinces reported receiving more funds than the amount reported as sent by the POE, while some other schools reported less than the amount reported by POE. For example, in 2015, schools in Battambang reported receiving 8 percent more than what was sent by the POE, while schools in Tbaung Khmum reported receiving 7 percent less than the amount of funds reported sent by the POE (Figure 8). Over-reporting and under-reporting of funds were still found in 2016, suggesting that poor recording of finances is an issue that needs to be addressed at the school level. Out of the study’s ten provinces, only schools in two provinces (Kampong Chhnang and Preah Vihear) were found to have reported their funds correctly (Figure 9). Figure 9: SOB Funds Sent and Reported in Fiscal Year 2016 Cambodia 99% Prey Veng 107% Pursat 103% Kampong Speu 101% Preah Vihear 100% Kampong Chhnang 100% Kampong Thom 99% Phnom Penh 98% Tbaung Khmum 97% Kampong Cham 95% Battambang 80% 0% 20% 40% 60% 80% 100% 120% No single pattern of poor financial recordings by school type or location can be identified from 2015 to 2016. There are no indications that secondary schools do better than primary schools, nor do schools in urban areas do better than schools in rural and remote areas with regard to proper recording of their finances (Figure 10). The PETS survey showed that in 2015, primary schools tended to over-report their receipt of SOB funds, while upper secondary schools tended to under-report. However, in 2016, primary, lower secondary, and upper secondary schools under- reported their funds. The survey also showed that in 2015, schools in remote areas tended to over-report SOB funds while schools in rural and urban areas tend to under-report. In 2016, all schools regardless of location under-reported the funds. 42 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Figure 10: SOB Amounts Sent/Received by Level and Location of School Actual amount of SOB reported by school/Actual amount Actual amount of SOB reported by school/Actual amount reported by POE in 2015 reported by POE in 2016 110% 104% 105% 99% 105% 100% 100% 96% 100% 95% 95% 89% 90% 87% 90% 85% 85% 80% 80% 75% 75% 70% 70% 65% 65% 60% 60% Primary Lower Secondary Upper Secondary Primary Lower Secondary Upper Secondary Actual amount of SOB reported by school/Actual amount Actual amount of SOB reported by school/Actual amount reported by POE in 2015 reported by POE in 2015 120% 105% 111% 99% 110% 100% 98% 99% 95% 100% 94% 89% 90% 90% 85% 80% 80% 70% 75% 70% 60% 65% 50% 60% Rural Urban Remote Rural Urban Remote Although SOB funds have reached schools without significant leakages, the main challenge is the major delay in disbursement at the beginning of the fiscal year, which could result in inefficient funds usage. Schools expected to receive their first wave of funds in January but received it several months later—for example, in April/May 2015 and February/March 2016 (Figure 11). Without funds early in the school year and with most schools having limited or no reserves, school directors have limited options. They may simply delay purchases of oftentimes essential school supplies which disadvantages their students, or they may buy on credit or borrow funds. Both latter options increase the costs of purchases, often considerably. Local merchants who know they may wait several months for payment build the costs of the delay into their prices, while borrowing informally may mean interest charges of 2 percent or more per month. Another alternative is to use SIG funds, which normally arrive early in the first quarter. Some purchases can be legitimate under SIG, while others may mean transfers between SOB and SIG accounts to balance the spending and the books. Unifying these funds as MoEYS is planning, into one SIF with a single account, may improve this situation. Education revenues and funding | 43 Figure 11: Average Months of Delay from Start of Quarter in POE Request for SOB Transfers 3.9 Months of delay 1.9 1.8 1.1 0.7 0.8 0.6 0.4 Q1 Q2 Q3 Q4 FY 2015 3.9 1.1 1.8 0.8 FY 2016 1.9 0.7 0.6 0.4 Quarter The significant delay in fund transfers to schools at the beginning of each fiscal year is likely due to systemic problems. The longest delays in making fund requests from POEs to provincial treasuries in 2015 were in Phnom Penh and Pursat provinces, where the POEs made the fund requests to their respective provincial treasuries on May 5th and 7th. For the other eight provinces, POEs made the fund requests to their provincial treasuries in April—also a long delay given that funds for the first quarter should have been processed in January 2015. The delay in 2015 was due in part to the introduction of full program budget, which required some adjustment. The situation improved in 2016, although there was still an average delay of 1.9 months (Figure 11). The delay can be attributed in part to budget request and approval processes. The procedures for fund request and release at the Ministry and provincial levels involve making a request to the provincial treasury. However, in the first quarter, the POE must wait until they receive the budget book and Prakas from the MEF and the latest enrollment data from all schools in the province for fund adjustments due to enrollment changes. Once the requests are made, fund releases are relatively quick, and the complete cycle takes only a few days from request to receipt (Figures 12 and 13). 44 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Figure 12: SOB Fund Processing Times for POE to PT and PT to POE 6.1 5.5 Days elapsed 3.7 3.1 3.4 3.1 2.9 2.7 2.3 1.7 1.8 1.3 1.1 0.5 1.4 0.9 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 2015 2016 Dates POE to PT days elapsed PT to POE processing and return funs Figure 13: Turnaround Time from POE back to POE 8.6 7.9 7.0 Days elapsed 6.5 5.5 4.6 5.0 4.2 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 2015 2016 Dates The delay in fund disbursement also occurred in the third quarter of the year, which is the third tranche of SOB fund releases. The delay was observed in all ten provinces in the study’s sampled provinces. Although the quarter commenced on July 1st, the POEs in 8 provinces9 only made fund requests to their respective provincial treasuries in August. The POEs in Phnom Penh and Tbaung Khmum did not make fund requests until September. It is unclear why the process of making fund requests from POEs to provincial treasuries for the third quarter of 2015 did not start in June given that POEs do not need to wait for the necessary documents and information (e.g., the budget book and enrollment data) to process fund requests to provincial treasuries. Previous PETS identified rigidity in SOB subaccounts, which remains an ongoing challenge. There are now 12 subaccounts in two SOB chapters (60 and 61) in which government funds flow to schools and in which the schools have to account to their respective POE. Most schools receive their funds allocated to the 9 Battambang, Kampong Cham, Kampong Chhnang, Kampong Speu, Kampong Thom, Preah Vihear, Prey Veng, and Pursat Education revenues and funding | 45 subaccounts even if these amounts do not reflect the needs of specific schools (e.g., they may receive funds for paying for electricity even though the school is not connected). Moving funds between these lines is necessary to spend the budget on identified needs at the school level. However, this is difficult and time-consuming, taking many weeks or even months. Flow of SIG funds The process for SIG fund releases differs from that of SOB funds (as noted in Figure 1). As described in the manual, SIG procedures involve a direct flow of funds from the center to the school account, initiated by the Finance Department of the MoEYS. Disturbingly, however, the provincial offices reported a number of release routes. While some indicated that the funds flowed from the MoEYS Finance Department as per the procedures, others said funds came through the provincial treasury or POE account. Overall, schools received the full amount of funds to which they were entitled. However, as with SOB, there are issues of poor recording at schools. For the 2014-15 school year, some schools over-reported or under-reported the amount of SIG funds (Table 16). Schools in Pursat, Prey Veng, Kampong Cham, and Battambang provinces over-reported the amount of SIG funds they received from the MoEYS Department of Finance, as their records showed they received more than 100 percent of the funds compared to the amount of funds recorded and provided by the POEs. Schools in Kampong Chhnang, Tbaung Khmum, Phnom Penh, and Kampong Speu under-reported the amount of SIG funds they received, as their records showed that they received less than 100 percent of the funds compared to the amounts recorded and provided by the POEs. The pattern of differential recording was repeated in the 2015-16 school year, with a similar overall pattern and most provinces reporting in a similar manner (Table 17). Similar to SOB, there is also no clear pattern of poor recording in terms of school type and location for SIG funds. Table 16: Average SIG Amounts per School Reported by POE and School 2014/15 Average amount of Average amount of SIG SIG reported by school Province reported by POE accountant 2014-2015 2014-2015 Battambang 678 689 Kampong Cham 876 901 Kampong Chhnang 777 771 Kampong Speu 1,044 1,004 Kampong Thom 827 827 Phnom Penh 2,160 2,124 Preah Vihear 733 731 Prey Veng 869 905 Pursat 660 690 Tbaung Khmum 1,087 1,072 46 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Table 17: Average SIG Amounts per School Reported by POE and School 2015/16 Average amount of Average amount of SIG SIG reported by school Province reported by POE accountant 2015-2016 2015-2016 Battambang 685 685 Kampong Cham 867 889 Kampong Chhnang 752 747 Kampong Speu 1,008 1,012 Kampong Thom 824 820 Phnom Penh 2,135 2,128 Preah Vihear 751 729 Prey Veng 859 886 Pursat 700 703 Tbaung Khmum 1,077 1,059 As with SOB, disbursement of SIG was also delayed. Schools are supposed to receive SIG funds in November, the beginning of the school year. Schools in some provinces received their SIG funds in December, while others received their SIG funds in January, February, April, May, or as late as July. There was little or no improvement in the timeliness of fund receipts between school year 2014-2015 and 2015-2016 (Table 18). Table 18: SIG Receipt Date 2014-15 and 2015-16 Schools received Schools received Province SIG funds SIG funds Difference 2014-2015 2015-2016 Battambang April 2015 July 2016 3 months later Kampong Cham December 2014 January 2016 1 month later Kampong Chhnang December 2014 January 2016 1 month later Kampong Speu February 2015 March 2016 1 month later Kampong Thom January 2015 February 2016 1 month later Phnom Penh January 2015 January 2016 Same Preah Vihear May 2015 February 2016 3 months earlier Prey Veng May 2015 January 2016 4 months earlier Pursat February 2015 February 2016 Same Tbaung Khmum December 2014 January 2016 1 month later   5 Quality of service delivery in education Assessing the impacts of SOB and SIG spending on quality of education at the service delivery point of the school is complex. First, it is necessary to consider how quality in schools can be measured. Quality has numerous aspects, including the general state of the school, quality of the teachers and their teaching, and management of the school, including its finances. Spending can also be assessed in various ways, including the total amount available to the school, amount available per student, and amount spent on quality-related goods and services such as teaching aids, library resources, and environmental aspects. Linking spending to quality outcomes is rarely a simple or straightforward exercise. School spending Two approaches to assessing quality of service delivery are used in this analysis. The first is related to total amount of fund available to school. The second is related to the amount spent on quality. The total amount available to the school in SOB and SIG is a significant determinant of what the school can do to affect educational outcomes. School and POE perceptions are that the use of SOB funds is more inflexible than that of SIG funds. The perception is that SOB funds are predetermined in the budget lines issued to schools and can only be changed after lengthy and formal approaches. On the other hand, SIG funds are perceived as lump sums that can be used flexibly according to the school’s needs and against a list of permitted uses—broadly, administration, access, and quality. Nonetheless, the total funding available is important, since even if SOB funds are more inflexible, SIG funds can be spent to make up for shortfalls or cover necessary school improvements. 48 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey The total operational funds available to schools (SOB and SIG combined) ranged from USD 248 to USD 25,833. Both SOB and SIG funding calculations include lump sums plus per capita components designed to offset the disadvantages suffered by small schools, while preserving uniform per capita payments. Half of all schools are classified as disadvantaged and thus receive extra operational funding. The average across all schools in the sample was USD 2,629 (Table 19). For analytical purposes, the 400 schools were assigned to quintiles based on their total reported operational funds received. Schools in the lowest quintile received USD 765 on average and were all primary schools. Schools in the highest quintile were mainly secondary schools and received an average of USD 6,715 in 2015-16. Standalone primary schools received an average of USD 1,278, while those with attached ECEs received an average of USD 2,270. Junior secondary schools received a total of USD 2,946 on average, and full secondary schools received USD 7,910 (Table 19). Table 19: Average Total Operational Funds (SOB+SIG) Received in 2015-16, by Quintile and Level Primary Secondary Operational fund Stand All schools With ECE Lower Full quintiles alone Quintile 1 $714 $891 $765 Quintile 2 $1,342 $1,332 $1,353 $1,338 Quintile 3 $1,733 $1,803 $1,926 $1,730 $1,815 Quintile 4 $2,509 $2,495 $2,536 $2,648 $2,512 Quintile 5 $5,683 $5,002 $8,401 $6,715 All schools $1,278 $2,270 $2,946 $7,910 $2,629 Note: 385 schools had sufficient data to be included. For most quality issues, the total amount available is the key factor. This is related to the ability to invest in school quality both in terms of physical environment and supporting learning outcome. The average total operational funds available to a school is strongly linked to school level and location. Primary schools in a remote area received the lowest average operational funding while secondary schools in urban area received the highest average operational funding (Figure 14). Quality of service delivery in education | 49 Figure 14: Average Operational Funds Received by Type and Level of School PRIMARY REMOTE $994 PRIMARY RURAL $1,356 PRIMARY URBAN $1,614 PRIMARY EC REMOTE $1,677 PRIMARY EC RURAL $1,776 LOWER SEC REMOTE $2,190 LOWER SEC RURAL $2,731 PRIMARY EC URBAN $3,433 LOWER SEC URBAN $4,504 SECONDARY REMOTE $4,662 SECONDARY RURAL $6,130 SECONDARY RUBAN $11,044 $0 $2,000 $4,000 $6,000 $8,000 $10,000 $12,000 The proportion of SIG funds spent on quality aspects is also important in assessing management and community intentions to improve the school. Both school level and location were closely related to the percentage of SIG funds allocated to quality aspects. Total SIG funds reported to be spent on quality aspects in 2015-16 ranged from USD 17 to USD 7,359. The average across all schools was USD 757. While this may be a relatively small proportion of total operational spending in some schools, it supports activities that directly contribute to improving the quality of education.10 Across all schools, the average proportion of SIG funds spent on quality aspects was 70 percent. This varied from 59 percent spent on quality by standalone remote primary schools to an average of 86 percent spent by rural secondary schools (Figure 15). Figure 15: Percent of SIG Funds Spent on Quality Aspects by School Type and Location SECONDARY RURAL 86 SECONDARY REMOTE 85 LOWER SEC URBAN 79 PRIMARY URBAN 74 LOWER SEC RURAL 74 LOWER SEC REMOTE 73 PRIMARY EC URBAN 73 ALL SCHOOLS 70 PRIMARY EC REMOTE 69 PRIMARY RURAL 69 PRIMARY EC RURAL 66 PRIMARY REMOTE 59 0 20 40 60 80 100 10 SIG quality funds can be spent on measures such as: (1) increase teaching and learning materials in the classroom; (2) introduce experimental and corner subject materials in the school; (3) increase library materials and rooms; (4) improve the school environment and playground to be clean and green; (4) initiate life skill practices, including agriculture; (5) support children with disabilities and vulnerable groups; (6) support slow learners and dropout prevention; (7) staff training and workshops; and (8) strengthen the quality and efficiency of school management. Unfortunately, the interview forms did not include these aspects separately so it was not possible to analyze the impacts of SIG quality spending in detail. 50 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Choosing to spend a higher percentage of SIG funds on quality is a reflection of the value placed on quality aspects by the school. The percentage of SIG funding allocated to quality aspects is a school choice and is the responsibility of the SSC. SIG funds may also be spent on administration or on improving access to school. The impact of choice on spending is mediated by the total amount of SIG funds available to the school—a large proportion of a small amount may still be much smaller in total than a smaller proportion of a large amount. In practice, schools with smaller SIG amounts tended to allocate a smaller percentage of their funds to quality aspects, while schools with larger amounts of SIG funds allocated a larger percentage to quality.11 Figure 16: Average SIG Amount Spent on Quality Aspects, by Quintile and School Type $3,000 $2,500 $2,000 $1,500 $1,000 $500 $0 Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Primary Stand alone $118 $235 $412 $541 $875 Primary With ECE $224 $382 $560 $902 $1,875 Secondary Lower $196 $402 $526 $874 $1,368 Secondary Full $919 $1,459 $1,816 $1,156 $2,887 The overall pattern of SIG quality spending by school level and quintile is similar to that for the receipt of operational funds. Schools were again divided into quintile categories based on the total SIG funds they spent on quality aspects (Figure 16). Spending by quintile ranged from USD 187 for the lowest quintile to USD 1,998 for the highest. Standalone primary schools spent the least on average (USD 357) while full secondary schools spent the most (USD 2,439). Schools in the lowest quintile spent much less on quality inputs than corresponding schools in the highest quintile. It seems clear that even with a high level of management skills and small school size, it is difficult to make an impact on educational quality when total spending is an average of USD 118 as it is with standalone primary schools in quintile 1, compared with the same type of school in quintile 5 which spent on average USD 875 for the same purpose (Figure 16). Development of quality indices The survey forms administered included several areas focused on quality aspects of Cambodian schooling. For example, the director form asked a range of questions about the school, its financial management, and human resource issues. The classroom observation form, which recorded observations in Grades 3, 5 or 8, and the school grounds contained more data that could be used to assess quality issues in the school. Four quality indices were developed to measure general school 11 There was a significant [0.000] correlation of 0.52 between the total SIG amount and the percentage allocated to quality. Quality of service delivery in education | 51 quality, financial management quality, classroom quality, and environmental quality. It was hypothesized that the measures of school spending outlined in Section 5.1 would predict or correlate strongly with the quality indices. School quality index Ultimately, school quality—as measured by the educational achievement and personal growth of students completing any stage in their education—is a function of both the characteristics of the school and the socio-economic status of the community from which the students are drawn. Characteristics of the school that are important include management, teaching, and school environment. The index developed here refers mainly to operational quality across a number of observable indicators that are amenable to a single-visit survey approach. The school quality index has sixteen items and was developed from the director’s form, which covered many areas related to school functioning.12 It is recognized that some of the conditions related to school quality are not fully under the control of the school. For example, although electricity and drinking water are crucial to a safely functioning school of reasonable quality, they are not uniformly available in all districts of Cambodia. While school-based provision of these services through generators or solar systems or through bores and pumps is possible, it is not usually affordable, especially for small rural schools where the majority of the students’ households do not have them. However, other items such as recordkeeping are well within the reach of even very small schools. Table 20: Items in the School Quality Index 1. Has School Development Plan 9. Library open to students 2. Has Teacher Attendance book 10. Has computers for students 3. Has Student Attendance book 11 Teacher absences not a problem 4. Has Library inventory 12. Adequate textbooks 5. One shift 13. Adequate learning materials 6. Has safe drinking water 14. Adequate student furniture 7. Has reliable electricity 15. Adequate maintenance resources 8. Has library 16. Adequate library resources Common features in school quality related responses include availability of teacher attendance books and SDPs in most school, operation of more than one shifts, inadequate safe drinking water, and very limited availability of student computers. Teacher attendance book and SDPs are available in 84 percent and 87 percent of all schools respectively. Sixty two percent of all schools operated more than one shift. More than two out of three primary schools operated more than one shift, compared to only one in four lower secondary schools. Thirty seven percent of the lower secondary schools had safe drinking water. Only 6 percent of all schools had student computers (Table 21). 12 It covers several aspects of the school including recordkeeping (items 1, 2, 3, 4), human resources (items 2, 11, 5), physical aspects (items 6, 7, 14, 15), educational resources (items 10, 12, 13), and library resources (items 4, 8, 9, 16) (Table 20). These constitute a basic set of conditions and resources that all schools should ideally possess. 52 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Table 21: School Quality Items, by School Type Primary Lower Full Primary school + Secondary Secondary All School Quality Index items school ECE school school schools 1 Has student computers 0% 3% 13% 33% 6% 2 Has adequate student furniture 28% 33% 32% 56% 33% 3 Has one shift 34% 24% 75% 58% 38% 4 Reliable electricity 26% 56% 48% 75% 47% 5 Has safe drinking water 45% 64% 37% 47% 52% 6 Students have library access 36% 75% 48% 69% 58% 7 Has adequate learning materials 59% 60% 49% 69% 59% 8 Has library inventory book 57% 68% 51% 58% 61% 9 Has library 37% 81% 54% 78% 62% 10 Has adequate maintenance funds 64% 69% 52% 78% 66% 11 Has Student Attendance book 69% 69% 57% 67% 67% 12 Has adequate textbooks 64% 68% 67% 81% 68% 13 Has adequate library resources 67% 72% 73% 83% 72% 14 Teacher absence not a problem 79% 85% 62% 75% 79% 15 Has Teacher Attendance book 75% 88% 84% 94% 84% 16 Has School Development Plan 79% 90% 92% 94% 87% Responses to the school quality index items also differed significantly by area, namely between urban and remote schools (Table 22). Urban schools were more likely to have computers for students, have a library, give students access to the library, and have an SDP. However, the libraries may not have adequate resources. Remote schools were much less likely to have safe drinking water and reliable electricity, have a library and a library inventory, and have an SDP. Location did not seem to be a strong factor in the school having adequate student furniture, maintenance funds, and teaching materials; having one shift; keeping Teacher and Student Attendance books; and having few problems with teacher attendance. Table 22: School Quality Items, by School Location All School Quality Index items Urban Rural Remote schools 1 Has student computers 13% 4% 3% 6% 2 Has adequate student furniture 32% 34% 31% 33% 3 Has one shift 32% 41% 37% 38% 4 Reliable electricity 78% 44% 17% 47% 5 Has safe drinking water 69% 50% 37% 52% 6 Students have library access 78% 57% 36% 58% 7 Has adequate learning materials 57% 60% 56% 59% 8 Has library inventory book 70% 62% 47% 61% 9 Has library 82% 61% 40% 62% 10 Has adequate maintenance funds 68% 65% 67% 66% 11 Has Student Attendance book 73% 64% 67% 67% 12 Has adequate textbooks 58% 70% 71% 68% 13 Has adequate library resources 60% 75% 76% 72% 14 Teacher absence not a problem 84% 78% 73% 79% 15 Has Teacher Attendance book 92% 82% 81% 84% 16 Has School Development Plan 94% 88% 77% 87% Quality of service delivery in education | 53 Financial quality index The financial quality index has 12 items and covers items in the director’s form and SOB records (Table 23). These items are part of the standard Ministry financial procedures and activities, and responsible school staff should have received training and manuals to cover them. Thus, all schools should be able to receive a high score on financial management practices. The survey did not investigate how well or poorly the school carried out these practices, so the index only reflects the extent to which the school could demonstrate its adherence to standard procedures.13 Table 23: Items in the Financial Quality Index 1. School has bank account 4. School follows MoEYS financial guidelines 2. School has budget process 5. School has consolidated budget proposal 3. School operating budget: 6. School has an approved SOB budget 7. School has consolidated quarterly a. Provided to School Support Committee financial reports b. Provided to community members on request 8. School has consolidated school performance checklist c. Publicly posted d. Posted in available format to public 9. School knows SOB entitlement It is recognized that there are reasons why some schools find it difficult to meet these standards. The principal reason is probably lack of training due to replacement of trained staff, as directors and senior school staff retire, are transferred, or promoted to other schools. Without a comprehensive plan to identify and train newcomers, accumulated knowledge and training slowly dissipates through attrition. Even with training, staff sometimes fail to implement the procedures they have been taught. As might be predicted, all types of schools had positive responses on the items in the financial quality index. Indeed, for items 6 to 12 on the list, the scores by school type were uniformly high (Table 24). The lowest scores overall were for the accountability items in the index. These items were related to the availability and clarity of the school budget to the community and its representatives on the SSC. Only two in five schools of whatever type made the budget available to the community on request—the lowest response for any item in this index. About two-thirds made the school budget available to the SSC, who were supposed to be involved in developing it. About four in five posted the budget in a public place, although only slightly more than half of the lower secondary schools did so. 13 The Financial Quality index could thus be described as a measure of observed financial practices linked to good financial management as set out in standard government operating procedures. 54 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Table 24: Financial Quality Items, by School Type Primary Lower Full Primary school + Secondary Secondary All Quality Index items school ECE school school schools 1 SOB budget available to community on request 42% 45% 37% 45% 43% 2 SOB budget provided to SSC 61% 56% 67% 75% 61% 3 SOB budget posted in a public place 72% 87% 56% 83% 77% 4 SOB budget presented in understandable format 88% 86% 71% 75% 83% 5 Has consolidated school performance checklist 85% 93% 86% 89% 89% 6 Has consolidated budget proposal for SOB 90% 89% 87% 92% 90% 7 Has consolidated report on SOB 2016 91% 94% 89% 92% 92% 8 Follows MoEYS guidelines 98% 97% 98% 97% 98% 9 Has consolidated quarterly report on SOB 99% 100% 97% 97% 99% by subchapter 10 Has budget process 98% 100% 100% 100% 100% 11 Has bank account 100% 99% 100% 100% 100% 12 Knows entitlement for SOB 99% 100% 100% 100% 100% Similarly, compliance for most of the procedural items (6-12) in the financial quality index was high across all locations; however, remote schools were much less likely to be accountable to their community than urban schools (items 1-3 in Table 25). This is a concern given that remote communities are already likely to be disadvantaged in terms of resources and are less likely to be closely supervised by provincial and district staff due to travel difficulties. Table 25: Financial Quality Items, by School Location All Quality Index items Urban Rural Remote schools 1 SOB budget available to community on request 52% 42% 33% 43% 2 SOB budget provided to SSC 61% 63% 53% 61% 3 SOB budget posted in a public place 89% 76% 63% 77% 4 SOB budget presented in understandable format 86% 82% 84% 83% 5 Has consolidated school performance checklist 93% 89% 85% 89% 6 Has consolidated budget proposal for SOB 95% 87% 91% 90% 7 Has consolidated report on SOB 2016 95% 91% 91% 92% 8 Follows MoEYS guidelines 98% 98% 94% 98% 9 Has consolidated quarterly report on SOB by subchapter 98% 100% 99% 99% 10 Has budget process 100% 99% 100% 100% 11 Has bank account 99% 100% 100% 100% 12 Knows entitlement for SOB 100% 100% 100% 100% Quality of service delivery in education | 55 Classroom quality index The classroom quality index has ten items, and the data was gathered during a class observation session in either a Grade 3, Grade 5, or Grade 8 class. The first three items reflect the presence of a prepared teacher with a class, while the remaining seven reflect the presence of essential items for quality education such as textbooks, learning material, and suitable and adequate furniture for teachers and students (Table 26). The data collectors selected a class from the desired grade randomly and observed the whole lesson, during which they noted the presence of—and counted where necessary—the various items on the list. Table 26: Items in the Classroom Quality Index 1. Teacher has lesson plan 6. Storage cupboard present 2. Teacher presence 7. Wall displays of teaching aids/student work 3. Percent class attendance 8. Student desks in good repair [pct.] 4. Proportion of textbooks 9. Student chairs in good repair [pct.] 5. Teacher furniture 10. Other books and reference material present Most school types had positive responses to the items on the classroom quality list (Table 27). Good condition of desks and chairs for students; display of student’s work or teaching aid; presence of teacher; and presence of a desk and a chair for teacher can be found in more than 90 percent of the all schools. For items 1 and 2 in Table 27 (presence of a storage cupboard and of other books and reference material), the overall score was low. There was a clear distinction between primary and secondary schools, with the latter much less likely to have either of the two items in the classrooms inspected. Table 27: Classroom Quality Items, by School Type Primary Lower Full Primary school + secondary secondary All Quality Index items school ECE school school schools 1 Other books and reference material present 42% 45% 29% 25% 40% 2 Storage cupboard for materials 43% 54% 19% 3% 41% 3 More than 75% have textbooks 74% 79% 60% 51% 72% 4 Teacher has lesson plan 85% 80% 84% 81% 82% 5 More than 75% attendance 78% 86% 87% 78% 83% 6 More than 75% of desks in good 89% 91% 89% 97% 91% condition 7 More than 75% of student chairs in 93% 94% 92% 97% 93% good condition 8 Wall displays of teaching aids or student 95% 97% 87% 92% 94% work 9 Teacher present in classroom 95% 98% 87% 94% 95% 10 There is a desk and chair for the teacher 97% 99% 100% 100% 99% 56 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey The presence or absence of textbooks (item 3) differed between school levels, with secondary school classes less likely to have 75 percent or more of the students with a textbook in class for that lesson. Three-quarters or more of the primary classes had 75 percent of their students with textbooks, compared to only half of the full secondary school classes, which is a significant difference. School location differences could be seen for the first five items of the classroom quality index (Table 28). More than half the urban schools had storage cupboards and extra reference material present in the classroom, compared to only one-third or less of the remote schools. A similar distinction could be seen in attendance: 91 percent of those on the class roll were present in the urban schools compared to 71 percent for remote school (Table 28, item 5). For items 6-10 in the index, the results were broadly similar across urban, rural, and remote classrooms observed by the data collectors. Table 28: Classroom Quality Items, by School Location All Quality Index items Urban Rural Remote schools 1 Other books and reference material present 51% 37% 35% 40% 2 Storage cupboard for materials 53% 39% 29% 41% 3 More than 75% have textbooks 79% 70% 71% 72% 4 Teacher has lesson plan 82% 84% 77% 82% 5 More than 75% attendance 91% 83% 71% 83% 6 More than 75% of desks in good condition 94% 90% 91% 91% 7 More than 75% of student chairs in good condition 97% 92% 94% 93% 8 Wall displays of teaching aids or student work 96% 93% 96% 94% 9 Teacher present in classroom 97% 95% 93% 95% 10 There is a desk and chair for the teacher 99% 99% 97% 99% Environmental quality index The environmental quality index has nine items, which refer to both safety and utility. The data collectors were told to walk around the school grounds and note the issues assessed in this index. They included items to do with safety (items 1, 2, 4), amenity (items 3, 5, 6), and utility (items 7, 8, 9) (Table 29). Quality of service delivery in education | 57 Table 29: Items in the Environmental Quality Index 1. Grounds secured by fence and gate 6. Toilets in good repair with doors 2. Play area with safe equipment 7. Garden area for teaching purposes 3. Available drinking water for students 8. Area neat, tidy with plants and trees 4. Secure bike parking area 9. Well-defined and tidy assembly area 5. Toilets for both male and female students   Across all school types, schools were least likely to have safe drinking water available on the school grounds and a garden area for teaching associated with life skills (Table 30). They were most likely to have toilets for both boys and girls and for the toilets to be in good shape with doors. Across all the environmental items, standalone primary schools were less likely to have the designated items than primary schools that included an ECE. The same pattern was observed with lower secondary schools and full secondary schools. Without further analysis, it seems likely that size may play a role in this pattern. Table 30: Environmental Quality Items, by School Type Primary Lower Full Primary school + secondary secondary All Quality Index items school ECE school school schools 1 Safe drinking water in grounds 34% 47% 13% 32% 37% 2 There is a garden area for teaching purposes 24% 39% 44% 60% 37% 3 Safe play area 37% 61% 40% 81% 52% 4 Secure bike parking area 50% 72% 58% 92% 65% 5 Secure grounds fenced with gate 52% 80% 71% 84% 70% 6 There is a well-defined and tidy assembly 55% 78% 74% 89% 71% area 7 The grounds are neat with plants and trees 61% 80% 84% 89% 76% 8 Toilets for both boys and girls 80% 93% 81% 87% 86% 9 Toilets are in good shape with doors 84% 96% 94% 92% 91% Location has a clear association with all of the environmental quality items (Table 31). In every case, the percentage of urban schools with the desired environmental characteristic is higher than that of rural schools, and in turn, the percentage of rural schools is higher than that of remote schools. Remote schools are doubly disadvantaged in that they are on average smaller as well as relatively isolated. One-quarter or less of the sampled remote schools have safe drinking water and a safe play area. In addition, only one-quarter of them have a garden for teaching purposes, although it seems that this would be fairly easy to set up in a remote school. Nearly twice as many urban schools have a garden, despite being pressed for room in most cases and generally having fully covered compounds. 58 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Table 31: Environmental Quality Items, by School Location All Quality Index items Urban Rural Remote schools 1 Safe drinking water in grounds 40% 38% 25% 37% 2 There is a garden area for teaching purposes 46% 37% 26% 37% 3 Safe play area 73% 53% 22% 52% 4 Secure bike parking area 91% 62% 39% 65% 5 Secure grounds fenced with gate 88% 70% 49% 71% 6 There is a well-defined and tidy assembly area 90% 72% 46% 71% 7 The grounds are neat with plants and trees 87% 77% 57% 76% 8 Toilets for both boys and girls 94% 87% 74% 86% 9 Toilets are in good shape with doors 93% 93% 86% 91% 6 Quality aspects School quality aspects School quality scores ranged from 2 to 18 out of a possible 19. The lowest score was a rural primary school that was categorized as disadvantaged, while the highest score of 18 was an urban lower secondary school that was not disadvantaged. Phnom Penh and Kampong Cham provinces were clearly ahead of the remaining provinces on this measure (Figure 17), while Preah Vihear, a remote province in northeastern Cambodia, was well behind the other provinces on overall school quality measures. Figure 17: School Quality Index Scores by Province PHNOM PENH 11.7 KAMPONG CHAM 11.0 KAMPONG THOM 10.3 PREY VENG 10.3 BATTAMBANG 10.2 KAMPONG SPEU 9.9 TBAUNG KHMUM 9.8 PURSAT 9.8 KAMPONG CHHNANG 9.3 PREAH VIHEAR 8.1 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 Full secondary schools scored best on the school quality index, along with urban primary schools with an ECE. At the other extreme were rural and remote standalone primary schools, which scored less than half on the index (Figure 18). 60 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Figure 18: Average School Quality Index Scores, by School Type and Location SECONDARY URBAN 12.6 SECONDARY RURAL 12.3 SECONDARY REMOTE 12.0 PRIMARY EC URBAN 11.7 PRIMARY EC RURAL 10.5 PRIMARY URBAN 10.4 LOWER SEC REMOTE 9.8 LOWER SEC URBAN 9.6 LOWER SEC RURAL 9.6 PRIMARY EC REMOTE 9.4 PRIMARY RURAL 8.9 PRIMARY REMOTE 7.5 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 The total amount of operational funding (SOB funds plus SIG funds)14 was divided into quintiles to assess the consequences for school quality. The one-fifth of schools in the sample that received the lowest amount of operational funds received between USD 248 and USD 1,092. In contrast, schools in the top quintile each received more than USD 3,103, with the top school receiving more than USD 25,000 (Table 32). Table 32: Quintile Divisions of Total Operational Funds Received by Schools in 2016-17 Quintile 1 $248 - $1,092 Quintile 2 $1,093 - $1,578 Quintile 3 $1,579 - $2,172 Quintile 4 $2,173 - $3,102 Quintile 5 $3,103 - $25,832 All schools in the lowest operational funds quintiles were primary schools while secondary schools (including lower secondary) are more commonly found in higher quintiles. Nearly half of those receiving the highest level of operational funds (quintile 5) were full secondary schools (Table 33). The composition of each quintile in terms of school type varies considerably. In terms of location, nine out of ten schools in the lowest quintile were remote or rural schools, while less than half of those in the top quintile were in remote or rural areas. Only 6 percent of remote schools were in the top quintile for operational receipts, while 45 percent of urban schools were in this quintile. Table 33: Operational Fund Receipt Quintiles, by School Level 14 Total funding was used rather than the per capita amount because it was judged that for many quality aspects, the total funds available were more significant—quality aspects such as school fencing, teacher training, or remedial classes were not dependent on per capita funding. In addition, per capita funding was found to be usually inversely related to the indices. This is because the highest per capita funding goes to small, disadvantaged schools. Even with high per capita funding, due to their size, these schools usually do not have large enough sums to make significant improvements to the school. Quality aspects | 61 Operational fund quintiles   School type 1 2 3 4 5 Total Primary School 55 25 16 18 0 114 Primary School with ECE 22 43 43 39 24 171 Lower Secondary School 0 9 17 18 19 63 Secondary School 0 0 1 2 34 37 All schools 77 77 77 77 77 385 On average, schools that received the most operational funds had the highest school quality index scores. At the same time, those schools that received the least funds had the lowest mean school quality index scores (Figure 19). However, schools in quintile 3, which received USD 1,815 on average, had lower school quality index scores than those in quintile 2, which received USD 1,338 on average. The small disadvantaged primary schools in quintile 1 receive much higher per capita operational funds from both SOB and SIG, but this does not and probably cannot compensate for the limitations imposed on them by their size. Even if they had unlimited funds, the small numbers of children restrict competition and classroom interaction, and the poor situation of their parental homes and staffing difficulties would still work against high-quality outcomes. Figure 19: School Quality Spending by Operational Spending Quintile Mean school quality index 11.1 11.5 10.2 9.9 8.2 QUINTILE 1 QUINTILE 2 QUINTILE 3 QUINTILE 4 QUINTILE 5 Total operational spending The percentage of SIG funds spent on quality was divided into quintiles to enable a comparison of effects on school quality. The 20 percent of schools that spent the lowest percentage of their SIG funds on quality aspects (quintile 1) spent between 14-57 percent of their SIG funds on quality-related areas. The schools in the highest quintile (quintile 5) spent 85 percent or more on quality aspects (Table 34). 62 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Table 34: Quintile Divisions of SIG Quality Spending Percentages Quintile 1 14 - 57 percent Quintile 2 58 - 69 percent Quintile 3 70 - 76 percent Quintile 4 77 - 84 percent Quintile 5 85 - 100 percent Spending more on quality is strongly associated with a higher school quality index score. Schools that spent about half or less of their SIG funds on quality scored less than half of the possible school quality index score. Average school quality index scores rise progressively across each quintile band, and schools that spent the largest proportion of their SIG funds on quality-related areas scored an average 11.3 on the school quality index (Figure 20). Figure 20: Average School Quality Index, by Quality Spending Quintile 11.31 10.29 Mean School Quality Index 10.05 9.65 8.97 1[14-57%] 2[58-69%] 3[70-76%] 4[77-84%] 5[85%+] Quality spending quintile Financial quality aspects Financial quality scores ranged from 5 to 12 out of a possible 12. More than 20 percent of all schools scored the maximum of 12 points on financial management, showing that a substantial minority of schools could meet all the financial management requirements. Prey Veng and Kampong Speu provinces averaged the highest in financial management, with Preah Vihear again being the lowest-scoring province (Figure 21). All provinces had fairly high average scores on financial management. The lowest-scoring school was an urban lower secondary school that was not disadvantaged. Quality aspects | 63 Phnom Penh Figure 21: Average Financial Quality Index Score, by Province PREY VENG 10.7 KAMPONG SPEU 10.7 PHNOM PENH 10.5 PURSAT 10.3 BATTAMBANG 10.2 KAMPONG CHHNANG 10.2 KAMPONG CHAM 10.1 TBAUNG KHMUM 10.0 KAMPONG THOM 9.8 PREAH VIHEAR 9.4 8.5 9.0 9.5 10.0 10.5 11.0 Remote and rural schools saw lower financial management scores, while urban schools were more likely to have higher scores (Figure 22). As this index depends mostly on knowledge of and implementation of standard financial procedures covered in the manuals that every school should possess, the results suggest that either rural schools have had less training or that their school directors and financial officers have less capacity than those in urban areas. Phnom Penh Figure 22: Average Financial Quality Index Score, by School Type and Location SECONDARY REMOTE 12.00 SECONDARY URBAN 10.85 PRIMARY URBAN 10.82 PRIMARY EC URBAN 10.63 PRIMARY EC RURAL 10.25 PRIMARY RURAL 10.22 LOWER SEC URBAN 10.00 PRIMARY EC REMOTE 9.94 SECONDARY RURAL 9.90 LOWER SEC RURAL 9.80 PRIMARY REMOTE 9.73 LOWER SEC REMOTE 9.64 0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 The proportion of SIG spent on quality aspects is weakly related to financial quality scores for primary schools and is not at all related for secondary schools.15 The mean financial quality scores were similar across SIG quality percent quintiles one to four (Figure 23). Only those schools that spent 85 percent or more of their SIG on quality outputs had a higher mean financial quality score. Province and location in the remote-urban dimension appear to be a stronger determinant of whether a school has good financial management as measured by the financial quality index. 15 For all schools, the Pearson correlation is weak at 0.102 [significance 0.045]. For primary schools it is 0.15 [significance 0.011], and for secondary schools the correlation is 0.095 [significance 0.351]. 64 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Phnom Penh Figure 23: Mean Financial Quality Score, by SIG Quality Percent Quintiles Mean financlal quality score 10.1 10.1 10.1 10.2 10.4 QUINTILE 1 QUINTILE 2 QUINTILE 3 QUINTILE 4 QUINTILE 5 SIG quality percent quintiles The total amount of operational funding does not have a significant association with the financial quality index in the sampled schools. There is very little difference between the average scores of schools that receive the lowest level of SOB and SIG funding and those that receive the highest level (Figure 24). This finding suggests that school practices in financial operations for their operational funds are not related to the size of the funds managed by the school. Even schools operational with smaller amounts of Phnom Penh funds to spend and account for can manage the tasks sufficiently. Figure 24: Mean financial quality index by operational funding quintile Mean Financlal quality score 10.1 10.2 10.3 10.1 10.4 QUINTILE 1 QUINTILE 2 QUINTILE 3 QUINTILE 4 QUINTILE 5 Operational funding quintiles Quality aspects | 65 Classroom quality aspects On the classroom quality index, 56 schools achieved the highest possible score of 13. Most of the high-scoring schools were primary schools (16 standalone primary schools and 36 primary schools with ECE). Lower and full secondary schools had relatively low average scores on classroom quality (Figure 25). Most of the components of this index were based on classroom resources related to education quality, and it appears that secondary schools are less likely to have adequate classroom furniture and teaching resources such as textbooks and teaching aids. This is a change in pattern, as secondary schools scored more highly on both Phnom Penh the school and financial quality indices. Figure 25: Mean Classroom Quality Index, by School Type and Location PRIMARY EC URBAN 11.8 PRIMARY URBAN 11.7 PRIMARY EC RURAL 11.3 LOWER SEC URBAN 11.2 PRIMARY RURAL 11.2 PRIMARY EC REMOTE 11.0 PRIMARY REMOTE 10.9 LOWER SEC RURAL 10.5 SECONDARY RURAL 10.4 SECONDARY URBAN 10.3 LOWER SEC REMOTE 10.0 SECONDARY REMOTE 9.5 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 Differences on this measure among the sampled provinces were small, and schools with lower scores were scattered geographically. Phnom Penh had the highest mean score on this classroom measure, along with Kampong Chhnang province (Figure 26). Again, Preah Vihear had the lowest mean score of 10.5. Most provinces had a mean score across their 40 schools of 11 or more out of a possible 13. Poor schools on this measure were scattered rather than located in a few provinces, suggesting that addressing this issue will be a task for POEs and DOEs. 66 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Phnom Penh Figure 26: Mean Classroom Quality Index, by Province PHNOM PENH 11.6 KAMPONG CHHNANG 11.6 KAMPONG SPEU 11.3 KAMPONG CHAM 11.2 TBAUNG KHMUM 11.2 KAMPONG THOM 11.1 BATTAMBANG 10.8 PURSAT 10.8 PREY VENG 10.8 PREAH VIHEAR 10.5 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 The scores revealed some surprising findings on the relationship between spending on quality and classroom quality. Perhaps unsurprisingly, the 20 percent of schools that spent the lowest proportion of their SIG funds on quality had the lowest mean classroom quality scores. However, quintile 2 had a much higher mean score for classroom quality than might be expected from spending relatively less on quality aspects. Moreover, the relationship between spending on quality and classroom quality then reverses and falls as schools spend a higher proportion of Phnom their SIG funds on quality Penh 27). (Figure Figure 27: Mean Classroom Quality Index Score, by SIG Quality Spending Quintile 11.3 Mean classroom quality score 11.2 11.1 11.0 10.7 QUINTILE 1 QUINTILE 2 QUINTILE 3 QUINTILE 4 QUINTILE 5 SIG quality spending quintiles Quality aspects | 67 Surprisingly, the 20 percent of schools that received the highest average operating funds had the lowest average classroom quality score. Average classroom quality scores were high and rose consistently through the first four quintiles, but then fell for the fifth and highest quintile (Figure 28). Two-thirds of the schools in this quintile were secondary schools, and all received at least USD 3,000 in operating funds from their SOB and SIG transfers. All of the schools in quintile 1, which were the lowest 20 percent of schools in terms of their operating funds from SOB and SIG, were Phnom primary schools. Penh Figure 28: Mean Classroom Quality Index Score, by Operational Funding Quintile 11.3 11.3 Mean classroom quality score 11.1 11.0 10.9 QUINTILE 1 QUINTILE 2 QUINTILE 3 QUINTILE 4 QUINTILE 5 Operational funding quintiles Environmental quality aspects Phnom Penh had the highest mean score and Preah Vihear the lowest, with its 40 schools scoring only half of the possible score for environmental quality on average (Figure 29). The environmental quality index has nine items that are scored equally, so the range of scores is from 0 to 9. The index focuses on the physical environment of the school grounds, which means that the data collectors saw little of quality on the grounds of the schools that scored zero. One in ten schools scored a full nine points, meaning that they were satisfactory on all the items. Ten schools failed to score any points in the assessment of their environmental quality. 68 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Figure 29: Mean Environmental Quality Index Scores, by Province PHNOM PENH 7.1 KAMPONG SPEU 6.7 KAMPONG THOM 6.6 KAMPONG CHAM 6.4 PREY VENG 5.9 KAMPONG CHHNANG 5.8 PURSAT 5.5 TBAUNG KHMUM 5.3 BATTAMBANG 4.9 PREAH VIHEAR 4.4 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 Remote schools did very poorly on the assessment of the quality of their school environment. Primary and secondary remote schools had the lowest mean scores on this index, with standalone remote primary schools averaging only 3.6 on environmental quality, and remote full secondary schools only 4.0. Full secondary rural and urban schools did much better with average environmental quality scores of 7.0 and 7.6, respectively. Urban and rural primary schools with ECE also scored very well, with average scores of 7.2 and 6.4, respectively (Figure 30). Phnom Penh Figure 30: Mean Environmental Quality Index, by School Type and Location SECONDARY URBAN 7.6 PRIMARY EC URBAN 7.2 SECONDARY RURAL 7.0 PRIMARY EC RURAL 6.4 PRIMARY URBAN 6.4 LOWER SEC URBAN 6.1 LOWER SEC RURAL 5.5 LOWER SEC REMOTE 5.5 PRIMARY RURAL 5.1 PRIMARY EC REMOTE 4.6 SECONDARY REMOTE 4.0 PRIMARY REMOTE 3.6 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 A strong positive relationship was found between the amount received by schools as operational funding (SOB plus SIG) and the environmental quality score. Schools that received the lowest amounts of operational funding had the lowest average scores—these were all primary schools and mainly located in rural and remote areas (Figure 31). Their average score was not much more than half the possible score, indicating a poor state of the playground and associated utilities such as water and toilets. Addressing most of these environmental aspects—fencing the Quality aspects | 69 school grounds, providing sufficient safe toilets, establishing a bore for safe water in rural and remote regions—can be costly. It is difficult to see how schools in the Phnom Penh lowest quintile will be able to address such issues with their limited operating funds. Figure 31: Mean Environmental Quality Score by Operational Funding Quintile Mean environmental quality score 7.2 6.0 6.0 5.6 4.9 QUINTILE 1 QUINTILE 2 QUINTILE 3 QUINTILE 4 QUINTILE 5 Operational funding quintiles Schools that spend the lowest proportion of their SIG funds on quality outputs have the lowest mean environmental quality scores. Although not as strong as the relationship between environmental quality scores and total operational funding (probably because the amounts are much smaller), it is still clearly a positive relationship (Figure 32). Unfortunately, the data collected did Phnom Penh not include the individual quality outputs that the schools addressed in their SIG budgeting. Figure 32: Mean Environmental Score, by SIG Quality Spending Quintile Mean environmental quality score 6.8 6.1 5.6 5.8 5.0 QUINTILE 1 QUINTILE 2 QUINTILE 3 QUINTILE 4 QUINTILE 5 SIG quality spending quintiles 70 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Summary quality aspects Most quality aspects of the schools in the sample—and in Cambodia— are linked to school size (Table 35). School size largely determines funding, as a major component of both SOB and SIG disbursements is related to per capita payments. Teachers and classrooms are also linked to student numbers, although not directly. Student numbers are affected by the practice of shifts, in which the school population is divided into two or even three sets who use the same facilities and are mostly taught by the same teachers. Within any one level of schooling, the quality indices are mostly directly related to location, with urban schools of whatever type having higher quality scores than rural, and rural having higher quality scores than remote locations. Again, this is related fairly directly to student numbers, as average school size is influenced by the population densities implied by urban, rural, and remote classifications. Table 35: Summary Quality, Funding, and School Characteristics, by Level and Location Characteristics Indices SIG Total Class Class Environ Level Location SOB quality operational Schools Students Teachers rooms Schools Finance rooms ment Secondary Urban $7,157 $3,383 $11,044 14 1,827 76 41 12.6 10.8 10.3 7.6 Rural $3,969 $1,885 $6,130 21 975 34 20 12.3 9.9 10.4 7.0 Remote $2,743 $1,642 $4,662 2 689 26 14 12.0 12.0 9.5 4.0 All $5,109 $2,439 $7,910 37 1,282 50 28 12.4 10.3 10.3 7.1 schools Lower Urban $3,047 $1,221 $4,504 11 619 42 15 9.6 10.0 11.2 6.1 Secondary Rural $1,886 $643 $2,731 41 316 19 7 9.6 9.8 10.5 5.5 Remote $1,461 $559 $2,190 11 227 12 5 9.8 9.6 10.0 5.5 All $2,015 $730 $2,946 63 353 22 8 9.7 9.8 10.6 5.6 schools Primary+EC Urban $2,092 $1,098 $3,433 53 836 27 21 11.7 10.6 11.8 7.2 Rural $1,054 $503 $1,776 101 350 9 9 10.5 10.3 11.3 6.4 Remote $936 $538 $1,677 19 273 7 7 9.4 9.9 11.0 4.6 All $1,362 $687 $2,270 173 490 14 12 10.7 10.3 11.4 6.5 schools Primary Urban $1,009 $385 $1,356 12 388 14 11 10.4 10.8 11.7 6.4 Rural $975 $506 $1,614 75 235 6 7 8.9 10.2 11.2 5.1 Remote $1,136 $255 $994 40 135 4 6 7.5 9.7 10.9 3.6 All $1,043 $357 $1,278 127 217 6 7 8.6 10.1 11.1 4.8 schools 7 Mathematics and physics achievement Mathematics and physics tests MoEYS policy provides for testing of subject achievement and learning outcomes. The Education Quality Assurance Department (EQAD) of MoEYS conducts the national assessment test every year for one grade in a cycle of three years for Grade 3, Grade 6, and Grade 8. The latest national assessment test was conducted in 2016. The PETS/QSDS team replicated the test in those schools that were not in the 2016 EQAD sample. The selection of students was done in two steps. For the first step, the data collectors had lists of 8th grade students in 2016 for each of the selected schools not tested in 2016. They used this list to select 15 students randomly. For the second step, after obtaining the random sample, the data collectors called the students by name when they arrived at school to do the test. If the sampled students were absent, replacements were made. A total of 1,495 students took the tests—595 in 2016 and 900 in 2017. Students’ achievements in mathematics and physics varied considerably, with weak results overall. On the mathematics test, the scores ranged from 6 to 94 percent with a mean score of 39 percent. For the physics test, the scores ranged from 0 to 87 percent with a mean score of 41 percent (Figure 33). Only 204 (14 percent) of the 1,495 students who took the mathematics test scored 60 percent or more, indicating some mastery of the subject. The results were worse for physics, with only 140 students (9 percent) scoring 60 percent or more. 72 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Phnom Penh Figure 33: Math and Physics Test Percentage Distribution 300 250 200 Frequency 150 100 50 0 0-5 6-10 11-1516-2021-25 26-3031-35 36-4041-4546-50 51-5556-6061-6566-70 71-75 76-80 81-8586-9091-95 96- 100 Mathematics 0 10 53 83 198 227 210 98 148 117 50 90 60 34 41 25 28 11 5 0 Physics 1 6 16 87 85 228 160 249 133 207 77 106 37 51 20 24 3 5 0 0 Percent Mathematics Physics Few schools averaged over 50 percent on either test. Only 11 schools (5 urban and 6 rural) had an average mathematics score of 50 percent or more, and just 8 schools (2 urban and 6 rural) averaged 50 percent or more in physics. Only five schools out of the 100 in the sample averaged 50 percent or more on both tests, and just one of these schools was urban. Mathematics results Urban full secondary schools had the strongest performance on the mathematics test, while remote lower secondary schools had the weakest performance. The urban full secondary schools, which had an average score of 45.4 percent (Figure 34), are the largest secondary schools in the sample and reputedly have the best teachers. The remote lower secondary schools and rural full secondary schools had the worst test results at 37.0 percent and 37.1 percent, respectively. Overall, urban schools had better results than rural schools in mathematics. Mathematics and physics achievement | 73 Phnom Penh Figure 34: Mean Mathematics Test Scores, by Type and Location of School SECONDARY URBAN 45.4 LOWER SECONDARY URBAN 41.0 SECONDARY REMOTE 39.3 LOWER SECONDARY RURAL 38.7 SECONDARY RURAL 37.1 LOWER SECONDARY REMOTE 37.0 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 50.0 Test scores varied across provinces. Some provinces such as Phnom Penh (average 43.7 percent) and Pursat (average 43.0 percent) had students who did better than the sample average (Figure 35). Even the students from the best provinces did not average 50 percent on the mathematics test. Other provinces such as Kampong Speu, which had the lowest average test score at 33.4 percent, did much worse. Most of the provinces clustered between 37 to 40 percent. Figure 35: Mean Mathematics Test Scores, by Province Phnom Penh PHNOM PENH 43.7 PURSAT 43.0 BATTAMBANG 42.8 KAMPONG CHAM 39.7 KAMPONG THOM 39.7 TBAUNG KHMUM 39.6 KAMPONG CHHNANG 37.7 PREY VENG 37.2 PREAH VIHEAR 37.1 KAMPONG SPEU 33.4 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 50.0 While no strong associations were found, there was a positive association between overall school quality and the mathematics results (Figure 36). The mathematics test results were divided into quintiles, from the lowest 20 percent of results to the highest 20 percent of results. Quality index scores were calculated for each quintile to check the level of association and whether it was significant— i.e. was the quality index predictive of the mathematics or physics results. Average school quality scores were low for students in the bottom 20 percent of mathematics scores and higher for those in the top 20 percent of scores. However, even though the association was positive, it was not strong. 74 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Phnom Penh Figure 36: Mean School Quality Score, by Math Test Quintile 10.6 10.8 10.6 11.0 Mean school quality score 10.4 QUINTILE 1 QUINTILE 2 QUINTILE 3 QUINTILE 4 QUINTILE 5 Math test quintiles Average classroom quality scores were almost identical across the whole range of quintile results in mathematics. Assessed classroom quality and assessed mathematics ability in the same school appeared to have almost no relationship. The same pattern was found with regard to environmental quality. There was a slight negative relationship between the mathematics quintiles and the environmental scores, but the relationship was small. No direct link was found between the measures of quality set out here as measures of service delivery and the results in the mathematics test. Province and school size appear to have some connection, but further analysis (perhaps multivariable) can be pursued to test how funding and quality aspects of schools contribute to mathematics achievement. Physics results Physics test results also varied among provinces, although apart from Battambang, the highest-achieving provinces were different from those in the mathematics test (Figure 37). Surprisingly, Phnom Penh—which had the best test results in mathematics—was only ahead of Preah Vihear and Kampong Speu, which had the lowest test results for physics. Mathematics and physics achievement | 75 Figure 37: Mean Physics Test Results, by Province Phnom Penh BATTAMBANG 45.6 KAMPONG CHAM 44.5 KAMPONG THOM 43.4 TBAUNG KHMUM 43.0 PURSAT 41.2 KAMPONG CHHNANG 40.4 PREAH VIHEAR 39.6 PHNOM PENH 39.0 PREY VENG 37.7 KAMPONG SPEU 33.6 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 50.0 Location and type of school were quite strongly associated with physics results (Figure 38). Overall, full secondary schools had better results than lower secondary schools, and within these groupings, urban schools had better results than rural, which in turn had better results than remote schools. One strong exception to this pattern was rural lower secondary school students, who had much better results than other lower secondary students and secondary students. Phnom Penh Figure 38: Mean Physics Test Results, by Location LOWER SECONDARY RURAL 42.5 SECONDARY URBAN 42.3 SECONDARY REMOTE 40.9 SECONDARY RURAL 39.1 LOWER SECONDARY URBAN 38.3 LOWER SECONDARY REMOTE 38.3 36.0 37.0 38.0 39.0 40.0 41.0 42.0 43.0 There is little evidence of an association between school quality scores and physics test results (Figure 39). The physics test results were divided into quintiles, from the lowest 20 percent of results to the highest 20 percent of results. Quality index scores were calculated for each quintile to check the level of association and whether it was significant—i.e. was the quality index predictive of the physics results. The lowest 20 percent of test results in physics were in schools where the average school quality result was 10.7, while the highest 20 percent of results were from schools where the school quality index was 10.5. Similarly, apart from the lowest 20 percent of physics scores, no real relationship was found between physics test results and financial quality scores (Figure 40). 76 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Figure 39: Mean School. Quality Scores, by Figure 40: Mean Financial Quality Scores, by Physics Test Quintile Physics Test Quintile Phnom Penh Phnom Penh 10.7 10.7 10.5 10.7 10.5 10.3 Average school quality score 9.9 9.9 9.9 Average financial quality score 9.8 QUINTILE 1 QUINTILE 2 QUINTILE 3 QUINTILE 4 QUINTILE 5 QUINTILE 1 QUINTILE 2 QUINTILE 3 QUINTILE 4 QUINTILE 5 Physics test quintiles Physics test quintiles The relationship between classroom quality and physics test results mirrors that for financial quality (compare Figures 40 and 41). The lowest-scoring 20 percent of physics students were in schools that had a higher average classroom quality score than the remaining physics students, whose classroom quality scores all averaged around 10.4. The association between physics results and environmental quality is stronger, but it is a negative association (Figure 42). The average environmental quality score for the lowest-scoring 20 percent of physics students is 6.5, compare to 5.7 for the highest-scoring 20 percent. Figure 41: Mean Classroom Quality Score, by Figure 42: Mean Environmental Quality Score, Physics Test Quintile by Physics Test Quintile Phnom Penh Phnom Penh 6.5 6.2 Average environmental quality score 6.2 5.7 10.8 5.7 Average classroom quality score 10.4 10.5 10.4 10.4 QUINTILE 1 QUINTILE 2 QUINTILE 3 QUINTILE 4 QUINTILE 5 QUINTILE 1 QUINTILE 2 QUINTILE 3 QUINTILE 4 QUINTILE 5 Physics test quintiles Physics test quintiles 8 Summary of findings and policy implications Summary of findings on fund flows Across the 400 sampled schools, the amounts stated to be sent to schools by Treasury or MoEYS were received in full by the schools. The flow of SOB and SIG is generally good from the provincial treasury or MoEYS HQ to POE then to school accounts. POEs received SOB funds in full and generally very soon after requesting the funds. There are generally only days between the initial POE request and the response from the provincial treasury, and the amounts are equivalent—i.e. what is requested is received. On average, fund flows to schools are also complete. Even with the anticipated change to SIF funding, which will see the two amounts combined at the national level to flow through the provincial treasury to school accounts, fund flows are expected to continue to be satisfactory. Timeliness of funds delivery to schools is a serious problem. Even if funds are received in full, delays in fund availability affect the efficiency of school operations and hence educational quality. Across all provinces, there are clear delays in SOB funding in the first quarter. SIG funds are also delayed but received by schools much earlier in the academic year, usually in January or February, which means that schools can use the funds to bridge the gap before their SOB funds arrive. POEs do not request first quarter SOB payments until April or May, which means SOB payments to schools are not received until well into the academic year. The POEs wait until they are informed by Treasury staff that the funds are in hand for payment. Fund requests are often made late in the third quarter, as well—instead of in June (immediately before the start of the quarter), requests were made in July or August. In addition, several POEs were often more delayed than average in the process of requesting or disbursing funds. 78 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Financial recordkeeping is poor in many schools. Discrepancies were found between what POE recorded as sent and what schools recorded as received, with school records showing an amount lower or higher than what POE recorded as sent. This suggests that recordkeeping at the school level is not sufficiently accurate. By province, schools reported receiving the full amount of SIG funds, but individually, there were similar recording problems as with SOB funds. Some schools under- reported, while others over-reported. The poor recordkeeping may be due to low capacity at the school level (school directors and SSC members) or lack of targeted support by DOEs, or a combination of both. Summary of findings on service delivery quality Quality aspects were analyzed relative to total operational funding and to percentage SIG spending on quality using four quality indices developed from questions and observations on several school-based survey forms. These were considered to be key determinants of the school’s ability to spend on quality- related areas, as the total amount of operational funding was important in terms of high-cost quality aspects. The percentage of SIG spending on quality was an indication of willingness to consider quality issues as important in school spending. Phnom Penh ranked at the top on three out of four indices, while Preah Vihear ranked last on all four. There were clear differences among provinces in most of the quality indices, suggesting that it would be useful initially to identify and target lower-quality provinces to get rapid gains in quality, rather than to simply address individual schools across all provinces. Secondary schools, with some variability by location, had higher average school and environmental quality scores and lower average classroom quality scores than primary schools. Urban schools generally scored higher than rural and remote schools on financial quality. Secondary schools and non-urban schools were less likely to provide classroom storage and to have other books and reference materials present in the classroom. This might suggest a difference in values and culture between these different types of schools. Schools were consistent in following financial procedures that made up most of the financial quality index. However, the three measures of community accountability involving distribution of the school budget were consistently low across all school levels and locations. This points to a need for further targeted interventions by the MoEYS Finance Unit to address accountability issues, especially for school directors. The total amount of operational funds (SOB plus SIG) received by schools related positively to all quality indices except classroom quality. More funds meant better quality. The proportion of SIG spent on quality also related positively to all indices except classroom quality. Plans for SIF expenditure on schools to 2021 as part of the new Sida funding proposal indicate a sharp rise in per school operational funds in this period commencing in 2018, with an almost doubling of school operational funds in SIF over the current combined SOB and SIG funds to Summary of findings and policy implications | 79 schools. As total operational funds increase, greater amounts could be available for spending on quality-related areas. Therefore, a rise in funding should mean better quality outcomes across all school types and locations. Poor results on mathematics and physics tests reflect quality of school outcomes. With only 14 percent scored 60 percent or more on mathematics test and 9 percent scored 60 percent or more on physics test, the challenge for quality improvement is huge. Policy recommendations Improve the timeliness of fund flows and align procedures for more efficiency in the management of school funds, as SOB and SIG funds flow together as SIF. Currently, the separate provision of SIG funds through a different disbursement process means that they bridge the gap in funding caused by the delay in the first quarter of SOB disbursement. Consideration could be given to providing authorization for expenditure in quarter one to mirror that in quarter four of the previous year, with any rectification made in quarter two to balance the books. Ease the rigidity of SOB subaccounts to enable schools to execute budget in a way that matches their needs. Rigidity of SOB subaccounts, which was identified in previous PETS, remains an ongoing challenge. Addressing this will require allowing SOB expenditures based on a school’s actual needs, without the constraint of line items imposed in the 12 subaccounts of two SOB chapters (60 and 61). Schools can report on actual expenditures, while reporting of expenditures following line items could still be retained by consolidating such expenditures at the provincial level. This can create efficacy in the use of SOB funds and improve development at the school level. Consider providing SOB payments to small schools in full early in the fiscal year. About 42 percent of schools received SOB funds less than USD 1,000, and in general, these schools rank low on the quality indices. Receiving one-quarter of a small amount four times a year precludes small schools from easily making major expenditures, such as those commonly needed to enhance the school environment. Address quality of learning outcome. Some measures to improve quality outcomes do not involve extra funding.16 Providing the intended number of instructional hours can be a powerful improver of outcomes. In Cambodia, the strongest determinant of this is the prevalence of double-shifting in schools, which reduces overall instructional hours. School management measures such as providing the full number of instructional days and the full number of instructional hours may be needed. 16 There may be some costs if this involves ensuring that replacement teachers are available during regular teachers’ absences, and the provision of such replacements is much more difficult in remote and small schools than in large and urban schools. 80 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Place top priority for school funding on boosting the quality of teaching and learning to improve substantive outcomes in terms of student achievement. The items could include support for slow learner students from week one of the school year, coaching of less experienced teachers by more experienced teachers in the school or from neighboring schools, teaching and learning materials, and enough drinking water and toilets for boys and girls. Improve community knowledge and involvement in school budgets and spending. This could be achieved through (i) improving compliance checks with the required disclosure of budget to the school, committee, local community through public display, and anyone requesting school budget information during provincial and district staff visits to schools and (ii) regularizing disclosure of the budget and actual expenditures as part of opening and closing parental meetings at all schools. Address widespread poor financial recordkeeping at the school level. Regular refresher training could be organized on a regional basis, grouping together geographically close provinces to provide enough numbers each year and to make training courses cost-effective. Refresher training can help address skill shortages, especially as attrition removes trained staff and replacements have no training. Soft skills should be included in the regular management training, in addition to hard skills such as accounting and recordkeeping. 9 Annexes Correlational aspects of the indices and school characteristics The discussion in Sections 6 and 7 made observations on the relations between the four indices and total school operational funding and SIG quality allocations by the school. However, it did not make quantified conclusions in terms of correlations, nor did it attempt to bring all the observations into one set so readers could compare the relative strength of the associations. One hypothesis that was not tested by this approach was whether what was observed in terms of the relationships between the indices and the funds available to the school was simply a school size effect. In other words, the larger the school, the larger the funds available because a significant proportion of both SOB and SIG funding is per capita based, and this accounts for the observed results. However, the relationship between funding and size is not direct. Each school receives a lump sum grant (Tables 1 and 2) as well as per capita funding, so as a consequence, small schools have higher actual per capita funding when all the funds are distributed over all the students. In addition, many schools are classified as disadvantaged, which further raises the actual per capita funding of small disadvantaged schools. A third factor that may further reduce the direct relationship between size and funding is the ability of the school community to choose what proportion of their SIG funds is spent on quality aspects of their school plan versus administration or access uses. To test whether the observed relationships were simply a function of the number of students, a correlation matrix was calculated including the four indices, three aspects of funding (amount of SIG funds devoted to quality, total operational funds, and total operational funds per student), and the total number of students in the school. This was done separately for the 300 primary schools (Table A1) and the 100 secondary schools in the study (Table A2). In addition, all of these were correlated with the mean test scores per school for mathematics and physics (Table A3). 82 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Primary correlations For primary schools, all of the pairs of correlations except that between the Classroom Quality index and the Financial Quality index were significant (Table A1). The great majority (24 out of 27) were highly significant (0.01 or less), which means that the observed relationships were very unlikely to be the result of chance. Most were positive—increases in one measure were paralleled by increases in the other. The exception was the set of correlations involving actual per student funding, which was negatively correlated with all of the other measures. This is also a size effect, as small schools (especially disadvantaged small schools) had larger base amounts relative to their student population. Thus, smaller schools had more per student funds. Table A1: Primary correlation matrix School School measures quality index School quality index Pearson Correlation 1.00 Financial Sig (2-tailed) quality index Financial quality Pearson Correlation .316** 1.00 Classroom index Sig (2-tailed) .000 quality index Classroom quality Pearson Correlation .222** .065 1.00 Environment index Sig (2-tailed) .000 .271 quality index Environment quality Pearson Correlation .471** .237** .315** 1.00 SIG index Sig (2-tailed) .000 .000 .000 quality funds SIG quality funds Pearson Correlation .299** .153** .206** .298** 1.00 Total Sig (2-tailed) .000 .009 .000 .000 operational funds Total operational Pearson Correlation .324** .148* .188** .304** -.978** 1.00 TOF per funds Sig (2-tailed) .000 .013 .002 .000 .000 capita TOF per capita Pearson Correlation -.367** -.118* -.122* -.294** -.305** -.308** 1.00 Total Sig (2-tailed) .000 .049 .042 .000 .000 .000 students Total students Pearson Correlation .339** .179** .176** .327** .961** .972** -.359** 1.00 Sig (2-tailed) .000 .002 .002 .000 .000 .000 .000 **.Correlation is significant at the 0.01 level (2-tailed). *.Correlation is significant at the 0.05 level (2-tailed). Although nearly all the relationships were statistically significant (27 out of 28), the strength of the relationships was mostly moderate. If two items are positively correlated to a significant degree, the score on the second item can be predicted by knowing the first since the relationship is not a chance one. However, the size of the correlation is a measure of how well one can predict the second by knowing the first. Correlations of the order of 0.2 to 0.4, which is where most of the correlations lie, indicate fairly weak predictive power and hence fairly small explanatory value. There are three exceptions in the matrix—all show a correlation of 0.95 or above and thus are very strongly related. Total operational funds are strongly correlated (0.978) with the amount of SIG funds allocated to quality aspects, but this is expected as SIG quality funds are part of the total operational funds as defined for the study. Total student numbers are also strongly correlated with both total operational funds and SIG quality allocations. Again, because a significant amount of both these funds is linked to per capita aspects of total SOB and SIG funds, the result is expected. It also Annexes | 83 suggests that the current ratio between lump sum funding and per capita funding is not strong enough to overcome the disadvantage of small school size. Secondary correlations The pattern of associations between the same variables for the 100 secondary schools is much weaker than for primary schools (Table A2). Out of 28 pairs, 21 of them have a significant relationship, and only 16 of these have a very strong relationship (0.01 or less). This was much less than observed with the primary schools. The classroom practices and resources summarized in the Classroom Quality index are not correlated significantly with any other measure apart from the Financial Quality index, and this relationship is weak with a correlation of 0.209 and a significance of 0.044. This is very different from the pattern observed for primary schools. Much of the content of the classroom index was based on observable resources in the actual classroom visited by the field workers (Tables 26 and 27). Primary classes are largely based in one room through the day with one generalist teacher. It is in the teacher’s interest to accumulate resources in that classroom, as it makes his or her workday better and helps the children learn. In contrast, teachers in secondary schools often move from class to class, and the classes from room to room. Thus, accumulating resources in one room is of much less benefit to any individual teacher. The Financial Quality index, which summarizes observed financial practices based on standard operating procedures, also has a much weaker set of correlations with other variables for secondary schools compared to primary schools. The other correlations in the matrix are generally stronger than the corresponding ones for primary schools. Table A2: Secondary correlation matrix School School measures quality index School quality Pearson Correlation 1.00 Financial index Sig (2-tailed) quality index Financial quality Pearson Correlation .270** 1.00 Classroom index Sig (2-tailed) .007 quality index Classroom quality Pearson Correlation -.118 .209* 1.00 Environment index Sig (2-tailed) .253 .044 quality index Environment Pearson Correlation .422** .215* .146 1.00 SIG quality index Sig (2-tailed) .000 .034 .156 quality funds SIG quality funds Pearson Correlation .445** .244* -.054 .465** 1.00 Total Sig (2-tailed) .000 .015 .599 .000 operational funds Total operational Pearson Correlation .423** .255* .011 .495** -.949** 1.00 TOF per funds Sig (2-tailed) .000 .011 .913 .000 .000 capita TOF per capita Pearson Correlation .409** -.185 -.086 -.350** -.469** -.404** 1.00 Total Sig (2-tailed) .000 .068 .406 .000 .000 .000 students Total students Pearson Correlation .484** .252* .009 .506** .940** .958** -.565** 1.00 Sig (2-tailed) .000 .012 .929 .000 .000 .000 .000 **.Correlation is significant at the 0.01 level (2-tailed). *.Correlation is significant at the 0.05 level (2-tailed). 84 | Cambodia Education Sector - Public Expenditure Tracking and Quality of Service Delivery Survey Correlations of Mathematics and Physics test results One measure of quality of service delivery is student achievement on standardized tests. Mathematics and physics scores were available for nearly 1,500 students in the sampled secondary schools. The students took the tests in either Grade 8 in 2016 or Grade 9 in 2017. The mean score by school for each of the tests was correlated with the same set of variables to test the strength of the relationships. It was hypothesized that the indices, the funding, or the school size would predict the test results (Table A3). However, only 3 of the 16 correlations were significant, and none of them were particularly strong. Table A3: Test results correlation matrix School Financial Classroom SIG Total quality quality quality Environment quality operational TOF per Total Exam subject Correlation index index index quality index funds funds capita students Pearson .180 -.041 -.066 -.062 .191 .228* -.080 .214* Mean maths Correlation percent Sig (2-tailed) .074 .689 .521 .542 .057 .022 .432 .032 Pearson -.001 -.109 -.178 -.269** -.081 -.057 .077 -.061 Mean physics Correlation percent Sig (2-tailed) .993 .286 .083 .007 .421 .572 .450 .544 **.Correlation is significant at the 0.01 level (2-tailed). *.Correlation is significant at the 0.05 level (2-tailed). Annexes | 85 Bibliography Advocacy and Policy Institute and CACHRD. 2013. Public Expenditure Tracking Survey in Early Childhood Education: Cambodia 2012. Advocacy and Policy Institute and CACHRD. Amin, Samia, Jishnu Das and Markus Goldstein [eds]. 2008. Are you being served? New Tools for measuring service delivery. Washington DC: The World Bank. Ampratwum, Edward, Daniel Armah-Attoh and Maxwell Ashon. 2012. Research Paper 20: Tracking possible leakages in the supply and distribution of textbooks in Public Primary Schools in Ghana. CDD-Ghana. Bolton, Laura. 2014. Education Expenditure Reviews. HEART Helpdesk Report. de Graaf, Kees. 2005. Public Expenditure Tracking (PET) in Tanzania at district- level: effects on local accountability: Participatory Research of and by Tanzanian NGOs, European Conference of African Studies Panel 58. Cattaneo, M et al. 2016. The More, the Better? The Impact of Instructional Time on Student Performance, IZA Discussion Paper 9797. de Jong, Romina, Leng Theavy and Gordon Conochie. 2013. Primary School Budgets in Cambodia: A Public Expenditure Tracking Survey. Phnom Penh: NGO Education Partnership. Gauthier, Bernard and R.Reinikka. 2007. Methodological Approaches to the Study of Institutions and Service Delivery: A Review of PETS, QSDS and CRCS, n.p. Gauthier, Bernard. 2006. PETS-QSDS in Sub-Saharan Africa: a stocktaking study. Washington DC: The World Bank. KIND and ANSA-EAP. 2013. Public Expenditure Tracking Survey Report Flow of Textbook Delivery for Upper Secondary Schools in Cambodia. KIND and ANSA-EAP. Koziol, Margaret and Courtney Tolmie. 2010. Using Public Expenditure Tracking Surveys to monitor projects and small-scale programs. Washington DC: The World Bank. Lavy, V. 2014. The Effect of Instructional Time on Achievement in Math, Science and Reading: Evidence from Developed and Developing Countries. www. bsg.ox.ac.uk Ministry of Economy and Finance (MEF) and Ministry of Education Youth and Sport (MoEYS), Cambodia. 2013. Joint Prakas No. 508. Phnom Penh: MEF and MoEYS. MoEYS, Cambodia. 2010. School Improvement Grant (SIG) Financial Management Manual. Phnom Penh: MoEYS. Ministry of Planning and Investment, Vietnam, n.d. , Public Expenditure Tracking Survey Manual: A social audit tool to monitor the progress of Vietnam’s social and economic development plan. UNICEF. Reinikka, Ritva and Jakob Svensson, n.d. The power of information: evidence from public expenditure tracking surveys, Global Corruption Report. Transparency International. Reinikka, Ritva and Nathaniel Smith. 2004. Public Expenditure Tracking Surveys in education. IIEP. Sundet, Geir. 2008. Following the Money: Do Public Expenditure Tracking Surveys matter? Anti-Corruption Resource Centre. The World Bank. 2005. Cambodia – Public Expenditure Tracking Survey (PETS) in Primary Education. The World Bank. van den Berg, Servas et al. 2010. Tracking public expenditure and assessing service quality in Early Childhood Development in South Africa. Department of Basic Education, Department of Social Development, South Africa. Winkler, Donald. n.d. Policy Brief: Public Expenditure Tracking in Education, EQUIP2. USAID. The World Bank Cambodia Country Office Exchange Square Building Floor 10th IBRD and 11th IFC Streets 51-61 and Streets 102 -106 Sangkat Wat Phnom, Khan Daun Penh Phnom Penh, Cambodia Website: www.worldbank.org/cambodia